NUREG-1576
EPA402-B-01-003
NTISPB2001-106745
%
\
I
Multi-Agency Radiological
Laboratory Analytical Protocols Manual
PLANNING
MARLAP
ASSESSMENT
MPLEMENTATON
uses NIST
Draft for Public Comment
August 2001
-------
ABSTRACT
The Multi-Agency Radiological Laboratory Analytical Protocols (MARLAP) manual provides
guidance for the planning, implementation, and assessment of projects that require the laboratory
analysis of radionuclides. MARLAP's basic goal is to provide guidance and a framework for
project planners, managers, and laboratory personnel to ensure that radioanalytical laboratory
data will meet a project's or program's data requirements. To attain this goal, the manual is
intended to provide the guidance necessary for national consistency in the form of a performance-
based approach for meeting a project's data requirements. The guidance in MARLAP is designed
to help ensure the generation of radioanalytical data of known quality, appropriate for its
intended use.
MARLAP was developed by a workgroup that included representatives from the U.S. Environ-
mental Protection Agency (EPA), Department of Energy (DOE), Department of Defense (DOD),
Nuclear Regulatory Commission (NRC), National Institute of Standards and Technology (NIST),
U.S. Geological Survey (USGS), and U.S. Food and Drug Administration (FDA). State participa-
tion in the development of the manual involved contributions from representatives from the
Commonwealth of Kentucky and the State of California. Contractors to EPA, DOE, and NRC,
and members of the public, have been present during the open meetings of the MARLAP
workgroup.
Examples of data collection activities that MARLAP supports include site characterization, site
cleanup and compliance demonstration, decommissioning of nuclear facilities, remedial and
removal actions, effluent monitoring of licensed facilities, environmental site monitoring,
background studies, and waste management activities.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT III DO NOT CITE OR QUOTE
-------
NOTICE
This draft manual being released for simultaneous public and peer review, and technical
comments are solicited as described below. MARLAP has not been approved for use in part or in
whole and should not be used, cited, or quoted except for the purposes of providing comments as
requested by the agencies participating in its development.
MARLAP was developed by a workgroup that included representatives from the U.S. Environ-
mental Protection Agency (EPA), Department of Energy (DOE), Department of Defense (DOD),
Nuclear Regulatory Commission (NRC), National Institute of Standards and Technology (NIST),
U.S. Geological Survey (USGS), and U.S. Food and Drug Administration (FDA). State participa-
tion in the development of the manual involved contributions from representatives from the
Commonwealth of Kentucky and the State of California. Contractors to EPA, DOE, and NRC,
and members of the public, have been present during the open meetings of the MARLAP
workgroup.
Although Federal Government personnel are involved in the preparation of this document, the
draft manual does not yet represent the official position of any participating agency. This review
is a necessary step in the development of a multi-agency consensus manual. References within
this manual to any specific commercial product, process, or service by trade name, trademark,
manufacturer, or otherwise does not necessarily constitute or imply its endorsement or
recommendation by the United States Government.
Members of the public are invited and encouraged to submit comments to the following website
http://www.eml.doe.gov/marlap/. Comments may also be submitted to either:
U.S. Environmental Protection Agency
ATTN: Air and Radiation Docket, Mail Stop 6102
Docket Number A-2001-16, Room M1500
401 M Street, SW
Washington, DC 20460
or
Chief, Rules and Directives Branch
Division of Administrative Services
Mail Stop T6D59
U.S. Nuclear Regulatory Commission
Washington, DC 20555-0001
MARLAP JULY 2001
DO NOT CITE OR QUOTE IV DRAFT FOR PUBLIC COMMENT
-------
Notice
All comments received will be reviewed by the entire MARLAP workgroup. Comments received
by the date published in the Federal Register Notice announcing the availability of the document
for public review will be considered. Comments received after that date will be considered if it is
practical to do so, but no assurance can be given for consideration of late comments.
Copies of the draft MARLAP manual and all comments received may be examined or copied for
a fee at the EPA Docket Room M1500, Docket Number A-2001-16, First Floor Waterside Mall,
401 M Street, SW, Washington, DC 20460; and the NRC Public Document Room, at U.S.
Nuclear Regulatory Commission, Public Document Room, Washington, DC 20555. The
document is also available through the National Technical Information Service (NTIS). The
NTIS document number is PB2001-106745, and the NTIS Sales Desk can be reached between
8:30 a.m. and 6:00 p.m. Eastern Time, Monday through Friday at 1-800-553-6847; TDD (hearing
impaired only) at (703) 487-4639.
In addition to providing comments on individual chapters and appendices, reviewers are also
requested to address the following questions while reviewing the draft manual:
(1) Is the performance-based approach used in MARLAP for the planning, implementation,
and assessment phases of projects technically sound, and is the approach reasonable in terms
of ease of implementation by project managers and laboratories? Does the approach
effectively link the three phases of a project, and is the guidance on quality control
appropriate and supportive of a performance-based approach?
(2) Is the guidance on laboratory operations in Part II (Chapters 10-20) technically accurate
and useful?
(3) Are the concepts covered under measurement statistics—specifically measurement
uncertainty, detection and quantification capability—presented accurately and appropriately?
(4) Is the information understandable and presented in logical sequence? How can the
presentation of material be modified to improve the manual?
(5) Does MARLAP provide benefits that are not currently available through other
approaches? What are the costs associated with implementing the guidance in MARLAP in
comparison with currently available alternatives?
Commentors are encouraged to use the website, http://www.eml.doe.gov/marlap, for their
review. The website has detailed instructions on how to submit comments and has several
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT V DO NOT CITE OR QUOTE
-------
Notice
features that should aid the review process. Commentors also may submit written comments to
either of the addresses listed on page iv of this Notice using the same general approach described
in the MARLAP website. Comments should be accompanied by supporting details, rationale, or
data. To ensure efficient and complete comment resolution, commentors are requested to
reference the page number and the line number to which the comment refers. Comments
corresponding to an entire chapter, section, or table should be referenced to the line number for
the title of the chapter (always line number 1), section, or table. Comments on footnotes should
be referenced to the line in the text where the footnote appears (footnotes do not have line
numbers). Comments on figures should be referenced to the page on which the figure appears
(figures do not have line numbers) and figure number. Comments on the entire manual should be
referenced to the title page.
MARLAP JULY 2001
DO NOT CITE OR QUOTE VI DRAFT FOR PUBLIC COMMENT
-------
ACKNOWLEDGMENTS
The origin of the Multi-Agency Radiological Laboratory Analytical Protocols (MARLAP)
manual can be traced to the recognition by a number of agencies for the need to have a nationally
consistent approach to producing radioanalytical data that meet a program's or project's needs. A
multi-agency workgroup was formed with representatives from the U.S. Environmental
Protection Agency (EPA), Department of Energy (DOE), Nuclear Regulatory Commission
(NRC), Department of Defense (DOD), U.S. Geological Survey (USGS), National Institute of
Standards and Technology (NIST), and Food and Drug Administration (FDA) to develop
guidance for the planning, implementation, and assessment of projects that require the laboratory
analysis of radionuclides. Representatives from the Commonwealth of Kentucky and the State of
California also contributed to the development of the manual.
Of particular importance to the workgroup is that the guidance needs to be both scientifically
rigorous and flexible enough to be applied to a diversity of projects and programs. The draft
MARLAP manual is the result of a cooperative effort with these goals in mind.
MARLAP would not have been possible without the workgroup members who contributed their
time, talent, and efforts to develop this guidance document:
John Griggs*, EPA, Chair
EPA: H.Benjamin Hull DOE: EmileBoulos*
Marianne Lynch* Carl Gogolak
Keith McCroan* Pam Greenlaw*
Eric Reynolds Catherine Klusek*
Jon Richards Stan Morton*
Colin Sanderson*
Stephanie Woolf*
DOD: CPT Andrew Scott (Army) NRC: Rateb (Boby) Abu Eid
Ronald Swatski* (Army) Tin Mo
Jan Dunker (Army Corps of Engineers) George Powers
Troy Blanton (Navy)
CAPT David Farrand (Navy) USGS: Ann Mullin*
Dale Thomas (Air Force)
NIST: Kenneth GW. Inn* FDA: Edmond Baratta
* These workgroup members also served as chapter chairs.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT VII DO NOT CITE OR QUOTE
-------
Acknowledgments
Special recognition is given to John Volpe, Commonwealth of Kentucky, and Penny Leinwander,
State of California, for their contributions to the development of the MARLAP manual. The
following Federal Agency contractors provided assistance in developing the MARLAP manual:
EPA: N. Jay Bassin (Environmental Management Support, Inc.)
Diane Dopkin (Environmental Management Support, Inc.)
U. Hans Behling ( S. Cohen & Associates, Inc.)
Richard Blanchard (S. Cohen & Associates, Inc.)
Harry Chmelynski (S. Cohen & Associates, Inc.)
Scott Hay (S. Cohen & Associates, Inc.)
Patrick Kelly (S. Cohen & Associates, Inc.)
Robert Litman (S. Cohen & Associates, Inc.)
Charles (Chick) Phillips (S. Cohen & Associates, Inc.)
William Richardson in (S. Cohen & Associates, Inc.)
Steven Schaffer (S. Cohen & Associates, Inc.)
DOE: David McCurdy (Duke Engineering & Services)
John Maney (Environmental Measurements Assessments)
Stan Blacker (MACTEC, Inc.)
Pat Harrington (MACTEC, Inc.)
Mike Miller (MACTEC, Inc.)
Lisa Smith (Argonne National Laboratory)
NRC: Eric W. Abelquist (ORISE)
Dale Condra (ORISE)
The MARLAP Workgroup was greatly aided in the development of the manual by the
contributions and support provided by the individuals listed below.
David Bottrell (DOE) David Friedman (EPA) Kevin Miller (DOE)
Lloyd Currie (NIST) LCDR Lino Fragoso (Navy) Jim Mitchell (EPA)
Mike Carter (EPA) Richard Graham (EPA) Colleen Petullo (EPA)
Mary Clark (EPA) Patricia Gowland (EPA) Steve Pia (EPA)
Ron Colle (NIST) Larry Jensen (EPA) Phil Reed (NRC)
Mark Doehnert (EPA) Jim Kotton (NRC) Cheryl Trottier (NRC)
Steve Domotor (DOE) Ed Messer (EPA) John Warren (EPA)
Joan Fisk (EPA)
MARLAP JULY 2001
DO NOT CITE OR QUOTE VIII DRAFT FOR PUBLIC COMMENT
-------
CONTENTS
Page
Abstract in
Notice IV
Acknowledgments VII
Acronyms and Abbreviations XLVII
1 Introduction to MARLAP 1-1
1.1 Overview 1-1
1.2 Purpose of the Manual 1-2
1.3 Use and Scope of the Manual 1-3
1.4 Key MARLAP Concepts and Terminology 1-4
1.4.1 Data Life Cycle 1-4
1.4.2 Directed Planning Process 1-5
1.4.3 Performance-Based Approach 1-6
1.4.4 Analytical Process 1-7
1.4.5 Analytical Protocol 1-8
1.4.6 Analytical Method 1-8
1.4.7 Uncertainty and Error 1-8
1.4.8 Precision, Bias, and Accuracy 1-10
1.4.9 Performance Objectives: Data Quality Objectives and Measurement Quality
Objectives 1-11
1.4.10 Analytical Protocol Specifications 1-12
1.4.11 The Assessment Phase 1-13
1.5 The MARLAP Process 1-14
1.6 Structure of the Manual 1-15
1.6.1 Overview of Part I 1-17
1.6.2 Overview of Part II 1-17
1.6.3 Overview of the Appendices 1-19
1.7 References 1-20
2 Project Planning Process 2-1
2.1 Introduction 2-1
2.2 The Importance of Directed Project Planning 2-2
2.3 Directed Project Planning Processes 2-4
2.3.1 A Graded Approach to Project Planning 2-4
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT IX DO NOT CITE OR QUOTE
-------
Contents
Page
2.3.2 Guidance on Directed Planning Processes 2-4
2.3.3 Elements of Directed Planning Processes 2-6
2.4 The Project Planning Team 2-7
2.4.1 Team Representation 2-7
2.4.2 The Radioanalytical Specialists 2-8
2.5 Direct Planning Process and Role of the Radioanalytical Specialists 2-9
2.5.1 Define the Problem 2-12
2.5.2 Identify the Decision 2-13
2.5.2.1 Action Level 2-13
2.5.2.2 Scale of the Decision 2-14
2.5.2.3 Inputs and Boundaries to the Decision 2-15
2.5.2.4 Data Needs 2-15
2.5.3 Specify the Decision Rule and the Tolerable Decision Error Rates 2-15
2.5.4 Optimize the Strategy for Obtaining Data 2-17
2.5.4.1 Analytical Protocol Specifications 2-18
2.5.4.2 Measurement Quality Objectives 2-18
2.6 Results of the Directed Planning Process 2-19
2.6.1 Output Required by the Radioanalytical Laboratory: The Analytical Protocol
Specifications 2-20
2.6.2 Chain of Custody 2-21
2.7 Project Planning and Project Implementation and Assessment 2-21
2.7.1 Documenting the Planning Process 2-21
2.7.2 Obtaining Analytical Services 2-22
2.7.3 Selecting Analytical Protocols 2-23
2.7.4 Assessment Plans 2-23
2.7.4.1 Data Verification 2-24
2.7.4.2 Data Validation 2-24
2.7.4.3 Data Quality Assessment 2-24
2.8 References 2-25
3 Key Analytical Planning Issues and Developing Analytical Protocol Specifications 3-1
3.1 Introduction 3-1
3.2 Overview of the Analytical Process 3-2
3.3 General Analytical Planning Issues 3-2
3.3.1 Develop Analyte List 3-4
3.3.2 Identify Concentration Ranges 3-6
MARLAP JULY 2001
DO NOT CITE OR QUOTE X DRAFT FOR PUBLIC COMMENT
-------
Contents
Page
3.3.3 Identify and Characterize Matrices of Concern 3-6
3.3.4 Determine Relationships Between the Radionuclides of Concern 3-7
3.3.5 Determine Available Project Resources and Deadlines 3-8
3.3.6 Refine Analyte List and Matrix List 3-8
3.3.7 Method Performance Characteristics and Measurement Quality Objectives . . . 3-9
3.3.7.1 Develop MQOs for Select Method Performance Characteristics 3-11
3.3.7.2 The Role of MQOs in the Protocol Selection and Evaluation Process . . . 3-16
3.3.7.3 The Role of MQOs in the Project's Data Evaluation Process 3-16
3.3.8 Determine Any Limitations on Analysis Options 3-17
3.3.8.1 Gamma Spectrometry 3-18
3.3.8.2 Gross Alpha and Beta Analysis 3-19
3.3.8.3 Radiochemical Nuclide-Specific Analysis 3-19
3.3.9 Determine Method Availability 3-19
3.3.10 Determine the Type and Frequency of, and Evaluation Criteria for, Quality
Control Samples 3-20
3.3.11 Determine Sample Tracking and Custody Requirements 3-21
3.3.12 Determine Data Reporting Requirements 3-21
3.4 Matrix-Specific Analytical Planning Issues 3-22
3.4.1 Solids 3-23
3.4.1.1 Homogenization and Subsampling 3-24
3.4.1.2 Removal of Unwanted Materials 3-24
3.4.2 Liquids 3-25
3.4.3 Filters and Wipes 3-26
3.5 Assembling the Analytical Protocol Specifications 3-26
3.6 Level of Protocol Performance Demonstration 3-27
3.7 Project Plan Documents 3-30
3.8 References 3-31
4 Project Plan Documents 4-1
4.1 Introduction 4-1
4.2 The Importance of Project Plan Documents 4-2
4.3 A Graded Approach to Project Plan Documents 4-3
4.4 Project Plan Documents 4-4
4.4.1 Guidance on Project Plan Documents 4-4
4.4.2 Approaches to Project Plan Documents 4-5
4.5 Elements of Project Plan Documents 4-6
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT XI DO NOT CITE OR QUOTE
-------
Contents
Page
4.5.1 Content of Project Plan Documents 4-7
4.5.2 Plan Documents Integration 4-9
4.5.3 Plan Content for Small Projects 4-10
4.6 Linking the Project Plan Documents and the Project Planning Process 4-10
4.6.1 Planning Process Report 4-15
4.6.2 Data Assessment 4-16
4.6.2.1 Data Verification 4-16
4.6.2.2 Data Validation 4-16
4.6.2.3 Data Quality Assessment 4-17
4.7 References 4-18
5 Obtaining Laboratory Services 5-1
5.1 Introduction 5-1
5.2 Importance of Writing a Technical and Contractual Specification Document 5-2
5.3 Statement of Work—Technical Requirements 5-2
5.3.1 Analytes 5-3
5.3.2 Matrix 5-3
5.3.3 Measurement Quality Objectives 5-3
5.3.4 Unique Analytical Process Requirements 5-4
5.3.5 Quality Control Samples and Participation in External Performance Evaluation
Programs 5-4
5.3.6 Laboratory Radiological Holding and Turnaround Times 5-5
5.3.7 Number of Samples and Schedule 5-5
5.3.8 Quality System 5-6
5.3.9 Laboratory's Proposed Methods 5-6
5.4 Request for Proposal—Generic Contractual Requirements 5-7
5.4.1 Sample Management 5-7
5.4.2 Licenses, Permits and Environmental Regulations 5-8
5.4.2.1 Licenses 5-8
5.4.2.2 Environmental and Transportation Regulations 5-9
5.4.3 Data Reporting and Communications 5-9
5.4.3.1 Data Deliverables 5-9
5.4.3.2 Software Verification and Control 5-10
5.4.3.3 Problem Notification and Communication 5-10
5.4.3.4 Status Reports 5-11
5.4.4 Sample Re-Analysis Requirements 5-11
MARLAP JULY 2001
DO NOT CITE OR QUOTE XII DRAFT FOR PUBLIC COMMENT
-------
Contents
Page
5.4.5 Subcontracted Analyses 5-11
5.5 Laboratory Selection and Qualification Criteria 5-12
5.5.1 Technical Proposal Evaluation 5-12
5.5.1.1 Scoring and Evaluation Scheme 5-12
5.5.1.2 Scoring Elements 5-13
5.5.2 Pre-Award Proficiency Evaluation 5-15
5.5.3 Pre-Award Assessments and Audits 5-15
5.6 References 5-16
5.6.1 Cited References 5-16
5.6.2 Other Sources 5-17
6 Selection and Application of an Analytical Method 6-1
6.1 Introduction 6-1
6.2 Method Definition 6-2
6.3 Life Cycle of Method Application 6-6
6.4 Generic Considerations for Method Development and Selection 6-10
6.5 Project-Specific Consideration for Method Selection 6-13
6.5.1 Matrix and Analyte Identification 6-13
6.5.1.1 Matrices 6-13
6.5.1.2. Analytes and Potential Interferences 6-15
6.5.2 Process Knowledge 6-16
6.5.3 Radiological Holding and Turnaround Times 6-17
6.5.4 Unique Process Specifications 6-18
6.5.5 Measurement Quality Objectives 6-19
6.5.5.1 Method Uncertainty 6-19
6.5.5.2 Quantification Capability 6-20
6.5.5.3 Detection Capability 6-21
6.5.5.4 Applicable Analyte Concentration Range 6-23
6.5.5.5 Method Specificity 6-23
6.5.5.6 Method Ruggedness 6-24
6.5.5.7 Bias Considerations 6-24
6.6 Method Validation 6-25
6.6.1 Laboratory's Method Validation Protocol 6-26
6.6.2 Tiered Approach to Validation 6-27
6.6.2.1 Existing Methods Requiring No Additional Validation 6-29
6.6.2.2 Use of a Validated Method for Similar Matrices 6-30
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT XIII DO NOT CITE OR QUOTE
-------
Contents
Page
6.6.2.3 New Application of a Validated Method 6-30
6.6.2.4 Newly Developed or Adapted Methods 6-32
6.6.4 Method Validation Documentation 6-32
6.7 Analyst Qualifications and Demonstrated Proficiency 6-33
6.8 Method Control 6-33
6.9 Continued Performance Assessment 6-34
6.10 Documentation To Be Sent to the Project Manager 6-36
6.11 References 6-37
7 Evaluating Methods and Laboratories 7-1
7.1 Introduction 7-1
7.2 Evaluation of Proposed Analytical Methods 7-2
7.2.1 Documentation of Required Method Performance 7-2
7.2.1.1 Method Validation Documentation 7-3
7.2.1.4 Method Experience, Previous Projects, and Clients 7-5
7.2.1.5 Internal and External Quality Assurance Assessments 7-5
7.2.2 Performance Requirements of the SOW—Analytical Protocol Specifications . 7-6
7.2.2.1 Matrix and Analyte Identification 7-6
7.2.2.2 Process Knowledge 7-7
7.2.2.3 Radiological Holding and Turnaround Times 7-7
7.2.2.4 Unique Processing Specifications 7-9
7.2.2.5 Measurement Quality Objectives 7-9
7.2.2.6 Bias Considerations 7-15
7.3 Initial Evaluation of a Laboratory 7-17
7.3.1 Review of Quality System Documents 7-17
7.3.2 Adequacy of Facilities, Instrumentation, and Staff Levels 7-19
7.3.3 Review of Applicable Prior Work 7-19
7.3.4 Review of Performance Indicators 7-20
7.3.4.1 Review of Internal QC Results 7-20
7.3.4.2 External PE Program Results 7-21
7.3.4.3 Internal and External Quality Assessment Reports 7-21
7.3.5 Initial Audit 7-22
7.4 Ongoing Evaluation of the Laboratory's Performance 7-22
7.4.1 Quantitative Measures of Quality 7-23
7.4.1.1 MQO Compliance 7-24
7.4.1.2 Other Parameters 7-30
MARLAP JULY 2001
DO NOT CITE OR QUOTE XIV DRAFT FOR PUBLIC COMMENT
-------
Contents
Page
7.4.2 Operational Aspects 7-31
7.4.2.1 Desk Audits 7-31
7.4.2.2 Onsite Audits 7-33
7.5 References 7-36
Radiochemical Data Verification And Validation 8-1
8.1 Introduction 8-1
8.2 Data Assessment Process 8-2
8.2.1 Planning Phase of the Data Life Cycle 8-2
8.2.2 Implementation Phase of the Data Life Cycle 8-3
8.2.2.1 Project Objectives 8-4
8.2.2.2 Documenting Project Activities 8-4
8.2.2.3 QA/QC 8-4
8.2.3 Assessment Phase of the Data Life Cycle 8-5
8.3 Validation Plan 8-7
8.3.1 Technical and Quality Objectives of the Project 8-8
8.3.2 Validation Tests 8-9
8.3.3 Data Qualifiers 8-9
8.3.4 Reporting and Documentation 8-11
8.4 Other Essential Elements 8-11
8.4.1 Statement of Work 8-12
8.4.2 Verified Data Deliverables 8-12
8.5 Data Verification and Validation Process 8-13
8.5.1 The Sample Handling and Analysis System 8-14
8.5.1.1 Sample Descriptors 8-15
8.5.1.2 Aliquant Size 8-16
8.5.1.3 Dates of Sample Collection, Preparation, and Analysis 8-16
8.5.1.4 Preservation 8-17
8.5.1.5 Tracking 8-18
8.5.1.6 Traceability 8-18
8.5.1.7 QC Types and Linkages 8-18
8.5.1.8 Chemical Separation (Yield) 8-19
8.5.1.9 Self-Absorption (Residue) 8-20
8.5.1.10 Efficiency, Calibration Curves, and Instrument Background 8-20
8.5.1.11 Spectrometry Resolution 8-20
8.5.1.12 Dilution and Correction Factors 8-21
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT XV DO NOT CITE OR QUOTE
-------
Contents
Page
8.5.1.13 Counts and Count Time (Duration) 8-22
8.5.1.14 Result of Measurement, Uncertainty, Minimum Detectable Concentration,
and Units 8-22
8.5.2 Quality Control Samples 8-22
8.5.2.1 Method Blank 8-24
8.5.2.2 Laboratory Control Samples 8-25
8.5.2.3 Laboratory Replicates 8-25
8.5.2.4 Matrix Spikes and Matrix Spike Duplicates 8-26
8.5.3 Tests of Detection and Unusual Uncertainty 8-26
8.5.3.1 Detection 8-26
8.5.3.2 Detection Capability 8-27
8.5.3.3 Large or Unusual Uncertainty 8-28
8.5.4 Final Qualification and Reporting 8-29
8.6 Validation Report 8-30
8.7 Other Sources of Information 8-32
9 Data Quality Assessment 9-1
9.1 Introduction 9-1
9.2 Assessment Phase 9-2
9.3 Graded Approach to Assessment 9-3
9.4 The Data Quality Assessment Team 9-4
9.5 Data Quality Assessment Plan 9-4
9.6 Data Quality Assessment Process 9-6
9.6.1 Review of Project Documents 9-8
9.6.1.1 TheProjectDQOsandMQOs 9-8
9.6.1.2 The DQA Plan 9-9
9.6.1.3 Summary of the DQA Review 9-9
9.6.2 Sample Representativeness 9-10
9.6.2.1 Review of the Sampling Plan 9-10
9.6.2.2 Sampling Plan Implementation 9-13
9.6.2.3 Data Considerations 9-14
9.6.3 Data Accuracy 9-16
9.6.3.1 Review of the Analytical Plan 9-19
9.6.3.2 Analytical Plan Implementation 9-21
9.6.4 Decisions and Tolerable Error Rates 9-22
9.6.4.1 Statistical Evaluation of Data 9-23
MARLAP JULY 2001
DO NOT CITE OR QUOTE XVI DRAFT FOR PUBLIC COMMENT
-------
Contents
Page
9.6.4.2 Evaluation of Decision Error Rates 9-26
9.7 Data Quality Assessment Report 9-27
9.8 References 9-29
9.8.1 Cited Sources 9-29
9.8.2 Other Sources 9-29
10 Field and Sampling Issues That Affect Laboratory Measurements 10-1
Part I: Generic Issues 10-1
10.1 Introduction 10-1
10.1.1 The Need for Establishing Channels of Communication 10-2
10.1.2 Developing Field Documentation 10-2
10.2 Field Sampling Plan: Non Matrix Specific Issues 10-3
10.2.1 Determination of Analytical Sample Size 10-3
10.2.2 Field Equipment and Supply Needs 10-3
10.2.3 Selection of Sample Containers 10-4
10.2.3.1 Container Material 10-4
10.2.3.2 Container Opening and Closure 10-5
10.2.3.3 Sealing Containers 10-5
10.2.3.4 Precleaned and Extra Containers 10-5
10.2.4 Container Label and Sample Identification Code 10-5
10.2.5 Field Data Documentation 10-6
10.2.6 Field Tracking, Custody, and Shipment Forms 10-8
10.2.7 Chain of Custody 10-9
10.2.8 Field Quality Control 10-9
10.2.9 Decontamination of Field Equipment 10-10
10.2.10 Packing and Shipping 10-11
10.2.11 Worker Health and Safety Plan 10-12
10.2.11.1 Physical Hazards 10-13
10.2.11.2 Biohazards 10-15
Part II: Matrix-Specific Issues That Impact Field Sample Collection, Processing, and
Preservation 10-16
10.3 Liquid Samples 10-17
10.3.1 Liquid Sampling Methods 10-18
10.3.2 Liquid Sample Preparation: Filtration 10-18
10.3.2.1 EPA Guidance for Samples/Filtration 10-19
10.3.2.2 Filters 10-21
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT XVII DO NOT CITE OR QUOTE
-------
Contents
Page
10.3.3 Field Preservation of Liquid Samples 10-22
10.3.3.1 Sample Acidification 10-22
10.3.3.2 Non-Acid Preservation Techniques 10-23
10.3.4 Liquid Samples: Special Cases 10-26
10.3.4.1 Radon-222 in Water 10-26
10.3.4.1 Milk 10-27
10.3.5 Non-aqueous Liquids and Mixtures 10-27
10.4 Solids 10-29
10.4.1 Soils 10-29
10.4.1.1 Soil Sample Preparation 10-30
10.4.1.2 Sample Ashing 10-31
10.4.2 Sediments 10-31
10.4.2.1 Initial Mixing and Transport Dispersion of Radionuclides Discharged to
Water 10-31
10.4.2.2 Sediment Effect 10-32
10.4.2.3 Sample Preparation/Preservation 10-32
10.4.3 Other Solids 10-32
10.4.3.1 Structural Materials 10-32
10.4.3.2 Biota: Samples of Plant and Animal Products 10-33
10.5 Air Sampling 10-37
10.5.1 Sampler Components 10-37
10.5.2 Filter Selection Based on Destructive Versus Non-destructive Analysis ... 10-39
10.5.3 Sample Preservation and Storage 10-39
10.5.4 Special Cases: Collection of Gaseous and Volatile Air Contaminants 10-40
10.5.4.1 Radioiodines 10-40
10.5.4.2 Gases 10-41
10.5.4.3 Tritium Air Sampling 10-41
10.5.5 Radon 10-42
10.5.5.1 Radon Sampling Methods 10-43
10.5.5.2 Selecting a Radon Sampling Method Based on Data
Quality Objectives 10-46
10.6 Wipe Sampling for Assessing Surface Contamination 10-47
10.6.1 Sample Collection Methods 10-48
10.6.1.1 Dry Wipes 10-48
10.6.1.2 Wet Wipes 10-48
10.6.2 Sample Handling 10-50
MARLAP JULY 2001
DO NOT CITE OR QUOTE XVIII DRAFT FOR PUBLIC COMMENT
-------
Contents
Page
10.7 References 10-50
11 Sample Receipt, Inspection, and Tracking 11-1
11.1 Introduction 11-1
11.2 General Considerations 11-3
11.2.1 Communication Before Sample Receipt 11-3
11.2.2 Standard Operating Procedures 11-3
11.2.3 Laboratory License 11-4
11.2.4 Sample Chain-of-Custody 11-5
11.3 Sample Receipt 11-6
11.3.1 Package Receipt 11-6
11.3.2 Radiological Screening 11-7
11.3.3 Corrective Action 11-9
11.4 Sample Inspection 11-9
11.4.1 Physical Integrity of Package and Sample Containers 11-9
11.4.2 Sample Identity Confirmation 11-10
11.4.3 Confirmation of Field Preservation 11-11
11.4.4 Presence of Hazardous Materials 11-11
11.4.5 Corrective Action 11-12
11.5 Laboratory Sample Tracking 11-12
11.5.1 Sample Log-In 11-13
11.5.2 Sample Tracking During Analyses 11-13
11.5.3 Storage of Samples 11-14
11.6 References 11-14
12 Laboratory Sample Preparation 12-1
12.1 Introduction 12-1
12.2 General Guidance for Sample Preparation 12-2
12.2.1 Potential Sample Losses During Preparation 12-2
12.2.1.1 Losses as Dust or Particulates 12-2
12.2.1.2 Losses Through Volatilization 12-3
12.2.1.3 Losses Owing to Reactions Between Sample and Container 12-4
12.2.2 Contamination from Sources in the Laboratory 12-6
12.2.2.1 Airborne Contamination 12-6
12.2.2.2 Contamination of Reagents 12-7
12.2.2.3 Contamination of Glassware/Equipment 12-7
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT XIX DO NOT CITE OR QUOTE
-------
Contents
Page
12.2.2.4 Contamination of Facilities 12-8
12.2.3 Cleaning of Labware, Glassware, and Equipment 12-8
12.2.3.1 Labware and Glassware 12-8
12.2.3.2 Equipment 12-10
12.3.1 General Procedures 12-14
12.3.1.1 Exclusion of Material 12-14
12.3.1.2 Principles of Drying Techniques 12-14
12.3.1.3 Obtaining a Constant Weight 12-24
12.3.1.4 Subsampling 12-26
12.3.2 Soil/Sediment Samples 12-30
12.3.2.1 Soils 12-31
12.3.2.2 Sediments 12-31
12.3.3.1 Biological Samples 12-32
12.3.3.2 Food 12-32
12.3.3.3 Vegetation 12-32
12.3.3.4 Bone and Tissue 12-33
12.3.4 Other Samples 12-33
12.4 Filters 12-33
12.5 Wipe Samples 12-34
12.6 Liquid Samples 12-35
12.6.1 Conductivity 12-35
12.6.2 Turbidity 12-36
12.6.3 Filtration 12-36
12.6.4 Aqueous Liquids 12-36
12.6.5 Nonaqueous Liquids 12-37
12.6.6 Mixtures 12-38
12.6.6.1 Liquid-Liquid Mixtures 12-38
12.6.6.2 Liquid-Solid Mixtures 12-39
12.7 Gases 12-39
12.8 Bioassay 12-40
12.9 References 12-41
12.9.1 Cited Sources 12-41
12.9.2 Other Sources 12-47
13 Sample Dissolution 13-1
13.1 Introduction 13-1
MARLAP JULY 2001
DO NOT CITE OR QUOTE XX DRAFT FOR PUBLIC COMMENT
-------
Contents
Page
13.2 The Chemistry of Dissolution 13-2
13.2.1 Solubility and the Solubility Product Constant, Ksp 13-2
13.2.2 Chemical Exchange, Decomposition, and Simple Rearrangement Reactions . 13-3
13.2.3 Oxidation-Reduction Processes 13-4
13.2.4 Complexation 13-5
13.2.5 Equilibrium: Carriers and Tracers 13-6
13.3 Fusion Techniques 13-6
13.3.1 Alkali-Metal Hydroxide Fusions 13-10
13.3.2 Boron Fusions 13-11
13.3.3 Fluoride Fusions 13-12
13.4 Wet Ashing and Acid Dissolution Techniques 13-13
13.4.1 Acids and Oxidants 13-13
13.4.2 Acid Digestion Bombs 13-23
13.4.3 Is it Dissolved? 13-23
13.5 Microwave Digestion 13-24
13.5.1 Focused Open-Vessel Systems 13-25
13.5.2 Low-Pressure, Closed-Vessel Systems 13-25
13.5.3 High-Pressure, Closed-Vessel Systems 13-26
13.6 Special Matrix Considerations 13-26
13.6.1 Liquid Samples 13-26
13.6.1.1 Aqueous Samples 13-26
13.6.1.2Nonaqueous Samples 13-27
13.6.2 Solid Samples 13-27
13.6.3 Filters 13-27
13.6.4 Wipe Samples 13-28
13.6.5 Liquid Scintillation Samples 13-28
13.6.5.1 Wet Oxidation 13-28
13.6.5.2 Dry Oxidation 13-29
13.7 Total Dissolution and Leaching 13-29
13.7.1 Acid Leaching 13-30
13.7.2 Total Dissolution through Fusion 13-31
13.7.3 Acid Digestion — Fusion Combined Approach 13-32
13.8 Examples of Decomposition Procedures 13-32
13.9 References 13-33
13.9.1 Cited References 13-33
13.9.2 Other Sources 13-36
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT XXI DO NOT CITE OR QUOTE
-------
Contents
Pas
14 Separation Techniques 14-1
14.1 Introduction 14-1
14.2 Oxidation/Reduction Processes 14-2
14.2.1 Introduction 14-2
14.2.2 Oxidation-Reduction Reactions 14-3
14.2.3 Common Oxidation States 14-6
14.2.4 Oxidation State in Solution 14-11
14.2.5 Common Oxidizing and Reducing Agents 14-12
14.2.6 Oxidation State and Radiochemical Analysis 14-14
14.3 Complexation 14-19
14.3.1 Introduction 14-19
14.3.2 Chelates 14-21
14.3.3 The Formation (Stability) Constant 14-24
14.3.4 Complexation and Radiochemical Analysis 14-25
14.3.4.1 Extraction of Laboratory Samples and Ores 14-25
14.3.4.2 Separation by Solvent Extraction and Ion-Exchange Chromatography
14-25
14.3.4.3 Formation and Dissolution of Precipitates 14-26
14.3.4.4 Stabilization of Ions in Solution 14-27
14.3.4.5 Detection and Determination 14-27
14.4 Solvent Extraction 14-27
14.4.1 Extraction Principles 14-27
14.4.2 Distribution Coefficient 14-28
14.4.3 Extraction Technique 14-30
14.4.4 Solvent Extraction and Radiochemical Analysis 14-33
14.4.5 Solid-Phase Extraction 14-35
14.4.5.1 Extraction Chromatography Columns 14-36
14.4.5.2 Extraction Membranes 14-37
14.4.6 Advantages and Disadvantages of Solvent Extraction 14-38
14.4.6.1 Advantages 14-38
14.4.6.2 Disadvantages 14-38
14.5 Volatilization and Distillation 14-39
14.5.1 Introduction 14-39
14.5.2 Volatilization Principles 14-40
14.5.3 Distillation Principles 14-42
MARLAP JULY 2001
DO NOT CITE OR QUOTE XXII DRAFT FOR PUBLIC COMMENT
-------
Contents
Page
14.5.4 Separations in Radiochemical Analysis 14-43
14.5.5 Advantages and Disadvantages of Volatilization 14-44
14.5.5.1 Advantages 14-44
14.5.5.2 Disadvantages 14-44
14.6 Electrodeposition 14-45
14.6.1 Electrodeposition Principles 14-45
14.6.2 Separation of Radionuclides 14-46
14.6.3 Preparation of Counting Sources 14-47
14.6.4 Advantages and Disadvantages of Electrodeposition 14-47
14.6.4.1 Advantages 14-47
14.6.4.2 Disadvantages 14-47
14.7 Chromatography 14-48
14.7.1 Chromatographic Principles 14-48
14.7.2 Gas-Liquid and Liquid-Liquid Phase Chromatography 14-49
14.7.3 Adsorption Chromatography 14-50
14.7.4 Ion-Exchange Chromatography 14-50
14.7.4.1 Principles of Ion Exchange 14-50
14.7.4.2 Resins 14-53
14.7.5 Affinity Chromatography 14-59
14.7.6 Gel-Filtration Chromatography 14-59
14.7.7 Chromatographic Laboratory Methods 14-60
14.7.8 Advantages and Disadvantages of Chromatographic Systems 14-61
14.8 Precipitation and Coprecipitation 14-61
14.8.1 Introduction 14-61
14.8.2 Solutions 14-62
14.8.3 Precipitation 14-64
14.8.3.1 Solubility and the Solubility Product Constant, Ksp 14-65
14.8.3.2 Factors Affecting Precipitation 14-70
14.8.3.3 Optimum Precipitation Conditions 14-75
14.8.4 Coprecipitation 14-76
14.8.4.1 Coprecipitation Processes 14-77
14.8.4.2 Water as an Impurity 14-81
14.8.4.3 Postprecipitation 14-81
14.8.4.4 Coprecipitation Methods 14-82
14.8.5 Colloidal Precipitates 14-86
14.8.6 Filterability of Precipitates 14-88
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT XXIII DO NOT CITE OR QUOTE
-------
Contents
Page
14.8.7 Advantages and Disadvantages of Precipitation and Coprecipitation 14-90
14.9 Carriers and Tracers 14-91
14.9.1 Introduction 14-91
14.9.2 Carriers 14-91
14.9.2.1 Isotopic Carriers 14-92
14.9.2.2 Nonisotopic Carriers 14-93
14.9.2.3 Common Carriers 14-94
14.9.2.4 Holdback Carriers 14-98
14.9.2.5 Yield (Recovery) of Isotopic Carriers 14-98
14.9.3 Tracers 14-99
14.9.3.1 Characteristics of Tracers 14-101
14.9.3.2 Coprecipitation 14-103
14.9.3.3 Deposition on Nonmetallic Solids 14-103
14.9.3.4 Radiocolloid Formation 14-103
14.9.3.5 Distribution (Partition) Behavior 14-105
14.9.3.6 Vaporization 14-105
14.9.3.7 Oxidation and Reduction 14-106
14.10 Radiochemical Equilibrium 14-107
14.10.1 Basic Principles of Equilibrium 14-107
14.10.2 Oxidation State 14-110
14.10.3 Hydrolysis 14-111
14.10.4 Polymerization 14-113
14.10.5 Complexation 14-114
14.10.6 Radiocolloid Interference 14-114
14.10.7 Isotope Dilution Analysis 14-115
14.10.8 Masking and Demasking 14-116
14.10.9 Review of Specific Radionuclides 14-120
14.10.9.1 Americium 14-120
14.10.9.2 Cesium 14-125
14.10.9.3 Cobalt 14-130
14.10.9.4 Iodine 14-136
14.10.9.5 Plutonium 14-144
14.10.9.6 Radium 14-153
14.10.9.7 Strontium 14-161
14.10.9.8 Technetium 14-167
14.10.9.9 Thorium 14-173
MARLAP JULY 2001
DO NOT CITE OR QUOTE XXIV DRAFT FOR PUBLIC COMMENT
-------
Contents
Page
14.10.9.10 Tritium 14-180
14.10.9.11 Uranium 14-185
14.10.9.12 Zirconium 14-196
14.11 References 14-204
14.12 Selected Bibliography 14-222
14.12.1 Inorganic and Analytical Chemistry 14-222
14.12.2 General Radiochemistry 14-223
14.12.3 Radiochemical Methods of Separation 14-223
14.12.4 Radionuclides 14-223
14.12.5 Separation Methods 14-225
15 Nuclear Counting Instrumentation 15-1
15.1 Introduction 15-1
15.2 Alpha Counting 15-2
15.2.1 Introduction 15-2
15.2.2 Detectors for Alpha Counting 15-3
15.2.2.1 lonization Chambers 15-3
15.2.2.2 Proportional Counters 15-3
15.2.2.3 Scintillation Counters 15-4
15.2.2.4 Liquid Scintillation Counters 15-5
15.2.2.5 Semiconductor Detectors 15-6
15.3 Beta Counting 15-7
15.3.1 Introduction 15-7
15.3.2 Proportional Counter 15-7
15.3.3 Liquid Scintillation 15-8
15.3.4 Solid Organic Scintillators 15-9
15.3.5 Beta Particle Counter 15-10
15.3.6 Associated Electronic Equipment 15-11
15.4 Gamma Counting 15-12
15.4.1 Introduction 15-12
15.4.2 Energy Efficiency Relationship 15-16
15.4.3 Sodium Iodide Detector Assembly 15-18
15.4.4 High Resolution Germanium Detectors 15-19
15.4.5 Low Background High Resolution Germanium Detectors 15-19
15.4.6 High Resolution Detectors for Low Energy Spectrometry 15-20
15.4.7 CsI(Tl) Detectors 15-20
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT XXV DO NOT CITE OR QUOTE
-------
Contents
Page
15.4.8 CdZnTe Detectors 15-20
15.4.9 BGO Detectors 15-21
15.5 Spectrometry Systems 15-21
15.5.1 Alpha/Gamma Coincidence Systems 15-21
15.5.2 Beta/Gamma Coincidence Systems 15-21
15.5.3 Gamma/Gamma Coincidence Systems 15-21
15.5.4 Photon-Electron Rejecting Alpha Li quid Scintillation Systems 15-22
15.6 Special Instruments 15-22
15.6.1 4-Ti Counter 15-22
15.6.2 Low-Geometry Counters 15-23
15.6.3 Internal Gas Counters 15-23
15.7 Spectrometers and Energy-Dependent Detectors 15-24
15.7.1 Anti-Coincidence Counters 15-29
15.7.2 Coincidence Counters 15-30
15.8 Shielding 15-31
15.9 Instrument Calibration 15-31
15.10 Other Considerations 15-32
15.10.1 Alpha 15-32
15.10.1.1 Troubleshooting 15-32
15.10.1.2 Calibration 15-35
15.10.1.3 Costs 15-35
15.10.1.4 Quality Control 15-37
15.10.2 Beta 15-40
15.10.2.1 Introduction 15-40
15.10.2.2 Alpha Particle Interference and Beta Energy Resolution 15-41
15.10.2.3 Liquid Scintillation Quenching 15-42
15.10.2.4 Beta Particle Attenuation 15-42
15.10.2.5 Calibration 15-44
15.10.2.6 Costs 15-44
15.10.2.7 Quality Control 15-46
15.10.3 Gamma 15-46
15.10.3.1 Troubleshooting 15-46
15.10.3.2 Calibration 15-48
15.10.3.3 Software 15-49
15.10.3.4 Costs 15-50
15.10.3.5 Quality Control 15-50
MARLAP JULY 2001
DO NOT CITE OR QUOTE XXVI DRAFT FOR PUBLIC COMMENT
-------
Contents
Page
15.10.4 Non-Nuclear Instrumentation 15-51
15.10.4.1 ICP-Mass Spectrometry 15-51
15.10.4.2 Laser 15-53
15.10.4.3 Radionuclides Analyzed By Neutron Activation 15-54
15.11 References 15-55
15.11.1 Cited References 15-55
15.11.2 Other Sources 15-61
Attachment 15A Field Measurements 15-63
15A.1 Introduction 15-63
15A.2 Analytical Level of Measurements 15-63
15A.3 Documentation of Methodology 15-65
15A.4 Instrument Operating Conditions 15-66
15A.5 Site Conditions/Limitations 15-66
15A.6 Interferences 15-67
15A.7 Calibration 15-67
15A.8 Minimum Detectable Concentrations 15-68
15A.9 Precision 15-69
15A. 10Accuracy 15-69
15 A. 11 Representativeness 15-69
15A.12Completeness 15-70
15 A. 13 Comparability 15-71
15A. 14Reference Measurements 15-71
15A.15Record Keeping 15-72
15A. 16Quality Improvement 15-72
15A. 17Management Assessment 15-73
15 A. ISCombined Laboratory and Field Measurements 15-73
15A.19References 15-73
16 Instrument Calibration and Test Source Preparation 16-1
16.1 Introduction 16-1
16.2 Instrument Calibration 16-2
16.2.1 Standards 16-3
16.2.2 Correspondence 16-3
16.2.3 Homogeneity 16-4
16.2.4 Uncertainty 16-4
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT XXVII DO NOT CITE OR QUOTE
-------
Contents
Page
16.3 General Test Source Characteristics 16-4
16.3.1 Geometrical Arrangement 16-5
16.3.2 Uniformity of Test Source Material 16-5
16.3.3 Self-Absorption and Scattering 16-6
16.3.4 CountingPlanchets 16-8
16.4 Test Source Preparation and Calibration for Alpha Measurements 16-8
16.4.1 Proportional Counters 16-9
16.4.1.1 Alpha Test Source Preparation 16-9
16.4.1.2 Proportional Counter Calibration — Alpha 16-10
16.4.2 ZnS(Ag) Scintillation Counter 16-11
16.4.3 Alpha Spectrometry With Semiconductor Detectors 16-12
16.4.4 Liquid-Scintillation Spectrometer 16-13
16.5 Characteristics of Sources for Beta Measurements 16-13
16.5.1 Proportional Counters 16-13
16.5.1.1 Beta Test Source Preparation 16-14
16.5.1.2 Proportional Counter Calibration — Beta 16-14
16.5.2 Liquid-Scintillation Spectrometers 16-15
16.5.2.1 Liquid Scintillation Test Source Preparation 16-16
16.5.2.1 Liquid-Scintillation Spectrometer Calibration 16-17
16.6 Characteristics of Sources for Gamma-Ray Measurements 16-18
16.6.1 Gamma Test Source Preparation 16-18
16.6.2 Gamma Spectrometer Calibration 16-20
16.7 Methods of Test Source Preparation 16-20
16.7.1 Electrodeposition 16-20
16.7.2 Coprecipitation 16-23
16.7.3 Evaporation 16-25
16.7.4 Thermal Volatilization/Sublimation 16-26
16.7.5 Preparing Sources to Measure Radioactive Gases 16-27
16.7.6 Preparing Air Filters for Counting 16-29
16.7.7 Preparing Swipes/Smears for Counting 16-29
16.8 References 16-30
17 Data Acquisition, Reduction, and Reporting 17-1
17.1 Introduction 17-1
17.2 Data Acquisition 17-2
17.2.1 Generic Counting Parameter Selection 17-3
MARLAP JULY 2001
DO NOT CITE OR QUOTE XXVIII DRAFT FOR PUBLIC COMMENT
-------
Contents
Page
17.2.1.1 Counting Duration 17-4
17.2.1.2 Counting Geometry 17-5
17.2.1.3 Software 17-5
17.2.2 Basic Data Reduction Calculations 17-6
17.3 Data Reduction on Spectrometry Systems 17-8
17.3.1 Gamma Spectrometry 17-8
17.3.1.1 Peak Search or Identification 17-10
Regions of Interest (ROI) Method 17-11
Gaussian Function Derivative Method 17-12
Channel Differential Method 17-12
Correlation Method 17-12
17.3.1.2 Singlet/Multiplet Peaks 17-13
17.3.1.3 Definition of Peak Centroid and Energy 17-14
17.3.1.4 Peak Width Determination 17-14
17.3.1.5 Peak Area Determination 17-16
17.3.1.6 Calibration Reference File 17-19
17.3.1.7 Activity and Concentration 17-19
17.3.1.8 Summing Considerations 17-20
17.3.1.9 Uncertainty Calculation 17-22
17.3.2 Alpha Spectrometry 17-23
17.3.2.1 Radiochemical Yield 17-27
17.3.2.2 Uncertainty Calculation 17-27
17.3.3 Liquid Scintillation Spectrometry 17-28
17.3.3.1 Overview of Liquid Scintillation Counting 17-28
17.3.3.2 Liquid Scintillation Spectra 17-29
17.3.3.3 Pulse Characteristics 17-29
17.3.3.4 Coincidence Circuitry 17-30
17.3.3.5 Quenching 17-30
17.3.3.6 Luminescence 17-30
17.3.3.7 Test Source Vials 17-31
17.3.3.8 Data Reduction for Liquid Scintillation Counting 17-31
17.4 Data Reduction on Non-Spectrometry Systems 17-32
17.5 Reporting Data 17-37
17.5.1 Sample and Analysis Method Identification 17-37
17.5.2 Units and Radionuclide Identification 17-38
17.5.3 Values, Uncertainly, and Significant Figures 17-38
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT XXIX DO NOT CITE OR QUOTE
-------
Contents
Page
17.5.4 Other Information to be Provided on Request 17-38
17.6 Data Packages 17-39
17.7 Electronic Data Deliverables 17-39
17.8 References 17-40
17.8.1 Cited References 17-40
17.8.2 Other Sources 17-42
18 Laboratory Quality Control 18-1
18.1 Introduction 18-1
18.1.1 Organization of Chapter 18-2
18.1.2 Format 18-2
18.2 Quality Control 18-3
18.3 Evaluation of Performance Indicators 18-4
18.3.1 Importance of Evaluating Performance Indicators 18-4
18.3.2 Statistical Means of Evaluating Performance Indicators — Control Charts .. 18-5
18.3.3 Measurement Uncertainty 18-7
18.4 Radiochemistry Performance Indicators 18-9
18.4.1 Method and Reagent Blank 18-9
18.4.2 Laboratory Replicates 18-13
18.4.3 Laboratory Control Samples, Matrix Spikes, and Matrix Spike Duplicates . 18-16
18.4.4 Certified Reference Materials 18-18
18.4.5 Chemical/Tracer Yield 18-21
18.5 Instrumentation Performance Indicators 18-25
18.5.1 Instrument Background Measurements 18-25
18.5.2 Efficiency Calibrations 18-27
18.5.3 Spectrometry Systems 18-31
18.5.3.1 Energy Calibrations 18-31
18.5.3.2 Peak Resolution and Tailing 18-34
18.5.4 Gas Proportional Systems 18-38
18.5.4.1 Voltage Plateaus 18-38
18.5.4.2 Self-Absorption, Backscatter, and Crosstalk 18-39
18.5.5 Liquid Scintillation 18-41
18.5.6 Summary 18-41
18.6 Related Concerns 18-43
18.6.1 Detection Capability 18-43
18.6.2 Secular Equilibrium 18-44
MARLAP JULY 2001
DO NOT CITE OR QUOTE XXX DRAFT FOR PUBLIC COMMENT
-------
Contents
Page
18.6.3 Half-Life 18-47
18.6.4 Interferences 18-48
18.6.5 Negative Results 18-50
18.6.6 Blind Samples 18-51
18.6.7 Calibration of Apparatus Used for Weight and Volume Measurements .... 18-54
18.7 References 18-55
18.7.1 Cited Sources 18-55
18.7.2 Other Sources 18-57
Attachment ISA: Control Charts 18-59
ISA. 1 Introduction 18-59
18A.2 X Charts 18-59
18A.3 Charts 18-63
18A.4 R Charts 18-64
18A.5 Control Charts for Instrument Response 18-65
18A.6 References 18-70
Attachment 18B: Statistical Tests for QC Results 18-71
18B.1 Introduction 18-71
18B.2 Tests for Excess Variance in the Instrument Response 18-71
18B.3 Instrument Background Measurements 18-78
18B.3.1 Detection of Background Variability 18-78
18B.3.2 Comparing a Single Observation to Preset Limits 18-80
18B.3.3 Comparing the Results of Consecutive Measurements 18-84
18B.4 Negative Activities 18-86
18B.5 References 18-86
19 Measurement Statistics 19-1
19.1 Overview 19-1
19.2 Statistical Concepts and Terms 19-2
19.2.1 Basic Concepts 19-2
19.2.2 Summary of Terms 19-5
19.3 Measurement Uncertainty 19-7
19.3.1 Measurement, Error, and Uncertainty 19-7
19.3.2 The Measurement Process 19-8
19.3.3 Analysis of Measurement Uncertainty 19-10
19.3.4 Corrections for Systematic Effects 19-11
19.3.5 Counting Uncertainty 19-11
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT XXXI DO NOT CITE OR QUOTE
-------
Contents
Page
19.3.6 Expanded Uncertainty 19-11
19.3.7 Significant Figures 19-12
19.3.8 Reporting the Measurement Uncertainty 19-13
19.3.9 Recommendations 19-14
19.3.10 Summary of Terms 19-14
19.4 Detection and Quantification Capability 19-17
19.4.1 Analyte Detection Decisions 19-17
19.4.2 The Minimum Detectable Concentration 19-19
19.4.3 Differences between the ISO and IUPAC Definitions 19-22
19.4.4 Other Detection Terminologies 19-23
19.4.5 The Minimum Quantifiable Concentration 19-24
19.4.6 Recommendations 19-25
19.4.7 Summary of Terms 19-25
19.5 Procedures for Estimating Uncertainty 19-27
19.5.1 Identifying Sources of Uncertainty 19-27
19.5.2 Evaluation of Standard Uncertainties 19-28
19.5.2.1 Type A Evaluations 19-28
19.5.2.2 Type B Evaluations 19-31
19.5.3 Combined Standard Uncertainty 19-33
19.5.4 The Estimated Covariance of Two Output Estimates 19-37
19.5.5 Nonlinear Models 19-38
19.5.5.1 Uncertainty Propagation 19-38
19.5.5.2 Bias 19-40
19.5.5.3 Nominal Values 19-42
19.6 Radiation Measurement Uncertainty 19-43
19.6.1 Radioactive Decay 19-43
19.6.2 Radiation Counting 19-44
19.6.3 Count Rate 19-47
19.6.3.1 Dead Time 19-48
19.6.3.2 A Confidence Interval for the Count Rate 19-49
19.6.4 Instrument Background 19-50
19.6.5 Counting Efficiency 19-52
19.6.6 Radionuclide Half-life 19-56
19.6.7 Gamma Spectrometry 19-56
19.6.8 Balances 19-56
19.6.9 Pipets and Other Volumetric Apparatus 19-58
MARLAP JULY 2001
DO NOT CITE OR QUOTE XXXII DRAFT FOR PUBLIC COMMENT
-------
Contents
Page
19.6.10 Digital Displays and Rounding 19-62
19.6.11 Subsampling 19-63
19.6.12 The Standard Uncertainty for a Hypothetical Measurement 19-64
19.7 Detection and Quantification Limits 19-65
19.7.1 Calculation of the Critical Value 19-65
19.7.1.1 Normally Distributed Signals 19-66
19.7.1.2 Poisson Counting 19-67
19.7.1.3 Reagent Blanks 19-70
19.7.2 Calculation of the Minimum Detectable Concentration 19-71
19.7.2.1 The Minimum Detectable Net Instrument Signal 19-72
19.7.2.2 Normally Distributed Signals 19-72
19.7.2.3 Poisson Counting 19-74
19.7.2.4 The MDC 19-75
19.7.2.5 Regulatory Requirements 19-76
19.7.2.6 Testing the MDC 19-77
19.7.3 Calculation of the Minimum Quantifiable Concentration 19-78
19.8 References 19-80
19.8.1 Cited Sources 19-80
19.8.2 Other Sources 19-84
Attachment 19A Distributions 19-85
19A.1 Introduction 19-85
19A.2 Normal Distributions 19-85
19A.3 Log-normal Distributions 19-86
19A.4 Chi-square Distributions 19-87
19A.5 T-Distributions 19-88
19A.6 Rectangular Distributions 19-90
19A.7 Trapezoidal and Triangular Distributions 19-91
19A.8 Exponential Distributions 19-92
19A.9 Binomial Distributions 19-92
19A.10 Poisson Distributions 19-93
19A.11 References 19-95
Attachment 19B Multicomponent Analyses 19-97
19B. 1 Matrix Equations 19-97
19B.2 Random Vectors and Matrices 19-98
19B.3 Linear Least Squares 19-99
19B.4 General Least Squares 19-101
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT XXXIII DO NOT CITE OR QUOTE
-------
Contents
Page
19B.5 The Covariance Matrix for a Least-Squares Solution 19-102
19B.6 Critical Values 19-103
19B.7 Detection and Quantification Limits 19-104
19B.8 References 19-104
Attachment 19C Estimation of Coverage Factors 19-105
19C.1 Introduction 19-105
19C.2 Procedure 19-105
19C.3 Poisson Counting Uncertainty 19-106
19C.4 References 19-107
Attachment 19D Low-Background Detection Limits 19-109
19D.1 Overview 19-109
19D.2 Calculation of the Critical Value 19-109
19D.2.1 Normally Distributed Signals 19-109
19D.2.2 Poisson Counting 19-110
19D.3 Calculation of the Minimum Detectable Concentration 19-123
19D.3.1 The Minimum Detectable Net Instrument Signal 19-123
19D.3.2 Normally Distributed Signals 19-123
19D.3.3 Poisson Counting 19-126
19D.4 References 19-132
Attachment 19E Example Calculations 19-135
19E.1 Overview 19-135
19E.2 Sample Collection and Analysis 19-135
19E.3 The Measurement Model 19-136
19E.4 The Combined Standard Uncertainty 19-138
19E.5 The Critical Net Count 19-140
19E.6 The Minimum Detectable Concentration 19-143
19E.7 The Minimum Quantifiable Concentration 19-148
Attachment 19F Tests for Normality 19-149
19F.1 Purpose 19-149
19F.2 Normal Probability Plots 19-149
19F.3 Filliben's Statistic 19-151
19F.4 References 19-155
Attachment 19G Balance Measurement Uncertainty 19-157
19G.1 Purpose 19-157
19G.2 Considerations 19-157
19G.3 Repeatability 19-157
MARLAP JULY 2001
DO NOT CITE OR QUOTE XXXIV DRAFT FOR PUBLIC COMMENT
-------
Contents
Page
19G.4 Environmental Factors 19-158
19G.5 Calibration 19-159
19G.6 Linearity 19-160
19G.7 Air Buoyancy Corrections 19-160
19G.8 Combining the Components 19-164
19G.9 References 19-165
20 Waste Management in a Radioanalytical Laboratory 20-1
20.1 Introduction 20-1
20.2 Types of Laboratory Wastes 20-1
20.3 Waste Management Program 20-2
20.3.1 Program Integration 20-3
20.3.2 Staff Involvement 20-3
20.4 Waste Minimization 20-4
20.5 Waste Determinations and Characterization 20-6
20.6 Specific Waste Management Requirements 20-7
20.6.1 Sample/Waste Exemptions 20-9
20.6.2 Storage 20-10
20.6.2.1 Container Requirements 20-11
20.6.2.2 Labeling Requirements 20-11
20.6.2.3 Time Constraints 20-11
20.6.2.4 Monitoring Requirements 20-12
20.6.3 Treatment 20-12
20.6.4 Disposal 20-13
20.7 Contents of a Laboratory Waste Management Plan/Certification Plan 20-14
20.7.1 Laboratory Waste Management Plan 20-14
20.7.2 Waste Certification Plan/Program 20-14
20.8 Useful Web Sites 20-16
20.9 References 20-17
20.9.1 Cited References 20-17
20.9.2 Other Sources 20-18
Glossary to be added follow ing public review
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT XXXV DO NOT CITE OR QUOTE
-------
Contents
Page
Appendices
Appendix A: Directed Planning Approaches A-l
A. IDirected Planning Approaches A-l
A.2Elements Common to Directed Planning Approaches A-l
A.SData Quality Objectives Process A-2
A.4Observational Approach A-3
A. 5 Streamlined Approach for Environmental Restoration A-4
A.6Technical Project Planning A-4
A.TExpedited Site Characterization A-5
A.SValue Engineering A-5
A.9Systems Engineering A-6
A. 10 Total Quality Management A-7
A. 11 Partnering A-7
A. 12 References A-8
A.12.1 Data Quality Objectives A-8
A. 12.2 Observational Approach A-10
A. 12.3 Streamlined Approach for Environmental Restoration (Safer) A-10
A. 12.4 Technical Project Planning A-l 1
A. 12.5 Expedited Site Characterization A-ll
A. 12.6 Value Engineering A-13
A. 12.7 Systems Engineering A-14
A. 12.8 Total Quality Management A-16
A. 12.9 Partnering A-17
Appendix B: The Data Quality Objectives Process B-l
Bl.O Introduction B-l
B2.0 Overview of the DQO Process B-2
B3.0 The Seven Steps of the DQO Process B-3
B3.1 DQO Process Step 1: State the Problem B-3
B3.2 DQO Process Step 2: Identify the Decision B-4
B3.3 DQO Process Step 3: Identify Inputs to the Decision B-5
B3.4 DQO Process Step 4: Define the Study Boundaries B-7
B3.5 Outputs of DQO Process Steps 1 to 4 Lead Into Steps 5 to 7 B-8
B3.6 DQO Process Step 5: Develop a Decision Rule B-8
B3.7 DQO Process Step 6: Specify the Limits on Decision Errors B-9
MARLAP JULY 2001
DO NOT CITE OR QUOTE XXXVI DRAFT FOR PUBLIC COMMENT
-------
Contents
Page
B3.8 DQO Process Step 7: Optimize the Design for Obtaining Data B-12
B3.9 References B-14
Attachment B-l Decision Error Rates And The Gray Region B-16
B-l.l Introduction B-16
B-1.2 The Region of Interest B-16
B-l.3 Measurement Uncertainty at the Action Level B-17
B-1.4 The Null Hypothesis B-18
Case 1: Assume The True Concentration is Over 1.0 B-18
Case 2: Assume The True Concentration is 0.9 B-20
B-l.5 The Critical Region B-20
B-l.6 The Gray Region B-21
Appendix C: Measurement Quality Objectives for Method Uncertainty And Detection and
Quantification Capability C-l
C.I Introduction C-l
C.2 Hypothesis Testing C-2
C.3 Development of MQOs for Analytical Protocol Selection C-4
C.4 The Role of the MQO for Method Uncertainty in Data Evaluation C-9
C.4.1 Uncertainty Requirements at Various Concentrations C-9
C.4.2 Acceptance Criteria for Quality Control Samples C-12
C.5 References C-19
Appendix D Content of Project Plan Documents D-l
Dl.O Introduction D-l
D2.0 Group A: Project Management D-3
D2.1 Project Management (Al): Title and Approval Sheet D-6
D2.2 Project Management (A2): Table of Contents D-7
D2.3 Project Management (A3): Distribution List D-7
D2.4 Project Management (A4): Project/Task Organization D-8
D2.5 Project Management (A5): Problem Definition/Background D-8
D2.6 Project Management (A6): Project/Task Description D-10
D2.7 Project Management (A7): Quality Objectives and Criteria for
Measurement Data D-12
D2.7.1 Project's Quality Objectives D-12
D2.7.2 Specifying Measurement Quality Objectives D-13
D2.7.3 Relation between the Project DQOs, MQOs, and QC Requirements . . . D-14
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT XXXVII DO NOT CITE OR QUOTE
-------
Contents
Page
D2.8 Project Management (A8): Special Training Requirements/Certification . . . D-14
D2.9 Project Management (A9): Documentation and Record D-14
D3.0 Group B: Measurement/Data Acquisition D-16
D3.1 Measurement/Data Acquisition (Bl): Sampling Process Design D-16
D3.2 Measurement/Data Acquisition (B2): Sampling Methods Requirements ... D-18
D3.3 Measurement/Data Acquisition (B3): Sample Handling and Custody
Requirements D-20
D3.4 Measurement/Data Acquisition (B4): Analytical Methods Requirements .. D-21
D3.5 Measurement/Data Acquisition (B5): Quality Control Requirements D-23
D3.6 Measurement/Data Acquisition (B6): Instrument/Equipment Testing, Inspection,
and Maintenance Requirements D-24
D3.7 Measurement/Data Acquisition (B7): Instrument Calibration and Frequency D-25
D3.8 Measurement/Data Acquisition (B8): Inspect!on/Acceptance Requirements for
Supplies and Consumables D-26
D3.9 Measurement/Data Acquisition (B9): Data Acquisition Requirements for Non-
Direct Measurement Data D-27
D3.10 Measurement/Data Acquisition (BIO): Data Management D-28
D4.0 Group C: Assessment/Oversight D-29
D4.1 Assessment/Oversight (Cl): Assessment and Response Actions D-29
D4.2 Assessment/Oversight (C2): Reports To Management D-30
D5.0 Group D: Data Validation and Usability D-31
D5.1 Data Validation and Usability (Dl): Verification and Validation
Requirements D-31
D5.2 Data Validation and Usability (D2): Verification and Validation Methods . D-32
D5.2.1 Data Verification D-32
D5.2.2 Data Validation D-33
D5.3 Data Validation and Usability (D3): Reconciliation with Data Quality
Objectives D-34
D6.0 References D-35
Appendix E: Contracting Laboratory Services E-l
E. 1 Introduction E-l
E.2 Procurement of Services E-5
E.2.1 Request for Approval of Proposed Procurement Action E-6
E.2.2 Types of Procurement Mechanisms E-6
E.3 Request for Proposals—The Solicitation E-8
MARLAP JULY 2001
DO NOT CITE OR QUOTE XXXVIII DRAFT FOR PUBLIC COMMENT
-------
Contents
Page
E.3.1 Market Research E-9
E.3.2 Length of Contract E-10
E.3.3 Subcontracts E-10
E.4Proposal Requirements E-l 1
E.4.1 RFP and Contract Information E-l 1
E.4.2 Personnel E-14
E.4.3 Instrumentation E-17
E.4.3.1 Type, Number, and Age of Laboratory Instruments E-18
E.4.3.2 Service Contract E-18
E.4.4 Narrative to Approach E-18
E.4.4.1 Analytical Methods or Protocols E-18
E.4.4.2Meeting Contract Measurement Quality Objectives E-19
E.4.4.3 Data Package E-19
E.4.4.4 Schedule E-19
E.4.4.5 Sample Storage and Disposal E-20
E.4.5 Quality Manual E-21
E.4.6 Licenses and Accreditations E-22
E.4.7 Experience E-22
E.4.7.1 Previous or Current Contracts E-23
E.4.7.2 Quality of Performance E-23
E.5 Proposal Evaluation and Scoring Procedures E-23
E.5.1 Evaluation Committee E-24
E.5.2 Ground Rules — Questions E-24
E.5.3 Scoring/Evaluation Scheme E-24
E.5.3.1 Review of Technical Proposal and Quality Manual E-26
E.5.3.2 Review of Laboratory Accreditation E-28
E.5.3.3 Review of Experience E-28
E.5.4 Pre-Award Proficiency Samples E-28
E.5.5 Pre-Award Audit E-29
E.5.6 Comparison of Prices E-33
E.5.7 Debriefing of Unsuccessful Vendors E-34
E.6The Award E-34
E.7For the Duration of the Contract E-35
E.7.1 Managing a Contract E-35
E.7.2 Responsibility of the Contractor E-36
E.7.3 Responsibility of the Agency E-36
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT XXXIX DO NOT CITE OR QUOTE
-------
Contents
Page
E.7.4 Anomalies and Nonconformance E-36
E.7.5 Laboratory Assessment E-37
E.7.5.1 Performance and Quality Control Samples E-37
E.7.5.2 Laboratory Performance Evaluation Programs E-38
E.7.5.3 Laboratory Evaluations Performed During the Contract Period E-39
E.8 Contract Completion E-40
E.9References E-41
Appendix F Laboratory Subsampling F-l
F.I Introduction F-l
F.2 Basic Concepts F-2
F.3 Sources of Measurement Error F-4
F.3.1 Sampling Bias F-4
F.3.2 Fundamental Error F-5
F.3.3 Grouping and Segregation Error F-7
F.4 Implementation of the Particulate Sampling Theory F-10
F.4.1 The Fundamental Variance F-10
F.4.2 Scenario 1 -Natural Radioactive Minerals F-ll
F.4.3 Scenario 2 - Hot Particles F-12
F.4.4 Scenario 3 -Particle Surface Contamination F-14
F.5 Summary F-16
F.6 References F-17
G Statistical Tables G-l
MARLAP JULY 2001
DO NOT CITE OR QUOTE XL DRAFT FOR PUBLIC COMMENT
-------
Contents
Page
List of Figures
Figure 1.1 The Data Life Cycle 1-5
Figure 1.2 Typical Components of an Analytical Process 1-7
Figure 1.3 The MARLAP Process 1-16
Figure 3.1 Typical components of an analytical process 3-3
Figure 3.2 Analytical protocol specifications 3-28
Figure 3.3 Example analytical protocol specifications 3-29
Figure 6.1 Analytical process 6-3
Figure 6.2 Method application life cycle 6-6
Figure 6.3 Expanded Figure 6.2 addressing the laboratory's method evaluation process .... 6-7
Figure 7.1 Considerations for the initial evaluation of a laboratory 7-18
Figure 8.1 The Assessment Process 8-6
Figure 9.1 Using physical samples to measure a characteristic of the population representatively 1
Figure 9.2 Types of sampling and analytical errors 9-18
Figure 10.1 Example of chain-of-custody record 10-10
Figure 11.1 Overview of sample receipt, inspection, and tracking 11-2
Figure 12.1 Degree of error in laboratory sample preparation (Scwedt, 1997) 12-1
Figure 12.2 Laboratory Sample Preparation Flowchart (for Solid Samples) 12-13
Figure 14.1 Ethylenediaminetetraacetic Acid (1)(EDTA) 14-22
Figure 14.2 Crown ethers 14-23
Figure 14.3 The behavior of elements in concentrated hydrochloric acid on cation-exchange
resins 14-57
Figure 14.4 behavior of elements in concentrated hydrochloric acid on anion-exchange resinsJ-58
Figure 14.5 The electrical double layer 14-86
Figure 15.1 Gamma-ray Interactions with Germanium 15-12
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT XLI DO NOT CITE OR QUOTE
-------
Contents
Page
Figure 15.2 Gamma-ray Spectra of 60Co 15-13
Figure 15.3 Energy Spectrum of 22Na 15-15
Figure 15.4 Efficiency vs. Gamma-ray Energy 15-16
Figure 15.5 Standard Cryostat FtPGe Background Spectrum 15-20
Figure 15.6 Low Background Cryostat HPGe Background Spectrum 15-20
Figure 15.7 Nal(Tl) Energy Spectrum of 137Cs 15-26
Figure 15.8 HPGe Energy Spectrum of 137Cs 15-27
Figure 15.9 Spectrum of 210Pb, 210Bi, and 210Po 15-29
Figure 15.10 Range vs. Energy for Alpha Particles in Air 15-35
Figure 15.11 Range vs. Energy for Beta Particles in Air and Water 15-43
Figure 15.12 Beta Detector Efficiency Curve for 131I vs. Weight 15-43
Figure 15.13 Beta-gamma coincidence efficiency curve for 129I 15-55
Figure 17.1 Gamma-ray spectrum 17-9
Figure 17.2 Gamma-ray analysis sequence 17-11
Figure 17.3 Low-energy tailing 17-16
Figure 17.4 Photopeak baseline continuum 17-17
Figure 17.5 Photopeak baseline continuum-step function 17-18
Figure 17.6 Alpha spectrum 17-23
Figure 18.1 Control chart for daily counting of a standard reference source, with limits corrected
for decay 18-7
Figure 18.2 Three general categories of blank changes 18-12
Figure 18.3 Failed performance indicator: replicates 18-15
Figure 18.4 Failed performance indicator: chemical yield 18-23
Figure 19.1 A symmetric distribution 19-4
Figure 19.2 An asymmetric distribution 19-4
Figure 19.3 The critical value xc and minimum detectable value XD
of the net state variable 19-20
Figure 19.4 Expected fraction of atoms remaining at time t 19-44
Figure 19.5 A normal distribution 19-85
Figure 19.6 A log-normal distribution 19-86
Figure 19.7 Chi-square distributions 19-88
Figure 19.8 The ^-distribution with 3 degrees of freedom 19-89
Figure 19.9 A rectangular distribution 19-91
MARLAP JULY 2001
DO NOT CITE OR QUOTE XLII DRAFT FOR PUBLIC COMMENT
-------
Contents
Page
Figure 19.10 A trapezoidal distribution 19-91
Figure 19.11 An exponential distribution 19-92
Figure 19.12 Type I error rate for the Poisson-normal approximation (tB = ts) 19-113
Figure 19.13 Type I error rates for Formula A 19-115
Figure 19.14 Type I error rates for Formula B 19-116
Figure 19.15 Type I error rates for Formula C 19-118
Figure 19.16 Type I error rates for the Stapleton approximation 19-119
Figure 19.17 Type I error rates for the nonrandomized exact test 19-121
Figure 19.18 Example: Normal probability plot 19-154
Figure Bl Seven steps of the DQO process B-2
Figure B2(a) Decision performance goal diagram null hypothesis: the parameter exceeds the
action level B-l 1
Figure B2(b) Decision performance goal diagram null hypothesis: the parameter is less than the
action level B-l 1
Figure B3 How Proximity to the action level determines what is an acceptable level of
uncertainty B-13
Figure C.I Required Analytical Standard Deviation (aReq) C-10
Figure E-l General Sequence Initiating and Later Conducting Work with a Contract LaboratoBy-4
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT XLIII DO NOT CITE OR QUOTE
-------
Contents
Page
List of Tables
Table 2.1 Summary of the directed planning process and radioanalytical specialists
participation 2-10
Table 3.1 Matrix-specific analytical planning issues 3-23
Table 4.1 Elements of project plan documents 4-7
Table 4.2 Crosswalk between project plan document elements and directed planning process4-ll
Table 6.1 Tiered method validation approach 6-28
Table 7.1 Cross reference of information available for method evaluation 7-4
Table 9.1 Summary of the DQA process 9-6
Table 10.1 Summary of sample preservation techniques 10-25
Table 11.1 Typical topics addressed in standard operating procedures related to sample receipt,
inspection, and tracking 11-4
Table 12.1 Examples of volatile radionuclides 12-4
Table 12.2 Properties of sample container materials 12-5
Table 12.3 Examples of dry-ashing temperatures (platinum container) 12-23
Table 13.1 Common fusion fluxes 13-7
Table 13.2 Examples of acids used for wet ashing 13-14
Table 13.3 Standard reduction potentials of selected half-reactions at 25 °C 13-14
Table 14.1 Oxidation states of elements 14-9
Table 14.2 Stable oxidation states of selected elements 14-10
Table 14.3 Redox reagents for radionuclides 14-14
Table 14.5 Radioanalytical methods employing solvent extraction 14-35
Table 14.6 Radioanalytical methods employing extraction chromatography 14-36
Table 14.7 Elements separable by volatilization as certain species 14-41
Table 14.8 Typical functional groups of ion-exchange resins 14-54
MARLAP JULY 2001
DO NOT CITE OR QUOTE XLIV DRAFT FOR PUBLIC COMMENT
-------
Contents
Page
Table 14.9 Common ion-exchange resins 14-55
Table 14.10 General solubility behavior of some cations of interest 14-63
Table 14.11 Summary of methods for utilizing precipitation from
homogeneous solution 14-74
Table 14.12 Influence of precipitation conditions on the purity of precipitates 14-76
Table 14.13 Common coprecipitating agents for radionuclides 14-83
Table 14.14 Coprecipitation behavior of plutonium and neptunium 14-85
Table 14.15 General properties of common filter papers 14-89
Table 14.16 Atoms and mass of select radionuclides equivalent to 500 dpm 14-91
Table 14.17 Masking agents for ions of various metals 14-117
Table 14.18 Masking agents for anions and neutral molecules 14-119
Table 15.1 Typical percent gamma-ray efficiencies for a 55 percent high-purity germanium
detector with various counting geometries 15-17
Table 15.2 Nuclides for gamma-ray spectrometer calibration 15-48
Table 16.1 Nuclides for alpha calibration 16-10
Table 16.2 Nuclides for beta calibration 16-15
Table 17.1 Units for data reporting 17-38
Table 18.1 Problems leading to loss of analytical control 18-3
Table 18.2a Certified Massic activities for natural radionuclides
with a normal distribution of measurement results 18-20
Table 18.2b Certified Massic activities for anthropogenic radionuclides with a Weibull
distribution of measurement results 18-20
Table 18.2c Uncertified Massic activities 18-20
Table 18.3 Instrument background evaluation 18-27
Table 18.4 Root cause analysis of performance check results 18-37
Table 18.5 Instrument calibration: example frequency and performance criteria 18-41
Table 18A.I Bias-correction factor for the experimental standard deviation 18-60
Table 19.1 Applications of the uncertainty propagation formula 19-34
Table 19.2 Density of air-free water 19-60
Table 19.3 95% confidence interval for a Poisson mean 19-94
Table 19.4 Critical gross count (well-known blank) 19-111
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT XLV DO NOT CITE OR QUOTE
-------
Contents
Page
Table 19.5 Bias factor for the experimental standard deviation 19-125
Table 19.6 Estimated and true values of SD (tB = tS) 19-131
Table 19.7 Input estimates and standard uncertainties 19-139
Table 20.1 Examples of laboratory-generated wastes 20-2
Table Dl QAPP groups and elements D-2
Table D2 Comparison of project plan contents D-3
Table D3 Content of the three elements that constitute the project description D-9
Table E.I Examples of procurement options to obtain materials or services E-7
Table E.2 SOW checklists for the agency and proposer E-13
Table E.3 Laboratory technical supervisory personnel listed by position title and examples for
Table suggested minimum qualifications E-15
Table E.4 Laboratory technical personnel listed by position title and examples for suggested
minimum qualifications and examples of optional staff members E-16
Table E.5 Laboratory technical staff listed by position title and examples for suggested
minimum qualifications E-16
Table E.6 Example of a proposal evaluation plan E-26
Table G.I Quantiles of the standard normal distribution G-l
Table G.2 Quantiles of Student's t distribution G-3
Table G.3 Quantiles of chi-square G-5
Table G.4 Critical values for the nonrandomized exact test G-7
Table G.5 Critical values of Filliben's statistic G-l 1
Table G.6 Summary of probability distributions G-12
MARLAP JULY 2001
DO NOT CITE OR QUOTE XLVI DRAFT FOR PUBLIC COMMENT
-------
ACRONYMS AND ABBREVIATIONS
Note: Bracketed numbers following each definition represent the first chapter in which the acronym appears.
ADC analog to digital convenor [18]
AEA Atomic Energy Act [20]
AL action level [C]
ANSI American National Standards Institute [1]
AOAC Association of Official Analytical Chemists [3]
APHA American Public Health Association [6]
APS analytical protocol specification [1]
ARARs applicable or relevant and appropriate requirements (CERCLA/Superfund) [D]
ASL analytical support laboratory [15]
ASQC American Society for Quality Control [2]
ASTM American Society for Testing and Materials [1]
AID alpha track detector [10]
BOA basic ordering agreement [4]
CAA Clean Air Act [20]
CBD Commerce Business Daily [E]
CC charcoal canisters [10]
CEDE committed effective dose equivalent [2]
CERCLA .... Comprehensive Environmental Response, Compensation, and Liability Act of
1980 (Superfund) [2]
CFM cubic feet per minute [16]
CFR Code of Federal Regulations [20]
CL central line (of a control chart) [15]
CMPO [octyl(phenyl)]-N,N-diisobutylcarbonylmethylphosphine oxide [14]
CMST Characterization, Monitoring, and Sensor Technology Program (DOE) [A]
COC chain of custody [2]
COR contracting officer's representative [5]
cpm counts per minute [12]
cps counts per second [15]
CRM continuous radon monitor [10]
CRM certified reference material [18]
CWA Clean Water Act [20]
CWLM continuous working level monitor [10]
JULY 2001
DRAFT FOR PUBLIC COMMENT
XL VII
MARLAP
DO NOT CITE OR QUOTE
-------
Acronyms and Abbreviations
DAAP diamylamylphosphonate [14]
DCGL derived concentration guideline level [2]
DIN di-isopropylnaphthalene [16]
DL discrimination limit [C]
DoD U.S. Department of Defense [1]
DOE U.S. Department of Energy [1]
DOELAP DOE Lab Accreditation Program [18]
DOT U.S. Department of Transportation [5]
DPM disintegrations per minute [12]
DPPP dipentylpentylphosphonate [14]
DQA data quality assessment [1]
DQI data quality indicators [3]
DQO data quality objective [1]
DTP A di ethyl ene triamine penta-acetic acid [10]
DVB divinylbenzene [14]
EDD electronic data deliverables [17]
EDTA ethyl ene diamine tetra acetic acid [10]
EGTA ethyleneglycol bis(2-aminoethylether)-tetraacetate [14]
EPA U.S. Environmental Protection Agency [1]
ERPRHVIS . . . Environmental Resources Program Management System (U.S. Air Force) [17]
ESC expedited site characterization [A]
eV electron volts [15]
FAR Federal Acquisition Regulations [E]
FDA U.S. Food and Drug Administration [1]
FWFDVI full width of a peak at half maximum [8]
FWTM full width of a peak at tenth maximum [18]
GC gas chromatography [14]
GLPC gas-liquid phase chromatography [14]
GM Geiger-Mueller detector [11]
GUM Guide to the Expression of Uncertainty in Measurement [1]
HDBP dibutylphosphoric acid [14]
HDEHP bis(2-ethylhexyl) phosphoric acid [16]
HDEHP diethylhexylphosphoric acid [14]
MARLAP
DO NOT CITE OR QUOTE
XL VIII
JULY 2001
DRAFT FOR PUBLIC COMMENT
-------
Acronyms and Abbreviations
HDPE high density polyethylene [10]
HPGe high-purity germanium [semiconductor] [15]
HPLC high-pressure liquid chromatography; high-performance liquid chromatography
[14]
HTRW hazardous, toxic and radioactive waste [10]
ICP-MS inductively coupled plasma-mass spectroscopy [14]
IPPD integrated product and process development [A]
ISO International Organization for Standardization [1]
IUPAC International Union of Pure and Applied Chemistry [1]
LAN local area network [17]
LBGR lower boundary of the gray region [B]
LCL lower control limit [18]
LCS laboratory control samples [3]
LDPE low density polyethylene [10]
LEGe low energy germanium [15]
LEVIS Laboratory Information Management System [17]
LLD lower limit of detection [19]
LLRW low-level radioactive waste [20]
LLRWPA .... Low Level Radioactive Waste Policy Act [20]
LOMI low oxidation-state transition-metal ion [10]
LPC liquid partition chromatography; liquid-phase chromatography [14]
LS liquid scintillation [15]
LSC liquid scintillation counting [15]
LWL lower warning limit [18]
MAPEP Mixed Analyte Performance Evaluation Program [DOE] [5]
MARSSEVI . . . Multi-Agency Radiation Survey and Site Investigation Manual [1]
MCA multichannel analyzer [15]
MDA minimum detection analysis [15]; minimum detectable amount [7]
MDC minimum detectable concentration [3]
MDL method detection limit [19]
MDC minimum detectable concentration [2]
MIBK methyl isobutyl ketone [14]
MQC minimum quantifiable concentration [3]
MQO measurement quality objective [1]
JULY 2001
DRAFT FOR PUBLIC COMMENT
XLIX
MARLAP
DO NOT CITE OR QUOTE
-------
Acronyms and Abbreviations
MS
MSD
MVRM
NELAC
NESHAP
NIST
NRC
NRTP
NTAorNTTA
NTU
NVLAP
OA
OFHC
OFPP
PARCC
PCB
PDF
PE
PFA
PIC
PT
PTFE
PUREX
PVC
QA
QAP
QAPP
QC
RCRA
REE
REGe
RFP
matrix spike [8]
matrix spike duplicate [8]
method validation reference material [5]
National Environmental Laboratory Accreditation Conference [5]
National Emission Standards for Hazardous Air Pollutants [12]
National Institute of Standards and Technology [1]
U.S. Nuclear Regulatory Commission [1]
NIST Radiochemistry Intercomparison Program [ 18]
nitrilotriacetate [14]
nephelometric turbidity units [10]
National Voluntary Laboratory Accreditation Program (NIST) [5]
observational approach [A]
oxygen-free high-conductivity [15]
Office of Federal Procurement Policy [E]
precision, accuracy, representativeness, completeness, and comparability [3]
polychlorinated biphenyl [20]
probability density function [19]
performance evaluation [5]
perfluoroalcoholoxil™ [13]
pressurized ionization chamber [15]
performance testing [5]
polytetrafluoroethylene [12]
plutonium uranium reduction extraction [14]
polyvinyl chloride [10]
quality assurance [2]
Quality Assessment Program (DOE) [5]
quality assurance project plan [1]
quality control [1]
Resource Conservation and Recovery Act [15]
rare earth elements [13]
reverse-electrode germanium [semiconductor] [15]
request for proposals [5]
MARLAP
DO NOT CITE OR QUOTE
JULY 2001
DRAFT FOR PUBLIC COMMENT
-------
Acronyms and Abbreviations
RFQ request for quotations [E]
RMDC required minimum detectable concentration [8]
ROI regions of interest [17]
RPD relative percent difference [7]
RPM Remedial Project Manager [2]
RSD relative standard deviation [19]
RSO Radiation Safety Officer [11]
SA spike activity [7]
SAFER streamlined approach for environmental restoration (DOE) [2]
SAM Site Assessment Manager [2]
SAP sampling and analysis plan [1]
SI international system of units [3]
SMO sample management office [2]
SOP standard operating procedure [4]
SOW Statement of Work [1]
SQC statistical quality control [15]
SR unspiked sample result [7]
SRM standard reference material [18]
SSR spiked sample result [7]
TAT turnaround time [7]
TBP tributyl phosphate [14]
TC to contain [glassware] [18]
TCLP toxicity characteristic leaching procedure [13]
TD to deliver [glassware] [18]
TEC technical evaluation committee [5]
TEDE total effective dose equivalent [2]
TES technical evaluation sheet (USGS) [5]
TFM tetrafluorometoxil™ [13]
TIOA tri-iso-octylamine [14]
TLD thermoluminescent dosimeter [10]
TOPO trioctylphosphinic oxide [14]
TPO Technical Project Officer [2]
TPP technical project planning [2]
TPU total propagated uncertainty [19]
TQM Total Quality Management [A]
JULY 2001
DRAFT FOR PUBLIC COMMENT
LI
MARLAP
DO NOT CITE OR QUOTE
-------
Acronyms and Abbreviations
TRUEX trans uranium extraction [14]
TSCA Toxic Substances Control Act [20]
TSDF treatment, storage, or disposal facility [20]
TTA thenoyltrifluoroacetone [14]
UBGR upper bound of the gray region [7]
UCL upper control limit [18]
USGS United States Geological Survey [1]
UWL upper warning limit [18]
V volts [15]
WCP waste certification plan [20]
XtGe extended-range germanium [semiconductor] [15]
MARLAP
DO NOT CITE OR QUOTE
LII
JULY 2001
DRAFT FOR PUBLIC COMMENT
-------
1 INTRODUCTION TO MARLAP
2 1.1 Overview
3 Each year, hundreds of millions of dollars are spent on projects and programs that rely, to varying
4 degrees, on radioanalytical data for decision-making. These decisions often have a significant
5 impact on human health and the environment. Of critical importance to informed decision-
6 making are data of known quality appropriate for its intended use. Making incorrect decisions
7 due to data inadequacies, such as failing to remediate a radioactively contaminated site,
8 necessitates the expenditure of additional resources, causes delays in project completions and,
9 depending on the nature of the project, can result in the loss of public trust and confidence. The
10 Multi-Agency Radiological Laboratory Analytical Protocols (MARLAP) Manual addresses the
11 need for a nationally consistent approach to producing radioanalytical laboratory data that meet a
12 project's or program's data requirements. MARLAP provides guidance for the planning,
13 implementation, and assessment phases of those projects that require the laboratory analysis of
14 radionuclides. The guidance provided by MARLAP is both scientifically rigorous and flexible
15 enough to be applied to a diversity of projects and programs. This guidance is intended for
16 project planners, managers, and laboratory personnel.
17 MARLAP is divided into two main parts. Part I is primarily for project planners and managers
18 and provides guidance on project planning with emphasis on analytical planning issues and
19 analytical data requirements. Part I also provides guidance on preparing project plan documents
20 and radioanalytical statements of work (SOWs), obtaining and evaluating radioanalytical
21 laboratory services, data validation, and data quality assessment. Part I of MARLAP covers the
22 entire life of a proj ect that requires the laboratory analysis of radionuclides from the initial
23 project planning phase to the assessment phase.
24 Part II of MARLAP is primarily for laboratory personnel and provides guidance in the relevant
25 areas of radioanalytical laboratory work. Part II offers information on the laboratory analysis of
26 radionuclides. It provides guidance on a variety of activities performed at radioanalytical
27 laboratories including sample preparation, sample dissolution, chemical separations, instrument
28 measurements, data reduction, etc. Note that Part II of the manual is not a compilation of
29 analytical procedures. While the chapters in Part n do not contain detailed step-by-step
30 instructions of how to perform certain laboratory tasks, they do provide information on many of
31 the options available for these tasks, and discuss advantages and disadvantages of each.
32 MARLAP was developed collaboratively by the following Federal agencies: the Environmental
33 Protection Agency (EPA), the Department of Energy (DOE), the Nuclear Regulatory
34 Commission (NRC), the Department of Defense (DOD), the National Institute of Standards and
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 1-1 DO NOT CITE OR QUOTE
-------
Introduction to MARLAP
35 Technology (NIST), the United States Geological Survey (USGS), and the Food and Drug
36 Administration (FDA). State participation in the development of MARLAP involved
37 contributions from representatives from the Commonwealth of Kentucky and the State of
38 California.
39 1.2 Purpose of the Manual
40 MARLAP's basic goal is to provide guidance and a framework for project planners, managers,
41 and laboratory personnel to ensure that radioanalytical laboratory data will meet a project's or
42 program's data requirements and needs. To attain this goal, MARLAP provides the necessary
43 guidance for national consistency in radioanalytical work in the form of a performance-based
44 approach for meeting a project's data requirements. In general terms, a performance-based
45 approach to laboratory analytical work involves clearly defining the analytical data needs and
46 requirements of a proj ect in terms of measurable goals during the planning phase of a proj ect.
47 These project-specific analytical data needs and requirements then serve as measurement
48 performance criteria for decisions as to exactly how the laboratory analysis will be conducted
49 during the implementation phase of a project. They are used subsequently as criteria for
50 evaluating analytical data during the assessment phase. Therefore, through a performance-based
51 approach, MARLAP provides guidance in the planning, implementation and assessment phases
52 for those projects that require the laboratory analysis of radionuclides. The manual focuses on
53 activities performed at radioanalytical laboratories, as well as activities and issues that direct,
54 affect, or can be used to evaluate activities performed at radioanalytical laboratories. The
55 guidance in MARLAP is intended to help ensure the generation of radioanalytical data of known
56 quality appropriate for its intended use.
57 Specific objectives of MARLAP include:
58 • Promoting a directed planning process for projects involving individuals from relevant
59 disciplines including radiochemistry;
60 • Highlighting common radioanalytical planning issues;
61 • Providing a framework and information resource for using a performance-based approach for
62 planning and conducting radioanalytical work;
63 • Providing guidance on linking project planning, implementation, and assessment;
64 • Providing guidance on obtaining and evaluating radioanalytical laboratory services;
MARLAP JULY 2001
DO NOT CITE OR QUOTE 1 -2 DRAFT FOR PUBLIC COMMENT
-------
Introduction to MARLAP
65 • Providing guidance for evaluating radioanalytical laboratory data, i.e., data verification, data
66 validation, and data quality assessment;
67 • Promoting high quality radioanalytical laboratory work; and
68 • Making collective knowledge and experience in radioanalytical work widely available.
69 As indicated by the list of objectives, MARLAP provides guidance to project planners, managers,
70 and laboratory personnel for a range of activities for those projects and programs that require the
71 laboratory analysis of radionuclides.
72 1.3 Use and Scope of the Manual
73 The guidance contained in MARLAP is for both governmental and private sectors. Users of
74 MARLAP include project planners, project managers, laboratory personnel, regulators, auditors,
75 inspectors, data evaluators, decision makers, and other end users of radioanalytical laboratory
76 data.
77 Since MARLAP uses a performance-based approach to laboratory measurements, the guidance
78 contained in the manual is applicable to a wide range of projects and activities that require
79 radioanalytical laboratory measurements. Examples of data collection activities that MARLAP
80 supports include:
81 • Site characterization activities;
82 • Site cleanup and compliance demonstration activities;
83 • License termination activities;
84 • Decommissioning of nuclear facilities;
85 • Remedial and removal actions;
86 • Effluent monitoring of licensed facilities;
87 • Environmental site monitoring;
88 • Background studies;
89 • Routine ambient monitoring; and
90 • Waste management activities.
91 MARLAP and the Multi-Agency Radiation Survey and Site Investigation Manual (MARS SIM,
92 2000) are complementary guidance documents in support of cleanup and decommissioning
93 activities. MARS SIM provides guidance on how to plan and carry out a study to demonstrate that
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 1 -3 DO NOT CITE OR QUOTE
-------
Introduction to MARLAP
94 a site meets appropriate release criteria. It describes a methodology for planning, conducting,
95 evaluating, and documenting environmental radiation surveys conducted to demonstrate
96 compliance with cleanup criteria. MARLAP provides guidance and a framework for both project
97 planners and laboratory personnel to ensure that radioanalytical data will meet the needs and
98 requirements of cleanup and decommissioning activities.
99 While MARLAP is designed to support a wide range of projects, some topics are not specifically
100 discussed in the manual. These include high-level waste, mixed waste, and medical applications
101 involving radionuclides. While they are not specifically addressed, much of MARLAP's
102 guidance may be applicable in these areas. Although the focus of the manual is to provide
103 guidance for the planning, implementation, and assessment phases of those projects that require
104 the laboratory analysis of radionuclides, much of the guidance on the planning and assessment
105 phases can be applied wherever the measurement process is conducted, for example, in the field.
106 In addition, MARLAP does not provide specific guidance on sampling design issues, sample
107 collection, field measurements, laboratory quality assurance issues, or laboratory health and
108 safety practices. However, a brief discussion of some aspects of these activities has been included
109 in the manual because of the effect these activities often have on the laboratory analytical
110 process.
in 1.4 Key MARLAP Concepts and Terminology
112 Some of the terms used in MARLAP were developed for the purpose of this manual, while
113 others are commonly used terms that have been adopted by MARLAP. Where possible, every
114 effort has been made to use terms and definitions from consensus-based organizations (e.g.,
115 International Organization for Standardization [ISO], American National Standards Institute
116 [ANSI], American Society for Testing and Materials [ASTM], International Union of Pure and
117 Applied Chemistry [IUPAC]).
118 The following sections are intended to familiarize the reader with the key terms and concepts
119 used in MARLAP. In general, each term or concept is discussed individually in each section
120 without emphasizing how these terms and concepts are linked. Section 1.5 ties these terms and
121 concepts together to provide an overview of the MARLAP process.
122 1.4.1 Data Life Cycle
123 The data life cycle (EPA, 2000) approach provides a structured means of considering the major
124 phases of projects that involve data collection activities (Figure 1.1). The three phases of the data
125 life cycle are planning, implementation, and assessment. MARLAP provides information on all
MARLAP JULY 2001
DO NOT CITE OR QUOTE 1 -4 DRAFT FOR PUBLIC COMMENT
-------
Introduction to MARLAP
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
three phases for two major types of
activities: those performed at radioanaly-
tical laboratories and those that direct,
affect, or evaluate activities performed at
radioanalytical laboratories (such as
project planning, development of plan
documents, data verification and data
validation). Consequently, MARLAP
provides guidance for project planners,
managers, and laboratory personnel.
One of the specific objectives of the
MARLAP Manual is to provide
guidance on, and to emphasize the
importance of, establishing the proper
linkages among the three phases of the
data life cycle—planning, implemen-
tation and assessment—thereby resulting
in an integrated and iterative process that
accurately translates the expectations
and requirements of data users into
DATA LIFE CYCLE
PROCESS
o
c
'c
c
ro
0.
Implementation
Assessment
Directed Planning
Process
Plan Documents
Contracting Services
Sampling
Analysis
Verification
Validation
Data Quality Assessment
PROCESS OUTPUTS
Development of Data Quality Objectives and
Measurement Quality Objectives (Including Optimized
Sampling and Analytical Design)
Project Plan Documents
Including Quality Assurance Project Plan (QAPP);
Work Plan or Sampling and Analysis Plan (SAP); Data
Validation Plan; Data Quality Assessment Plan
Statement of Work (SOW)
and Other Contractual Documents
Laboratory Samples
Laboratory Analysis
(Including QC Samples)
Complete Data Package
1
Verified Data
Data Verification Report
Validated Data
Data Validation Report
1
Assessment Report
Data of Known Quality Appropriate for the Intended Use
FIGURE 1.1 — The Data Life Cycle
measurement performance criteria for data suppliers. From an analytical perspective, the
integration of the three phases of the data life cycle is critical to ensure that the analytical data
requirements defined during the planning phase serve as measurement performance criteria
during the implementation phase and subsequently as criteria for data evaluation during the
assessment phase. The proper linkages and integration of the three phases of the data life cycle
should be established during the planning phase. Without the proper linkages and integration of
the three phases, there is a significant likelihood that the analytical data will not meet a project's
data requirements, and the data may be evaluated using criteria that have little relation to their
intended use. Therefore, failure to integrate and adequately link the three phases of the data life
cycle increases the likelihood of project cost escalation or project failure.
1.4.2 Directed Planning Process
MARLAP recommends the use of a directed or systematic planning process. A directed planning
process is an approach for setting well-defined, achievable objectives and developing a cost-
effective, technically sound sampling and analysis design that balances the data user's tolerance
for uncertainly in the decision process with the resources available for obtaining data to support a
JULY 2001
DRAFT FOR PUBLIC COMMENT
1-5
MARLAP
DO NOT CITE OR QUOTE
-------
Introduction to MARLAP
161 decision. While MARLAP recommends and promotes the use of a directed planning process, it
162 does not recommend or endorse any particular directed planning process. However, MARLAP
163 employs many of the terms and concepts associated with the data quality objective (DQO)
164 process (ASTM D5792, EPA, 2000). This was done to ensure consistent terminology throughout
165 the manual, and also because many of the terms and concepts of this process are familiar to those
166 engaged in environmental data collection activities.
167 1.4.3 Performance-Based Approach
168 MARLAP provides the necessary guidance for using a performance-based approach to meet a
169 project's analytical data requirements. In a performance-based approach, the project-specific
170 analytical data requirements that are determined during directed planning serve as measurement
171 performance criteria for analytical selections and decisions. The project-specific analytical data
172 requirements also are used for the initial, ongoing, and final evaluation of the laboratory's
173 performance and the laboratory's data. MARLAP provides guidance for using a performance-
174 based approach for all three phases—planning, implementation and assessment—of the data life
175 cycle for those projects that require radioanalytical laboratory data. This involves not only using a
176 performance-based approach for selecting an analytical protocol, but also using a performance-
177 based approach for other project activities, such as developing acceptance criteria for laboratory
178 quality control samples, laboratory evaluations, data verification, data validation, and data quality
179 assessment.
180 There are three major steps or processes associated with a performance-based approach. The first
181 is clearly and accurately defining the analytical data requirements for the project. This process is
182 discussed in more detail in Section 1.4.9 of this chapter. The second involves using an organized,
183 interactive process for selecting or developing analytical protocols to meet the specified
184 analytical data requirements and for demonstrating the protocol's ability to meet the analytical
185 data requirements. The last major activity involves using the analytical data requirements as
186 measurement performance criteria for the ongoing and final evaluation of the laboratory data,
187 which would include data verification, data validation, and data quality assessment. MARLAP
188 provides guidance in all three of these areas. Within the constraints of other factors, such as cost,
189 a performance-based approach allows for the use of any analytical protocol that meets the
190 project's analytical data requirements. For all relevant project activities, the common theme of a
191 performance-based approach is the use of project-specific analytical data requirements that are
192 developed during project planning and serve as measurement performance criteria for selections,
193 evaluations, and decision-making.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 1 -6 DRAFT FOR PUBLIC COMMENT
-------
Introduction to MARLAP
194 1.4.4 Analytical Process
195 Most environmental data collection
196 efforts center around two major
197 processes: the sampling process and
198 the analytical process. MARLAP
199 does not provide general guidance
200 on the sampling process, except for
201 brief discussions of certain activities
202 that often affect the analytical
203 process (field processing,
204 preservation, etc.). The analytical (or
205 measurement) process is a general
206 term used by MARLAP to refer to a
207 compilation of activities starting
208 from the time a sample is collected
209 and ending with the reporting of
210 data. These activities typically
211 include field sample preparation and
212 preservation, sample receipt and
213 inspection, laboratory sample
214 preparation, sample dissolution,
215 chemical separations, preparation of
216 samples for instrument measure-
217 ments, instrument measurements,
218 data reduction, data reporting, and
219 quality control of the process. Figure
220 1.2 illustrates the maj or components FIGURE 1.2 — Typical Components of an Analytical Process
221 of an analytical process. It should be noted that a particular analytical process for a project may
222 not include all of the activities listed. For example, if a project involves the analysis of tritium in
223 drinking water, then the analytical process for the project will not include sample dissolution and
224 the chemical separation of the radionuclide of concern. It is important to identify the relevant
225 activities of the analytical process for a particular project early in the planning phase. Once the
226 activities have been identified, the analytical requirements of the activities can be established,
227 which will ultimately lead to defining how the activities will be accomplished through the
228 selection or development of written procedures for the various activities.
Sample Tracking
i
r
Field Sample Preparation and
Preservation
i
Sample Receipt and Inspection
i
Laboratory Sample Preparation
i
Sample D ssolution
i
Chemical Separation of
Radionuclides of Concern
i
Preparation of Samples for
Instrument Measurements
i
Instrument Measurements
i
Data Reduction and Reporting
QA/QC
i
r
JULY 2001
DRAFT FOR PUBLIC COMMENT
1-7
MARLAP
DO NOT CITE OR QUOTE
-------
Introduction to MARLAP
229 The analytical process should not be confused with the written procedures necessary to perform
230 the associated activities of the analytical process. The analytical process (i.e., the compilation of
231 activities starting from the time a sample is collected and ending with the reporting of the data)
232 should be performed according to written procedures
233 1.4.5 Analytical Protocol
234 MARLAP uses the term "analytical protocol" to refer to a compilation of specific procedures and
235 methods that are performed in succession for a particular analytical process. For example, a
236 protocol for the analysis of drinking water samples for tritium would be comprised of the set of
237 procedures that describe the relevant activities, such as sample tracking, quality control, field
238 sample preparation and preservation, sample receipt and inspection, laboratory sample prepara-
239 tion (if necessary), preparing the samples for counting, counting the samples, and data reduction
240 and reporting. A written procedure may cover one or more of the activities, but it is unlikely that
241 a single procedure will cover all of the activities of a given analytical process. It should be noted
242 that with a performance-based approach, there may be a number of alternative protocols that
243 might be appropriate analytical protocols for a particular analytical process. Selecting or develop-
244 ing an analytical protocol requires knowledge of the particular analytical process, as well as an
245 understanding of the analytical data requirements developed during the project planning phase.
246 1.4.6 Analytical Method
247 A major component of an analytical protocol is the analytical method, which normally includes
248 written procedures for sample digestion, chemical separation (if required) and counting. It is
249 recognized that in many instances the analytical method may cover many of the activities of a
250 particular analytical process. Therefore attention is naturally focused on the selection or
251 development of an analytical method. However, many analytical methods do not address
252 activities such as field preparation and preservation, certain aspects of laboratory preparation,
253 laboratory subsampling, etc., which are often important activities within an analytical process.
254 The analytical protocol is generally more inclusive of the activities that make up the analytical
255 process than the analytical method. For this reason, MARLAP focuses on the selection,
256 implementation, and assessment of analytical protocols that cover the entire analytical process
257 for a particular project or program.
258 1.4.7 Uncertainty and Error
259 An important aspect of sampling and measurement is uncertainty. The term "uncertainty" has
260 different shades of meaning in different contexts, but generally the word always refers to a lack
MARLAP JULY 2001
DO NOT CITE OR QUOTE 1 -8 DRAFT FOR PUBLIC COMMENT
-------
Introduction to MARLAP
261 of complete knowledge about something of interest. In the context of metrology (the science of
262 measurement), the more specific term "measurement uncertainty" often will be used. "Uncertain-
263 ty (of measurement)" is defined in the Guide to the Expression of Uncertainty in Measurement
264 (ISO 1995—"GUM") as a "parameter, associated with the result of a measurement, that charac-
265 terizes the dispersion of values that could reasonably be attributed to the measurand." The
266 "measurand" is the quantity being measured. MARLAP recommends the terminology and
267 methods of GUM for describing, evaluating, and reporting measurement uncertainty. The
268 uncertainty of a measured value is typically expressed as an estimated standard deviation, called
269 a "standard uncertainty" (or "one-sigma uncertainty"). The standard uncertainty of a calculated
270 result usually is obtained by propagating the standard uncertainties of a number of other
271 measured values, and in this case, the standard uncertainty is called a "combined standard
272 uncertainty." The combined standard uncertainty may be multiplied by a specified factor called a
273 "coverage factor" (e.g., 2 or 3) to obtain an "expanded uncertainty" (a "two-sigma" or "three-
274 sigma" uncertainty), which describes an interval about the result that can be expected to contain
275 the true value with a specified high probability. MARLAP recommends that either the combined
276 standard uncertainty or an expended uncertainty be reported with every result. Chapter 19
277 discusses the terminology, notation, and methods of GUM in more detail and provides guidance
278 for applying the concepts to radioanalytical measurements.
279 While measurement uncertainty is a parameter associated with an individual result and is
280 calculated after a measurement is performed, MARLAP uses the term "method uncertainty" to
281 refer to the predicted uncertainty of a measured value that likely would result from the analysis of
282 a sample at a specified analyte concentration. Method uncertainty is a method performance
283 characteristic much like the detection capability of a method. Reasonable values for both
284 characteristics can be predicted for a particular method based on typical values for certain
285 parameters and on information and assumptions about the samples to be analyzed. These
286 predicted values can be used in the method selection process to identify the most appropriate
287 method based on a project's data requirements. Chapter 3 provides MARLAP's recommenda-
288 tions for deriving analytical protocol selection criteria based on the required method uncertainty
289 and other analytical requirements.
290 When a decision maker bases a decision on the results of measurements, the measurement
291 uncertainties affect the probability of making a wrong decision. When sampling is involved,
292 sampling statistics also contribute to the probability of a wrong decision. Since decision errors
293 are possible, there is uncertainty in the decision-making process. MARLAP uses the terms
294 "decision uncertainty" or "uncertainty of the decision" to refer to this type of uncertainty.
295 Decision uncertainty is usually measured by the estimated probability of a decision error under
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 1 -9 DO NOT CITE OR QUOTE
-------
Introduction to MARLAP
296 specified assumptions. Appendix B discusses decision uncertainty further in the context of the
297 DQO process.
298 A concept that should not be confused with uncertainty is error. In general, error refers to
299 something that deviates from what is correct, right or true. In terms of measurements such as
300 laboratory analyses, the difference between the measured result and the actual value of the
301 measurand is the error of the measurement. Since the actual value of the measurand is generally
302 not known, the measurement error cannot be determined. Therefore, the error of a measurement
303 is primarily a theoretical concept with little practical use. However, the measurement uncertainty,
304 which provides an estimated bound for the likely size of the measurement error, is very useful
305 and plays a key role in MARLAP's performance-based approach.
306 1.4.8 Precision, Bias, and Accuracy
307 Analytical data requirements often have been described in terms of precision and bias. Precision
308 is usually expressed as a standard deviation, which measures the dispersion of measured values
309 about their mean. It is more natural to speak of imprecision, since larger values of the standard
310 deviation indicate less precision and greater imprecision. Bias is a persistent difference between
311 the measured result and the true value of the quantity being measured, which does not vary if the
312 measurement is repeated. If the measurement process is in statistical control, then imprecision
313 may be reduced by averaging the results of many independent measurements of the same
314 quantity. Bias is unaffected by averaging.
315 A bias in a data set may be caused by measurement errors that occur in steps of the measurement
316 process that are not repeated, such as the determination of a half-life. Imprecision may be caused
317 by measurement errors in steps that are repeated many times, such as weighing, pipetting, and
318 radiation counting. However, distinguishing between bias and imprecision is complicated by the
319 fact that some steps in the process, such as instrument calibration or tracer preparation, are
320 repeated at frequencies less than those of other steps, and the measurement errors in seldom
321 repeated steps may affect large blocks of data. Consequently measurement errors that produce
322 apparent biases in small data sets might produce apparent imprecision in larger data sets.
323 Because the same type of measurement error may produce either bias or imprecision, depending
324 on one's point of view, the concept of measurement uncertainty, described in Section 1.3.7, treats
325 all types of measurement error alike and combines estimates of their magnitudes into a single
326 numerical parameter (i.e., combined standard uncertainty). The concepts of imprecision and bias
327 are useful in context when a measurement process or a data set consisting of many measurement
328 results is considered. When one considers only a single measurement result, the concept of
MARLAP JULY 2001
DO NOT CITE OR QUOTE 1-10 DRAFT FOR PUBLIC COMMENT
-------
Introduction to MARLAP
329 measurement uncertainty tends to be more useful than the concepts of imprecision and bias.
330 Therefore, it is probably best to consider imprecision and bias to be characteristics of the
331 measurement process or of the data set, and to consider measurement uncertainty to be an aspect
332 of each individual result.
333 Quality control samples are analyzed for the purpose of assessing imprecision and bias. Spiked
334 samples and method blanks are typically used to assess bias, and duplicates are used to assess
335 imprecision. Since a single measurement of a spike or blank cannot in principle distinguish
336 between imprecision and bias, a reliable estimate of bias requires a data set that includes many
337 such measurements.
338 Different authors have given the word accuracy different technical definitions, expressed in
339 terms of bias and imprecision. MARLAP avoids all of these technical definitions and uses the
340 term "accuracy" in its common, ordinary sense, which is consistent with its definition in the
341 International Vocabulary of'Basic and General Terms in Metrology(ISO, 1993). InMARLAP's
342 terminology, the result of a measurement is "accurate" if it is close to the true value of the
343 quantity being measured. Inaccurate results may be caused either by bias or imprecision in the
344 measurement process.
345 While it is recognized that the terms bias, imprecision, and accuracy are commonly used in data
346 collection activities, these terms are used somewhat sparingly in this manual. MARLAP
347 emphasizes and provides guidance in the use of measurement uncertainty as a means of
348 establishing analytical data requirements and in the evaluation of single measurement results.
349 1.4.9 Performance Objectives: Data Quality Objectives and Measurement Quality
350 Objectives
351 One of the outputs of a directed planning process is DQOs for a project or program. DQOs are
352 qualitative and quantitative statements that clarify the study objectives; define the most
353 appropriate type of data to collect; determine the most appropriate conditions from which to
354 collect the data; and specify tolerable limits on decision error rates (ASTM D5792; EPA, 2000).
355 DQOs apply to all data collection activities associated with a project or program, including
356 sampling and analysis. In particular, DQOs should encompass the "total uncertainty" resulting
357 from all data collection activities, including analytical and sampling activities.
358 From an analytical perspective, a process of developing the analytical data requirements from the
359 DQOs of a project is essential. These analytical data requirements serve as measurement perfor-
360 mance criteria or objectives of the analytical process. MARLAP refers to these performance
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 1-11 DO NOT CITE OR QUOTE
-------
Introduction to MARLAP
361 objectives as "measurement quality objectives" (MQOs). The MARLAP Manual provides
362 guidance on developing the MQOs from the overall project DQOs (Chapter 3). MQOs can be
363 viewed as the analytical portion of the DQOs and are therefore project-specific. MARLAP
364 provides guidance on developing MQOs during project planning for select method performance
365 characteristics, such as method uncertainty at a specified concentration; detection capability;
366 quantification capability; specificity, or the capability of the method to measure the analyte of
367 concern in the presence of interferences; range; ruggedness, etc. An MQO is a statement of a
368 performance objective or requirement for a particular method performance characteristic. Like
369 DQOs, MQOs can be quantitative and qualitative statements. An example of a quantitative MQO
370 would be a statement of a required method uncertainty at a specified radionuclide concentration,
371 such as the action level—i.e., "a method uncertainty of 3.7 Bq/kg (0.10 pCi/g) or less is required
372 at the action level of 37 Bq/kg (1.0 pCi/g)." An example of a qualitative MQO would be a
373 statement of the required specificity of the analytical protocol—the ability to analyze for the
374 radionuclide of concern given the presence of interferences—i.e., "the protocol must be able to
375 quantify the amount of 226Ra present given high levels of 235U in the samples."
376 The MQOs serve as measurement performance criteria for the selection or development of
377 analytical protocols and for the initial evaluation of the analytical protocols. Once the analytical
378 protocols have been selected and evaluated, the MQOs serve as criteria for the ongoing and final
379 evaluation of the laboratory data, including data verification, data validation, and data quality
380 assessment. In a performance-based approach, analytical protocols are either selected or rejected
381 for a particular project, to a large measure, based on their ability or inability to achieve the stated
382 MQOs. Once selected, the performance of the analytical protocols is evaluated using the project-
383 specific MQOs.
384 1.4.10 Analytical Protocol Specifications
385 MARLAP uses the term "analytical protocol specifications" (APSs) to refer to the output of a
386 directed planning process that contains the project's analytical data requirements in an organized
387 concise form. In general, there will be an APS developed for each analysis type, and since most
388 projects require that a number of different analyses be performed, several APSs will normally be
389 developed for a particular project. These specifications serve as the basis for the evaluation and
390 selection of the analytical protocols that will be used for a particular project. In accordance with a
391 performance-based approach, the APSs contains only the minimum level of specificity required
392 to meet the project's analytical data requirements without dictating exactly how the requirements
393 are to be met. At a minimum, the APSs should indicate the analyte of interest, the matrix of
394 concern, the type and frequency of quality control (QC) samples, and provide the required MQOs
395 and any specific analytical process requirements, such as chain-of-custody for sample tracking.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 1-12 DRAFT FOR PUBLIC COMMENT
-------
Introduction to MARLAP
396 Depending on the particular project, a number of specific analytical process requirements may be
397 included. For example, if project or process knowledge indicates that the radionuclide of interest
398 exists in a refractory form, then the APSs may require a fusion step for sample digestion.
399 However, the level of specificity in the APSs should be limited to those requirements that are
400 considered essential to meeting the project's analytical data requirements. In most instances, a
401 particular APS document would be a one-page form (see Chapter 3, Figure 3.2). For a particular
402 project, APSs would be developed for each analysis required.
403 Within the constraints of other factors, such as cost, MARLAP's performance-based approach
404 allows the use of any analytical protocol that meets the requirements contained in the APSs. The
405 requirements in the APSs, in particular the MQOs, are used for the selection and evaluation of
406 the analytical protocols. Once the analytical protocols have been selected and evaluated, the
407 APSs then serve as criteria for the ongoing and final evaluation of the laboratory data, including
408 data verification, data validation, and data quality assessment.
409 1.4.11 The Assessment Phase
410 As noted, the MARLAP Manual provides guidance for the assessment phases for those projects
411 that require the laboratory analysis of radionuclides. The guidance on the assessment phase of
412 projects focuses on three major activities: data verification, data validation, and data quality
413 assessment.
414 Data verification assures that laboratory conditions and operations were compliant with the
415 statement of work and any appropriate project plan documents (e.g., Quality Assurance Project
416 Plan), which may reference laboratory documents such as laboratory standard operating
417 procedures) Verification compares the material delivered by the laboratory to these requirements
418 (compliance) and checks for consistency and comparability of the data throughout the data
419 package, correctness of calculations, and completeness of the results to ensure that all necessary
420 documentation is available. The verification process produces a report identifying which
421 requirements are not met. The verification report is used to determine payment for laboratory
422 services and to identify problems that should be investigated during data validation. Verification
423 works iteratively and interactively with the generator (i.e., laboratory) to assure receipt of all
424 available, necessary data. Although the verification process identifies specific problems, the
425 primary function should be to apply appropriate feedback resulting in corrective action
426 improving the analytical services before the work is completed.
427 Validation addresses the reliability of the data. The validation process begins with a review of the
428 verification report and laboratory data package to screen the areas of strength and weakness of
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 1-13 DO NOT CITE OR QUOTE
-------
Introduction to MARLAP
429 the data set. The validator evaluates the data to determine the presence or absence of an analyte
430 and the uncertainty of the measurement process for contaminants of concern. During validation,
431 the technical reliability and the degree of confidence in reported analytical data are considered.
432 Validation "flags" (i.e., qualifiers) are applied to data that do not meet the acceptance criteria
433 established to assure data meet the needs of the project. The product of the validation process is a
434 validation report noting all data sufficiently inconsistent with the validation acceptance criteria in
435 the expert opinion of the validator. The appropriate data validation tests should be established
436 during the project planning phase.
437 Data quality assessment (DQA), the third and final step of the assessment phase, is defined as the
438 "scientific and statistical evaluation of data to determine if data are of the right type, quality, and
439 quantity to support their intended use." DQA is more global in its purview than the previous
440 verification and validation steps. DQA, in addition to reviewing the issues raised during verifica-
441 tion and validation, may be the first opportunity to review other issues, such as field activities
442 and their impact on data quality and usability. DQA should consider the combined impact of all
443 project activities in making a data usability determination, which is documented in a DQA report.
444 1.5 The MARLAP Process
445 An overarching objective of the MARLAP Manual is to provide a framework and information
446 for the selection, development, and evaluation of analytical protocols and the resulting laboratory
447 data. The MARLAP process is a performance-based approach that develops APSs and uses these
448 requirements as criteria for the analytical protocol selection, development and evaluation
449 processes, and for the evaluation of the resulting laboratory data. This process, which spans the
450 three phases of the data life cycle for a project—planning, implementation and assessment—is
451 the basis for achieving MARLAP's basic goal of ensuring that radioanalytical data will meet a
452 project's data requirements. A brief overview of this process, which is referred to as the
453 MARLAP process and is the focus of Part I of the manual, is provided below.
454 The MARLAP process starts with a directed planning process. Within a directed planning
455 process, key analytical issues based on the project's particular analytical processes are discussed
456 and resolved. The resolution of these key analytical issues produces the APSs, which include the
457 MQOs. The APSs are documented in project plan documents (e.g., Quality Assurance Project
458 Plans, Sampling and Analysis Plans). A SOW is then developed that contains the APSs. The
459 laboratories receiving the SOW respond with proposed analytical protocols based on the require-
460 ments of the APSs and provide evidence that the proposed protocols meet the performance
461 criteria in the APSs. The proposed analytical protocols are initially evaluated by the project
462 manager or designee to determine if they will meet the requirements in the APSs. If the proposed
MARLAP JULY 2001
DO NOT CITE OR QUOTE 1-14 DRAFT FOR PUBLIC COMMENT
-------
Introduction to MARLAP
463 analytical protocols are accepted, the project plan documents are updated by the inclusion or
464 referencing of the actual analytical protocols to be used. During analyses, resulting sample and
465 QC data will be evaluated primarily using MQOs from the respective APSs. Once the analyses
466 are completed, an evaluation of the data will be conducted, including data verification, data
467 validation, and data quality assessment with the respective MQOs serving as criteria for
468 evaluation. The role of the APSs (particularly the MQOs, which make up an essential part of the
469 APSs) in the selection, development, and evaluation of the analytical protocols and the laboratory
470 data is to provide a critical link between the three phases of the data life cycle of a project. This
471 linkage helps to ensure that radioanalytical laboratory data will meet a project's data require-
472 ments, and that the data are of known quality appropriate for their intended use. The MARLAP
473 process is illustrated in Figure 1.3. Although the diagram used to represent the MARLAP Process
474 is presented in a linear fashion, it is important to note that the process is an iterative one, and
475 there can be many variations on this stylized diagram.
476 1.6 Structure of the Manual
477 MARLAP is divided into two main parts. Part I provides guidance on implementing the
478 MARLAP process as described in Section 1.5. This part of the manual focuses on the sequence
479 of steps involved when using a performance-based approach for projects requiring radioanalytical
480 laboratory work starting with a directed planning process and ending with DQA. Part I provides
481 the overall guidance for using a performance-based approach for all three phases of a project. A
482 more detailed overview of Part I is provided in Section 1.6.1.
483
484 Part II of the manual provides information on the laboratory analysis of radionuclides to support
485 a performance-based approach. Part II provides guidance and information on the various
486 activities performed at radioanalytical laboratories, such as sample preparation, sample
487 dissolution, chemical separations, preparing sources for counting, nuclear counting, etc. Using
488 the overall framework provided in Part I, the material in Part n can be used to assist project
489 planners, managers, and laboratory personnel in the selection, development, evaluation, and
490 implementation of analytical protocols for a particular project or program. A more detailed
491 overview of Part n is provided in Section 1.6.2. In addition to Part I and Part II, MARLAP has
492 several appendices that support both Part I and Part n of the manual. An overview of the
493 appendices is provided in Section 1.6.3 of this chapter.
494 Because of the structure and size of the manual, most individuals will naturally focus on those
495 chapters that provide guidance in areas directly related to their work. Therefore, to help ensure
496 that key concepts are conveyed to the readers, there is some material is repeated, often in very
497 similar or even the same language, throughout the manual.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 1-15 DO NOT CITE OR QUOTE
-------
Introduction to MARLAP
Directed Planning Process (Chapter 2)
• Key Analytical Issues (Chapter 3)
• Development of Analytical Protocol Specifications
• Includes MQOs (Chapter 3)
Develop Plan Documents That Incorporate
Analytical Protocol Specifications (Chapter 4)
(e.g., QAPP, SAP, Data Validation Plan)
Development of SOW (Chapter 5)
• Includes Analytical Protocol Specifications (MQOs)
Laboratory Responds with Analytical Protocols
(Chapter 6)
• Selected to Meet Requirements of Analytical Protocol Specifications
• Performance Data Provided
Project Manager
Protocols
Rejected
Initial Evaluation of Analytical Protocols and Laboratory
(Chapter 7)
• Review of Performance Data
• Performance Evaluation (PE) Samples/Certified Reference
Materials (CRMs) Analyzed
• Quality Systems Audit
Protocols Accepted
Update Plan Documents
(Chapter 4)
• Include or Reference Accepted Analytical Protocols
Project Planning Team
Corrective
Actions
Start Analysis of Samples
Ongoing Evaluation of Laboratory Performance
(Chapter 7)
• Evaluation of QC and PE Sample Results
• Laboratory Audits
• Evaluation of Sample-Specific Parameters (i.e., yield)
Analyses Completed
Data Evaluation and Assessment
• Data Verification (Chapter 8)
• Data Validation (Chapter 8)
• Data Quality Assessment (Chapter 9)
Planning
Phase
Implementation
— Phase
Assessment
Phase
Data of Known Quality for Decision Making
FIGURE 1.3 — The MARLAP Process
MARLAP
DO NOT CITE OR QUOTE
1-16
JULY 2001
DRAFT FOR PUBLIC COMMENT
-------
Introduction to MARLAP
498 1.6.1 Overview of Part I
499 Part I begins with Chapter 2, Project Planning Process, which provides an overview of the
500 directed planning process and discusses important analytical outputs of the planning process.
501 Chapter 3, Key Analytical Planning Issues and Developing APSs., provides an overview of key
502 analytical planning issues that need to be addressed during a directed planning process and
503 provides guidance on developing APSs, which are outputs of the planning process. These outputs
504 are incorporated into plan documents (e.g., work plans, quality assurance project plans, sampling
505 and analysis plans), which are covered in Chapter 4, Project Plan Documents. Chapter 4 provides
506 an overview of different types of project plan documents and provides guidance on the linkage
507 between project planning and project plan documents. Information from the plan documents is
508 then incorporated into a SOW, which is covered in Chapter 5, Obtaining Laboratory Services.
509 Chapter 5 provides guidance on developing a SOW that incorporates the APSs. Chapter 6,
510 Selection and Application of an Analytical Method, provides guidance on selecting or developing
511 analytical protocols that will meet the MQOs and other requirements as outlined in the APSs.
512 Chapter 7, Evaluating Protocols and Laboratories, provides guidance on the initial and ongoing
513 evaluation of analytical protocols and also provides guidance on the overall evaluation of
514 radioanalytical laboratories. Chapter 8, Radiochemical Data Verification and Validation,
515 provides an overview of the data evaluation process, provides general guidelines for data
516 verification and validation, and provides "tools" for data validation. The last chapter of Part I,
517 Chapter 9, Data Quality Assessment, provides an overview of data quality assessment and
518 provides guidance on linking data quality assessment and the planning process.
519 Figure 1.3, the MARLAP Process, illustrates the sequence of steps that make up the framework
520 of a performance-based approach for the planning, implementation, and assessment phases of
521 projects that require the laboratory analysis of radionuclides. The primary audience for Part I is
522 project planners and managers. However, Chapter 6, Selection and Application of an Analytical
523 Method, is intended primarily for laboratory personnel. This is because, under a performance-
524 based approach, a laboratory would be able to use any analytical protocol that meets the
525 analytical requirements as defined by the APSs. Other factors, such as cost, also will play a role
526 in the selection of analytical protocols. While the primary audience for Part I is project planners
527 and managers, other groups, such as laboratory personnel, can benefit from the guidance in Part I.
528 1.6.2 Overview of Part II
529 The chapters in Part II are intended to provide information on the laboratory analysis of
530 radionuclides. The chapters provide information on many of the options available for analytical
531 protocols, and discuss common advantages and disadvantages of each. The chapters highlight
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 1-17 DO NOT CITE OR QUOTE
-------
Introduction to MARLAP
532 common analytical problems and ways to identify and correct them. The chapters also serve to
533 educate the reader by providing a detailed explanation of the typical activities performed at a
534 radioanalytical laboratory. Consistent with a performance-based approach, the chapters in Part II
535 do not contain detailed step-by-step instructions on how to perform certain laboratory tasks, such
536 as the digestion of a soil sample. The chapters do contain information and guidance intended to
537 assist primarily laboratory personnel in deciding on the best approach for a particular laboratory
538 task. For example, while the chapter on sample dissolution does not contain step-by-step
539 instructions on how to dissolve a soil sample, it does provide information on acid digestion,
540 fusion techniques, and microwave digestion, which is intended to help the reader select the most
541 appropriate technique or approach for a particular project.
542 The primary audience for Part II is laboratory personnel and the chapters generally contain a
543 significant amount of technical information. While the primary target audience is laboratory
544 personnel, other groups, such as project planners and managers, can benefit from the guidance in
545 Part II. Listed below are the chapters that make up Part n of the manual. It should be noted that
546 Part II of the manual does not provide specific guidance for some laboratory activities that are
547 common to all laboratories, such as laboratory quality assurance, and laboratory health and safety
548 practices. This is primarily due to the fact that these activities are not unique to radioanalytical
549 laboratories and considerable guidance in these areas already exists.
550 Chapter 10 Requirements When Collecting, Preserving, and Shipping Samples That
551 Require Analytical Measurement
552 Chapter 11 Sample Receipt, Inspection and Tracking
553 Chapter 12 Laboratory Sample Preparation
554 Chapter 13 Sample Dissolution
555 Chapter 14 Separation Techniques
556 Chapter 15 Nuclear Counting Instrumentation
557 Chapter 16 Instrument Calibration and Source Preparation
558 Chapter 17 Nuclear Counting and Data Reduction and Reporting
559 Chapter 18 Laboratory Quality Control
560 Chapter 19 Measurement Statistics
561 Chapter 20 Waste Disposal
562 Chapters 10 through 17 provide information on the typical components of an analytical process
563 in the order in which activities that make up an analytical process are normally performed. While
564 not providing step-by-step procedures for activities such as sample preservation, sample
565 digestion, nuclear counting, etc., the chapters do provide an overview of options available for the
566 various activities and importantly, provide information on the appropriateness of the assorted
MARLAP JULY 2001
DO NOT CITE OR QUOTE 1-18 DRAFT FOR PUBLIC COMMENT
-------
Introduction to MARLAP
567 options under a variety of conditions. Chapter 18, Laboratory Quality Control, provides
568 guidance on monitoring key laboratory performance indicators as a means of determining if a
569 laboratory's measurement processes are in control. The chapter also provides information on
570 likely causes of excursions for select laboratory performance indicators, such as chemical yield,
571 instrument background, quality control samples, etc. Chapter 19, Measurement Statistics,
572 provides information on statistical principles and methods applicable to radioanalytical
573 measurements, calibrations, data interpretation, and quality control. Topics covered in the chapter
574 include detection and quantification, measurement uncertainty, and procedures for estimating
575 uncertainty. Chapter 20, Waste Disposal provides an overview of many of the regulations for
576 waste disposal and provides guidance for managing wastes in a radioanalytical laboratory.
577 1.6.3 Overview of the Appendices
578 MARLAP includes several appendices to both Part I and Part II of the manual to provide
579 additional guidance on specific topics. Brief descriptions of the appendices are provided below.
580 • Appendix A, Directed Planning Approaches, provides an overview of a number of directed
581 planning processes and discusses some common elements of the different approaches.
582 • Appendix B, The Data Quality Objective Process, provides an expanded discussion of the
583 Data Quality Objectives Process including detailed guidance on setting up a "gray region"
584 and establishing tolerable decision error rates.
585 • Appendix C, Measurement Quality Objectives for Method Uncertainty and Detection and
586 Quantification Capability, provides the rationale and guidance for developing MQOs for
587 select method performance characteristics.
588 • Appendix D, Content of Project Plan Documents, provides guidance on the appropriate
589 content of plan documents.
590 • Appendix E, Contracting Laboratory Services, contains detailed guidance on contracting
591 laboratory services.
592 • Appendix F, Laboratory Subsampling, provides information on improving and evaluating
593 laboratory subsampling techniques.
594 • Appendix G, Statistical Tables, provides a compilation of statistical tables.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 1-19 DO NOT CITE OR QUOTE
-------
Introduction to MARLAP
595 1.7 References
596 American Society for Testing and Materials (ASTM). D5792. Standard Practice for Generation
597 of Environmental Data Related to Waste Management Activities: Development of Data
598 Quality Objectives, 1995.
599 International Organization for Standardization (ISO). 1993. International Vocabulary of Basic
600 and General Terms in Metrology. ISO, Geneva, Switzerland.
601 International Organization for Standardization (ISO). 1995. Guide to the Expression of
602 Uncertainty in Measurement. ISO, Geneva, Switzerland.
603 MARSSEVI. 2000. Multi-Agency Radiation Survey and Site Investigation Manual, Revision 1.
604 NUREG-1575 Rev 1, EPA 402-R-97-016 Revl, DOE/EH-0624 Revl. August. Available
605 from http://www.epa.gov/radiation/marssim/filesfm.htm.
606 U.S. Environmental Protection Agency (EPA). 2000. Guidance for the Data Quality Objective
607 Process (EPA QA/G-4). EPA/600/R-96/055, Washington, DC. available from www.epa.gov/
608 qualityl/qa_docs.html.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 1 -20 DRAFT FOR PUBLIC COMMENT
-------
2 PROJECT PLANNING PROCESS
2 2.1 Introduction
3 Efficient environmental data collection activities depend on successfully identifying the type,
4 quantity, and quality of data needed, as well as how the data will be used to support the decision
5 making process. MARLAP recommends the use of a directed or systematic planning process.
6 These planning processes provide a logic and framework for setting well-defined, achievable
7 objectives and developing a cost-effective, technically sound and defensible sampling and
8 analysis design that balances the data user's tolerance for uncertainty in the decision process and
9 the available resources for obtaining data to support a decision. MARLAP has chosen to use the
10 term "directed planning" to emphasize that the planning process, in addition to having a
11 framework or structure (i.e., it is systematic), is focused of defining the data needed to achieve an
12 informed decision for a specific project.
13 The objective of this MARLAP chapter is to promote:
14 1. Directed project planning as a tool for project management to identify and document the data
15 quality objectives (DQOs)—that is, qualitative and quantitative statements that define the
16 project objectives and the tolerable rate of making decision errors that will be used as the
17 basis for establishing the quality and quantity of data needed to support the decision—and the
18 measurement quality objectives (MQOs) that define the analytical data requirements
19 appropriate for decision making;
20 2. The involvement of technical experts in particular radioanalytical specialists in the planning
21 process; and
22 3. Integration of the outputs from the directed planning process into the implementation and
23 assessment phases of the project through documentation in project plan documents, the
24 analytical SOW, and the data assessment plans (e.g., for data validation, data verification, and
25 data and data quality assessment—DQA).
26 MARLAP will use the terms "DQOs" and "MQOs," as defined above and in Chapter 1,
27 throughout this document because of their widespread use in environmental data collection
28 activities. These concepts may be expressed by other terms, such as "decision performance
29 criteria" or "project quality objectives" for DQOs and "measurement performance criteria" or
30 "data quality requirements" for MQOs.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 2-1 DO NOT CITE OR QUOTE
-------
Project Planning Process
31 This chapter provides an overview of the directed planning process. Additional discussion on the
32 planning process in Chapter 3, Key Analytical Planning Issues and Developing Analytical
33 Protocol Specifications, will focus on project planning from the perspective of the analytical
34 process and the development of Analytical Protocol Specifications (APSs). Section 2.2 will
35 discuss the importance of directed project planning. The approach, guidance and common
36 elements of directed planning are discussed in Section 2.3. The project planning team is
37 discussed in Section 2.4, and the role of the radioanalytical specialists is highlighted in Section
38 2.5. The results of the planning process are discussed in Section 2.6. Section 2.7 presents the next
39 steps of the planning phase of the project, which will document the results of the planning
40 process and will link the results of the planning process to the implementation and assessment
41 phases of data collection activities.
42 The environmental data collection process consists of a series of elements: planning, developing,
43 and updating project plan documents; contracting for services; sampling; analysis; data
44 verification; data validation; and data quality assessment (see Section 1.4.7, "Data Life Cycle," of
45 Chapter 1, Introduction to MARLAP}. These elements are interrelated (sampling and analysis
46 cannot be performed efficiently or resources allocated effectively without first identifying data
47 needs during planning). Linkage and integration of the data collection process elements are
48 essential to the success of the environmental data collection activity.
49 2.2 The Importance of Directed Project Planning
50 A directed planning process has several notable strengths. It brings together the stakeholders (see
51 box), decision makers, and technical experts at the beginning of the project to gain commitment
52 to the project and a consensus on the nature of the problem and the desired decision. MARLAP
53 recognizes the need for a directed planning process that involves radioanalytical and other
54 technical experts as principals to ensure the decision makers' data requirements and the results
55 from the field and radioanalytical laboratory are linked effectively. Directed planning enables
56 each participant to play a constructive role in clearly defining:
57 • The problem that requires resolution;
58 • What type, quantity, and quality of data the decision maker needs to resolve that problem;
59 • Why the decision maker needs that type and quality of data;
60 • What are the tolerable decision error rates; and
61 • How the decision maker will use the data to make a defensible decision.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 2-2 DRAFT FOR PUBLIC COMMENT
-------
Project Planning Process
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
Example of Stakeholders for a Cleanup Project
A stakeholder is anyone with an interest in the outcome of an activity. For a cleanup
project, some of the stakeholders could be:
• Federal, regional, State, and tribal environmental agencies with regulatory
interests (e.g., NRC and EPA).
• States with direct interest in transportation, storage and disposition of wastes,
and a range of other issues.
• City and County Governments with interest in the operations and safety at sites
as well as economic development and site transition.
• Site Advisory Boards, citizens groups, licensees, special interest groups, and
other members of the public with interest in cleanup activities at the site.
A directed planning process encourages efficient planning by providing a framework for
organizing complex issues. The process promotes timely, open, and effective communication
among the stakeholders resulting in well-conceived and documented plans. Because of the
emphasis on documentation, directed planning also provides project management with a more
efficient and consistent transfer of knowledge to new project members.
A directed planning process focuses on collection of only those data needed to address the
appropriate questions and support defensible decisions. Directed planning helps to eliminate poor
or inadequate sampling and analysis designs that require analysis of (1) too few or too many
samples, (2) samples that will not meet the needs of the project, or (3) inappropriate QC samples.
During directed planning, which is an iterative process, the sufficiency of existing data is
evaluated, and the need for additional data to fill the gaps, as well as the desired quality of the
additional data, is determined. By defining the MQOs, directed planning provides input for
obtaining appropriate radioanalytical services, which balance constraints and the required data
quality.
The time invested in preliminary planning can greatly reduce resource expenditure in the more
resource-intensive execution phase of the project. Less overall time (and money) is expended
when early efforts are focused on defining (and documenting) the project's objectives (DQOs),
technically based, project-specific analytical data needs (MQOs and any specific analytical
JULY 2001
DRAFT FOR PUBLIC COMMENT
2-3
MARLAP
DO NOT CITE OR QUOTE
-------
Project Planning Process
91 process requirements), and measures of performance for the assessment phase of the data
92 collection activity.
93 2.3 Directed Project Planning Processes
94 The recognition of the importance of project planning has resulted in the development of a
95 variety of directed planning approaches. MARLAP does not endorse any one planning approach.
96 Users of this manual are encouraged to consider the available approaches and choose a directed
97 planning process that is appropriate to their project and agency. Appendix A, Directed Planning
98 Approaches, provides brief descriptions of several directed planning processes.
99 A graded approach to project planning will be discussed in Section 2.3.1. Standards and guidance
100 on project planning are presented in Section 2.3.2. An overview of common elements of project
101 planning is discussed in Section 2.3.3. The elements of project planning will be discussed in
102 detail in Section 2.5.
103 2.3.1 A Graded Approach to Project Planning
104 The sophistication, the level of QC and oversight, and the resources applied should be approp-
105 riate to the project (i.e., a "graded approach"). Directed planning for small or less complex
106 projects follows the logic of the process but will proceed faster and involve fewer people. The
107 goal still will be to (1) plan properly to collect only the data needed to meet the objectives of the
108 project and (2) establish the measures of performance for the implementation and assessment
109 phases of the data life cycle of the proj ect.
110 2.3.2 Guidance on Directed Planning Processes
ill The following national standards related to directed project planning for environmental data
112 collection are available:
113 • Standard Practice (D5792) for Generation of Environmental Data Related to Waste
114 Management Activities: Development of Data Quality Objectives (American Society for
115 Testing and Materials (ASTM, 1995a), which addresses the process of development of data
116 quality objectives for the acquisition of environmental data. This standard describes the DQO
117 process in detail.
118 • Standard Provisional Guide (PS85) for Expedited Site Characterization of Hazardous Waste
119 Contaminated Sites (ASTM, 1996a), which describes the Expedited Site Characterization
MARLAP JULY 2001
DO NOT CITE OR QUOTE 2-4 DRAFT FOR PUBLIC COMMENT
-------
Project Planning Process
120 (ESC) process used to identify all relevant contaminant migration pathways and determine
121 the distribution, concentration and fate of the contaminants for the purpose of evaluating risk,
122 determining regulatory compliance, and designing remediation systems.
123 • Standard Guide (D5730) Site Characteristics for Environmental Purposes with Emphasis on
124 Soil, Rock, the Vadose Zone and Ground Water (ASTM, 1996b), which covers a general
125 approach to planning field investigations using the process of defining one or more
126 conceptual site models that is useful for any type of environmental reconnaissance or
127 investigation plan with a primary focus on the surface and subsurface environment.
128 • Standard Guide (D5612) Quality Planning and Field Implementation of a Water Quality
129 Measurements Program (ASTM, 1994), which defines criteria and identifies activities that
130 may be required based on the DQOs.
131 • Standard Guide (D5851) Planning and Implementing a Water Monitoring Program (ASTM,
132 1995b), which provides a procedural flowchart for planning the monitoring of point and non-
133 point sources of pollution of water resources (surface or ground water, rivers, lakes or
134 estuaries).
135 Several directed planning approaches have been implemented by the federal sector for
136 environmental data collection activities. MARLAP does not endorse a single planning approach
137 and project planners should be cognizant of their agency's requirements for planning. The
138 following guidance is available:
139 • EPA developed the DQO Process (EPA, 2000) and has tailored DQO Process guidance for
140 specific programmatic needs of project planning under the Comprehensive Environmental
141 Response, Compensation, and Liability Act of 1980 (CERCLA/Superfund) (EPA, 1993) and
142 for site-specific remedial investigation feasibility study activities (EPA, 2000).
143 • The U. S. Army Corps of Engineers (ACE) Technical Project Planning (TPP) Process (ACE,
144 1998) was developed for technical projects planning for hazardous, toxic and radioactive
145 waste sites.
146 • DOE has developed the Streamlined Approach for Environmental Restoration (SAFER)
147 (DOE, 1993) for its environmental restoration activities.
148 • Planning guidance, including decision frameworks, for projects demonstrating compliance
149 with a dose- or risk-based regulation is available for final status radiological surveys
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 2-5 DO NOT CITE OR QUOTE
-------
Project Planning Process
150 (MARSSIM, 2000) and radiological criteria for license termination (NRC, 1998a; NRC,
151 1998b).
152 Additional information on the DQO Process (ASTM, 1995a; EPA, 2000) is presented in
153 Appendix B, The Data Quality Objectives Process.
154 2.3.3 Elements of Directed Planning Processes
155 Environmental data collection activities require planning for the use of data in decision making.
156 The various directed planning approaches, when applied to environmental data collection
157 activities, address common planning considerations. Some common elements of the planning
158 processes are:
159 1. Define the problem: Identifying the problem(s) facing the stakeholder/customer that requires
160 attention, or the concern that requires streamlining.
161 2. Identify the Decision: Defining the decision(s) or the alternative actions that will address the
162 problem(s) or concern and satisfy the stakeholder/customer, and determine if new data are
163 required to make the decision.
164 3. Specify the Decision Rule and the Tolerable Decision Error Rates: Develop a decision rule to
165 get from the problem or concern to the desired decision and define the limits on the decision
166 error rates that will be acceptable to the stakeholder/customer. The decision rule can take the
167 form of "if ...then..." statements for choosing among decisions or alternative actions.
168 4. Optimize the Strategy for Obtaining Data: Determine the optimum, cost-effective way to
169 reach the decision while satisfying the desired quality of the decision. Define the quality of
170 the data that will be required for the decision by establishing specific, quantitative and
171 qualitative analytical performance measures (e.g, MQOs). Define the process and criteria to
172 evaluate the suitability of the data to support their intended use (DQA).
173 The objective of the directed project planning process for environmental data collection activities
174 is to reach consensus among the stakeholders on defining the problem, the full range of possible
175 solutions, the desired decision, the optimal data collection strategy, and performance measures
176 for implementation and assessment phases of the project. If a cursory job is done defining the
177 problem or the desired results, the consequence will be the development of a design that may be
178 technically sound but answers the wrong question, may answer the question only after the
MARLAP JULY 2001
DO NOT CITE OR QUOTE 2-6 DRAFT FOR PUBLIC COMMENT
-------
Project Planning Process
179 collection of significant quantities of unnecessary data, or may collect insufficient data to answer
180 the question.
181 The key outputs of the directed planning process are DQOs: qualitative and quantitative
182 statements that define the project objectives and the tolerable decision error rates that will be
183 used as the basis for establishing the quality and quantity of data needed to support the decision.
184 The MQOs and the decisions on key analytical planning issues will provide the framework for
185 Analytical Protocol Specifications. The MQOs and the tolerable decision error rates will provide
186 the basis for the data assessment phase (data validation and DQA). The elements of project
187 planning will be discussed in detail in Section 2.5 from the perspective of the radioanalytical
188 specialists after introducing the concepts of the project planning team and radioanalytical
189 specialists in Section 2.4. Key analytical planning issues and Analytical Protocol Specifications
190 are discussed in Chapter 3, Key Analytical Planning Issues and Developing Analytical Protocol
191 Specifications.
192 2.4 The Project Planning Team
193 Participants in the project planning process will vary depending on the nature of the project, but
194 in most cases a multi-disciplinary team will be required. The project planning team should
195 consist of all the parties who have a vested interest or can influence the outcome (stakeholders).
196 A key to successful directed planning of environmental projects is getting the data users and data
197 suppliers to work together early in the process to understand each other's needs and require-
198 ments, to agree on the desired end product, and to establish lines of communication. Equally
199 important is having integrated teams of operational and technical experts. These experts will
200 determine whether the problem has been sufficiently defined and if the desired outcomes are
201 achievable. With technical expert input early in the planning process, efforts are focused on
202 feasible solutions, and resources are not wasted pursuing unworkable solutions.
203 2.4.1 Team Representation
204 Thus, members of the project planning team may include program and project managers,
205 regulators, public representatives, project engineers, health and safety advisors, and specialists in
206 statistics, health physics, chemical analysis, radiochemical analysis, field sampling, quality
207 assurance/quality control (QA/QC), data assessment, contract and data management, field
208 operation, and other technical specialists. The program or project manager(s) may be a Remedial
209 Project Manager (RPM), a Site Assessment Manager (SAM), or a Technical Project Officer
210 (TPO). Some systematic planning processes, such as Expedited Site Characterization, utilize a
211 core technical team supported as needed by members of larger technical and operational teams.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 2-7 DO NOT CITE OR QUOTE
-------
Project Planning Process
212 Throughout this document, the combined group of decision makers and technical experts is
213 referred to as the "project planning team."
214 The duration of service for the project planning team members can vary, as can the level of
215 participation required of each member during the various planning phases. While the project
216 planning team may not meet as frequently once the project objectives and the sampling and
217 analysis design have been established, a key point to recognize is that the project planning team
218 should not disband. Rather, the team or a "core group" of the team (including the project
219 manager and other key members) should continue to meet at agreed upon intervals to review the
220 project's progress and to deal with actual project conditions that require changes to the original
221 plan. The availability of a core team also provides the mechanism for the radioanalytical
222 laboratory to receive needed information to clarify questions as they arise.
223 A key concept built into directed planning approaches is the ability to revisit previous decisions
224 after the initial planning is completed (i.e., during the implementation phases of the
225 environmental data collection process). Even when objectives are clearly established by the
226 project planning team and contingency planning was included in the plan development, the next
227 phases of the project may uncover new information or situations, which require alterations to the
228 data collection strategy. For example, finding significantly different levels of analytes or different
229 analytes than were anticipated based on existing information may require changes in the process.
230 To respond to unexpected events, the project planning team (or the core group) should remain
231 accessible during other phases of the data collection process to respond to questions raised,
232 revisit and revise project requirements as necessary, and communicate the basis for previous
233 assumptions.
234 2.4.2 The Radioanalytical Specialists
235 Depending on the size and complexity of the project, MARLAP recognizes that a number of key
236 technical experts should participate on the project planning team and be involved throughout the
237 project as needed. When the problem or concern involves radioactive analytes, it is important
238 that the radioanalytical specialist(s) are part of the project planning team, in addition to radiation
239 health and safety specialists. MARLAP recommends that the radioanalytical specialists be a part
240 of the integrated effort of the project planning team. Throughout this manual, the term
241 "radioanalytical specialists" will be used to refer to the radioanalytical expertise needed.
242 Radioanalytical specialists may provide expertise in (1) radiochemistry and radiation/nuclide
243 measurement systems and (2) the knowledge of the chemical characteristics of the analyte of
244 concern. In particular, the radioanalytical specialist plays a key role in the development of
MARLAP JULY 2001
DO NOT CITE OR QUOTE 2-8 DRAFT FOR PUBLIC COMMENT
-------
Project Planning Process
245 MQOs. The radioanalytical specialists may also provide knowledge about sample transportation
246 issues, preparation, preservation, sample size, subsampling, available analytical protocols and
247 achievable analytical data quality. If more than one person is needed, the specialists members
248 need not be from the same organization. The radioanalytical specialists need not be from the
249 contractual radioanalytical laboratory. The participation of the radioanalytical specialists is
250 critical to the success of the planning process and the effective use of resources available to the
251 project.
252 2.5 Direct Planning Process and Role of the Radioanalytical Specialists
253 The importance of technical input in a directed planning process becomes apparent when one
254 examines the common difficulties facing the radioanalytical laboratory. Without sufficient input,
255 there is often a disconnect in translating the project planning team's analytical data requirements
256 into laboratory requirements and products. Radioanalytical advice and input during planning,
257 however, help to assure that the analytical protocol(s) selected will satisfy the data requirements,
258 including consideration of time, cost and relevance to the data requirements and budget. The role
259 of the radioanalytical specialists during the early stage of the directed planning process is to focus
260 on whether the desired radionuclides can be measured and the practicality of obtaining the
261 desired analytical data. During the latter part of the process, the radioanalytical specialists can
262 provide specific direction and fine tuning for defining the analytical performance requirements
263 (MQOs) and other items of the Analytical Protocol Specifications.
264 Planning with input from radioanalytical specialists can help ensure that the data received by the
265 data users will meet the project's DQOs. Common areas that are improved with radioanalytical
266 specialists' participation in project planning include:
267 • The correct radionuclide is measured;
268 • MQOs are adequately established and achievable;
269 • Consideration is given to the impact of half-life and parent/progeny factors;
270 • The data analysis is not compromised by interferences;
271 • Unnecessary or overly sophisticated analytical techniques are avoided in favor of analytical
272 techniques appropriate to the required level of measurement uncertainty;
273 • Optimum radioanalytical variables, such as count time and sample volume, are considered;
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 2-9 DO NOT CITE OR QUOTE
-------
Project Planning Process
274
275
276
277
278
279
• Environmental background levels are considered;
• Chemical speciation is addressed; and
• Consideration is given to lab operations (e.g., turnaround time, resources).
These improvements result in an appropriate data collection design with specified MQOs and any
specific analytical process requirements to be documented in the project plan documents and
SOWs.
280 The following sections, using the common planning elements outlined in Section 2.3.3, will
281 discuss the process and results of directed planning in more detail and emphasize the input of
282 radioanalytical specialists. Table 2.1 provides a summary of (1) the information needed by the
283 project planning team, (2) how the radioanalytical specialists participate, and (3) the output or
284 product for each element of the directed planning process. It must be emphasized that a directed
285 planning process is an iterative, rather than step-wise, process. Although the process is presented
286 in discrete sections, the project planning may not progress in such an orderly fashion. The
287 planning team will more precisely define decisions and data needs as the planning progresses and
288 use new information to modify or change earlier decisions until the planning team has
289 determined the most resource effective approach to the problem. The common planning elements
290 are used for ease of presentation and to delineate what should be covered in planning, not the
291 order of discussion.
292
293
294
295
TABLE 2.1 Summary of the Directed Planning Process and Radioanalytical Specialists Participation
Element
1. State the
problem
Information Needed by The
Project Planning Team
Key stakeholders and their
concerns.
Facts relevant to current
situation (e.g., site history,
ongoing studies).
Analytes of concern or
analytes driving risk.
Matrix of concern.
Regulatory requirements and
related issues.
Existing data and the
reliability of the information.
Known sampling constraints.
Resources and relevant
deadlines.
Radioanalytical Specialists
Particip ation/Input
Evaluate existing radiological data
for use in defining the issues (e.g.,
analytes of concern).
Assure that the perceived problem
is really a concern by reviewing
the underlying data that is the
basis for the problem definition.
Consider how resource limitations
and deadlines will impact
measurement choices.
Use existing data to begin to
define the analyte of concern and
the potential range of
concentrations.
Output/Product
Define the problem with
specificity.
Identify the primary
decision maker, the
available resources, and
constraints.
MARLAP
DO NOT CITE OR QUOTE
2-10
JULY 2001
DRAFT FOR PUBLIC COMMENT
-------
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
Project Planning Process
Element
2a. Identify
the
decision(s)
2b. Identify
inputs to
the
decision(s)
2c. Define the
decision
boundaries
3a. Develop a
decision
rule
3b. Specify
limits on
decision
error rates
Information Needed by The
Project Planning Team
• Analytical aspects related to
the decision.
• Possible alternative actions.
• Sequence and priority for
addressing the problem.
• All useful existing data.
• The general basis for
establishing an action level.
• Acquisition strategy options
(if new data is needed).
• Sampling or measurement
timeframe.
• Sampling areas and
boundaries.
• Subpopulations.
• Practical constraints on data
collection (season,
equipment, turnaround time,
etc.).
• Available protocols.
• Statistical parameter to be
used to describe the
parameter of interest and to
be compared to the action
level.
• The action level
(quantitative).
• The scale of decision
making.
• Potential consequences of
making wrong decisions.
• Possible range of the
parameter of interest.
• Allowable differences
between the action level and
the actual value.
• Acceptable level of decision
Radioanalytical Specialists
Particip ation/Input
• Provide focus on what analytes
need to be measured considering
analyte relationships and
background.
• Begin to address the feasibility of
different analytical protocols.
• Begin to identify the items of the
Analytical Protocol
Specifications.
• Begin to determine how sample
collection and handling will affect
MQOs.
• Review the quality and sufficiency
of the existing radiological data.
• Identify alternate analytes.
• Identify temporal trends and
spatial heterogeneity using
existing data.
• With the sampling specialists,
identify practical constraints that
impact sampling and analysis.
• Determine feasibility of obtaining
new data with current
methodology.
• Identify limitations of available
protocols.
• Potentially useful methods.
• Estimates of measurement
uncertainty and detection limits of
available analytical protocols.
• Assess variability in existing data
for decisions on hypothesis testing
or statistical decision theory.
• Evaluate whether the tolerable
decision error rates can be met
with available laboratory
protocols or the error tolerance
needs to be relaxed or new
Output/Product
• Statements that link the
defined problem to the
associated decision(s)
and alternative actions.
• Defined list of needed
new data.
• Define the characteristic
or parameter of interest
(analyte/matrix).
• Define the action level.
• Identify estimated.
concentration range for
analyte(s) of interest.
• Temporal and spatial
boundaries.
• The scale of decision.
• A logical, sequential
series of steps
("if... then") to resolve
the problem.
• Definition of the
baseline condition (null
hypothesis) and quanti-
tative estimates of
acceptable decision
error rates.
• Define the range of
possible parameter
JULY 2001
DRAFT FOR PUBLIC COMMENT
2-11
MARLAP
DO NOT CITE OR QUOTE
-------
Project Planning Process
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
Element
4. Optimize
the
Strategy
for
Obtaining
Data
Information Needed by The
Project Planning Team
errors or confidence.
All outputs from all previous
elements including
parameters (analytes and
matrix) of concern, action
levels, anticipated range of
concentration, tolerable
decision error rates,
boundaries, resources and
practical constraints.
Available protocols for
sampling and analysis.
Radioanalytical Specialists
Particip ation/Input
methods developed.
With sampling specialists, consider
the potential combinations of
sampling and analytical methods, in
relation to:
Sample preparation, compositing,
subsampling.
Available protocols.
Method requirement by
regulations (if any).
Detection and quantitation
capability.
MQOs achievable by method,
matrix and analyte.
Quality control sample types,
frequencies, and evaluation
criteria.
Sample volume, field processing,
preservatives, and container
requirements.
Assure that the MQOs for sample
analysis are realistic.
Assure that the parameters for the
Analytical Protocol Specifications
are complete.
Resources and time frame to
develop and validate new
method(s). if required.
Output/Product
values where the
consequence of a Type
II decision error is
relatively minor (gray
region).
The most resource-
effective sampling and
analysis design that
meets the established
constraints (i.e., number
of samples needed to
satisfy the DQOs and
the tolerable decision
error rates).
A method for testing the
hypothesis.
The MQOs and the
statement(s) of the
Analytical Protocol
Specifications.
The process and criteria
for data assessment.
342
2.5.1 Define the Problem
343 The first and most important step of the project planning process is a clear statement of the
344 fundamental issue to be addressed by the project. Correctly implemented, directed planning
345 ensures that a clear definition of the problem is developed before any additional resources are
346 committed. The project planning team should understand clearly the conditions or circumstances
347 that are causing the problem and the reason for making a decision (e.g., threat to human health or
348 environment).
349 Many projects present a complex interaction of technical, economic and political factors. The
350 problem definition should include a summary of the study objectives, regulatory context, funding
351 and other resources available, relevant deadlines, previous study results, and any obvious data
MARLAP
DO NOT CITE OR QUOTE
2-12
JULY 2001
DRAFT FOR PUBLIC COMMENT
-------
Project Planning Process
352 collection design constraints. By participating in the initial stages of the project planning, the
353 radioanalytical specialists will understand the context of the facts and logic used to define the
354 problem and begin to formulate information on applicable protocols based on the projects's
355 resources (time and budget).
356 Existing data (e.g., monitoring data, radioactive materials license, emergency actions, site permit
357 files, operating records) may provide specific details about the identity, concentrations, and
358 geographic, spatial, or temporal distribution of analytes. However, these data should be examined
359 carefully. Conditions may have changed since the data were collected. For example, additional
360 waste disposal may have occurred, the contaminant may have been released or migrated, or
361 decontamination may have been performed. In some cases, a careful review of the historical data
362 by the project planning team will show that a concern is not a problem or the problem can be
363 adequately addressed using the available data.
364 2.5.2 Identify the Decision
365 The project planning team will define the decision(s) to be made (or the question the project will
366 attempt to resolve) and the inputs and boundaries to the decision. There may also be multiple
367 decision criteria that have to be met and each should be clearly defined. For example, the
368 decision may be for an individual survey area rather than the site as a whole, or a phase of the site
369 closure project (scoping, characterization, operation or final status survey) rather than the project
370 as a whole because of the different objectives and data requirements.
371 The decision should be clear and unambiguous. It may be useful to state specifically what
372 conclusions may and may not be drawn from the data. If the study is to be designed, for example,
373 to investigate whether or not a site may be released for use by the general public, then the project
374 planning team may want to specifically exclude other possible uses for the data.
375 2.5.2.1 Action Level
376 The term "action level" is used in this document to denote the numerical value that will cause the
377 decision maker to choose one of the alternative actions. The action level may be a derived
378 concentration guideline level, background level, release criteria, regulatory decision limit, etc.
379 The action level is often associated with the type of medium, analyte and concentration limit.
380 Some action levels, such as the release criteria for license termination, are expressed in terms of
381 dose or risk. The release criterion is typically based on the total effective dose equivalent
382 (TEDE), the committed effective dose equivalent (CEDE), risk of cancer incidence (morbidity)
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 2-13 DO NOT CITE OR QUOTE
-------
Project Planning Process
383 or risk of cancer death (mortality) and generally cannot be measured directly. For example, in site
384 cleanup, a radionuclide-specific predicted concentration or surface area concentration of specific
385 nuclides that can result in a dose (TEDE or CEDE) or specific risk equal to the release criterion
386 is called the "derived concentration guideline level" (DCGL). A direct comparison can be made
387 between the project's analytical measurements and the DCGL (MARSSEVI, 2000). For drinking
388 water analysis, an example of an action level would probably be a radionuclide specific
389 concentration based on the Maximum Contaminant Level under the Safe Drinking Water Act.
390 The project planning team should also determine possible alternative actions that may be taken.
391 Consideration should also be given to the option of taking no action, as this option is frequently
392 overlooked (e.g., no technology available, too costly, relocation will create problems).
393 During these discussions of the directed planning process, the role of the radioanalytical
394 specialists is to ensure that the analytical aspects of the project have been clearly defined and
395 incorporated into the decision(s). The radioanalytical specialists focus on defining: (1) the
396 parameter (analyte/matrix) of interest; (2) what analytical information could resolve the problem;
397 and (3) the practicality of obtaining the desired field and laboratory data. Sections 3.3.1 through
398 3.3.7 of Chapter 3 discuss in more detail the analytical aspects of the decision (or question) and
399 determining the characteristic or parameter of concern. This information is incorporated into the
400 Analytical Protocol Specifications.
401 2.5.2.2 Scale of the Decision
402 The project planning team clearly should define the geographical area(s) to which the decision
403 will apply. The scale of the decision selected should be the smallest, most appropriate subset of
404 the population for which decisions will be made based on the spatial or temporal boundaries. For
405 example, at a remediation site, a survey unit is generally formed by grouping contiguous site
406 areas with a similar use history and the same classification of potential concentration of the
407 analyte of interest. The survey unit will be defined with a specified size and shape for which a
408 separate decision will be made as to whether the unit attains the site-specific reference-based
409 cleanup standard for the designated analyte of interest (MARSSEVI, 2000; NRC, 1998c).
410 The survey unit is established to delineate areas or volumes of similar composition and history
411 for which a single decision can be made based on the statistical analysis of the data. The
412 variability in the measurement data for a survey unit is a combination of the imprecision of the
413 measurement process and the real spatial and temporal variability of the analyte concentration. If
414 the measurement data include a background contribution, the spatial variability of the
415 background adds to the overall measurement variability.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 2-14 DRAFT FOR PUBLIC COMMENT
-------
Project Planning Process
416 2.5.2.3 Inputs and Boundaries to the Decision
417 The project planning team determines the specific information and data required for decision
418 making. The statistical parameter (e.g., mean) that will be used in the comparison to the action
419 level should be established. Typically, the study boundaries are discussed when the project
420 planning team defines the problem. Changing conditions (e.g., weather, temperature, humidity)
421 that could impact the success of sampling or analysis or data interpretation should be considered
422 as well. The radioanalytical specialists can provide input during the determination of the
423 appropriate action level and the appropriate parameter of interest (e.g., mean concentration).
424 2.5.2.4 Data Needs
425 The project planning team should develop a list of the specific data (number and type) and data
426 requirements (quality). An estimate of the expected variability of the data will be needed.
427 Existing data, experience and scientific judgement can be used to establish the estimate.
428 Information on environmental background levels and variability may be needed (see Chapter 3
429 for a discussion of background). The project planning team establishes whether the existing data
430 are sufficient or whether new data are needed to resolve the problem.
431 2.5.3 Specify the Decision Rule and the Tolerable Decision Error Rates
432 A decision statement or rule is developed by combining the decisions and the alternative actions.
433 The decision rule presents the strategy or logical basis for choosing among the alternative
434 decisions, generally by use of a series of "if. ..then" statements. For a complex problem, it may be
435 helpful to develop a logic flow diagram (called a decision tree or decision framework), arraying
436 each element of the issue in its proper sequence along with the possible actions. The decision
437 rule identifies (1) the action level that will be a basis for decision and (2) the statistical parameter
438 that is to be compared to the action level.
439
440
441
Example of a Decision Rule:
If the mean concentration in the survey unit is less than the action level, then the
survey unit is in compliance with the release criterion.
442 The radioanalytical specialists play a key role in the development of alternative technical actions
443 that are realistic and quantifiable and that satisfy the programmatic and regulatory needs. The
444 results of the technical actions must be measurable: the protocols suggested will be able to detect
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 2-15 DO NOT CITE OR QUOTE
-------
Project Planning Process
445 the radionuclide of interest, (see Chapter 3, Critical Analytical Planning Issues and Developing
446 Analytical Protocol Specifications, for additional discussion on background.)
447 For each proposed alternative technical action, the radioanalytical specialists can:
448 • Focus the project planning team on what radionuclides will need to be measured and what
449 types of analytical techniques are available;
450 • Address whether it is feasible to obtain the necessary analytical results;
451 • Present the technical limitations (i.e., the minimum detectable concentrations—MDCs) of
452 available measurement systems; and
453 • Address how sample collection and handling will affect what measurement techniques can be
454 used.
455 The project planning team also assesses the potential consequences of making a wrong decision.
456 While the possibility of a decision error can never be totally eliminated, it can be controlled. The
457 potential consequences of a decisions error are used to establish tolerable limits on the
458 probability that the data will mislead the decision maker into making an incorrect decision, (see
459 Appendix B for a discussion of hypothesis testing, action levels, and Type I and Type n decision
460 errors). The decision rule and decision makers' limits on the decision error rates are used to
461 establish performance criteria for a data collection design.
462 In developing the tolerable decision error rate, the team needs to look at alternative measurement
463 approaches, the sources of error in field and laboratory handling of samples and analysis, factors
464 that would influence the likelihood of a Type I or Type II error, estimates of the cost of analysis,
465 and judicious use of resources. Determining realistic levels of tolerable decision error rates for
466 the decision rule will reduce or eliminate attempts by the project planning team in developing
467 and optimizing the sampling and analysis design that later will have to be re-designed to attain
468 more realistic decision error rates.
469 2.5.4 Optimize the Strategy for Obtaining Data
470 During the process of developing and optimizing the options for the sampling and analysis of
471 data, the technical team members should determine the most resource effective analytical
472 protocols and associated quality control that will meet all the requirements (desired outputs)
MARLAP JULY 2001
DO NOT CITE OR QUOTE 2-16 DRAFT FOR PUBLIC COMMENT
-------
Project Planning Process
473 established by the project planning team. Optimizing the data collection design generally requires
474 extensive coordination between the radioanalytical specialists and the sampling specialists.
475 Typical issues that require consideration in the development of the analysis design include the
476 number of samples required, the analytical protocol specifications, which include the MQOs
477 (e.g., a statement of the required method uncertainty) required of the analytical procedures. The
478 analytical protocol specifications, which include the MQOs, will be discussed in Sections 2.5.4.1
479 and 2.5.4.2 below. In general, the more certainty required in the DQOs, the greater the number of
480 samples or the more precise and unbiased the measurements need to be. During planning, the
481 costs and time for field and analytical procedures must be balanced against the level of certainty
482 that is needed to arrive at an acceptable decision.
483 The radioanalytical specialists are involved in evaluating the technical options and their effect on
484 the sources of decision error, their resource requirements and the ability to meet the project's
485 objectives. The radioanalytical specialists can identify an array of potential analytical methods,
486 which can be combined in analytical protocols to meet the defined data needs and MQOs.
487 Working with the sampling specialists, potential sampling methods are identified based on the
488 sample requirements of the potential analytical protocols and other sampling constraints. The
489 planning team specialists need to consider sources of bias and imprecision that will impact the
490 representativeness of the samples and the accuracy of the data collected. Appropriate
491 combinations of sampling methods, analytical protocols and sampling constraints can then be
492 assessed with regard to resource effectiveness.
493 It may be useful at this point for the project planning team to perform a sensitivity analysis on the
494 input parameters that contribute to the final analytical result. The final analytical result directly
495 impacts the decision, so this sensitivity analysis will allow the project planning team to identify
496 the portions of the analytical protocols, which potentially have the most impact on the decision.
497 Once identified, these portions of the analytical protocols can be targeted to receive a propor-
498 tionally larger share of the resources available for developing the protocols.
499 2.5.4.1 Analytical Protocol Specifications
500 Requirements of the desired analytical protocol(s) should be based on the intended use of the
501 data. That is, project-specific critical parameters should be considered, including the type of
502 radioactivity and the nuclides of concern, the anticipated range of concentrations, the media type
503 and complexity, regulatory required methods and customer method preferences, the measurement
504 uncertainty required at some activity concentration, detection limits required, necessary chemical
505 separation, qualification or quantification requirements, QC requirements and turnaround time
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 2-17 DO NOT CITE OR QUOTE
-------
Project Planning Process
506 needed. MQOs are a key component of the Analytical Protocol Specifications and are discussed
507 in Section 2.5.4.2. Chapter 3, Key Analytical Planning Issues and Developing Analytical
508 Protocol Specifications, contains more detailed discussion on some of the key decisions and
509 needed input to successfully optimize the sampling and analysis design and develop Analytical
510 Protocol Specifications. Chapter 6 discusses the selection of an analytical protocol from the
511 laboratory's perspective.
512 The project planning team should ensure that there are analytical methods available to provide
513 acceptable measurements. If analytical methods do not exist, the project planning team will need
514 to consider the resources needed to develop a new method, reconsider the approach for providing
515 input data, or perhaps reformulate the decision statement.
516 2.5.4.2 Measurement Quality Objectives
517 When additional data are to be obtained, the project planning process should establish measures
518 of performance for the analysis (MQOs) and evaluation of the data. Without these measures of
519 performance, data assessment is difficult and arbitrary.
520 A MQO is a statement of a performance objective or requirement for a particular method
521 performance characteristic such as the required method uncertainty at some concentration. MQOs
522 can be both quantitative and qualitative performance objectives. Quantitative and qualitative
523 MQOs are used for real-time compliance monitoring by field and lab staff and during subsequent
524 assessments and data usability determinations. Quantitative MQOs provide numerical criteria for
525 field and laboratory QC samples or procedure performance (e.g., specifications for MDC, yield,
526 efficiency, laboratory control sample precision and recovery, blank levels, lab duplicate
527 precision, collocated sample precision). Precision, bias, completeness, and sensitivity are
528 common data quality indicators for which quantitative MQOs could be developed during the
529 planning process (ANSI/ASQC, 1994). Thus, quantitative MQOs are statements that contain
530 specific units of measure, such as: x percent recovery, x percent relative standard uncertainty, a
531 standard deviation of x Bq/L, or a MDC of x Bq/g. The specificity of the MQOs allows specific
532 comparisons of the data to an MQO. Chapter 3 provides detailed guidance on developing MQOs
533 for select method performance characteristics.
534 A graded approach should be taken to the selection of the MQOs. For example, from a project
535 viewpoint, it is highly practical and economical to establish MQOs on a graded basis that are in
536 concert with the anticipated range of the analytes concentration compared to the action level. For
537 example, the required method uncertainty, when the analyte concentration is much greater than
MARLAP JULY 2001
DO NOT CITE OR QUOTE 2-18 DRAFT FOR PUBLIC COMMENT
-------
Project Planning Process
538 the action level, can be less restrictive than when the analyte concentration approaches the action
539 level. These decisions are extremely important in the protocol selection process.
540 The MQOs for the analytical data should be documented in the project plan documents (e.g., the
541 QA Project Plan). MQOs are also the basis for the data verification and validation criteria (see
542 Appendix D, Section 2.7, for discussion of MQOs and QA Project Plans).
543 2.6 Results of the Directed Planning Process
544 By the end of the directed planning process, the project planning team has established their
545 priority of concerns, the definition of the problem, the decision(s) or outcome to address the
546 posed problem, the inputs and boundaries to the decision(s), and the tolerable decision error
547 rates. They have also agreed on decision rules that incorporate all this information into a logic
548 statement about what must be done to obtain the desired answer. The key output of the planning
549 process is the DQOs: qualitative and quantitative statements that clarify study objectives, define
550 the appropriate type of data, and specify the tolerable rate of making decision errors that will be
551 used as the basis for establishing the quantity and quality of data needed to support the decisions
552 and the criteria for data assessment.
553 If new data are required, then the project planning team has defined the desired analytical quality
554 of the data (MQOs). That is, the project planning team has determined the type, quantity, and
555 quality of data needed to support a decision. The directed planning process has clearly linked
556 sampling and analysis efforts to an action and a decision. This linkage allows the project
557 planning team to determine when enough data have been collected.
558 If new data are to be obtained, the project planning team has developed the most resource-
559 effective sampling and analysis design that will provide adequate data for decision making.
560 Based on the DQOs, the project planning team specifies the sampling collection design and
561 Analytical Protocol Specifications, including:
562 • The type and quantity of samples to be collected;
563 • Where, when, and under what conditions they should be collected;
564 • What radionuclides are to be measured; and
565 • The MQOs to ensure that the analytical errors are controlled sufficiently to meet the tolerable
566 decision error rates specified in the DQOs.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 2-19 DO NOT CITE OR QUOTE
-------
Project Planning Process
567 2.6.1 Output Required by the Radioanalytical Laboratory: The Analytical Protocol
568 Specifications
569 As a result of directed planning, the description of the DQOs for the project and the Analytical
570 Protocol Specifications, which contain the MQOs and any specific analytical process require-
571 ments for additional data will provide the radioanalytical laboratory with a clear and definitive
572 description of the desired data, as well as the purpose and use of the data. This information will
573 be provided to the project implementation team through the SOW and the project plan
574 documents. Precise statements of analytical needs may prevent the radioanalytical laboratory
575 from:
576 • Having to make a "best guess" as to what data are really required;
577 • Using the least costly or most routine protocol, which may not meet the needed data quality;
578 • Independently developing solutions for unresolved issues without direction from the project
579 planning team; and
580 • Having "moving targets" and "scope creep" that stem from ambiguous statements of work.
581 The output of the planning process, from the perspective of the radioanalytical laboratory, is the
582 Analytical Protocol Specifications. The Analytical Protocol Specifications should contain the
583 minimum level of specificity required to meet the project data requirements. In accordance with a
584 performance based measurement approach the laboratory will use this information to select or
585 develop (specific) analytical protocols that will meet the MQOs. The Analytical Protocol
586 Specifications should present the resolution of the project planning team on both general issues
587 and matrix-specific issues. Chapter 3, Key Analytical Planning Issues and Developing Analytical
588 Protocol Specifications., addresses some of the common radioanalytical planning issues.
589 The Analytical Protocol Specifications should include, but not be limited to:
590 • The radionuclide(s) of concern;
591 • The media of concern with information on chemical, explosive and other hazardous
592 components;
593 • The anticipated concentration range (estimate, maximum or detection capability);
594 • The MQOs desired for the radionuclides of concern;
595 • The sample preparation and preservation requirements (laboratory and field);
596 • The type and frequency of QC samples required of each radionuclide of concern;
597 • The sample transport, tracking and custody requirements;
598 • The required analytical turnaround time for the project and the anticipated budget for the
599 analysis; and
MARLAP JULY 2001
DO NOT CITE OR QUOTE 2-20 DRAFT FOR PUBLIC COMMENT
-------
Project Planning Process
600 • The data reporting requirements.
601 2.6.2 Chain of Custody
602 Requirements for formal Chain of Custody (COC) should be specified in the Analytical Protocol
603 Specifications if required. COC procedures provide the means to trace possession and handling
604 of the sample from collection to data reporting. The data report requires a number of items, not
605 all of which can be listed here. COC will impact how the field and lab components handle the
606 sample. COC is discussed in Chapter 10 and Chapter 11.
607 2.7 Project Planning and Project Implementation and Assessment
608 A directed planning process generally is considered complete with the approval of an optimal
609 data collection design approach or when historical data are deemed sufficient to support the
610 desired decision. However to complete the process, the project planning team clearly should
611 document the results of the planning process and link DQOs and MQOs to the implementation
612 and assessment processes. The directed planning process is the first activity in the project's
613 planning phase (see Figure 1.1, "The Data Life Cycle"). The planning process outputs are key
614 inputs to the implementation and assessment processes of the data collection activities. That is,
615 the outputs of the directed planning process are the starting point for developing plan documents,
616 obtaining analytical services, selecting specific analytical protocols and assessing the data
617 collected. This section will provide an overview of the next steps of the planning phase and the
618 linkage to the implementation and assessment phases and to other chapters in MARLAP, Part I.
619 2.7.1 Documenting the Planning Process
620 A concept inherent in directed planning approaches is the establishment of a formal process to
621 document both the decisions and supporting logic established by the team during the project
622 planning process. Establishing this documentation process is not only good management practice,
623 but also tends to prevent situations where new team members recreate the past logic for activities
624 being performed upon the departure of their predecessors. As actual field conditions or other
625 situations force changes to the original plans, the documentation can then be updated through a
626 change control process to continue to maintain the technically defensible basis for the actions
627 being taken.
628 When properly documented, the directed planning process:
629 • Provides a background narrative of the project;
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 2-21 DO NOT CITE OR QUOTE
-------
Project Planning Process
630 • Defines the necessary input needed (nuclides, matrices, estimate of concentration range, etc.);
631 • Defines the constraints and boundaries within which the project would have to operate;
632 • Defines the decision rule, which states the action level that will be the basis for the decision
633 and the parameter that is to be compared to the action level;
634 • Identifies the tolerable decision error rates;
635 • Identifies MQOs for new analytical data; and
636 • Identifies processes and criteria for usability of the data.
637 The results of the project planning process are also needed for the development of project plan
638 documents required for implementing the sampling and analysis activities. These project plan
639 documents may include a Quality Assurance Project Plan (QAPP), Work Plan, or Sampling and
640 Analysis Plan (SAP). The format and naming of plan documents are usually a function of the
641 authoring organization's experience, the controlling federal or state regulations, or the controlling
642 Agency. Project plan documents are discussed in Chapter 4, Project Plan Documents, and in
643 Appendix D, Content of Project Plan Documents. The project plan documents will rely on the
644 planning process outputs, including the MQOs, to describe in comprehensive detail the necessary
645 QA, QC, and other technical activities that must be implemented to ensure that the results of the
646 work performed will satisfy the stated DQOs. The project plan documents should also document
647 the processes and criteria developed for data assessment. MARLAP recommends that the
648 planning process rationale is documented and the documentation integrated with the project plan
649 documents. Documentation of the planning process can be incorporated directly in the project
650 plan documents or through citation to a separate report on the planning process.
651 2.7.2 Obtaining Analytical Services
652 If contractual laboratory services are required, the contracting office or Sample Management
653 Office (SMO) should rely on the planning process statements of required data and data quality,
654 the Analytical Protocol Specifications, to develop the Statement of Work (SOW) for the
655 laboratory. The SOW is the contractual agreement, which describes the project scope and
656 requirements (i.e., what work is to be accomplished). Contracting laboratory services is discussed
657 in Chapter 5, Obtaining Laboratory Services, and Chapter 7, Evaluating Methods and
658 Laboratories. MARLAP recommends that a SOW be developed even if a contract is not
659 involved, for example, when an agency employs one of its own labs.
660 2.7.3 Selecting Analytical Protocols
661 From an analytical perspective, one of the most important functions of a directed planning
662 process is the identification and resolution of key analytical planning issues for a project. A key
MARLAP JULY 2001
DO NOT CITE OR QUOTE 2-22 DRAFT FOR PUBLIC COMMENT
-------
Project Planning Process
663 analytical planning issue may be defined as one that has the potential to be a significant contribu-
664 tor of uncertainty to the analytical process and ultimately the resulting data. Identifying key
665 analytical issues for a particular process requires a clear understanding of the analytical process.
666 It is the role of the radioanalytical specialist on the project planning team to ensure that key
667 analytical planning issues have been clearly defined and articulated and incorporated into the
668 principal decision or principal study question. Chapter 3 discusses the key analytical planning
669 issues.
670 The selection of radioanalytical protocols by the laboratory is made in response to the Analytical
671 Protocol Specifications (for each analyte/matrix) developed by the project planning team as
672 documented in the SOW. Unless required by regulatory policy, rarely will a radioanalytical
673 method be specifically stated. A number of radioanalytical methods are available but no one
674 method provides a general solution; all have advantages and disadvantages. The selection of a
675 method is related to a broad range of consideration, including analyte and matrix characteristics,
676 technical complexity and practicality of the method, quality requirements, availability of
677 equipment, facility and staff resources, regulatory and economic considerations, and practicality
678 and previous use of the method. Chapter 6 discusses the selection of a protocol, as well as, the
679 modification of an existing protocol to account for changes in sample substrate.
680 2.7.4 Assessment Plans
681 Concurrent with the development of MQOs and other specifications of the optimized analytical
682 design, is the development of the data assessment plans. Data assessment is difficult and
683 arbitrary when attempted at the end of the project without planning and well defined, project
684 specific criteria. The development of these plans during the project planning process should
685 ensure that the appropriate documentation will be available for assessment and that those
686 implementing and assessing data will be aware of how the data will be assessed. Assessment of
687 environmental data consists of three separate and identifiable phases: data verification, data
688 validation, and data quality assessment (DQA). Verification and validation pertain to evaluation
689 of analytical data generated by the laboratory. DQA considers all sampling, analytical, and data
690 handling details, and other historical project data when determining the usability of data in the
691 context of the decisions to be made. The focus of verification and validation is on the analytical
692 process and a data point by data point review, while DQA considers the entire data collection
693 process and the entire data set as it assesses data quality. Verification, validation, and DQA
694 assure the technical strengths and weaknesses of the overall project data are known, and
695 therefore, establishes the technical defensibility of the data. Assessment plan documents are
696 discussed in detail in Chapters 8 and 9.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 2-23 DO NOT CITE OR QUOTE
-------
Project Planning Process
697 2.7.4.1 Data Verification
698 The data verification process should be defined during the project planning process and
699 documented in a data verification plan or the project plan documents (e.g., the QAPP). The
700 verification plan should specify the types of documentation needed for verification. Analytical
701 data verification assures that laboratory conditions and operations were compliant with the
702 contractual SOW and project plan (i.e., SAP or QAPP). The contract for analytical services and
703 the project plan determine the procedures the laboratory must use to produce data of acceptable
704 quality (MQOs) and the content of the analytical data package. Verification compares the
705 material delivered by the laboratory to these requirements and checks for consistency of the data
706 throughout the data package, correctness of calculations, and completeness of the results to
707 ensure all documentation is available. Compliance, exceptions, missing documentation and the
708 resulting inability to verify compliance must be recorded in the data verification report. Data
709 verification is discussed in more detail in Chapter 8, Radiological Data Verification and
710 Validation.
711 2.7.4.2 Data Validation
712 Performance objectives and criteria for data validation should be developed during the project
713 planning process and documented in a separate plan or included in the project plan documents
714 (e.g., QAPP). Guidance on Data Validation Plans is provided in Chapter 8, Radiological Data
715 Verification and Validation. After the data are collected, data validation activities will rely on the
716 planning process statements of the MQOs to confirm whether the obtained data meet the
717 requirements of the project.
718 2.7.4.3 Data Quality Assessment
719 The DQA process evaluates whether the quality and quantity of data will support their intended
720 use. The DQA process determines whether the data meet the assumptions under which the DQOs
721 and the data collection design were developed and whether the analytical uncertainty in the data
722 will allow the decision maker to use the data to support the decision within the tolerable decision
723 error rates established during the directed planning process. Guidance on the DQA Process and
724 plan development is provided in Chapter 9, Data Quality Assessment. The process and criteria to
725 be used for DQA process should be developed by the project planning team and documented in
726 the project plan documents or in a stand alone plan that is cited or appended to the project plan
727 documents.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 2-24 DRAFT FOR PUBLIC COMMENT
-------
Project Planning Process
728
729
730
731
732
733
Summary of Recommendations
• MARLAP recommends the use of a directed project planning process.
• MARLAP recommends that the radioanalytical specialists be a part of the integrated effort
of the project planning team.
• MARLAP recommends that the planning process rationale be documented and the
documentation integrated with the project plan documents.
734 2.8 References
735 American National Standards Institute and the American Society for Quality Control
736 (ANSI/ASQC). 1994. Specifications and Guidelines for Quality Systems for Environmental
737 Data Collection and Environmental Technology Programs, National Standard E-4.
738 American Society for Testing and Materials (ASTM). 1994. Standard Guide for Quality
739 Planning and Field Implementation of a Water Quality Measurements Program, D5612.
740 American Society for Testing and Materials (ASTM). 1995a. Standard Practice for Generation
741 of Environmental Data Related to Waste Management Activities: Development of Data
742 Quality Objectives, D5792-95.
743 American Society for Testing and Materials (ASTM). 1995b. Standard Guide for Planning and
744 Implementing a Water Monitoring Program, D5851.
745 American Society for Testing and Materials (ASTM). 1996a. Standard Provisional Guidance for
746 Expedited Site Characterization of Hazardous Waste Contaminated Sites, PS85-96.
747 American Society for Testing and Materials (ASTM). 1996b. Standard Guide for Site
748 Characterization for Environmental Purposes with Emphasis on Soil, Rock, the Vadose Zone
749 and Ground Water, D5730-96.
750 MARSSEVL 2000. Multi-Agency Radiation Survey and Site Investigation Manual, Revision 1.
751 NUREG-1575 Rev 1, EPA 402-R-97-016 Revl, DOE/EH-0624 Revl. August. Available
752 from http://www.epa.gov/radiation/marssim/filesfm.htm.
753 U.S. Army Corps of Engineers (ACE). 1998. Technical Project Planning (TPP) Process.
754 Engineer Manual EM-200-1-2.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 2-25 DO NOT CITE OR QUOTE
-------
Project Planning Process
755 U.S. Department of Energy (DOE). December 1993. Remedial Investigation/Feasibility Study
756 (RI/FS) Process, Elements and Techniques Guidance, Module 7 Streamlined Approach for
757 Environmental Restoration., Office of Environmental Guidance, RCRA/CERCLA Division
758 and Office of Program Support, Regulatory Compliance Division Report DOE/EH-
759 94007658.
760 U.S. Environmental Protection Agency (EPA). September 1993. Data Quality Objective Process
761 for Superfund. EPA/540/G-93/071, Washington, DC.
762 U.S. Environmental Protection Agency (EPA). 2000. Guidance for the Data Quality Objective
763 Process (EPA QA/G-4). EPA/600/R-96/055, Washington, DC. available from www.epa.gov/
764 qualityl/qa_docs.html.
765 U.S. Environmental Protection Agency (EPA). 1998. Guidance for the Quality Assurance
766 Project Plans (EPA QA/G-5). EPA/600/R-98/018, Washington, DC.
767 U.S. Environmental Protection Agency (EPA). 2000. Data Quality Objectives Process for
768 Hazardous Waste Site Investigations (Quality Assurance/G-4HW), EPA 600/R-00/007,
769 Washington, DC.
770 U.S. Nuclear Regulatory Commission (NRC). 1998a. Decision Methods for Dose Assessment to
111 Comply with Radiological Criteria for License Termination. NUREG-1549 (Draft).
772 U.S. Nuclear Regulatory Commission (NRC). 1998b. Demonstrating Compliance with the
773 Radiological Criteria for License Termination. Regulatory Guide DG-4006.
774 U.S. Nuclear Regulatory Commission (NRC). 1998c. A Nonparametric Statistical Methodology
775 for the Design and Analysis of Final Status Decommissioning Surveys. NUREG-1505, Rev. 1.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 2-26 DRAFT FOR PUBLIC COMMENT
-------
i 3 KEY ANALYTICAL PLANNING ISSUES
2 AND DEVELOPING ANALYTICAL PROTOCOL
3 SPECIFICATIONS
4 3.1 Introduction
5 This chapter provides an overview of key analytical planning issues that should be addressed and
6 resolved during a directed planning process (see Chapter 2). The resolution of these issues results
7 in the development of Analytical Protocol Specifications (APSs). A key analytical planning issue
8 may be defined as one that has a significant effect on the selection and development of analytical
9 protocols, or one that has the potential to be a significant contributor of uncertainty to the
10 analytical process and, ultimately, the resulting data. It should be noted that a key analytical
11 planning issue for one project may not be a key issue for another project. From an analytical
12 perspective, one of the most important functions of a directed planning process is the
13 identification and resolution of these key issues for a project.
14 In accordance with a performance-based approach, APSs only should contain the minimum level
15 of specificity required to meet the project or program data requirements and resolve the key
16 analytical planning issues. Identification and resolution of these issues should be an integral part
17 of a directed planning process, and the APSs should be an output or product of that process. This
18 chapter provides a focused examination of analytical planning issues and the development of
19 APSs.
20 In order to assist the project planning team in identifying key issues, this chapter provides a list
21 of potential key analytical planning issues. Neither the list nor discussion of these potential issues
22 is an exhaustive examination of all possible issues for a project. However, this chapter does
23 provide a framework and a broad base of information that can assist in the identification of key
24 analytical planning issues for a particular project during a directed planning process.
25 Analytical planning issues can be divided into two broad categories—those that tend to be
26 matrix-specific and those that are more general in nature. While there is certainly some overlap
27 between these two broad categories, MARLAP divides analytical planning issues along these
28 lines because of the structure and logic it provides in developing APSs. This approach involves
29 identifying key analytical planning issues from the general (non-matrix-specific) issues first and
30 then proceeding on to the matrix-specific issues. Examples of non-matrix-specific analytical
31 planning issues include sample tracking and custody issues. These general issues are discussed in
32 detail in Section 3.3. Examples of matrix-specific issues include filtration and preservation issues
33 of water samples. Matrix-specific analytical planning issues will be discussed in detail in Section
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 3-1 DO NOT CITE OR QUOTE
-------
Key Analytical Planning Issues...
34 3.4. Section 3.5 provides guidance on assembling the APSs from the resolution of these issues.
35 Section 3.6 discusses defining the level of protocol performance demonstration required for a
36 particular project, and Section 3.7 discusses incorporating the APSs into the project plan
37 documents.
38 3.2 Overview of the Analytical Process
39 Identifying key analytical issues for a particular project requires a clear understanding of the
40 analytical process. The analytical process as described in Chapter 1 includes all activities, starting
41 with field sample preparation and preservation, followed by sample receipt and inspection,
42 laboratory sample preparation; sample dissolution; chemical separations; instrument measure-
43 ments, data reduction and reporting, and sample tracking and quality control of the process.
44 Figure 3.1 illustrates the analytical process. It should be noted that a particular project's ana-
45 lytical process may not include all of the activities listed above. For example, if the project's
46 analytical process involves performing gamma spectrometry on soil samples, sample dissolution
47 and chemical separation activities normally are not required. Each step of a particular analytical
48 process contains potential planning issues that may be key analytical planning issues depending
49 on the nature and data requirements of the project. Therefore, it is important to identify the
50 relevant activities of the analytical process for a particular project early in the directed planning
51 process. Once the analytical process for a particular project has been established, key analytical
52 planning issues, including both general and matrix-specific ones, can be identified.
53 3.3 General Analytical Planning Issues
54 This section discusses a number of general analytical planning issues that are common to many
55 types of projects and are often key planning issues, depending on the nature and data
56 requirements of the project. (Section 6.5 of Chapter 6 also discusses a number of these planning
57 issues to provide context on the method selection process.) This section presents each planning
58 issue as an activity to be accomplished during a directed planning process and also identifies the
59 expected outcome of the activity in general terms. The resolution of these general analytical
60 planning issues, particularly those that are key planning issues for a project, provides the basic
61 framework of the APSs and, therefore, should be identified and resolved before proceeding to
62 matrix-specific planning issues. Normally the resolution of these issues results, at a minimum, in
63 an analyte list, identified matrices of concern, measurement quality objectives (MQOs), and
64 established frequencies and acceptance criteria for quality control (QC) samples. The resolution
65 of matrix-specific issues, particularly those that are key issues for a project, normally provides
66 the necessary additions and modifications to the basic framework of the APSs needed to
67 complete and finalize the specifications. MARLAP recommends that any assumptions made
MARLAP JULY 2001
DO NOT CITE OR QUOTE 3 -2 DRAFT FOR PUBLIC COMMENT
-------
Key Analytical Planning Issues..
Sample Tracking
Field Sample Preparation
and Preservation
Sample Receipt and Inspection
Laboratory Sample Preparation
Sample Dissolution
Chemical Separation of
Radionuclides of Concern
Preparation of Samples for
Instrument Measurements
Instrument Measurements
Data Reduction and Reporting
QA/QC
Figure 3.1 — Typical components of an analytical process
JULY 2001
DRAFT FOR PUBLIC COMMENT
5-3
MARLAP
DO NOT CITE OR QUOTE
-------
Key Analytical Planning Issues...
68 during the resolution of key analytical planning issues are documented, and that these
69 assumptions are incorporated into the appropriate narrative sections of project plan documents.
70 Documenting these assumptions may help answer questions or help make decisions during the
71 implementation and assessment phases of the project.
72 3.3.1 Develop Analyte List
73 From an analytical perspective, one of the most important planning issues that should be
74 addressed very early in a directed planning process by the project planning team is the target
75 analyte list—the radionuclides of concern for the project. For many projects, data are available
76 from previous activities for this purpose. Four possible sources of information are (1) historical
77 data, (2) process knowledge, (3) previous studies, and (4) information obtained from conducting
78 a preliminary survey or characterization study. Although discussed separately in Section 3.3.3,
79 the identification and characterization of matrices of concern is often done concurrently with the
80 development of an analyte list.
81 Historical data are one source of existing information. Many activities associated with
82 radioactive materials have been well-documented. For example, activities licensed by the
83 Nuclear Regulatory Commission (NRC) or NRC Agreement States normally generate much
84 documentation. Chapter 3 of MARS SUV! (2000) provides guidance on obtaining and evaluating
85 historical site data.
86 Another source of existing information is process knowledge. Some sites are associated with a
87 specific activity or process that involved radioactive material, where the process was well-
88 defined and the fate of the radioactive material in the process was known or controlled. Examples
89 include uranium and rare earth ore processing, operations at Department of Energy (DOE)
90 weapons facilities, and operations at commercial nuclear power plants. (See Section 6.5.2 of
91 Chapter 6 for additional discussion on process knowledge.)
92 A third source of existing information is previous studies. Similar projects or studies of related
93 topics can provide valuable information during a directed planning process. Previous studies may
94 provide useful information on background radiation. Many radionuclides are present in measur-
95 able quantities in the environment. Natural background radiation is due both to primordial and
96 cosmogenic radionuclides. Anthropogenic background includes radionuclides that are ubiquitous
97 in the environment as a result of such human activities as the atmospheric testing of nuclear
98 weapons. Natural and anthropogenic backgrounds can be highly variable even within a given site.
99 It may be important to consider the background and its variability when choosing an action level
MARLAP JULY 2001
DO NOT CITE OR QUOTE 3 -4 DRAFT FOR PUBLIC COMMENT
-------
Key Analytical Planning Issues...
100 and when establishing the MQOs. Every effort should be made to obtain as much existing infor-
101 mation as possible priorto initiating a directed planning process.
102 Sometimes there are little or no historical data that can help identify radionuclides or the
103 concentration range of potential concern, or the existing data may be of inadequate quality. In
104 these cases, it may be necessary to perform preliminary analyses to identify the radionuclides of
105 concern or their concentration range. A fourth source of information is generated by conducting a
106 preliminary survey or characterization study. The design of preliminary surveys or characteriza-
107 tion studies should be part of the project planning process. The need for fast turnaround and
108 lower costs at this stage of the project may lead to different data quality objectives (DQOs) and
109 MQOs that are less restrictive than those used for the primary phase of the project. However, it is
110 important that analytical requirements for the survey or study be established during the project
ill planning process. Gross alpha, gross beta, and gamma spectrometry analyses often are used for
112 preliminary survey or characterization studies.
113 The benefits of performing these types of measurements include:
114 • Rapid analysis and short turnaround time;
115 • Relatively low analytical costs; and
116 • Detecting the presence of a wide range of radionuclides in a variety of media.
117 There are also limitations on the use of these analyses. These limitations include:
118 • No specific identification for pure alpha- or pure beta-emitting radionuclides and low-energy,
119 gamma-emitting radionuclides are generally not identified; and
120 • Failing to identify the presence of several radionuclides (e.g., 3H and other volatile
121 radionuclides; 55Fe and other radionuclides that decay by electron capture).
122 OUTPUT: An initial list of radionuclides of potential concern including a brief narrative explain-
123 ing why each radionuclide is on the list as well as an explanation of why certain radionuclides
124 were considered but not listed. This list may be modified as more project-specific information
125 becomes available. It is better to include radionuclides on the initial list even if the probability
126 that they significantly contribute to the addressed concerns is small. The consequence of
127 discovering an additional radionuclide of concern late in a project generally outweighs the effort
128 of evaluating its potential during planning.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 3-5 DO NOT CITE OR QUOTE
-------
Key Analytical Planning Issues...
129 3.3.2 Identify Concentration Ranges
130 Once the radionuclides of concern have been identified, the expected concentration range for
131 each radionuclide should be determined. Historical data, process knowledge, and previous
132 studies, if available, can be used to determine the expected concentration range for each analyte.
133 While most analytical protocols are applicable over a fairly large concentration range for the
134 radionuclide of concern, performance over a required concentration range can serve as an MQO
135 for the protocol selection process and some analytical protocols may be eliminated if they cannot
136 accommodate the expected concentration range. In addition, the expected concentration ranges of
137 all of the radionuclides of concern can provide useful information about possible chemical and
138 spectral interferences. For example, while an analytical protocol for a particular radionuclide may
139 be able to accommodate the expected concentration range for that radionuclide, the concentra-
140 tions of other radionuclides may present interference problems, thus eliminating the use of that
141 analytical protocol.
142 OUTPUT: The expected concentration range for each radionuclide of concern as well as the
143 expected concentration range of any potential chemical or radiological interference.
144 3.3.3 Identify and Characterize Matrices of Concern
145 During a directed project planning process, the matrices of concern should be clearly identified.
146 For many projects, typical matrices may include surface soil, subsurface soil, sediment, surface
147 water, groundwater, drinking water, air particulates, biota, structural materials, metals, etc.
148 Historical data, process knowledge, previous studies, conceptual site models, transport models,
149 and other such sources generally are used to identify matrices of concern. It is critical to be as
150 specific as possible when identifying a matrix.
151 From an analytical perspective, information on the chemical and physical characteristics of a
152 matrix is extremely useful. Therefore, in addition to identifying the matrices of concern, every
153 effort should be made to obtain any information available on the chemical and physical charac-
154 teristics of the matrices. This information is particularly important when determining the required
155 specificity of the analytical protocol, i.e., the ability to accommodate possible interferences. It is
156 also important to identify any possible hazards associated with the matrix, such as the presence
157 of explosive or other highly reactive chemicals. Issues related to specific matrices, such as filtra-
158 tion of water samples and removal of foreign material, are discussed in more detail in Section 3.5
159 and in Section 6.5.1.1 of Chapter 6.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 3 -6 DRAFT FOR PUBLIC COMMENT
-------
Key Analytical Planning Issues...
160 OUTPUT: A list of the matrices of concern along with any information on the chemical and
161 physical characteristics of the matrices and any information on possible hazards associated with
162 them. As previously noted, the list of matrices of concern and the analyte list often are developed
163 concurrently. In some cases, one analyst list is applicable to all the matrices of concern, and in
164 other cases there are variations in the analyte lists for each matrix.
165 3.3.4 Determine Relationships Between the Radionuclides of Concern
166 Known or expected relationships among radionuclides can be used to establish "alternative"
167 radionuclides that may be easier and less costly to measure. In most cases, an "easy-to-measure"
168 radionuclide is analyzed, and the result of this analysis is used to estimate the concentration of
169 one or more radionuclides that may be difficult to measure or costly to analyze.
170 One of the best known and easiest relationships to establish is between a parent radionuclide and
171 its associated progeny. Once equilibrium conditions have been established, the concentration of
172 any member of the decay series can be used to estimate the concentration of any other member of
173 the series. For example, the thorium decay series contains 12 radionuclides. If each radionuclide
174 in this series is analyzed separately, the analytical costs can be very high. However, if equilib-
175 rium conditions for the decay series have been established, a single analysis using gamma spec-
176 trometry may be adequate for quantifying all of the radionuclides in the series simultaneously.
177 Similarly, process knowledge can be used to predict relationships between radionuclides. For
178 example, in a nuclear power reactor, steel may become irradiated, producing radioactive isotopes
179 of the elements present in the steel. These isotopes often include 60Co, 63Ni, and 55Fe. 60Co decays
180 by emission of a beta particle and two high-energy gamma rays, which are easily measured using
181 gamma spectrometry. 63Ni also decays by emission of a beta particle but has no associated
182 gamma rays. 55Fe decays by electron capture and has several associated X-rays with very low
183 energies. Laboratory analysis of 63Ni and 55Fe typically is time-consuming and expensive.
184 However, since all three radionuclides are produced by the same mechanism from the same
185 source material, there is an expected relationship at a given time in their production cycle. Once
186 the relationship between these radionuclides has been established, the 60Co concentration can be
187 used to estimate the concentration of 63Ni and 55Fe.
188 The uncertainty in the concentration ratio between radionuclide concentrations used in the alter-
189 nate analyte approach should be included as part of the combined standard uncertainty of the
190 analytical protocol in the measurement process. Propagation of uncertainties is discussed in
191 Chapter 19.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 3 -7 DO NOT CITE OR QUOTE
-------
Key Analytical Planning Issues...
192 OUTPUT: A list of known radionuclide relationships (e.g., those based on parent-progeny rela-
193 tionships or previous study results) and a list of potential radionuclide relationships (i.e., based
194 on process knowledge). A preliminary study to determine the project-specific radionuclide
195 relationships may be necessary, and additional measurements may be required to confirm the
196 relationship used during the project. This information may be used to develop a revised analyte
197 list.
198 3.3.5 Determine Available Project Resources and Deadlines
199 The available project resources can have a significant impact on the selection or development of
200 analytical protocols, as well as the number and type of samples to be analyzed. In addition,
201 project deadlines, and, in particular, required analytical turnaround times (see Section 6.5.3), can
202 be important factors in the selection and development of analytical protocols for a particular
203 project. During a directed planning process, radioanalytical specialists can provide valuable
204 information on typical costs and turnaround times for various types of laboratory analyses.
205 OUTPUT: A statement of the required analytical turnaround times for the radionuclides of concern
206 and the anticipated budget for the laboratory analysis of the samples.
207 3.3.6 Refine Analyte List and Matrix List
208 As additional information about a project is collected, radionuclides may be added to or removed
209 from the analyte list. There may be one analyte list for all matrices or separate lists for each
210 matrix. Developing an analyte list is an iterative process, however. The list should become more
211 specific during the project planning process.
212 Radionuclides might be added to the analyte list when subsequent investigations indicate that
213 additional radionuclides were involved in a specific project. In some cases, radionuclides may be
214 removed from the analyte list. When the initial analyte list is compiled, there may be significant
215 uncertainty associated with the presence of specific radionuclides. These radionuclides may be
216 included on the analyte list to be conservative, even when there is only a small probability they
217 may be present. Subsequent investigations may determine if specific radionuclides are actually
218 present and need to be considered as part of the project. For example, a research laboratory was
219 licensed for a specific level of activity from all radionuclides with atomic numbers between 2 and
220 87. Even limiting the analyte list to radionuclides with a half-life greater than six months
221 provides several dozen radionuclides. A study may be designed to identify the actual
222 radionuclides of concern through the use of historical records and limited analyses to justify
223 removing radionuclides from the analyte list.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 3 -8 DRAFT FOR PUBLIC COMMENT
-------
Key Analytical Planning Issues...
224 OUTPUT: A revised analyte list. Radionuclides can always be added to or removed from the
225 analyte list, but justification for adding or removing radionuclides should be included in the
226 project documentation.
227 3.3.7 Method Performance Characteristics and Measurement Quality Objectives
228 The output of a directed planning process includes DQOs for a project. DQOs apply to all data
229 collection activities associated with a project, including sampling and analysis. In particular,
230 DQOs for data collection activities describe the overall level of uncertainty that a decisionmaker
231 is willing to accept for project results. This overall level of uncertainty is made up of
232 uncertainties from sampling and analysis activities.
233 Since DQOs apply to both sampling and analysis activities, what are needed from an analytical
234 perspective are performance objectives specifically for the analytical process of a particular
235 project. MARLAP refers to these performance objectives as measurement quality objectives. The
236 MQOs can be viewed as the analytical portion of the overall project DQOs. In a performance-
237 based approach, the MQOs are used initially for the selection and evaluation of analytical
238 protocols and are subsequently used for the ongoing and final evaluation of the analytical data.
239 In MARLAP, the development of MQOs for a project depends on the selection of an action level
240 and gray region for each analyte during the directed planning process. The term "action level" is
241 used to denote the numerical value that will cause the decisionmaker to choose one of the
242 alternative actions. The "gray region" is a set of concentrations close to the action level, where
243 the project planning team is willing to tolerate a high decision error rate (see Chapter 2 and
244 Appendices B and C for a more detailed discussion of action levels and gray region). MARLAP
245 recommends that an action level and gray region be established for each analyte during the
246 directed planning process.
247 MARLAP provides guidance on developing MQOs for select method performance characteristics
248 such as:
249 • The method uncertainty at a specified concentration (expressed as an estimated standard
250 deviation);
251 • The method's detection capability (expressed as the minimum detectable concentration, or
252 MDC);
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 3 -9 DO NOT CITE OR QUOTE
-------
Key Analytical Planning Issues...
253 • The method's quantification capability (expressed as the minimum quantifiable
254 concentration, or MQC);
255 • The method's range, which defines the method's ability to measure the analyte of concern
256 over some specified range of concentration;
257 • The method's specificity, which refers to the ability of the method to measure the analyte of
258 concern in the presence of interferences; and
259 • The method's ruggedness, which refers to the relative stability of method performance for
260 small variations in method parameter values.
261 An MQO is a statement of a performance objective or requirement for a particular method per-
262 formance characteristic. An example MQO for the method uncertainty at a specified concentra-
263 tion, such as the action level, would be: "A method uncertainty of 0.01 Bq/g or less is required at
264 the action level of 0.1 Bq/g." A qualitative example of an MQO for method specificity would be
265 "The method must be able to quantify the amount of 226Ra present, given elevated levels of 235U
266 in the samples." MQOs may be quantitative or qualitative in nature.
267 The list provided in this section is not intended to be an exhaustive list of method performance
268 characteristics, and for a particular project, other method performance characteristics may be
269 important and should be addressed during the project planning process. In addition, one or more
270 of the method performance characteristics listed may not be important for a particular project.
271 From an analytical perspective, a key activity during project planning is the identification of
272 important method performance characteristics and the development of MQOs for the method
273 performance characteristics.
274 In addition to developing MQOs for method performance characteristics, MQOs may be estab-
275 lished for other parameters, such as data quality indicators (DQIs). DQIs are qualitative and
276 quantitative descriptors used in interpreting the degree of acceptability or utility of data. The
277 principal DQIs are precision, bias, representativeness, comparability, and completeness. These
278 five DQIs are also referred to by the acronym PARCC; the "A" stands for accuracy instead of
279 bias, although both indicators are included in discussions of the PARCC parameters (EPA,
280 1998). Since the distinction between imprecision and bias depends on context, and since a
281 reliable estimate of bias requires a data set that includes many measurements, MARLAP focuses
282 on developing an MQO for method uncertainty. Method uncertainty effectively combines
283 imprecision and bias into a single parameter whose interpretation does not depend on context.
284 This approach assumes that all potential sources of bias present in the analytical process have
MARLAP JULY 2001
DO NOT CITE OR QUOTE 3-10 DRAFT FOR PUBLIC COMMENT
-------
Key Analytical Planning Issues...
285 been considered in the estimation of the measurement uncertainty and, if not, that any appre-
286 ciable bias would only be detected after a number of measurements of QC and performance
287 evaluation samples have been performed. MARLAP provides guidance on the detection of bias,
288 for example, during analytical protocol validation and evaluation (Chapters 6 and 7). However,
289 the most likely time to detect, and possibly correct, an unanticipated bias is during data quality
290 assessment (see Chapter 9).
291 While MARLAP does not provide specific guidance on developing MQOs for the DQIs, estab-
292 lishing MQOs for the DQIs may be important for some projects. EPA Guidance for Quality
293 Assurance Project Plans (EPA, 1998) contains more information on DQIs. MARLAP provides
294 guidance on developing MQOs for method performance characteristics in the next section.
295 3.3.7.1 Develop MQOs for Select Method Performance Characteristics
296 Once the important method performance characteristics for an analytical process have been iden-
297 tified, the next step is to develop MQOs for them. This section provides guidance on developing
298 MQOs for the method performance characteristics listed in the previous section. As noted, other
299 method performance characteristics may be important for a particular analytical process, and
300 MQOs should be developed for them during project planning.
301 METHOD UNCERTAINTY
302 While measurement uncertainty is a parameter associated with an individual result and is calcu-
303 lated after a measurement is performed, MARLAP uses the term "method uncertainty" to refer to
304 the predicted uncertainty of a measured value that would likely result from the analysis of a
305 sample at a specified analyte concentration. Method uncertainty is a method performance charac-
306 teristic much like the detection capability of a method. Reasonable values for both characteristics
307 can be predicted for a particular method based on typical values for certain parameters and on
308 information and assumptions about the samples to be analyzed. These predicted values can be
309 used in the method selection process to identify the most appropriate method based on a project's
310 data requirements. Because of its importance in the selection and evaluation of analytical proto-
311 cols and its importance in the evaluation of analytical data, MARLAP recommends that the
312 method uncertainty at a specified concentration (typically the action level) always be identified
313 as an important method performance characteristic, and that an MQO be established for it for
314 each analyte.
315 The MQO for the method uncertainty at a specified concentration plays a key role in MARLAP's
316 performance-based approach. It effectively links the three phases of the data life cycle: planning,
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 3-11 DO NOT CITE OR QUOTE
-------
Key Analytical Planning Issues...
317 implementation, and assessment. This MQO, developed during the planning phase, is used
318 initially in the selection and validation of an analytical method for a project (Chapter 6). This
319 MQO provides criteria for the evaluation of QC samples during the implementation phase
320 (Appendix C and Chapter 7). It also provides criteria for verification and validation during the
321 assessment phase (Chapter 8). The use of the project-specific MQOs for the method uncertainty
322 of each analyte in the three phases of the life of a project, as opposed to arbitrary non-project-
323 specific criteria, helps to ensure the generation of radioanalytical data of known quality
324 appropriate for its intended use.
325 The MQO for method uncertainty for an analyte at a specified concentration, normally the action
326 level, is related to the width of the gray region. The gray region has an upper bound and a lower
327 bound. The upper bound typically is the action level. The width of the gray region is represented
328 by the symbol A. See Appendix B for information on setting up a gray region.
329 Appendix C provides the rationale and detailed guidance on the development of MQOs for
330 method uncertainty. Outlined below is MARLAP's recommended guideline for developing
331 MQOs for method uncertainty when a decision is to be made about the mean of a population
332 represented by multiple samples. Appendix C provides additional guidelines for developing
333 MQOs for method uncertainty when decisions are to be made about individual items or samples.
334 If decisions are to be made about the mean of a sampled population, MARLAP recommends that
335 the method uncertainty (HMR) be less than or equal to the width of the gray region divided by 10
336 for sample concentrations at the upper bound of the gray region (typically the action level). If this
337 requirement cannot be met, the project planners should require at least that the method
338 uncertainty be less than or equal to the width of the gray region divided by 3 (Appendix C).
339
340
341
342
343
344
345
EXAMPLE
Suppose the action level is 0.1 Bq/g and the lower bound of the gray region is 0.02 Bq/g. If
decisions are to be made about survey units based on samples, then the required method uncer-
tainty (HMR) at 0.1 Bq/g is
_A_ _ 0.1 -0.02
10 10
= 0.008 Bq/g
If this uncertainty cannot be achieved, then a method uncertainty (HMR) as large as A / 3 =
0.027 Bq/g may be allowed if more samples are taken.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 3-12 DRAFT FOR PUBLIC COMMENT
-------
Key Analytical Planning Issues...
346 In the example above, the required method uncertainty (HMR) is 0.008 Bq/g. In terms of method
347 selection, this particular MQO calls for a method that can ordinarily produce measured results
348 with expected combined standard uncertainties (la) of 0.008 Bq/g or less at sample concentra-
349 tions at the action level (0.1 Bq/g in this example). Although individual measurement uncertain-
350 ties will vary from one measured result to another, the required method uncertainty is effectively
351 a target value for the individual measurement uncertainties.
352 OUTPUT: MQOs expressed as the required method uncertainty at a specified concentration for
353 each analyte.
354 DETECTION AND QUANTIFICATION CAPABILITY
355 For a particular project, the detection capability or the quantification capability may be identified
356 as an important method performance characteristic during project planning. If the issue is
357 whether an analyte is present in an individual sample and it is therefore important that the
358 method be able to reliably distinguish small amounts of the analyte from zero, then an MQO for
359 the detection capability should be established during project planning. If the emphasis is on being
360 able to make precise measurements of the analyte concentration for comparing the mean of a
361 sampled population to the action level, then an MQO for the quantification capability should be
362 established during project planning.
363 Detection Capability
364 When decisions are to be made about individual items or samples (e.g., drinking water samples),
365 and the lower bound of the gray region is at or near zero for the analyte of concern, the detection
366 capability of the method is an important method performance characteristic, and an MQO should
367 be developed for it. MARLAP recommends that the MQO for the detection capability be
368 expressed as a required MDC (Chapter 19).
369 Outlined below is MARLAP's recommended guideline for developing MQOs for detection
370 capability. Appendix C provides the rationale along with detailed guidance on the development
371 of MQOs for detection capability.
372 If the lower bound of the gray region is at or near zero and decisions are to be made about
373 individual items or specimens, choose an analytical method whose minimum detectable
374 concentration is no greater than the upper bound of the gray region.l
The MDC is defined as the analyte concentration at which the probability of detection is 1 - p.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 3-13 DO NOT CITE OR QUOTE
-------
Key Analytical Planning Issues...
375 Quantification Capability
376 When decisions are to be made about a sampled population and the lower bound of the gray
377 region is at or near zero for the analyte of concern, the quantification capability of the method is
378 an important method performance characteristic and an MQO should be developed for it.
379 MARLAP recommends that the MQO for the quantification capability be expressed as a required
380 MQC (see Chapter 19).
381 Outlined below is MARLAP's recommended guideline for developing MQOs for quantification
382 capability. The MQC, as used in the guideline, is defined as the analyte concentration at which
383 the relative standard uncertainty is 10 percent (see Chapter 19). Appendix C provides the ration-
384 ale along with detailed guidance on the development of MQOs for quantification capability.
385 // the lower bound of the gray region is at or near zero and decisions are to be made about a
386 sampled population, choose an analytical method whose minimum quantifiable concentration is
387 no greater than the upper bound of the gray region which is typically the action level.
388 If an MQO for method uncertainty has been established, then establishing an MQO for the
389 quantification capability in terms of a required MQC is somewhat redundant since an MQC is
390 defined in terms of a specified relative standard uncertainty. However, this method performance
391 characteristic is included in MARLAP for several reasons. First, it has been included to empha-
392 size the importance of the quantification capability of a method for those instances where the
393 issue is not whether an analyte is present or not—for example measuring 238U in soil where the
394 presence of the analyte is given—but rather how precisely the analyte can be measured. Second,
395 this method performance characteristic has been included so as to promote the MQC as an
396 important method parameter. And last, it has been included as an alternative to the overemphasis
397 on establishing required detection limits in those instances where detection (reliably distinguish-
398 ing an analyte concentration from zero) is not the key analytical question.
399 OUTPUT: If the lower bound of the gray region is at or near zero, and decisions are to be made
400 about a sample population, MQOs expressed as MQCs should be developed for each analyte. If
401 the lower bound of the gray region is zero and decisions are to be made about individual items or
402 specimens, MQOs expressed as MDCs should be developed for each analyte.
403 RANGE
404 Depending on the expected concentration range for an analyte (Section 3.3.2), the method's
405 range may be an important method performance characteristic. Most radioanalytical methods are
MARLAP JULY 2001
DO NOT CITE OR QUOTE 3-14 DRAFT FOR PUBLIC COMMENT
-------
Key Analytical Planning Issues..
406 capable of performing over a fairly large range of activity concentrations. However, if the
407 expected concentration range is large for an analyte, the method's range should be identified as
408 an important method performance characteristic and an MQO should be developed for it. The
409 radioanalytical specialist on the project planning team will determine when the expected concen-
410 tration range of an analyte warrants the development of an MQO for the method's range. Since
411 the expected concentration range for an analyte is based on past data which may or may not be
412 accurate, the MQO for the method's range should require that the method perform over a larger
413 concentration range than the expected range. This will help prevent the selection of methods
414 which cannot accommodate the actual concentration range of the analyte.
415 OUTPUT: MQOs for the method's concentration range for each analyte.
416 SPECIFICITY
417 Depending on the chemical and physical characteristics of the matrices, as well as the concen-
418 trations of analytes and the concentrations of other chemical constituents, the method's speci-
419 ficity may be an important method performance characteristic for an analytical process. Method
420 specificity refers to the ability of the method to measure the analyte of concern in the presence of
421 interferences. In order to determine if method specificity is an important method performance
422 characteristic, the radioanalytical specialist on the project planning team will need information on
423 expected concentration ranges of the analytes of concern and other chemical constituents in the
424 samples (Section 3.3.2), along with information on the chemical and physical characteristics of
425 the matrices (Section 3.3.3). If it is determined that method specificity is an important method
426 performance characteristic, then an MQO should be developed for it. The MQO can be qualita-
427 tive or quantitative in nature.
428 OUTPUT: MQOs for the method specificity for those analytes likely affected by interferences.
429 RUGGEDNESS
430 For a project which involves analyzing samples which are complex in terms of their chemical
431 and physical characteristics, the method's ruggedness may be an important method performance
432 characteristic. Method ruggedness refers to the relative stability of the method's performance
433 when small variations in method parameter values are made, such as a change in pH, a change in
434 amount of reagents used, etc. In order to determine if method ruggedness is an important method
435 performance characteristic, the radioanalytical specialist on the planning team needs detailed
436 information on the chemical and physical characteristics of the samples. If it is determined that
437 method ruggedness is an important method performance characteristic, then an MQO should be
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 3-15 DO NOT CITE OR QUOTE
-------
Key Analytical Planning Issues...
438 developed for it. The MQO may require performance data which demonstrates the method's
439 ruggedness for specified changes in select method parameters. The statistical manual of the
440 Association of Official Analytical Chemists (AOAC) and the Standard Guide for Conducting
441 Ruggedness Tests ASTM El 169 provides guidance on ruggedness testing.
442 OUTPUT: MQOs for method ruggedness for specified changes in select method parameters.
443 3.3.7.2 The Role of MQOs in the Protocol Selection and Evaluation Process
444 Once developed, the MQOs become an important part of the project's APSs and are subsequently
445 incorporated into project plan documents (Chapter 4) and into the analytical Statement of Work
446 (Chapter 5). In MARLAP, MQOs are used initially in the selection, validation, and evaluation of
447 analytical protocols (Chapters 6 and 7). In a performance-based approach, analytical protocols
448 are either accepted or rejected largely on their ability or inability to meet the project MQOs.
449 3.3.7.3 The Role of MQOs in the Project's Data Evaluation Process
450 Once the analytical protocols have been selected and implemented, the MQOs and—in
451 particular—the MQOs for method uncertainty, are used in the evaluation of the resulting
452 laboratory data relative to the project's analytical requirements. The most important MQO for
453 data evaluation is the one for method uncertainty at a specified concentration. It is expressed as
454 the required method uncertainty (HMR) at some concentration, normally the action level (for this
455 discussion, it is assumed that the action level is the upper bound of the gray region). When the
456 analyte concentration of a laboratory sample is less than the action level, the combined standard
457 uncertainty of the measured result should not exceed the required method uncertainty.
458 For example, if the required method uncertainty is 0.01 Bq/g or less at an action level of 0.1
459 Bq/g, then for any measured result less than 0.1 Bq/g, the laboratory's reported combined
460 standard uncertainty should be less than or equal to 0.01 Bq/g. When the concentration is greater
461 than the action level, the combined standard uncertainty of the measured result should not exceed
462 the relative value of the required method uncertainty. If the required method standard uncertainty
463 is 0.01 Bq/g or less at an action level of 0.1 Bq/g (10 percent of the action level), then for any
464 measured result greater than 0.1 Bq/g, the laboratory's reported combined standard uncertainty
465 should be no greater than 10 percent of the measured result. If an expanded uncertainty is
466 reported with each measured value, and the coverage factor is also specified, the combined
467 standard uncertainty may be calculated and checked against the required value. The check
468 described relies on the laboratory's estimate of its measurement uncertainty. Additional checks
469 are needed to ensure that the uncertainties are not seriously underestimated.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 3-16 DRAFT FOR PUBLIC COMMENT
-------
Key Analytical Planning Issues..
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
Appendix C provides guidance on developing criteria for QC samples based on the MQO for
method uncertainty. Specifically, Appendix C contains equations for determining warning and
control limits for QC sample results based on the project's MQO for method uncertainty.
The following example illustrates the use of the MQO for method uncertainty in evaluating QC
sample results. Chapter 8, Data Verification and Validation, provides guidance on developing
validation criteria based on the MQO for the required method uncertainty.
EXAMPLE
Suppose the upper bound of the gray region (the action level) is 0.1 Bq/g, and the required
method uncertainty (HMR) at this concentration is 0.01 Bq/g, or 10 percent. A routine laboratory
control sample (LCS) is prepared with an analyte concentration of 0.150 Bq/g. (For the
purpose of this example the uncertainty in the spike concentration is assumed to be negligible.)
The lab analyzes the LCS with a batch of samples and obtains the measured result 0.140 ±
0.008 Bq/g, where 0.008 Bq/g is the combined standard uncertainty (la).
Question: Is this LCS result acceptable?
Answer: The LCS result may be acceptable if it differs from the accepted true value by no
more than three times the required method uncertainty at that concentration. In this example
the required method uncertainty is 10 percent at 1.50 Bq/g. So, the LCS result is required to be
within 30 percent of 1.50 Bq/g, or in the range 0.105-0.195 Bq/g. Since 0.140 Bq/g is clearly
in the acceptance range, the data user considers the result acceptable. Note also that the
laboratory's reported combined standard uncertainty is less than the required method
uncertainty, as expected.
3.3.8 Determine Any Limitations on Analysis Options
With the outputs of the resolution of a number of key analytical planning issues, such as a refined
analyte list, MQOs for the analyte list, known relationships between radionuclides of concern, a
list of possible alternate analytes, required analytical turnaround times, the analytical budget, etc.,
the project planning team may choose to determine the analyses to be performed for the project
and thereby limit the analysis options available to the laboratory. It should be emphasized that
determining which analyses need to be performed is not the same as indicating that a particular
analytical protocol or analytical method has to be used. With the exception of gross alpha and
beta measurements and gamma spectrometry, MARLAP uses the term "analysis " to refer to a
JULY 2001
DRAFT FOR PUBLIC COMMENT
3-17
MARLAP
DO NOT CITE OR QUOTE
-------
Key Analytical Planning Issues...
500 radionuclide/matrix combination. Examples of analyses to be performed include 3H in water, 90Sr
501 in milk, 238Pu in soil, etc. Although determining the analyses to be performed during the planning
502 process may seem inconsistent with a performance-based approach, the project planning team
503 may determine the analyses to be performed or may decide to eliminate some analyses from
504 consideration. This decision may be based on information obtained during project planning, such
505 as the absence of equilibrium between the analyte and other radionuclides in its decay chain or
506 the presence of other radionuclides known to cause spectral interferences. However, in the
507 absence of such considerations, the project planning team should allow the laboratory the flexi-
508 bility of selecting the analyses which meet the analytical requirements as contained in the Ana-
509 lytical Protocol Specifications.
510 The role of the radioanalytical specialist is critical in determining if any limitations on analytical
511 options are necessary because of the many laboratory-related issues and factors involved. For
512 example, if several of the radionuclides of concern on the target analyte list are gamma-emitters,
513 the radioanalytical specialist can determine if gamma spectrometry is an appropriate analysis
514 given the required MQOs, matrices of concern, possible spectral interferences, etc. The radio-
515 analytical specialist may determine that not only is gamma spectrometry an appropriate analysis
516 for the gamma-emitting radionuclides of concern, but since there is evidence that equilibrium
517 conditions are present, the results for gamma spectrometry can be used for other radionuclides of
518 concern in the same decay chain as the gamma-emitting radionuclides. In other instances, such as
519 the use of gamma spectrometry to quantify 226Ra in the presence of elevated levels of 235U, the
520 radioanalytical specialist may determine that gamma spectrometry is not an appropriate analysis
521 due to possible spectral interferences. The following sections provide a brief overview of some
522 analysis procedures.
523 3.3.8.1 Gamma Spectrometry
524 In general, gamma spectrometry has many advantages over other choices. It is capable of
525 identifying and quantifying a large number of radionuclides. In comparison with other analyses, it
526 offers a fairly quick turnaround time and, since limited sample manipulation is involved, it is
527 relatively inexpensive, particularly compared to analyses which require sample dissolution and
528 chemical separations. It also allows for the use of relatively large sample sizes, thereby reducing
529 the measurement uncertainty associated with subsampling at the laboratory. However, given its
530 many advantages, gamma spectrometry cannot be used to analyze for all radionuclides. For
531 example, gamma spectrometry may not be able to achieve the project's MQOs, since some or all
532 of the radionuclides of concern may not be gamma-emitters, interfering radionuclides may
533 present problems, etc. The radioanalytical specialist on the planning team can evaluate the
MARLAP JULY 2001
DO NOT CITE OR QUOTE 3-18 DRAFT FOR PUBLIC COMMENT
-------
Key Analytical Planning Issues...
534 appropriateness of the use of gamma spectrometry for some or all of the radionuclides on the
535 analyte list or for alternate analytes.
536 3.3.8.2 Gross Alpha and Beta Analysis
537 Gross alpha and beta analysis provides information on the overall level of alpha- and beta-
538 emitting radionuclides present in a sample. The analysis has the advantage of a relatively quick
539 turnaround time and generally is inexpensive compared to other analyses. The analysis also has
540 significant limitations. It does not identify specific alpha- and beta-emitting radionuclides, so the
541 source of the overall alpha and beta radiation is not determined by the analysis. It does not detect
542 contribution from low-energy beta-emitting radionuclides such as 3H. The measurement uncer-
543 tainty of the analysis, particularly for matrices other than water, tends to be larger than the meas-
544 urement uncertainty of other analyses. However, even with these limitations, gross alpha and beta
545 analysis can be an important and appropriate analysis for a project.
546 3.3.8.3 Radiochemical Nuclide-Specific Analysis
547 In many instances, due to the project's MQOs, the lack of an appropriate alternate analyte, the
548 lack of equilibrium conditions, etc., radiochemical nuclide-specific analyses are required. This is
549 often true when radionuclides such as 3H, 14C, 90Sr, isotopes of Pu, "Tc, etc., are on the analyte
550 list. These analyses generally involve more manipulation of the samples than do gamma spec-
551 trometry and gross alpha and beta analysis. These analyses often require sample dissolution and
552 chemical separation of the radionuclides of concern. For liquid scintillation counting, distillation
553 is usually required for water samples, and some oxidative/combustion procedure is usually
554 required for solid samples. Because of this, these analyses generally have longer turnaround
555 times and are more expensive than other analyses.
556 Given the many analytical factors and considerations involved, the role of the radioanalytical
557 specialist is critical to determining if any limitations on analysis options are necessary.
558 OUTPUT: Any limitations on analysis options, if appropriate.
559 3.3.9 Determine Method Availability
560 After the required analyses have been determined along with the sample matrices, the required
561 MQOs, the analytical turnaround times, etc., the radioanalytical specialist should be able to
562 determine if there are analytical methods currently available to meet the project's requirements.
563 There are a number of sources of radioanalytical methods, including those published by the
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 3-19 DO NOT CITE OR QUOTE
-------
Key Analytical Planning Issues...
564 American Society of Testing and Materials (ASTM), Standard Methods for the Examination of
565 Water and Waste Water (APHA/AWWA, 1992), methods published in scientific journals,
566 methods published in laboratory procedure manuals, and those published by Federal and State
567 agencies.
568 If there are no known analytical methods that would meet the project's analytical requirements,
569 the project planning team must evaluate options. They may decide to reevaluate the analytical
570 data requirements, such as the MQOs, to see if they can be changed to allow the use of existing
571 methods or increase the analytical budget and project timeline to allow for method development.
572 OUTPUT: A statement of method availability.
573 3.3.10 Determine the Type and Frequency of, and Evaluation Criteria for, Quality Control
574 Samples
575 There are three main types of laboratory QC samples—blanks, replicates, and spikes. In addition,
576 there are different types of blanks, replicates, and spikes. For example, spikes can be matrix
577 spikes, laboratory control samples, external performance evaluation samples, etc. Chapter 18
578 contains a detailed discussion of the different types of QC samples and the information they pro-
579 vide. Since the results of the three main types of QC samples often are used to evaluate different
580 aspects of the analytical process, most projects should employ all three types as part of the QC
581 process.
582 The frequency of laboratory QC sampling for a project essentially represents a compromise
583 between the need to evaluate and control the analytical process and the resources available. In
584 addition, the nature of the project and the intended use of the data will play a role in determining
585 the frequency of QC samples required. For example, the frequency of QC samples for a project
586 involving newly developed methods for analytes in a complex matrix normally should be greater
587 than the frequency of QC samples for a project using more established methods on a simpler
588 matrix, assuming the intended use of the data is the same for both projects. The radioanalytical
589 specialists on the project planning team play a key role in determining the type and frequency of
590 QC samples for a project.
591 In order to adequately evaluate laboratory data, it is important that the QC samples be clearly
592 linked to a group of project samples. Typically, this is done by analyzing QC samples along with
593 a batch of samples and reporting the results together.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 3-20 DRAFT FOR PUBLIC COMMENT
-------
Key Analytical Planning Issues..
594 In addition to determining the type and frequency of QC samples, evaluation criteria for the QC
595 sample results should be developed during the directed planning process and incorporated into
596 the project's APSs. Appendix C provides guidance on developing criteria for QC samples and
597 contains equations that calculate warning and control limits for QC sample results based on the
598 proj ecf s MQO for method uncertainty.
599 OUTPUT: List of type and frequency of QC samples required and the criteria for evaluating QC
600 sample results.
601 3.3.11 Determine Sample Tracking and Custody Requirements
602 A procedural method for sample tracking should be in place for all projects so that the proper
603 location and identification of samples is maintained throughout the life of the project. Sample
604 tracking should cover the entire process from sample collection to sample disposal. For some
605 projects, a Chain-of-custody (COC) process is needed. COC procedures are particularly
606 important in demonstrating sample control when litigation is involved. In many cases, Federal,
607 State, or local agencies may require that COC be maintained for specific samples. Chapter 10,
608 Field and Sampling Issues that affect Laboratory Measurements, provides guidance on sample
609 tracking and COC. It is important that the requirements for sample tracking be clearly established
610 during project planning.
611 OUTPUT: Project sample tracking requirements.
612 3.3.12 Determine Data Reporting Requirements
613 The data reporting requirements should be established during project planning. This involves
614 determining not only what is to be reported but also how it is to be reported. Items that are
615 routinely reported are listed below. It should be noted that this is not a comprehensive list, and
616 some projects may require the reporting of more items while other projects may require the
617 reporting of fewer items:
618 • Field sample identification number
619 • Laboratory sample identification number
620 • Sample receipt date
621 • Analysis date
622 • Radionuclide
623 • Radionuclide concentration units
624 • Sample size (volume, mass)
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 3-21 DO NOT CITE OR QUOTE
-------
Key Analytical Planning Issues...
625 • Aliquant size (volume, mass)
626 • Radionuclide concentration at specified date
627 • Combined standard uncertainty or expanded uncertainty (coverage factor should be indicated)
628 • Sample-specific minimum detectable concentration
629 • Analysis batch identification
630 • Quality control sample results
631 • Laboratory instrument identification
632 • Specific analytical parameters (e.g., chemical yields, counting times, etc.)
633 • Analytical method/procedure reference
634 It is important that the required units for reporting specific items be determined during project
635 planning. MARLAP recommends that units of the International System of Units (SI) be used
636 whenever possible. However, since regulatory compliance levels are usually quoted in traditional
637 radiation units, it may be appropriate to report in both SI and traditional units, with one being
638 placed in parenthesis. MARLAP also recommends that all measurement results be reported
639 directly as obtained, including negative values, along with the measurement uncertainty—for
640 example 2a, 3a, etc. Additional guidance on data reporting, including a discussion of electronic
641 data deliverables, is provided in Chapter 17, Data Acquisition, Reduction, and Reporting, and in
642 Chapters, Obtaining Laboratory Services.
643 OUTPUT: Data reporting requirements for a project.
644 3.4 Matrix-Specific Analytical Planning Issues
645 This section discusses a number of matrix-specific analytical planning issues common to many
646 types of projects. For each matrix there is a discussion of several potential key analytical plan-
647 ning issues specific to that matrix. It should be noted that what may be a key analytical planning
648 issue for one project, may not be a key issue for another project. The list of potential matrix-
649 specific key analytical planning issues discussed in this section is summarized in Table 3.1.
650 Table 3.1 is not a comprehensive list, but rather is an overview of some common matrix-specific
651 planning issues.
652 This section is divided into solids, liquids, filters and wipes. While filters and wipes are solids,
653 they are discussed separately because of the unique concerns associated with them.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 3 -22 DRAFT FOR PUBLIC COMMENT
-------
Key Analytical Planning Issues..
654
655
656
657
658
TABLE 3.1 — Matrix-specific analytical planning issues
659
660
661
662
663
MATRIX
Solids (soil, sediment,
structural material,
biota, metal, etc.)
Liquids (drinking water,
groundwater,
precipitation, solvents,
oils, etc.)
Filters and Wipes
RECOMMENDED KEY ISSUES
Homogenization
Subsampling
Removal of unwanted material
Is filtering required?
Sample preservation
Should sample be filtered or preserved
first?
Filter material
Pore size
Sample volume or area wiped
POTENTIAL KEY ISSUES
Container type
Container material
Sample preservation
Screening samples for health and safety
Volatile compounds
Sample identification
Cross-contamination
Sample size
Compliance with radioactive materials license
Compliance with shipping regulations
Chemical and physical form of the substrate
Sample identification
Volume of sample
Immiscible layers
Precipitation
Total dissolved solids
Reagent background
Compliance with radioactive materials license
Compliance with shipping regulations
Sample identification
Compliance with radioactive materials license
Compliance with shipping regulations
Subsampling
Background from filter material
664
3.4.1 Solids
665 Solid samples consist of a wide variety of materials that include soil and sediment; plant and
666 animal tissue; concrete; asphalt; trash, etc. In general, most solid samples do not require preser-
667 vation (Chapter 10) but do require specific processing both in the field and in the laboratory. In
668 certain instances, some biota samples may require preservation, primarily in the form of lowered
669 temperatures, to prevent sample degradation and loss of water. Some common analytical
670 planning issues for solid samples include homogenization and Subsampling (Section 3.4.1.1) and
671 the removal of unwanted materials (Section 3.4.1.2). For certain types of biological samples,
672 removal and analysis of edible portions may be a key analytical planning issue.
673 Other issues that may represent key analytical issues for solids include container type and mate-
674 rial (Chapter 10); sample preservation (Chapter 10); sample drying—wet, dry, ashed weights and
675 ratios—(Chapter 10), screening samples for health and safety (Chapter 11); volatile compounds
JULY 2001
DRAFT FOR PUBLIC COMMENT
3-23
MARLAP
DO NOT CITE OR QUOTE
-------
Key Analytical Planning Issues...
676 (Chapter 10); sample identification (Chapters 10, 11, and 12); cross-contamination (Chapter 10);
677 sample size (Chapters 10, 11, and 12); compliance with the radioactive materials license and
678 shipping regulations (Chapter 11); and the chemical and physical form of the sample substrate
679 (Chapters 13 and 14).
680 3.4.1.1 Homogenization and Subsampling
681 For many types of analyses, a portion of the sample sent to the laboratory must be removed for
682 analysis. As with sampling in the field, this portion of the sample should be representative of the
683 entire sample. Adequate homogenization and proper subsampling techniques are critical to
684 obtaining a representative portion of the sample for analysis. Developing requirements for—and
685 measuring the adequacy of—homogenization processes and subsampling techniques can be
686 complicated for various types of solid matrices. General guidance on homogenization and sub-
687 sampling is provided in Chapter 12 and Appendix F. The input of the radioanalytical specialist as
688 a member of the project planning team is critical to developing requirements for homogenization
689 processes and subsampling techniques.
690 3.4.1.2 Removal of Unwanted Materials
691 When a solid sample is collected in the field, extraneous material may be collected along with the
692 "intended" sample. For example, when collecting a soil sample, rocks, plant matter, debris, etc.,
693 may also be collected. Unless instructed otherwise, samples received by the laboratory typically
694 are analyzed exactly as they are received. Therefore, it is important to develop requirements
695 regarding the treatment of extraneous materials. Ultimately, these guidelines should be based on
696 the project's DQOs. The requirements should clearly state what, if anything, is to be removed
697 from the sample and should indicate what is to be done with the removed materials. The
698 guidelines should indicate where the removal process should occur (in the field, in the laboratory
699 or at both locations) and the material to be removed should be clearly identified.
700 For soil samples, this may involve identifying rocks of a certain sieve size, plant matter, debris,
701 etc., as extraneous material to be removed, weighed, and stored at the laboratory. For sediment
702 samples, requirements for occluded water should be developed. In the case of biological samples,
703 if the entire sample is not to be analyzed, the analytical portion should be identified clearly.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 3 -24 DRAFT FOR PUBLIC COMMENT
-------
Key Analytical Planning Issues..
704 3.4.2 Liquids
705 Liquids include aqueous liquids (e.g., surface water, groundwater, drinking water, aqueous
706 process wastes, and effluents), nonaqueous liquids (e.g., oil, solvents, organic liquid process
707 wastes), and mixtures of aqueous and nonaqueous liquids.
708 A key analytical planning issue for most liquids is whether or not filtering is required or neces-
709 sary; this is discussed in Chapter 10. The question of whether or not to filter a liquid is generally
710 defined by the fundamental analytical question (Section 3.3.3). If the question is related to total
711 exposure from ingestion, the liquids are generally not filtered or the filters are analyzed
712 separately and the results summed. If the question is concerned with mobility of the analyte the
713 concentration in the liquid fraction becomes more important than the concentration in the sus-
714 pended solids (although some suspended solids may still be important to questions concerning
715 mobility of contamination). In many projects, all of the liquids are filtered and the question
716 becomes which filters need to be analyzed. Issues related to this decision include where and
717 when to filter (Chapter 10); homogenization and subsampling (Chapter 10); volatile compounds
718 (Chapter 10); screening for health and safety (Chapter 11); and cross-contamination (Chapter
719 10).
720 Another key analytical planning issue involves preservation of liquid samples, which is also dis-
721 cussed in Chapter 10. Sample preservation involves decisions about the method of preservation
722 (temperature or chemical, Chapter 10), container type and material (Chapter 10), and chemical
723 composition of the sample (Chapters 13 and 14). Preservation of radionuclides in liquids is
724 generally accomplished in the same manner as preservation of metals for chemical analysis.
725 There are of course exceptions such as for 3H and 129I.
726 A third key analytical issue results from the first two issues and involves the decision of which
727 issue should be resolved first. Should the sample be filtered and then preserved, or preserved first
728 and filtered later? This issue is also discussed in Chapter 10. In general, acid is used to preserve
729 liquid samples. Since acid brings many radionuclides into solution from suspended or undis-
730 solved material, filtering is generally performed in the field prior to preserving the sample with
731 acid.
732 Other analytical planning issues that may be important for a specific project include: sample
733 identification (Chapters 10, 11, and 12); volume of sample (Chapter 10); compliance with radio-
734 active materials license and shipping regulations (Chapter 11); immiscible layers (for mixtures of
735 aqueous and nonaqueous liquids, Chapter 12); precipitation between filtration and analysis
736 (Chapter 12); total dissolved solids (Chapter 12); and reagent background (Chapter 12).
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 3-25 DO NOT CITE OR QUOTE
-------
Key Analytical Planning Issues...
737 3.4.3 Filters and Wipes
738 Filters include a wide variety of samples, including liquid filters, air filters for suspended
739 particulates, and air filters for specific compounds. Once the decision to filter has been made,
740 there are at least three key analytical planning issues: filter material, pore size, and volume of
741 material to be filtered.
742 The selection of filter or wipe material can be very important. The wrong filter or wipe can
743 dissolve, break, or tear, thus invalidating the sample. Chapter 10 includes a discussion of the
744 various types of filter and wipe materials. Issues influencing this decision include the volume of
745 material to be filtered, the loading expected on the filter, and the chemical composition of the
746 material to be filtered.
747 The pore size is also important when preparing to filter. Too large a pore size will fail to collect
748 all of the material that is needed, while too small a pore size may lead to clogged filters and
749 reduced sample sizes. If an evaluation is being performed of respirable-size particles being
750 released by a process, the pore size of the filter should reflect this requirement.
751 The volume of material to be filtered, or area to be wiped, is generally determined by the detec-
752 tion requirements for the project. Lower detection limits require larger samples. Larger samples
753 may, in turn, result in problems with shipping samples or analytical problems where multiple
754 filters were required to meet the requested detection limits.
755 Other analytical planning issues that may be important for a specific project include sample
756 identification (Chapters 10, 11, and 12), compliance with radioactive materials license and ship-
757 ping regulations (Chapter 11), and background contributions from filter materials (Chapter 12).
758 3.5 Assembling the Analytical Protocol Specifications
759 After key general and matrix-specific analytical planning issues have been identified and
760 resolved, the next task of the project planning team is to organize and consolidate the results of
761 this process into APSs for the project. In general, there will be an APS for each type of analysis
762 (analyte-matrix combination). At a minimum, the APS should include the analyte list, the sample
763 matrix, possible interferences, the MQOs, any limitations on analysis options, the type and
764 frequency of QC samples along with acceptance criteria, and any analytical process requirements
765 (e.g., sample tracking requirements). The analytical process requirements should be limited to
766 only those requirements which are considered essential to meeting the project's analytical data
767 requirements. For example, if the analyte of concern is known to exist in a refractory form in the
MARLAP JULY 2001
DO NOT CITE OR QUOTE 3-26 DRAFT FOR PUBLIC COMMENT
-------
Key Analytical Planning Issues...
768 samples, then fusion for sample digestion may be included as an analytical process requirement.
769 However, in a performance-based approach, it is important that the level of specificity in the
770 Analytical Protocol Specifications should be limited to those requirements which are considered
111 essential to meeting the project's analytical data requirements. The APS should be a one- or
772 two-page form that summarizes the resolution of key analytical planning issues.
773 Figure 3.2 provides an example form for Analytical Protocol Specifications with references to
774 sections in this chapter as major headers on the form. Figure 3.3 provides for the purpose of an
775 example, an APS for 226Ra in soil for an information gathering project.
776 3.6 Level of Protocol Performance Demonstration
777 As discussed in Section 3.3.7.3, during project planning, the project planning team should deter-
778 mine what level of analytical performance demonstration or method validation is appropriate for
779 the project. The question to be answered is how the analytical protocols will be evaluated. There
780 are three parts of this overall evaluation process: (1) the initial evaluation, (2) the ongoing evalu-
781 ation, and (3) the final evaluation. This section briefly discusses the initial evaluation of protocol
782 performance. Chapters 7 and 8 provide guidance on the ongoing and final evaluation of protocol
783 performance, respectively.
784 The project planning team should determine what level of initial performance demonstration is
785 required from the laboratory to demonstrate that the analytical protocols the laboratory proposes
786 to use will meet the MQOs and other requirements in the APSs. The project planning team
787 should decide the type and amount of performance data required. For example, for the analysis of
788 3H in drinking water, the project planning team may decide that past performance data from the
789 laboratory, such as the results of internal QC samples for the analysis of 3H in drinking water, are
790 sufficient for the initial demonstration of performance for the laboratory's analytical protocols if
791 they demonstrate the protocol's ability to meet the MQOs. If the analysis is for 238Pu in a sludge,
792 the project planning team may decide that past performance data (if it exists) would not be
793 sufficient for the initial demonstration of performance. The planning team may decide that
794 satisfactory results on performance evaluation samples would be required for the initial
795 demonstration of analytical protocol performance. Section 6.6 provides detailed guidance on
796 protocol performance demonstration/method validation, including a tiered approach based on the
797 project analytical needs and available resources.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 3-27 DO NOT CITE OR QUOTE
-------
Key Analytical Planning Issues...
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
Analytical Protocol Specifications
Analyte List: (Section 3.3.1. 3.3.7 Analysis Limitations: (Sections 3.3.9)
Matrix: (Section 3.3.3) Possible Interferences: (Sections 3.3.3. 3.3.5)
Concentration Range: (Section 3.3.2) Action Level (Section 3.3.8)
MQOs:
(Section 3.3.8)
(Section 3.3.8)
(Section 3.3.8)
(Section 3.3.8)
QC Samples
Type
(Section 3. 3. 11)
(Section 3. 3. 11)
(Section 3. 3. 11)
(Section 3. 3. 11)
Frequency
(Section 3. 3. 11)
(Section 3. 3. 11)
(Section 3. 3. 11)
(Section 3. 3. 11)
Evaluation Criteria
(Section 3.3. 8.2)
(Section 3.3. 8.2)
(Section 3.3. 8.2)
(Section 3.3. 8.2)
Analytical Process Requirements*
Activity
Field Sample Preparation and Preservation
Sample Receipt and Inspection
Laboratory Sample Preparation
Sample Dissolution
Chemical Separations
Preparing Sources for Counting
Nuclear Counting
Data Reduction and Reporting
Sample Tracking Requirements
Other
Special Requirements
(Section 3. 4)
(Section 3. 4. 12)
(Section 3. 4)
(Section 3. 4)
(Section 3. 4)
(Section 3. 4)
(Section 3. 4)
(Section 3.3. 13)
(Section 3. 3. 12)
* Consistent with a performance-based approach, analytical process requirements should be kept to a minimum,
therefore none or N/A may be appropriate for many of the activities.
MARLAP
DO NOT CITE OR QUOTE
FIGURE 3.2 — Analytical protocol specifications
3-28
JULY 2001
DRAFT FOR PUBLIC COMMENT
-------
Key Analytical Planning Issues...
826
827
828
829
830
Analytical Protocol Specifications (Example)
Analyte List: 226Ra
Matrix: Soil
Analysis Limitations: Must perform direct measurement of
analyte or analysis of progeny allowed if equilibrium established at
laboratory
Possible Interferences: Elevated levels of 235U
831
Concentration Range: 0.01 to 1.50Bq/g Action Level: 0.5 Bq/g
832 MQOs:
833 A method uncertainty (»„,') of 0.04 Bq/g or less at 0.5 Bq/g
QC Samples
Type
Method blank
Duplicate
Matrix Spike
Frequency
1 per batch
1 per batch
1 per batch
Evaluation Criteria
See attachment B*
See attachment B*
See attachment B*
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
Analytical Process Requirements
Activity
Field Sample Preparation and Preservation
Sample Receipt and Inspection
Laboratory Sample Preparation
Sample Dissolution
Chemical Separations
Preparing Sources for Counting
Nuclear Counting
Data Reduction and Reporting
Sample Tracking Requirements
Other
Special Requirements
None
None
None
None
None
None
None
See attachment A*
Chain-of -Custody
Attachments A and B are not provided in this example
FIGURE 3.3 — Example analytical protocol specifications
JULY 2001
DRAFT FOR PUBLIC COMMENT
3-29
MARLAP
DO NOT CITE OR QUOTE
-------
Key Analytical Planning Issues...
853 3.7 Project Plan Documents
854 Once the APSs have been completed, they should be incorporated into the appropriate project
855 plan documents and, ultimately, into the analytical Statement of Work. Chapters 4 and 5 provide
856 guidance on the development of project plan documents and analytical Statements of Work,
857 respectively. While the APSs are concise compilations of the analytical data requirements, the
858 appropriate plan documents should detail the rationale behind the decisions made in the develop-
859 ment of the APSs.
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
Summary of Recommendations
MARLAP recommends that any assumptions made during the resolution of key analytical
planning issues are documented, and that these assumptions are incorporated into the
appropriate narrative sections of project plan documents.
MARLAP recommends that an action level and gray region be established for each analyte
during the directed planning process.
MARLAP recommends that the method uncertainty at a specified concentration (typically
the action level) always be identified as an important method performance characteristic,
and that an MQO be established for it for each analyte.
MARLAP recommends that the MQO for the detection capability be expressed as a
required minimum detectable concentration.
MARLAP recommends that the MQO for the quantification capability be expressed as a
required minimum quantifiable concentration.
MARLAP recommends that units of the International System of Units (SI) be used
whenever possible.
MARLAP recommends that all measurement results be reported directly as obtained,
including negative values, along with the measurement uncertainty.
MARLAP
DO NOT CITE OR QUOTE
3-30
JULY 2001
DRAFT FOR PUBLIC COMMENT
-------
Key Analytical Planning Issues..
87? 3.8 References
878 American Public Health Association, American Water Works Association, Water Environment
879 Federation (APHA/AWWA). Standard Methods for the Examination of Water and
880 Wastewater, 18th ed, 1992. American Public Health Association, Washington, DC.
881 American Society for Testing and Materials (ASTM) El 169. Standard Guide for Conducting
882 Ruggedness Test. 1989.
883 U.S. Environmental Protection Agency (EPA). 1998. Guidance for the Quality Assurance
884 Project Plans (EPA QA/G-5). EPA/600/R-98/018, Washington, DC.
885 MARSSEVI. 2000. Multi-Agency Radiation Survey and Site Investigation Manual, Revision 1.
886 NUREG-1575 Rev 1, EPA 402-R-97-016 Revl, DOE/EH-0624 Revl. August. Available
887 from http://www.epa.gov/radiation/marssim/filesfm.htm.
888 Youden, WJ. and E.H. Steiner. 1975. Statistical Manual of the Association of Official Analytical
889 Chemists. Association of Official Analytical Chemists International, Gaithersburg, MD.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 3-31 DO NOT CITE OR QUOTE
-------
4 PROJECT PLAN DOCUMENTS
2 4.1 Introduction
3 The project plan documents are a blueprint for how a particular project will achieve data of the
4 type and quality needed and expected by the project planning team. In the planning documents,
5 the data user's expectations and requirements, which are developed during the planning
6 process—including the Analytical Protocol Specifications and measurement quality objectives
7 (MQOs)—are documented along with the standard operating procedures (SOPs), health and
8 safety protocols, and quality assurance/quality control (QA/QC) procedures for the field and
9 laboratory analytical teams. The objectives of this chapter are to discuss:
10 • The importance of project plan documents;
11 • The elements of project plan documents; and
12 • The link between project planning and project plan documents, in particular the incorporation
13 of the analytical protocols.
14 The importance of project plan documents is discussed in Section 4.2. Section 4.3 discusses a
15 graded approach to project plan documents. The different types of planning documents and the
16 elements of the project plan documents are discussed in Sections 4.4 and 4.5, respectively. The
17 link between project planning and project plan documents is discussed in Section 4.6.
18 The project plan documents should be dynamic documents, used and updated over the life of the
19 project. Under a performance-based approach, the analytical protocols requirements in the project
20 plan documents initially may reflect the Analytical Protocol Specifications established by the
21 project planning team and issued in the statement of work (SOW) (or Basic Ordering Agreement
22 Task Order). When the analytical laboratory has been selected, the project plan documents should
23 be updated to reflect the actual protocols to be used. The protocols should be cited, or the SOPs
24 for the protocols should be included as appendices. (Analytical Protocol Specifications and the
25 relation to project measurement quality objectives (MQOs) have been discussed in Chapter 3 and
26 represented in Figure 3.2 and 3.3).
27 While this chapter will address the documentation of QA/QC used in project activities,
28 MARLAP is cognizant of, and fully endorses, the need for an organizational quality system and a
29 quality system, management plan, or quality manual. The development of the project plan
30 documents should be addressed in the quality system requirements documentation. The project
31 plan documents should reflect, and be consistent with, the organization's QA policies and
32 procedures. Guidance on elements of a quality system for environmental data collection activities
33 is available from several sources including ANSI/ASQC (1994) and ISO Standard 9001 (1994).
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 4-1 DO NOT CITE OR QUOTE
-------
Project Plan Documents
34 The QA requirements have been developed by several Federal Agencies and consensus standard
35 organizations including the following:
36 • 10 CFR 830.120
37 • 10 CFR 50, Appendix B
38 • ANSIN42.23-1996
39 • ASMENQA-1-1989
40 • DOE Order 4.14.1 on QA
41 • EPA Order 5360.1 on Quality Systems (1998c)
42 • DODQA requirement MIL-Q-9858A
43 4.2 The Importance of Project Plan Documents
44 Project plan documents are important in environmental data collection activities to ensure that
45 the type and quantity of data are sufficient for the decision to be made. Project plans document
46 the decisions made during the planning process and integrate the technical operations with the
47 management and quality system practices. Project plans also:
48 • Support data defensibility for environmental compliance;
49 • Can be used to defend project objectives and budget; and
50 • Are a tool for communication with stakeholders.
51 The development of project plan documents and the implementation of the project plan provide
52 the following benefits:
53 • Full documentation for legal, regulatory, and historical use of the information;
54 • Specification of data collection and quality control;
55 • Documentation of analytical requirements through the incorporation of an Analytical
56 Protocol Specifications;
57 • Implementation of planned data collection activities (through internal and external
58 assessment and oversight activities); and
59 • Meeting project-specific criteria (i.e., MQOs, DQOs) through data validation and usability
60 assessment.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 4-2 DRAFT FOR PUBLIC COMMENT
-------
Project Plan Documents
6i 4.3 A Graded Approach to Project Plan Documents
62 A graded approach is the process of basing the level of management controls applied to an item
63 or work on the intended use of the results and the degree of confidence needed in the quality of
64 the results (ANSI/ASQC, 1994). MARLAP recommends a graded approach to project plan
65 development because of the diversity of environmental data collection activities. This diversity in
66 the type of project and the data to be collected impacts the content and extent of the detail to be
67 presented in the plan document. The plan document development team should be flexible in their
68 application of guidance according to the nature of the work being performed and the intended use
69 of the data.
70 Under a graded approach, a mix of project-specific and site-based quality system documentation
71 may be relied upon to ensure quality. For example, the project specific plan may:
72
• Address design, work processes, and inspection; and
73 • Incorporate by citation site-wide plans that address records management, quality
74 improvement, procurement, and assessment.
75 A comprehensive and detailed project plan is required for some data collection activities because
76 of the need for legal and scientific defensibility of the data. A comprehensive and detailed plan
77 may also be desirable when Office of Management and Budget (OMB) clearance and approval is
78 needed to carry out the project (e.g., NRC/EPA proposed Publicly Owned Treatment Works
79 Survey).
80 Other environmental data collection activities, such as basic studies or small projects, may only
81 require a discussion of the experimental process and its objectives, which is often called a project
82 narrative statement. (Other titles used for project narrative statements are "QA narrative
83 statement" and "proposal QA plan" (EPA, 1998a). Basic studies and small projects generally are
84 of short duration or limited scope and could include proof of concept studies, exploratory
85 projects, small data collection tasks, feasibility studies, qualitative screens, or initial work to
86 explore assumptions or correlations. Although basic studies and small projects may be used to
87 acquire a better understanding of a phenomenon, they will not by themselves be used to make
88 significant decisions or establish policy. Further discussion on the content of plan documents for
89 basic studies and small projects is provided in Section 4.5.3.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 4-3 DO NOT CITE OR QUOTE
-------
Project Plan Documents
90 4.4 Project Plan Documents
91 The ANSI/ASQC (1994) definition for a QA Project Plan (QAPP), which is also applicable to
92 other integrated project plan documents, is "a formal document describing in comprehensive
93 detail the necessary QA, QC and other technical activities that must be implemented to ensure
94 that the results of the work performed will satisfy the stated performance criteria." The project
95 plan documents should contain this information in a clear and integrated manner so that all
96 implementation teams can understand their role and the project objectives.
97 Project plan documents vary in size and format and are referred to by a variety of names. The
98 size of the project plan documents tends to reflect the issuing agency's requirements, complexity,
99 and scope of the project activities. Some projects with multiple phases may have more than one
100 plan document. For example, separate plan documents may be developed for scoping surveys,
101 characterization, and the final status survey for the same site because of the different objectives
102 and data requirements. Available guidance on project plans will be discussed in Section 4.4.1,
103 and a general discussion of various approaches is discussed in Section 4.4.2.
104 4.4.1 Guidance on Project Plan Documents
105 National standards guidance on project plan documents is available in:
106 • ASTM Standard Practice (D5283) for Generation of Environmental Data Related to Waste
107 Management Activities: Quality Assurance and Quality Control Planning and
108 Implementation (ASTM, 1992);
109 • Standard Guide (D5612), Quality Planning and Field Implementation of a Water Quality
110 Measurements Program (ASTM, 1994); and
ill • Standard Provisional Guide (PS85) for Expedited Site Characterization of Hazardous Waste
112 Contaminated Sites (ASTM, 1996).
113 Guidance on project plans for environmental data collection activities in the federal sector is also
114 available (EPA, 1998a; 40 CFR 300.430; NRC, 1989; and USACE, 1994 and 1997). Other
115 Federal Agency guidance may follow EPA guidance for QAPPs (EPA, 1998a).
MARLAP JULY 2001
DO NOT CITE OR QUOTE 4-4 DRAFT FOR PUBLIC COMMENT
-------
Project Plan Documents
116 4.4.2 Approaches to Project Plan Documents
117 The approach and naming of project plan documents is usually a function of the authoring
118 organization's experience, any controlling Federal or state regulations, or the controlling Agency.
119 Project plan, work plan, QAPP, field sampling plan, sampling and analysis plan, and dynamic
120 work plan are some of the names commonly used for project plan documents. The names can
121 however often represent different documents to different agencies, states, companies and even to
122 different people within the same organization.
123 A work plan is often the primary and integrating plan document when the data collection activity
124 is a smaller supportive component of a more comprehensive project (for example, data collection
125 activity in support of an aspect of an environmental impact statement for a large multi-year
126 project). The QAPP is often the primary document when the data collection activity is a major
127 portion of the project (for example, data collection activity in support of an initial site
128 investigation). A National Contingency Plan (NCP) format (specified in 40 CFR 300.430) is
129 appropriate when data collection activities are in support of National Priorities List (NPL)
130 Superfund site projects. The NCP format has a sampling and analysis plan as the primary plan
131 document. The project documentation consists of two integrated documents: a field sampling
132 plan and a QAPP. Stand-alone health and safety plans are also developed.
133 Traditional site investigations are generally based on a phased engineering approach, which
134 collects samples based on a pre-specified grid pattern and does not provide the framework for
135 making changes in the plan in the field. The work plan (the project plan document) for the site
136 investigation typically will specify the number of samples to be collected, the location of each
137 sample and the analyses to be performed. A newer concept is to develop a dynamic work plan
138 (the project plan document), which, rather than specifying the number of samples to be collected
139 and the location of each sample, would specify the decision making logic that will be used in the
140 field to determine where the samples will be collected, when the sampling will stop, and what
141 analyses will be performed. Guidance on dynamic work plans is available in the Standard
142 Provisional Guide (PS85) for Expedited Site Characterization of Hazardous Waste Contaminated
143 Sites (ASTM, 1996).
144 MARLAP does not recommend a particular project plan document approach, title or arrange-
145 ment. Federal and state agencies have different requirements for the various environmental data
146 collection activities. In certain cases there are regulatory requirements. If an organization has
147 successful experience addressing the essential content of plan documents (Section 4.5) in a well
148 integrated, document format, it is usually unnecessary and wasteful of time and monies to change
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 4-5 DO NOT CITE OR QUOTE
-------
Project Plan Documents
149 a proven approach. The project plan document should reflect, and be consistent with, the
150 organization's QA policies and procedures.
151 MARLAP recommends a primary project plan document that includes other documents by
152 citation or as appendices. The primary project plan document serves to integrate the multi-
153 disciplinary sections, other management plans, and stand alone documents into a coherent plan.
154 Appropriate management plans may include the Health and Safety Plan, Waste Management
155 Plan, Risk Analysis Plan, Community Relations Plan, or Records Management Plan. If a detailed
156 discussion of the project already exists in another document, which is available to project
157 participants, then a brief description of site history and incorporation of the document into the
158 project plan document by reference may be appropriate. Incorporation by citation may also be
159 appropriate when the complexity of the project requires an extensive discussion of background
160 issues. Other documents that should be integrated, if available, are the report on the planning
161 process, the Data Validation Plan (Chapter 8), and the DQA Plan (Chapter 9). If stand alone
162 documents are not immediately available to project participants, they should be appended to the
163 (primary) project plan document.
164 4.5 Elements of Project Plan Documents
165 A project plan document must address a range of issues. The extent of the detail is dependent on
166 the type of project and the intended use of the results as previously discussed in applying a
167 graded approach to plan documents (Section 4.3). For all projects, the project plan document
168 must provide the project information and decisions developed during the project planning
169 process. Project plan documents should address:
170 • Theproject'sDQOsandMQOs;
171 • The sampling and analytical protocols that will be used to achieve the project objectives; and
172 • The assessment procedures and documentation that are sufficient to confirm that the data are
173 of the type and quality needed.
174 Content of plan documents is discussed in Section 4.5.1. The integration of project plan
175 documents is discussed in Section 4.5.2. Special consideration of project documentation for
176 small projects is discussed in Section 4.5.3.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 4-6 DRAFT FOR PUBLIC COMMENT
-------
Project Plan Documents
111
4.5.1 Content of Project Plan Documents
178 The plan document development team should remain flexible with regards to format and should
179 focus on the appropriate content of plan documents needed to address the elements listed above.
180 The content of plan documents, regardless of the title or format, will include similar information,
181 including:
182
The project description and objectives;
183
184
Identification of those involved in the data collection and their responsibilities and
authorities;
185
Enumeration of the QC procedures to be followed;
186
Reference to specific SOPs that will be followed for all aspects of the projects; and
187
Health and Safety protocols.
188 The project plan document(s) should present the document elements as integrated chapters,
189 appendices, and stand alone documents, and plans should be included by citation. Table 4.1
190 provides summary information on project plan elements for three different plan documents:
191 project plans, dynamic work plans, and QAPPs as provided in ASTM and EPA guidance. The
192 table also illustrates the similarity of project plan content.
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
TABLE 4.1—Elements of Project Plan Documents
Project Plan
(ASTM D5283,1992 and ASTM D5612,
1994)
Project Management
Identify individuals with designated res-
ponsibility and authority to: (1) develop
project documents; (2) select organizations
to perform the work; (3) coordinate com-
munications; and (4) review and assess
final data.
Background Information
Reasons for data collection.
Identify regulatory programs governing
data collection.
Dynamic Work Plan
(ASTM PS 85,1996)
1. Regulatory Framework
2. Site Descriptions and
History of Analyte Use and
Discovery
3. Analysis of Prior Data and
Preliminary Conceptual
Site Model
QAPP
(EPA, 1998a)
A. Project Management
Al Approval Sheet
A2 Table of Contents
A3 Distribution List
A4 Project Organization
A5 Problem Definition and
Background
JULY 2001
DRAFT FOR PUBLIC COMMENT
4-7
MARLAP
DO NOT CITE OR QUOTE
-------
Project Plan Documents
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
Project Plan
(ASTM D5283,1992 and ASTM D5612,
1994)
Project Objectives
Clearly define objectives of field and
laboratory work.
Define specific objectives for the
sampling location.
Describe intended use of data.
Sampling Requirements
Sample requirements are specified,
including:
Sampling locations.
Equipment and Procedures (SOPs).
Sample preservation and handling .
Analytical Requirements
The analytical requirements are specified,
including:
Analytical procedures (SOPs).
Analyte list.
Required method uncertainty.
Required detection limits.
Regulatory requirements and DQO
specifications are considered.
Quality Assurance and Quality Control
Requirements
QA/QC requirements are addressed for
both field and laboratory activities.
Type and frequency of QC samples will
be specified.
Control parameters for field activities
will be described.
Performance criteria for laboratory
analysis will be specified.
Data validation criteria (for laboratory
analysis) will be specified.
Dynamic Work Plan
(ASTM PS 85,1996)
Dynamic Technical Program
Essential questions to be
answered or specific
objectives.
Identify the investigation
methods and the areas in which
they may be applied.
Provide clear criteria for
determining when the project
objectives have been met.
Field Protocols and Standard
Operating Procedures (this
section may be attached as a
separate document)
[* see footnote]
Quality Assurance and Quality
Control Plan
QAPP
(EPA, 1998a)
A6 Project Description.
A7 Quality Objectives and Criteria
for Measurement Data.
A8 Special Training Require-
ments/Certifications .
A9 Documentation and Records.
B. Measurement/Data
Acquisition
B1 Sampling Process Designs.
B2 Sampling Method
Requirements.
B3 Sample Handling and Custody
Requirements.
B4 Analytical Methods
Requirements.
B5 Quality Control Requirements.
B6 Instrument/Equipment Testing
Inspection and Maintenance
Requirements.
B7 Instrument Calibration and
frequency.
B8 Inspection/Acceptance
Requirements for Supplies and
Consumables.
B9 Data Acquisition
Requirements for Non-direct
Measurements.
BIO Data Management.
MARLAP
DO NOT CITE OR QUOTE
4-8
JULY 2001
DRAFT FOR PUBLIC COMMENT
-------
Project Plan Documents
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
Project Plan
(ASTM D5283,1992 and ASTM D5612,
1994)
Project Documentation
All documents required for planning,
implementating, and evaluating the data
collection efforts are specified, may
include:
SOW, Work Plan, SAP, QAPP, H&S
Plan, Community Relations Plan.
Technical reports assessing data.
Requirements for field and analytical
records.
Dynamic Work Plan
(ASTM PS 85,1996)
1. Data Management Plan
2. Health and Safety Plan
3. Community Relations Plan
QAPP
(EPA, 1998a)
C. Assessment/Oversight
Cl Assessments and response
Actions.
C2 Reports to Management.
D. Data Validation and Usability
D1 Data Review, Verifications
and Validation Requirements.
Verification and Validation
Methods.
Reconciliation with DQO.
D2
D3
[* The combined Dynamic Technical Program section and Field Protocols and SOPs section is the functional
equivalent of a Field Sampling and Analysis Plan.]
Appendix D provides more detailed guidance on the content of project plan documents following
the outline developed by EPA requirements (EPA, 1998b) and guidance (EPA, 1998a) for
Quality Assurance Project Plans for environmental data operations. The EPA element identifiers
(Al, A2, etc.) and element titles are used in the tables and text of this chapter for ease of cross
reference to the appropriate section in Appendix D. The EPA elements for a QAPP are used to
facilitate the presentation and do not represent a recommendation by MARLAP on the use of a
QAPP as the project plan document format.
4.5.2 Plan Documents Integration
MARLAP strongly discourages the use of a number of stand-alone plan components of
equivalent status without integrating information and without a document being identified as a
primary document. For large project plan compilations, it is appropriate to issue stand-alone
portions of the plan that focus on certain activities such as sampling, analysis or data validation,
since it can be cumbersome for sampling and laboratory personnel to keep the entire volume(s)
of the project plan document readily available. However, each stand-alone component should
contain consistent project information, in addition to the component specific plan information,
such as the following:
• A brief description of the proj ect including pertinent history;
• A brief discussion of the problem to be solved or the question to be answered (DQO);
• An organizational chart or list of key contact persons and means of contact;
• The analyte(s) of interest; and
• The appropriate health and safety protocols and documentation requirements.
JULY 2001
DRAFT FOR PUBLIC COMMENT
4-9
MARLAP
DO NOT CITE OR QUOTE
-------
Project Plan Documents
274 In addition, a cross-referenced table is helpful in the primary document, which identifies where
275 project plan elements are located in the integrated plan document.
276 4.5.3 Plan Content for Small Projects
277 The project plan documents for small projects and basic studies (Section 4.3) generally consist of
278 three elements: the Title and Approval Sheet, the Distribution List, and a Project Narrative. The
279 Project Narrative should discuss in a concise manner the majority of issues that are normally
280 addressed in a project plan document, such as a QAPP. A typical Project Narrative (EPA, 1998b)
281 may be a concise and brief description of:
282 • Problem and site history (A5)
283 • Project/taskorganization(A4)
284 • Project tasks, including a schedule and key deliverables (A6)
285 • Anticipated use of the data (A5, A6)
286 • MQOs(A7)
287 • Sampling process design requirements and description (Bl)
288 • Sample type and sampling location requirements (B2)
289 • Sample handling and custody requirements (B3)
290 • Analytical protocols (B4)
291 • QC and calibration requirements for sampling and analysis (B5, B7)
292 • Inspection and maintenance of analytical instrumentation (B6)
293 • Plans for peer or readiness reviews prior to data collection (C1)
294 • Assessments to be conducted during actual operation (Cl)
295 • Procedure for data review (D2)
296 • Identification of any special reports on QA/QC activities, as appropriate (C2)
297 • Reconciliation with DQOs or other objectives (D3)
298 Table 4.2 or Appendix D gives information on what is addressed in each bullet above, using the
299 element identifier shown in parenthesis.
300 4.6 Linking the Project Plan Documents and the Project Planning Process
301 Directed planning processes (see Chapter 2 and Appendix B) yield many outputs, such as the
302 Analytical Protocol Specifications (Chapter 3), which must be captured in project plan
303 documents to ensure that data collection activities are implemented properly. MARLAP
304 recommends that the project plan documents integrate all technical and quality aspects for the life
305 cycle of the project, including planning, implementation, and assessment.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 4-10 DRAFT FOR PUBLIC COMMENT
-------
Project Plan Documents
306
307
308
309
310
311
312
313
TABLE 4.2-Crosswalk Between Project Plan Document Elements
and Directed Planning Process
314
315
316
ID
Project Plan Document
Elements
(QAPP, EPA QA/R-5,
1998b)*
Content
Directed Planning Process Input
PROJECT MANAGEMENT
Al
A2_
A3
A4
A5
A6
A7
Title and Approval Sheet
Table of Contents
Distribution List
Project/Task Organization
Problem Definition/
Background
Project/Task Description
Quality Objectives and
Criteria for Measurement
Data
Title and approval sheet.
Document control format.
Distribution list for the plan
document revisions and final
guidance.
1) Identify individuals or
organizations participating in the
project and discuss their roles and
responsibilities.
2) Provide an organizational chart
showing relationships and
communication lines.
1) State the specific problem to be
solved and decision to be made.
2) Include enough background to
provide a historical perspective.
Identify measurements, special
requirements, sampling and
analytical methods, action levels,
regulatory standards, required data
and reports, quality assessment
techniques, and schedules.
1) Identify DQOs, data use, type of
data needed, domain, matrices,
constraints, action levels, statistical
parameters, and acceptable decision
errors.
2) Establish MQOs that link
Include the members of the project
planning team and stakeholders.
The directed planning process:
• Identified the stakeholders, data
users, decision makers.
• Identified the core planning team and
the technical planning team members
who will often be responsible for
technical oversight.
• Will often identify the specific
persons/organizations that will be
responsible for project
implementation (sampling and
analysis).
Project planning team:
• Documented the problem, site
history, existing data, regulatory
concerns, background levels and
thresholds.
• Developed a decision statement.
Project planning team identified:
• Deadlines and other constraints that
can impact scheduling.
• Existing and needed data inputs.
Project planning team established:
• Action levels and tolerable decision
error rates that will be the basis for
the decision rule.
• The optimized sampling and
analytical design as well as quality
criteria.
Project planning team:
• Identified the regulatory standards
and the action level(s).
• Established the decision rule.
• Described the existing and needed
data inputs.
• Described practical constraints and
JULY 2001
DRAFT FOR PUBLIC COMMENT
4-11
MARLAP
DO NOT CITE OR QUOTE
-------
Project Plan Documents
317
318
319
320
321
ID
A8
A9
Project Plan Document
Elements
(QAPP, EPA QA/R-5,
1998b)*
Special Training
Requirements/
Certification
Documentation and Record
Content
analysis to the user's quality
objectives.
Identify and discuss special
training/certificates required to
perform work.
Itemize the information and records,
which must be included in a data
report package including report
format and requirements for storage
etc.
Directed Planning Process Input
the domain.
Established the statistical parameter
that is compared to the action level.
Established tolerable decision error
rates used to choose quality criteria.
Established quality criteria linked to
the optimized design.
Establish data verification, validation
and assessment criteria and
procedures.
Establish APS and MQOs.
Project planning team:
• Identified training, certification,
accreditation requirements for field
and laboratory.
• Identified Federal and state
requirements for certification for
laboratories.
• Identified Federal and state
requirements for activities, such as
disposal of field-generated residuals.
Project planning team:
• Indicated whether documents will be
controlled and the distribution list
incomplete.
• Identified documents that must be
archived.
• Specified period of time that
documents must be archived.
• Specified procedures for error
corrections (for hard copy and
electronic files).
MEASUREMENT/DATA ACQUISITION
Bl
B2
Sampling Process Designs
(Experimental Designs)
Sampling Methods
Requirements
(1) Outline the experimental design,
including sampling design and
rationale, sampling frequencies,
matrices, and measurement
parameter of interest.
(2) Identify non-standard methods
and validation process.
Describe sampling procedures,
needed materials and facilities,
decontamination procedures, waste
handling and disposal procedures,
Project planning team established the
rationale for and details of the
sampling design.
Project planning team specified the
preliminary details of the optimized
sampling method.
MARLAP
DO NOT CITE OR QUOTE
4-12
JULY 2001
DRAFT FOR PUBLIC COMMENT
-------
Project Plan Documents
322
323
324
325
326
327
ID
B3
B4
B5
B6
B7
B8
Project Plan Document
Elements
(QAPP, EPA QA/R-5,
1998b)*
Sample Handling and
Custody Requirements
Analytical Methods
Requirements
Quality Control
Requirements
Instrument/Equipment
Testing Inspection and
Maintenance Requirements
Instrument Calibration and
Frequency
Inspection/Acceptance
Requirements for Supplies
and Consumables
Content
and include a tabular description of
sample containers, sample volumes,
preservation and holding time
requirements.
Describe the provisions for sample
labeling, shipment, sample tracking
forms, procedures for transferring
and maintaining custody of samples.
Identify analytical methods and
procedures including needed
materials, waste disposal and
corrective action procedures.
(1) Describe QC procedures and
associated acceptance criteria and
corrective actions for each sampling
and analytical technique.
(2) Define the types and frequency
of QC samples should be defined
along with the equations for
calculating QC statistics.
1) Discuss determination of
acceptable instrumentation
performance.
2) Discuss the procedures for
periodic, preventive and corrective
maintenance.
(1) Identify tools, gauges and
instruments, and other sampling or
measurement devices that need
calibration.
(2) Describe how the calibration
should be done.
Define how and by whom the
sampling supplies and other
consumables will be accepted for
use in the project.
Directed Planning Process Input
Project planning team described the
regulatory situation and site history,
which can be used to identify the
appropriate sample tracking level.
Project planning team:
• Identified inputs to the decision
(nuclide of interest, matrix, etc.).
• Established the allowable
measurement uncertainty that will
drive choice of the analytical
protocols.
• Specified the optimized sampling
and analytical design.
Project planning team:
• Established the allowable
measurement uncertainty, which will
drive QC acceptance criteria.
• Established the optimized analytical
protocols and desired MQOs.
Project planning team established the
desired MQOs, which will drive
acceptance criteria for
instrumentation performance.
JULY 2001
DRAFT FOR PUBLIC COMMENT
4-13
MARLAP
DO NOT CITE OR QUOTE
-------
Project Plan Documents
ID
B9
Bl
0
Project Plan Document
Elements
(QAPP, EPA QA/R-5,
1998b)*
Data Acquisition
Requirements (Non-direct
Measurements)
Data Management
Content
Define criteria for the use of non-
direct measurement data such as
data that come from databases or
literature.
(1) Outline of data management
scheme including path of data, use
of storage and& record keeping
system. (2) Identify all data handling
equipment and procedures that will
be used to process, compile, analyze
the data, and correct errors.
Directed Planning Process Input
Project planning team:
• Identified the types of existing data
that are needed or would be useful.
• Established the desired MQOs that
would also be applicable to archived
data.
ASSESSMENT/OVERSIGHT
Cl
C2
Assessments and Response
Actions
Reports to Management
(1) Describe the number, frequency
and type of assessments needed for
the project.
(2) For each assessment: list
participants and their authority, the
schedule, expected information,
criteria for success and unsatis-
factory conditions and those who
will receive reports and procedures
for corrective actions.
Identify the frequency, content and
distribution of reports issued to keep
management informed.
• The project planning team
established the MQOs and
developed statements of the
Analytical Protocol Specifications,
which are used in the selection of the
analytical protocols and in the
ongoing evaluation of the protocols.
DATA VALIDATION AND USABILITY
Dl
D2
Data Review, Verification
and Validation Requirements
Verification and Validation
Methods
State the criteria including specific
statistics and equations, which will
be used to accept or reject data
based on quality.
Describe the process to be used for
validating and verifying data,
including COC for data throughout
the lifetime of the project.
• Project planning team established.
• Established the MQOs for the
sample analysis, and may also have
discussed completeness and
representativeness requirements that
will be the basis of validation.
• Established the action level(s)
relevant to the project DQOs.
• Established the data validation
criteria.
Project planning team:
• Determines appropriate level of
custody.
• May develop a Validation Plan.
328
329
330
331
332
333
334
335
336
MARLAP
DO NOT CITE OR QUOTE
4-14
JULY 2001
DRAFT FOR PUBLIC COMMENT
-------
Project Plan Documents
337
ID
D3
Project Plan Document
Elements
(QAPP, EPA QA/R-5,
1998b)*
Reconciliation With Data
Quality Objectives
Content
Describe how results will be
evaluated to determine if DQOs are
satisfied.
Directed Planning Process Input
Project planning team:
• Defined the necessary data input
needs.
• Defined the constraints and
boundaries with which the project
would have had to comply.
• Defined the decision rule.
• Identified the hypothesis and
tolerable decision error rates.
• Defined MQOs for achieving the
project DOOs.
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
[Adapted from: EPA, 1998a]
[* EPA QAPP elements are discussed in Appendix D]
The project plan should be a dynamic document, used and updated over the life of the project.
For example, the analytical methods requirements in the project plan documents (B4) will
initially reflect the Analytical Protocol Specifications established by the project planning team
(Chapter 3) and issued in the SOW or BOA Task Order (Chapter 5). When the analytical
laboratory has been selected (Chapter 7), the project plan document should be updated to reflect
the specific analytical protocols: the actual protocols to be used, which should be included by
citation or inclusion of the SOPs as appendices.
4.6.1 Planning Process Report
MARLAP recommends the inclusion, by citation or as an appendix, of the directed planning
process report in the project plan documents. If the planning process was not documented in a
report, MARLAP recommends that a summary of the planning process addressing, for example,
the assumptions and decisions, the established action levels, the DQO statement, and the
Analytical Protocol Specifications, which include the established MQOs and any specific
analytical process requirements, be included in the project plan document section on Problem
Definition/Background (A5). Additional detailed information on the analytical protocol
specifications including the MQOs will be presented in the project plan document sections on
Project/Task Description (A6), Quality Objectives and Criteria for Measurement Data (A7), and
Analytical Methods Requirements (B4).
MARLAP views the project plan documents as the principal product of the planning process. To
illustrate how to capture and integrate the outputs of the planning process into the plan docu-
JULY2001
DRAFT FOR PUBLIC COMMENT
4-15
MARLAP
DO NOT CITE OR QUOTE
-------
360 ment(s), Table 4.2 presents a crosswalk of the elements of the EPA QAPP Document and outputs
361 of a directed planning process.
362 4.6.2 Data Assessment
363 Assessment (Verification, Validation and Data Quality Assessment) is the last step in the
364 project's data life cycle and precedes the use of data. Assessment, and in particular DQA, are
365 designed to evaluate the suitability of project data to answer the underlying project question or
366 the suitability of project data to support the project decision. The project planners should define
367 the assessment process in enough detail that achievement or failure to meet goals can be
368 established upon project completion. An important output of the directed planning process to be
369 captured in the project plan document is the data verification, validation and assessment criteria
370 and procedures.
371 4.6.2.1 Data Verification
372 Analytical data verification assures that laboratory conditions and operations were compliant
373 with the contractual SOW and the project plan. Verification compares the data package to these
374 requirements (contract compliance) and checks for consistency and comparability of the data
375 throughout the data package and completeness of the results to ensure all necessary documen-
376 tation is available. Performance criteria for verification should be documented in the contract and
377 in the project plan document in the sections that address Data Review, Verification, and
378 Validation Requirements (Dl), and Verification and Validation Methods (D2).
379 4.6.2.2 Data Validation
380 Validation addresses the reliability of the data. During validation, the technical reliability and the
381 degree of confidence in reported analytical data are considered. Data validation criteria and
382 procedures should be established during the planning process and captured in the project plan
383 document (and the SOW for the validation contractor). Performance criteria for data validation
384 can be documented directly in the project plan document in Data Review, Verifications, and
385 Validation Requirements (Dl) and Verifications and Validation Methods (D2) or in a separate
386 plan, which is included by citation or as an appendix in the project plan document.
387 Guidance on Data Validation Plans is provided in Chapter 8, Section 8.3. The data validation
388 plan should contain the following information:
389 • A summary of the project, which provides sufficient detail about the project's Analytical
390 Protocol Specifications, including the MQOs;
MARLAP JULY 2001
DO NOT CITE OR QUOTE 4-16 DRAFT FOR PUBLIC COMMENT
-------
Project Plan Documents
391 • The set of data to be validated and whether all the raw data will be reviewed and in what
392 detail;
393 • The necessary validation criteria and the MQOs deemed appropriate for achieving project
394 DQOs;
395 • Specifications on what qualifiers are to be used and how final qualifiers are to be assigned;
396 and
397 • Information on the content of the validation report.
398 4.6.2.3 Data Quality Assessment
399 Data Quality Assessment consists of a scientific and statistical evaluation of project-wide
400 knowledge to determine if the data set is of the right type, quality and quantity to support its
401 intended use. The data quality assessor integrates the data validation report, field information,
402 assessment reports and historical project data and compares the findings to the original project
403 objectives and criteria (DQOs).
404 Performance criteria for data usability for the project should be documented in the project plan
405 documents in a section on DQA or reconciliation of the data results with DQOs (D3) or in a
406 separate plan, which is included by citation or as an appendix in the project plan document.
407 Guidance on DQA Plans is provided in Section 9.5. The DQA plan should contain the following
408 information:
409 • A summary of the project, which provides sufficient detail about the project's DQOs and
410 tolerable decision error rates;
411 • Identification of what issues will be addressed by the DQA;
412 • Identification of any statistical tests that will be used to evaluate the data;
413 • Description of how the representativeness of the data will be evaluated (for example, review
414 the sampling strategy, the suitability of sampling devices, subsampling procedures,
415 assessment findings);
416 • Description of how the accuracy of the data, including potential impact of non-measurable
417 factors (for example, subsampling bias) will be considered (for example, review the
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 4-17 DO NOT CITE OR QUOTE
-------
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
Analytical Protocol Specifications and the analytical plan, the suitability of analytical
protocols, subsampling procedures, assessment findings);
Description of how the MQOs will be used to determine the usability of measurement data
(that is, did the uncertainty in the data significantly affect confidence in the decision);
Identification of what will be included in the DQA report; and
Identification of who will receive the report and the mechanism for its archival.
Summary of Recommendations
• A graded approach to project plan writing because of the diversity of environmental data
collection activities.
• A primary integrating project plan that includes other documents by citation or as
appendices.
• Project plan documents that integrate all technical and quality aspects for the life-cycle of
the project, including planning, implementation, and assessment.
• Inclusion, by citation or as an appendix, of the report on the directed planning process in
the project plan documents. If the planning process was not documented in a report,
MARLAP recommends that a summary of the planning process addressing assumptions
and decisions, established action levels, DQO statement and established MQOs, and
Analytical Protocol Specifications be included in the project plan documents.
4.7 References
437 American National Standards Institute and the American Society for Quality Control
438 (ANSI/ASQC). 1994. Specifications and Guidelines for Quality Systems for Environmental
439 Data Collection and Environmental Technology Programs, National Standard E-4.
440 American National Standards Institute (ANSI). 1996. Measurement and Associated Instruments
441 Quality Assurance for Radioassay Laboratories, National Standard N42.23.
442 American Society of Mechanical Engineers (ASME). 1989. Quality Assurance Program
443 Requirements for Nuclear Facilities. NQA-1, ASME, New York, New York.
MARLAP
DO NOT CITE OR QUOTE
4-18
JULY 2001
DRAFT FOR PUBLIC COMMENT
-------
Project Plan Documents
444 American Society of Testing and Materials (ASTM). 1992. Standard Practice for Generation of
445 Environmental Data Related to Waste Management Activities: Quality Assurance and Quality
446 Control Planning and Implementation., D 5 2 8 3.
447 American Society of Testing and Materials (ASTM). 1994. Standard Guide for Quality Planning
448 and Field Implementation of a Water Quality Measurements Program, D5612.
449 American Society of Testing and Materials (ASTM). 1996. Standard Provisional Guidance for
450 Expedited Site Characterization of Hazardous Waste Contaminated Sites, PS85.
451 Code of Federal Regulations (CFR). 1999. 10 CFR 50 Appendix B, "Quality Assurance Criteria
452 for Nuclear Power Plants and Fuel Reprocessing Plants."
453 Code of Federal Regulations (CFR). 1994. 10 CFR 830.120, "Nuclear Safety Management -
454 Quality Assurance Requirements."
455 Code of Federal Regulations (CFR). 1997. 40 CFR 300.430, "National Oil and Hazardous
456 Substance Pollution Contingency Plan - Remedial Investigation/Feasibility Study and
457 Selection of Remedy."
458 International Organization for Standardization (ISO). 1994. Quality Systems -Modelfor Quality
459 Assurance in Design, Development, Installation and Servicing, ISO Standard 9001.
460 U.S. Army Corps of Engineers (USAGE). 1994. Requirements for the Preparation of Sampling
461 and Analysis Plans. Engineer Manual EM 200-1-3.
462 U.S. Army Corps of Engineers (USAGE). 1997. Chemical Quality Assurance for Hazardous,
463 Toxic and Radioactive Waste Projects. Engineer Manual EM 200-1-6.
464 U.S. Department of Defense (DOD). 1963. Quality Program Requirements. Military
465 Specification MIL-Q-9858A. Washington, DC.
466 U.S. Department of Energy (DOE). 1991. Quality Assurance. DOE Order 414.1 (Replaced DOE
467 Order 5 700. 6C), Washington, DC.
468 U.S. Environmental Protection Agency (EPA). 1998a. EPA Guidance for Quality Assurance
469 Project Plans (EPA QA/G-5). EPA/600/R-98/018, Washington, DC.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 4-19 DO NOT CITE OR QUOTE
-------
470 U.S. Environmental Protection Agency (EPA). 1998b. EPA Requirements for Quality Assurance
471 Project Plans for Environmental Data Operations. EPA QA/R-5, External Review Draft Final,
472 Washington, DC.
473 U.S. Environmental Protection Agency (EPA). 1998c. EPA Policy and Program Requirements
474 for the Mandatory Agency-Wide Quality System. EPA Order 5360.1, Washington, DC.
475 U.S. Nuclear Regulatory Commission (NRC). 1989. Standard Format and Content of
476 Decommissioning Plans for Licensees Under 10 CFR Parts 30, 40, and 70. Regulatory Guide
477 3.65.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 4-20 DRAFT FOR PUBLIC COMMENT
-------
5 OBTAINING LABORATORY SERVICES
2 5.1 Introduction
3
4 This chapter provides guidance on obtaining radioanalytical laboratory services. In particular,
5 this chapter discusses the broad items that should be considered in the development of a
6 procurement vehicle to obtain laboratory services. Throughout this chapter, MARLAP uses the
7 request for proposal (RFP) as an example of a procurement vehicle. Agencies and other
8 organizations may use a variety of procurement vehicles, depending upon circumstances and
9 policies. The RFP typically includes a statement of work (SOW), generic contractual
10 requirements, and the description of the laboratory qualification and selection process. It should
11 be noted that for some agencies or organizations, not all technical, quality, and administrative
12 aspects of a contract are specified in a SOW. Many technical, administrative, legal, and
13 regulatory items are specified in a RFP and eventually in a contract. More detailed guidance and
14 discussion for contracting issues (such as scoring proposals, etc.) can be found in Appendix E.
15 This chapter is written for contracting outside laboratory services, but the principal items and
16 information provided would be applicable to obtaining services not requiring a formal contract,
17 such as a service agreement within an Agency or organization. It should be noted that the
18 information and specifications of a SOW may appear in many contract vehicles other than a
19 formal contract resulting from a RFP. These include purchase and work orders, as well as a task
20 order under a Basic Ordering Agreement. MARLAP recommends that technical specifications be
21 prepared in writing in a single document designated as a SOW for all radioanalytical laboratory
22 services, regardless of whether the services are to be contracted out or performed by an
23 Agency's laboratory.
24
25 Analytical Protocol Specifications (APSs) should be compiled in the SOW in order for the
26 laboratory to propose the analytical protocols that the laboratory wishes to use for the project
27 (Chapter 6). The development of APSs, which includes the measurement quality objectives
28 (MQOs), was described in detail in Chapter 3, and the incorporation of these protocols into the
29 relevant project plan documents was covered in Chapter 4. These specifications should include
30 such items as the MQOs, the type and frequency of quality control (QC) samples, the level of
31 performance demonstration needed, number and type of samples, turnaround times, and type of
32 data package.
33
34 Section 5.3 of this chapter discusses the technical requirements of a SOW, Section 5.4 provides
35 guidance on generic contractual requirements, and Section 5.5 discusses various elements of the
36 laboratory selection and qualification criteria.
37
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 5-1 DO NOT CITE OR QUOTE
-------
Obtaining Laboratory Services
38 5.2 Importance of Writing a Technical and Contractual Specification
39 Document
40
41 One objective of the SOW and contractual documents is to provide the analytical requirements in
42 a concise format that will facilitate the laboratory's selection of the appropriate analytical
43 protocols. The authors of the SOW may be able to extract most, if not all, of the necessary
44 technical information from the project plan documents (Chapter 4) if they have been prepared
45 properly. If specific information is not available, the author should contact the planning team.
46 The preparation of a SOW can be viewed as a check to make sure that the project planning
47 documents contain all the information required for the selection and implementation of the
48 appropriate analytical protocols. One important aspect of writing the SOW is that it should
49 clearly identify the project laboratory's responsibility for documentation to be provided for
50 subsequent data verification, validation, and quality assessment—these project laboratory
51 requirements should be addressed in the assessment plans developed during directed planning
52 (Chapter 2).
53
54 5.3 Statement of Work — Technical Requirements
55
56 A review of the project plan documents (Chapter 4) should result in a summary list of the
57 technical requirements needed to develop a SOW. Much of this information, including the
58 project MQOs and any unique analytical process requirements, will be contained in the APSs.
59 When possible, a project summary of sufficient detail (i.e., process knowledge) to be useful to
60 the laboratory should be included in the SOW. The Project Planning Team is responsible for
61 identifying and resolving key analytical planning issues and for ensuring that the resolutions of
62 these issues are captured in the APSs. Consistent with a performance-based approach, the level
63 of specificity in the APSs is limited to those requirements that are essential to meeting the
64 project's analytical data requirements. In response to such project management decisions, the
65 laboratory may propose for consideration several alternative validated methods that meet the
66 MQOs under the performance-based approach (such as measurement of a decay progeny as an
67 alternate radionuclide). Chapter 7 provides guidance on the evaluation of a laboratory and
68 analytical methods.
69
70 The SOW should specify what the laboratory needs to provide in order to demonstrate its ability
71 to meet the technical specifications in the RFP. This should include documentation relative to the
72 method validation process to demonstrate compliance with the MQOs and information on
73 previous contracts for similar analytical work as well as performance in performance evaluation
74 (PE) programs using the proposed method. Any specific requirements on sample delivery
MARLAP JULY 2001
DO NOT CITE OR QUOTE 5 -2 DRAFT FOR PUBLIC COMMENT
-------
Obtaining Laboratory Services
75 (Section 5.3.7) should also be made clear to the laboratory. In addition, the requirements for the
76 laboratory's quality system should be discussed.
77
78 5.3.1 Analytes
79
80 Each APS should state the analyte of concern. The SOW should specify all analytes of concern
81 and, when possible, an analyte's expected chemical form and anticipated concentration range
82 (useful information for separating high activity samples from low activity samples) and potential
83 chemical or radiometric interferences (Chapter 3, Sections 3.3.1 and 3.3.2). In some instances,
84 because of process knowledge and information on the absence of equilibrium between analytes
85 and their parents and progeny, the SOW may require the direct measurement of an analyte rather
86 than allowing for the measurement of other radionuclides in the analyte's decay chain. In these
87 cases, the SOW should indicate the analyses to be performed. Examples of analyses include gross
88 alpha and beta, gamma spectrometry, and radionuclide/matrix specific combinations such as 3H
89 in water and 238Pu in soil.
90
91 5.3.2 Matrix
92
93 Each APS should state the sample matrix to be analyzed. The sample matrix for each
94 radionuclide or analysis type (e.g., gamma-ray spectrometry) should be listed and described in
95 detail where necessary. The matrix categories may include surface soil, sub-surface soil,
96 sediment, sludge, concrete, surface water, ground water, salt water, aquatic and terrestrial biota,
97 air, air sample filters, building materials, etc. Additional information should be provided for
98 certain matrices (e.g., the chemical form of the matrix for solid matrices) in order for the
99 laboratory to select the appropriate sample preparation or dissolution method (Chapter 3, Section
100 3.3.3).
101
102 5.3.3 Measurement Quality Objectives
103
104 The APSs should provide the MQOs for each analyte-matrix combination. The MQOs can be
105 viewed as the analytical portion of the overall project data quality objectives (DQOs). An MQO
106 is a statement of a performance objective or requirement for a particular method performance
107 characteristic. Examples of method performance characteristics include the method's uncertainty
108 at some concentration, detection capability, quantification capability, specificity, analyte
109 concentration range, and ruggedness. An example MQO for the method uncertainty at some
110 analyte concentration such as the action level would be, "A method uncertainty of 0.5 Bq/g is
111 required at the action level of 5.0 Bq/g" (Chapters 1, 3, and 19). The MQOs are a key part of a
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 5-3 DO NOT CITE OR QUOTE
-------
Obtaining Laboratory Services
112 project's APSs. Chapter 3 provides guidance on developing MQOs for select method
113 performance characteristics.
114
115 5.3.4 Unique Analytical Process Requirements
116
117 The APS should state any unique analytical processing requirement. The SOW should give any
118 matrix-specific details necessary for the laboratory to process the sample, such as type of soil,
119 type of debris to be removed, whether or not filtering a sample at the laboratory is required,
120 processing whole fish versus edible parts, drying of soils, information on any known or suspected
121 interferences, hazards associated with the sample, etc. (Chapter 3, Section 3.4). In some cases,
122 unique analytical process requirements or instructions should be specified that further delineate
123 actions to be taken in case problems occur during sample processing. For example, the SOW may
124 require that the laboratory reprocess another aliquant of the sample by a more robust technique
125 when a chemical yield drops below a stated value.
126
127 If necessary, special instructions should be provided as to how or when the analytical results are
128 to be corrected for radioactive decay or ingrowth. In some cases, the sample collection date may
129 not be the appropriate date to use in the decay or ingrowth equations.
130
131 5.3.5 Quality Control Samples and Participation in External Performance Evaluation
132 Programs
133
134 The SOW should state the type and frequency of internal QC samples needed as well as whether
135 they are to be included on a batch or some other basis. The quality acceptance limits for all types
136 of QC samples should be stated (see Appendix E for guidance on developing acceptance limits
137 for QC samples based on the MQO for method uncertainty). In addition, the SOW should state
138 when and how the project manager or the contracting officer's representative (COR) should be
139 notified about any nonconformity. In addition, the SOW should spell out the conditions under
140 which the laboratory will have to re-analyze samples due to a nonconformance.
141
142 The evaluation of the laboratory's ability to perform the required radiochemical analyses should
143 be based on the acceptability of the method validation documentation submitted by the
144 laboratory. The evaluation should also include the laboratory's performance in various external
145 PE programs administered by government agencies or commercial radioactive source suppliers
146 that are traceable to the National Institute of Standards and Technology (NIST; additional
147 information on evaluating a laboratory's performance is provided in Chapter 7). As such, the
148 RFP should request the laboratory's participation in a NIST-traceable PE program appropriate for
149 the analytes and matrices under consideration. In addition, the weighting factor (Appendix E)
MARLAP JULY 2001
DO NOT CITE OR QUOTE 5 -4 DRAFT FOR PUBLIC COMMENT
-------
Obtaining Laboratory Services
150 given to scoring the laboratory's performance in such a program should be provided to the
151 laboratory. Some examples of government programs include DOE's Quality Assessment
152 Program (QAP) and the Mixed Analyte Performance Evaluation Program (MAPEP) and the
153 NIST-administered National Voluntary Laboratory Accreditation Program (NVLAP)
154 Performance Testing (PT) providers.
155
156 5.3.6 Laboratory Radiological Holding and Turnaround Times
157
158 The SOW should include specifications on the required laboratory radiological holding time (i.e.,
159 the time between the date of sample collection and the date of analysis) and the sample
160 processing turnaround time (i.e., the time between the receipt of the sample at the laboratory to
161 the reporting of the analytical results). Such radiological holding and turnaround times, which are
162 usually determined by specific project requirements, are typically specified in terms of calendar
163 or working days. The SOW should state whether the laboratory may be requested to handle
164 expedited or rush samples. In some cases, time constraints become an important aspect of sample
165 processing (e.g., in the case of radionuclides that have short half-lives). Some analyses will call
166 for specific steps that take a prescribed amount of time. Requesting an analytical protocol that
167 requires several days to complete is obviously not compatible with a 24-hour turnaround time.
168 This highlights the need for input from radioanalytical specialists during the planning process.
169
170 In some cases, the required sample-processing turnaround times are categorized according to
171 generic headings such as routine, expedited or rush, and emergency sample processing. Under
172 these circumstances, the SOW should specify the appropriate category for the samples and
173 analyses.
174
175 5.3.7 Number of Samples and Schedule
176
177 Estimating the volume of work for a laboratory is commonly considered part of the planning
178 process that precedes the initiation of a project. Thus, the SOW should estimate the anticipated
179 amount of work and should spell out the conditions under which the laboratory will have to
180 reanalyze samples due to some non-conformance. Similarly, the estimate should allow the
181 laboratory to judge if its facility has the capacity to compete for the work. The estimate for the
182 number of samples is a starting point, and some revision to the volume of work may occur,
183 unless the laboratory sets specific limits on the number of samples to be processed.
184
185 The SOW should indicate whether samples will be provided on a regular basis, seasonally, or on
186 some other known or unknown schedule. It should also be specified if some samples may be sent
187 by overnight carrier for immediate analysis. Holidays may be listed when samples will not be
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 5-5 DO NOT CITE OR QUOTE
-------
Obtaining Laboratory Services
188 sent to the laboratory. The SOW should state if Saturday deliveries may be required.
189 Furthermore, it should specify whether samples will be sent in batches or individually, and from
190 one location or different locations.
191
192 The carrier used to ship samples to the laboratory should be experienced in the delivery of field
193 samples, provide next day and Saturday deliveries, have a package tracking system and be
194 familiar with hazardous materials shipping regulations.
195
196 5.3.8 Quality System
197
198 The RFP should require that a copy of the laboratory's Quality System documentation (such as a
199 Quality Manual), related standard operating procedures (including appropriate methods) and
200 documentation (such as a summary of the internal QC and external PE sample results) be
201 included with the proposal submittal, as necessary. Only those radioanalytical laboratories that
202 adhere to a well-defined quality system can ensure the appropriate quality of scientifically valid
203 and defensible data. The laboratory's Quality System (NELAC, 2000; ANSI N42.23; ISO/IEC
204 17025) for a radioanalytical laboratory should address at a minimum the following items:
205
206 • Organization and management;
207 • Quality system establishment, audits, essential quality controls and evaluation and data
208 verification;
209 • Personnel (qualifications and resumes);
210 • Physical facilities—accommodations and environment;
211 • Equipment and reference materials;
212 • Measurement traceability and calibration;
213 • Test methods and standard operating procedures (methods);
214 • Sample handling, sample acceptance policy and sample receipt;
215 • Records;
216 • Subcontracting analytical samples;
217 • Outside support services and supplies; and
218 • Complaints.
219
220 5.3.9 Laboratory's Proposed Methods
221
222 Under the performance-based approach to method selection, the laboratory will select and
223 identify a radioanalytical methods (Chapter 6) that will meet the MQOs and other performance
224 specifications of the SOW. MARLAP recommends that the laboratory submit the proposed
225 methods and required method validation documentation with the formal response. The SOW
MARLAP JULY 2001
DO NOT CITE OR QUOTE 5 -6 DRAFT FOR PUBLIC COMMENT
-------
Obtaining Laboratory Services
226 should state that the proposed methods and method validation documentation will be evaluated in
227 accordance with agency procedures by a Technical Evaluation Committee (TEC) based on
228 experience, expertise, and professional judgement. MARLAP uses the term TEC for the group
229 that performs this function. Agencies and other organizations may use various terms and
230 procedures for this process.
231
232 The TEC should provide their findings and recommendations to the organization's contracting
233 officer for further disposition. In some cases, the organization may inform a laboratory that the
234 proposed methods were deemed inadequate, and, if appropriate, request that the laboratory
235 submit alternative methods with method validation documentation within a certain time period.
236
237 When the methods proposed by the laboratories have been deemed adequate to meet the technical
238 specifications of the SOW, the TEC may want to rank the proposed methods (and laboratories)
239 according to various factors (e.g., robustness, performance in PE programs or qualifying samples,
240 etc.) as part of the contract scoring process.
241
242 5.4 Request for Proposal—Generic Contractual Requirements
243
244 Not all quality and administration aspects of a contract are specified in a SOW. Many quality
245 (e.g., requirement for a quality system), administrative, legal, and regulatory items need to be
246 specified in a RFP and eventually in the contract. Although not inclusive, the items or categories
247 discussed in the following sections should be considered as part of the contractual requirements
248 and specifications of a RFP.
249
250 5.4.1 Sample Management
251
252 The RFP should require the laboratory to have an appropriate sample management program that
253 includes those administrative and quality assurance aspects covering sample receipt, control,
254 storage and disposition. The RFP should require the laboratory to have adequate facilities,
255 procedures, and personnel in place for the following actions:
256
257 • Receive, log-in, and store samples in a proper fashion to prevent deterioration, cross-
258 contamination, and analyte losses;
259
260 • Verify the receipt of each sample shipment: compare shipping documentation with samples
261 actually received; notify the point of contact or designee by telephone within a prescribed
262 number of business days and subsequently provide details in all case narratives of any
263 discrepancies in the documentation;
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 5-7 DO NOT CITE OR QUOTE
-------
Obtaining Laboratory Services
264
265 • Sign, upon receipt of the samples, the sample receipt form or, if required, chain of custody
266 (COC) form(s) submitted with each sample release. Only authorized laboratory personnel
267 should sign the forms. The signature date on the COC form, if required, is normally the
268 official sample receipt date. All sample containers should be sealed prior to their removal
269 from the site; and
270
271 • Store unused portions of samples in such a manner that the analyses could be repeated or new
272 analyses requested, if required, for a certain specified time period following the submission
273 of an acceptable data package. Unused sample portions should be stored with the same
274 sample handling requirements that apply to samples awaiting analysis. Documentation should
275 be maintained pertaining to storage conditions and sample archival or disposal.
276
277 5.4.2 Licenses, Permits and Environmental Regulations
278
279 Various Federal, State, and local permits, licences and certificates (accreditation) may be
280 necessary for the operation of a radioanalytical laboratory. The RFP should require the laboratory
281 to have the necessary government permits, licenses, and certificates in place before the
282 commencement of any laboratory work for an awarded contract. The following sections provide a
283 partial list of those provisions that may be necessary. Some projects may require special
284 government permits in order to conduct the work and transport and analyze related samples. For
285 these cases, the necessary regulations or permits should be cited in the RFP.
286
287 5.4.2.1 Licenses
288
289 When required, the laboratory will be responsible for maintaining a relevant Nuclear Regulatory
290 Commission (NRC) or Agreement State License to accept low-level radioactive samples for
291 analyses. In certain circumstances, the laboratory may have to meet host nation requirements if
292 operating outside the United States (e.g., military fixed or deployed laboratories located
293 overseas).
294
295 When necessary, the laboratory should submit a current copy of the laboratory's radioactive
296 materials license with their proposal. Some circumstances may require a copy of the original
297 radioactive materials license. For more complete information on license requirements, refer to
298 either the NRC or State government offices in which the laboratory resides, or to 10 CFR 30.
299
MARLAP JULY 2001
DO NOT CITE OR QUOTE 5-8 DRAFT FOR PUBLIC COMMENT
-------
Obtaining Laboratory Services
300 5.4.2.2 Environmental and Transportation Regulations
301
302 Performance under a contract or subcontract must be in compliance with all applicable local,
303 State, Federal, and international laws and regulations. Such consideration must not only include
304 relevant laws and regulations currently in effect, but also revisions thereto or public notice that
305 has been given that may reasonably be anticipated to be effective during the term of the contract.
306
307 The laboratory may be required to receive (and in some cases ship) samples according to
308 international, Federal, State, and local regulations. In particular, the laboratory should be aware
309 of U.S. Postal Service and Department of Transportation (DOT) hazardous materials regulations
310 applicable to the requirements specified in the SOW and aware that appropriate personnel should
311 be trained in these regulations.
312
313 5.4.3 Data Reporting and Communications
314
315 The type of information, schedules and data reports required to be delivered by the laboratory, as
316 well as the expected communications between the appropriate staff or organizations, should be
317 delineated in the RFP. The required schedule and content of the various reports, including sample
318 receipt acknowledgment, chain of custody, final data results, data packages, QA/QC project
319 summaries, status reports, sample disposition, and invoices should be provided in the RFP. In
320 addition, the expected frequency and lines of communications should be specified.
321
322 In some cases, the RFP may request relevant information relative to the point-of-contact for
323 certain key laboratory positions such as the Laboratory Director, Project Manager, QA Officer,
324 Sample Manager, Record Keeping Supervisor, Radiation Safety or Safety Officer and
325 Contracting Officer. Contact persons should be identified along with appropriate telephone
326 numbers (office, FAX, pager), e-mail, and postal and courier addresses.
327
328 5.4.3.1 Data Deliverables
329
330 The SOW should specify what data are required for data verification, validation, and quality
331 assessment. A data package, the pages of which should be sequentially numbered, may include a
332 project narrative, the results in a specified format including units, a data review checklist, any
333 non-conformance memos resulting from the work, sample receipt acknowledgment or chain of
334 custody form (if required), sample and quality control sample data, calibration verification data,
335 and standard and tracer information. In addition, the date and time of analysis, instrument
336 identification, and analyst performing the analysis should be included on the appropriate
337 paperwork. At the inception of the project, initial calibration data may be required for the
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 5-9 DO NOT CITE OR QUOTE
-------
Obtaining Laboratory Services
338 detectors used for the work. When a detector is recalibrated, or a new detector is placed in
339 service, updated calibration data should be required whenever those changes could affect the
340 analyses in question. In some cases, only the summary or final data report may be requested. In
341 these cases, the name of the data reviewer, the sample identification information, reference and
342 analysis dates, and the analytical results along with the reported measurement uncertainties
343 should be reported.
344
345 The laboratory should be informed of the acceptable formats for electronic and hard copy
346 records. The SOW should state at what intervals the data will be delivered (batch, monthly, etc.).
347
348 5.4.3.2 Software Verification and Control
349
350 The policy for computer software verification, validation and documentation typically are
351 included in the laboratory's Quality Manual. If there are specific software verification and
352 validation requirements germane to the project, the RFP should instruct or specify such
353 requirements. ASTM E919, "Standard Specification for Software Documentation for a
354 Computerized System," describes computer program documentation that should be provided by a
355 software supplier. Other sources for software QC are ANSI ANS 10.3 "Documentation of
356 Computer Software" and IEEE Standard 1063, "IEEE Standard for Software User
357 Documentation."
358
359 5.4.3.3 Problem Notification and Communication
360
361 Communication is key to the successful management and execution of the contract. Problems,
362 schedule delays, potential overruns, etc., can be resolved quickly only if communication between
363 the laboratory and organization's representative is conducted promptly. The RFP should state
364 explicitly when, how, and in what time frame communication or notification is required by the
365 laboratory for special technical events, such as the inability to meet MQO specifications for a
366 sample or analyte, when a QC sample result is outside of an acceptance limit or some other non-
367 conformance and when—if required by the project manager—the laboratory fails to meet its
368 internal QC specifications.
369
370 The laboratory should document and report all deviations from the method and unexpected
371 observations that may be of significance to the data reviewer or user. Such deviations should be
372 documented in the narrative section of the data package produced by the contract laboratory.
373 Each narrative should be monitored closely to assure that the laboratory is documenting
374 departures from contract requirements or acceptable practice.
375
MARLAP JULY 2001
DO NOT CITE OR QUOTE 5-10 DRAFT FOR PUBLIC COMMENT
-------
Obtaining Laboratory Services
376 Communication from the organization's representative to the laboratory is also important. A key
377 element in managing a contract is the timely review of the data packages provided by the
378 laboratory. Early identification of problems allows for corrective actions to improve laboratory
379 performance and, if necessary, the cessation of laboratory analyses until solutions can be
380 instituted to prevent the production of large amounts of data that are unusable. Note that some
381 sample matrices and processing methods can be problematic for even the best laboratories. Thus,
382 the organization's technical representative must be able to discern between failures due to
383 legitimate reasons and poor laboratory performance.
384
385 5.4.3.4 Status Reports
386
387 The SOW may require the laboratory to submit, on a specified frequency, sample processing
388 status reports that include such information as the sample identification number, receipt date,
389 analyses required, expected analytical completion date and report date. Depending on the
390 project's needs, a status report may include the disposition of remaining portions of samples
391 following sample processing or sample processing wastes.
392
393 5.4.4 Sample Re-Analysis Requirements
394
395 There may be circumstances when samples should be re-analyzed due to questionable analytical
396 results or suspected poor quality as reflected by the laboratory's batch QC or external PT
397 samples. Specific instructions and contractual language should be included in the RFP that
398 address such circumstances and the resultant fiscal responsibilities (Appendix E).
399
400 5.4.5 Subcontracted Analyses
401
402 MARLAP recommends that the RFP state that subcontracting will be permitted only with the
403 contracting organization's approval. In addition, contract language should be included giving the
404 contracting organization the authority to approve proposed subcontracting laboratories. For
405 continuity or for quality assurance, the contract may require one laboratory to handle the entire
406 analytical work load. However, the need may arise to subcontract work to another laboratory
407 facility if the project calls for a large number of samples requiring quick turnaround times or
408 specific methodologies that are not part of the primary laboratory's support services. The use of
409 multiple service providers adds complexity to the organization's tasks of auditing, evaluating and
410 tracking services.
411
412 Any intent to use a subcontracted laboratory should be specified in the response to the RFP or
413 specific task orders. The primary laboratory should specify which laboratory(ies) are to be used,
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 5-11 DO NOT CITE OR QUOTE
-------
Obtaining Laboratory Services
414 should require that these laboratories comply with all contract or task order requirements, and
415 verify that their operations can and will provide data quality meeting or exceeding the SOW
416 requirements. Subcontract laboratories should be required to allow the contracting organization
417 full access to inspect their operations, although it should be understood that the primary
418 laboratory should maintain full responsibility for the performance of subcontract laboratories.
419
420 5.5 Laboratory Selection and Qualification Criteria
421
422 A description of the laboratory qualification and selection process should be stated in the RFP.
423 The initial stages of the evaluation process focus on the technical considerations only. Cost will
424 enter the selection process later. The organization's TEC will consider all proposals and then will
425 make an initial selection (see Figures E.6a and E.6b in Appendix E), whereby some laboratories
426 are eliminated based on the screening process. The laboratory selection process is based on
427 predetermined criteria that are related to the RFP and how a laboratory is technically able to
428 support the contract. A laboratory that is obviously not equipped to perform work according to
429 the RFP is certain to be dropped early in the selection process. In some cases, the stated ability to
430 meet the analysis request may be verified by the organization, through pre-award audits and
431 proficiency testing as described below. Letters notifying unsuccessful bidders may be sent at this
432 time.
433
434 5.5.1 Technical Proposal Evaluation
435
436 The RFP requires each bidding contractor laboratory to submit a technical proposal and a copy of
437 its Quality Manual. This Quality Manual is intended to address all of the technical and general
438 laboratory requirements. As noted previously, the proposal and Quality Manual are reviewed by
439 members of the TEC who are both familiar with the proposed project and are clearly
440 knowledgeable in the field of radiochemistry and laboratory management.
441
442 5.5.1.1 Scoring and Evaluation Scheme
443
444 The RFP should include information concerning scoring of proposals or weighting factors for
445 areas of evaluation. This helps a laboratory to understand the relative importance of specific
446 sections in a proposal and how a proposal will be evaluated or scored. This allows the laboratory
447 to focus on those areas of greater importance. If the laboratory submits a proposal that lacks
448 sufficient information to demonstrate support in a specific area, the organization can then
449 indicate how the proposal does not fulfill the need as stated in the request. Because evaluation
450 formats differ from organization to organization, laboratories may wish to contact the
451 organization for additional organization-specific details concerning this process. A technical
MARLAP JULY 2001
DO NOT CITE OR QUOTE 5-12 DRAFT FOR PUBLIC COMMENT
-------
Obtaining Laboratory Services
452 evaluation sheet (TES) may be used in conjunction with the Proposal Evaluation Plan as outlined
453 in the next section (see Figures E.6a and E.6b in Appendix E) to list the total weight for each
454 factor and to provide a space for the evaluator' s assigned rating. In the event of a protest, the TES
455 can be used to substantiate the selection process. The TES also provides areas to record the RFP
456 number, identity of the proposer, and spaces for total score, remarks, and evaluator's signature.
457 The scoring and evaluation scheme is based on additional, more detailed, considerations which
458 are discussed briefly in the Sections E.4 and E.5 in Appendix E.
459
460 Once all proposals are accepted by the organization, the TEC scores the technical portion of the
461 proposal. MARLAP recommends that all members of the TEC have a complete technical
462 understanding of the subject matter related to the proposed work. These individuals are also
463 responsible for responding to any challenge to the organization's selection for the award of the
464 contract. Their answers to such challenges are based on technical merit in relation to the
465 proposed work.
466
467 5.5.1.2 Scoring Elements
468
469 Although each organization may have a different scoring process to evaluate a laboratory's
470 response to a RFP, there are various broad categories or common elements that are typically
471 evaluated. For example, these may include the following:
472
473 • Technical merit;
474 • Adequacy and suitability of laboratory resources and equipment;
475 • Staff qualifications;
476 • Related experience and record of past performance; and
477 • Other RFP requirements.
478
479 Although each organization may score or weight these items differently, performance-based
480 contracting requires the weighting of past performance of the contractor as a significant technical
481 element. Each of these elements is considered in the following paragraphs. Outlined below are
482 the key elements that are discussed in more detail in Appendix E.
483
484 TECHNICAL MERIT
485
486 The response to the RFP should include details of the laboratory's Quality System and all the
487 analytical methods to be employed by the laboratory as well as the method validation
488 documentation (Section 6.6). The information provided should outline or demonstrate that the
489 methods proposed are likely to be suitable and meet the APSs. The methods should be evaluated
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 5-13 DO NOT CITE OR QUOTE
-------
Obtaining Laboratory Services
490 against the APSs and MQOs provided in the SOW. Chapter 7 provides guidance on the
491 evaluation of methods and laboratories. The laboratory's Quality Manual should be reviewed for
492 adequacy and completeness to ensure the required data quality.
493
494 ADEQUACY AND SUITABILITY OF LABORATORY RESOURCES AND EQUIPMENT
495
496 When requested, the laboratory will provide a listing of the available instrumentation or
497 equipment by analytical method category. In addition, the RFP may request information on the
498 available sample processing capacity and the workload for other clients during the proposed
499 contract period. The information provided should be evaluated by the TEC to determine if the
500 laboratory has the sample processing capacity to perform the work. The instrumentation and
501 equipment must be purchased, set-up, calibrated, and on-line before award of contract. In
502 addition, the laboratory should provide information relative to the adequacy and suitability of the
503 laboratory space available for the analysis of samples.
504
505 STAFF QUALIFICATIONS
506
507 The RFP should require the identification of the technical staff and their duties, along with their
508 educational background and experience in radiochemistry, radiometrology or laboratory
509 operations. The laboratory staff that will perform the radiochemical analyses should be employed
510 and trained prior to the award of the contract. Appendix E provides guidance on staff
511 qualifications.
512
513 RELATED EXPERIENCE AND RECORD OF PAST PERFORMANCE
514
515 The RFP should require the laboratory to furnish references in relation to its past or present work.
516 To the extent possible, this should be done with regard to contracts or projects similar in
517 composition, duration and number of samples to the proposed project. In some cases, the
518 laboratory's previous performance for the same Agency may be given special consideration.
519
520 OTHER RFP REQUIREMENTS
521
522 Within the response to the RFP, the laboratory should outline the various programs and
523 commitments (QA, safety, waste management, etc.) as well as submit various certifications,
524 licences and permits to ensure the requirements of the RFP will be met. The reasonableness of
525 the proposed work schedule, program and commitments should be evaluated by the TEC. In
526 addition, if accreditation is required in the RFP, the TEC should confirm the laboratory's
527 accreditation for radioanalytical services by contacting the organization that provided the
MARLAP JULY 2001
DO NOT CITE OR QUOTE 5-14 DRAFT FOR PUBLIC COMMENT
-------
Obtaining Laboratory Services
528 certification. If State accredited, a laboratory is typically accredited by the State in which it
529 resides. If the organization expects a laboratory to process samples from numerous States across
530 the United States, then additional accreditations for other States may be required. The TEC
531 should review and confirm the applicability and status of the licenses and permits with respect to
532 the technical scope and duration of the project.
533
534 5.5.2 Pre-Award Proficiency Evaluation
535
536 Some organizations may elect to send proficiency or PT samples (also referred to as "perfor-
537 mance evaluation" samples) to the laboratories that meet a certain scoring criteria in order to
538 demonstrate the laboratory's analytical capability. The composition and number of samples
539 should be determined by the nature of the proposed project. The PT sample matrix should be
540 composed of well-characterized materials. It is recommended that site specific PT matrix
541 samples or method validation reference material (MVRM, See Chapter 6) be used when
542 available.
543
544 Each competing lab should receive an identical set of PE samples. The RFP should specify who
545 will bear the cost of analyzing these samples as well as the scoring scheme, e.g., pass/fail or a
546 sliding scale. Any laboratory failing to submit results should be disqualified. The results should
547 be evaluated and each laboratory given a score. This allows the organization to make a second
548 cut—after which only two or three candidate laboratories are considered.
549
550 5.5.3 Pre-Award Assessments and Audits
551
552 The RFP should indicate that the laboratories with the highest combined scores for technical
553 proposals and proficiency samples may be given an on-site audit. A pre-award assessment or
554 audit may be performed to provide assurance that a selected laboratory is capable of fulfilling the
555 contract in accordance with the RFP (Appendix E). In other words, is the laboratory's represen-
556 tation on paper (i.e., proposal) realistic when compared to the actual facilities? To answer this
557 question, auditors should be looking to see that a candidate laboratory appears to have all the
558 required elements to meet the proposed contract's needs. Refer to Appendix E for details on the
559 pre-award assessments and audits.
562
563
564
565
566
Summary of Recommendations
MARLAP recommends that technical specifications be prepared in writing in a single
document designated as a SOW for all radioanalytical laboratory services, regardless of
whether the services are to be contracted out or performed by an Agency's laboratory.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 5-15 DO NOT CITE OR QUOTE
-------
MARLAP recommends that the laboratory submit the proposed methods and required
method validation documentation with the formal response.
MARLAP recommends that the RFP state that subcontracting will be permitted only with
the contracting organization's approval.
MARLAP recommends that all members of the TEC have a complete technical
understanding of the subject matter related to the proposed work.
Obtaining Laboratory Services
567
568
569
570
571
572
573
574
575
576
577
578 5.6 References
579
580 5.6.1 Cited References
581
582 American National Standard Institute (ANSI) N42.23. Measurement and Associated
583 Instrumentation Quality Assurance for Radioassay Laboratories. 1996.
584
585 American National Standard Institute (ANSI) ANS 10.3. Documentation of Computer Software.
586
587 American Society for Testing and Materials (ASTM) E919. Standard Test Methods for Software
588 Documentation for a Computerized System.
589
590 U.S. Environmental Protection Agency (EPA). 1998. Guidance on Quality Assurance Project
591 Plans (EPA QA/G-5). EPA/600/R-98/018, Washington, DC. Available from www.epa.gov/
592 quality l/qs-docs/g5-final.pdf.
593
594 International Electrical and Electronics Engineers (IEEE). Standard 1063. Software User
595 Documentation.
596
597 International Standards Organization/International Electrotechnical Commission (ISO/IEC)
598 17025. General Requirements for the Competence of Testing and Calibration Laboratories.
599 December 1999, 26 pp.
600
601 National Environmental Laboratory Accreditation Conference (NELAC). 2000. Quality Systems.
602 July. Available from: http://www.epa.gov/ttn/nelac/.
603
MARLAP JULY 2001
DO NOT CITE OR QUOTE 5-16 DRAFT FOR PUBLIC COMMENT
-------
Obtaining Laboratory Services
604 5.6.2 Other Sources
605
606 U.S. Department of Energy (DOE). Order 414.1-1: Implementation Guide for Use with Indepen-
607 dent and Management Assessment Requirements of 10 CFR Part 830.120 and DOE 5700.6c
608 Quality Assurance. August. Available from www.directives.doe.gov/pdfs/doe/doetext/
609 neword/414/g4141-l.
610
611 U. S. Department of Energy (DOE). 1997. Model Statement of Work for Analytical Laboratories.
612 Albuquerque Operations Office, Prepared by AGRA Earth and Environmental, Inc.,
613 Albuquerque, NM. March.
614
615 U.S. Nuclear Regulatory Commission (NRC). 1996. NRC Acquisition of Supplies and Services.
616 Directive 11.1, July 23.
617
618 U.S. Nuclear Regulatory Commission (NRC). 1994. NRC Procedures for Placement and
619 Monitoring of Work With the U. S. Department of Energy (DOE). Directive 11.7, May 3.
620
621 Office of Federal Procurement Policy (OFPP) 1997. Performance-Based Service Contracting
622 (PBSC) Solicitation/Contract/Task Order Review Checklist. August 8. Available from: http://
623 www.arnet.gov/Library/OFPP/PolicyDocs/pbscckls.html.
624
625
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 5-17 DO NOT CITE OR QUOTE
-------
i 6 SELECTION AND APPLICATION OF AN
2 ANALYTICAL METHOD
3 6.1 Introduction
4 This chapter provides guidance to both the project manager and the laboratory on the selection
5 and application of analytical method. It offers guidance to the project manager on the develop-
6 ment of the Analytical Protocol Specifications (APSs) from the laboratory's perspective on
7 method appropriateness and availability. It offers guidance to the laboratory on the key elements
8 to consider when selecting an analytical method (Chapter 1, Section 1.4.5) to meet the objectives
9 of the APSs contained in the Statement of Work (SOW). Assuming that the laboratory has
10 received a SOW, certain subsections of Section 6.5 provide guidance on how to review and
11 properly evaluate the APSs therein. However, Section 6.5 also provides guidance for the project
12 planning team on the important laboratory considerations needed to develop the Measurement
13 Quality Objectives (MQOs). Section 6.6 deals with method validation requirements and has been
14 written for both the project planners and the laboratory.
15 Because the method constitutes the major part of the analytical protocol (Chapter 1), this chapter
16 focuses on the selection of a method. However, other parts of the protocol should be evaluated
17 for consistency with the method (Figure 6.1). MARLAP recommends the performance-based
18 approach for method selection. Thus, the laboratory should be able to propose whichever method
19 meets the project's analytical data requirements (MQOs), within constraints of other factors such
20 as regulatory requirements, cost, and project deadlines. The selection of a method by the
21 laboratory is in response to the APSs (Chapter 3) that were formulated during the directed
22 planning process (Chapter 2) and documented in the SOW (Chapter 5). In most project plan
23 documents, the project manager or the project planning team has the authority and responsibility
24 for approving the methods proposed by the laboratory. The APSs will, at a minimum, document
25 the analytes, sample matrices, and the MQOs. A MQO is a statement of a performance objective
26 or requirement for a particular method performance characteristic. The MQOs can be viewed as
27 the analytical portion of the DQOs (Chapter 3).
28 Background material in Section 6.2.1 provides the reader with the subtleties of the performance-
29 based approach to method selection, contrasted with the use of prescribed methods and the
30 importance of the directed panning process and MQOs in the selection of the method. This
31 chapter does not provide a listing of existing methods with various attributes indexed to certain
32 applications. Analytical methods may be obtained from national standards bodies, government
33 laboratories and publications, and the open literature.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 6-1 DO NOT CITE OR QUOTE
-------
Selection and Application of an Analytical Method
34 In this chapter, method validation is defined as the demonstrated method applicability for a
35 particular project. MARLAP recommends that only methods validated for a project's application
36 be used. This recommendation should not be confused with the generic method validation that all
37 methods should undergo during method development. The laboratory should validate the method
38 to the APS requirements of a SOW for the analyte/matrix combination and provide the method
39 validation documentation to the project manager prior to the implementation of routine sample
40 processing (Section 6.6). If applicable, consideration should be given to the uncertainty of the
41 laboratory's protocol for subsampling (heterogeneity) of the received field sample when selecting
42 a method. Appendix F provides guidance on the minimization of subsampling uncertainty.
43 Section 6.3 provides an overview of the generic application of a method for a project and how a
44 laboratory meets the recommendations of the guidance provided in this and other chapters.
45 Generic considerations for the method selection process that a laboratory should evaluate are
46 provided in Section 6.4. Project-specific considerations for method selection relevant to APSs are
47 discussed in Section 6.5. Recommendations on the degree of method validation specified by the
48 project planning team are outlined in Section 6.6. Sections 6.7, 6.8, and 6.9 provide guidance on
49 analyst qualifications, method control, and continued laboratory performance assessment,
50 respectively. Section 6.10 outlines recommendations for the method proposal and validation
51 documentation that a laboratory should send to the project manager.
52 6.2 Method Definition
53 For this chapter, a laboratory "method" includes all physical, chemical, and radiometric processes
54 conducted at a laboratory in order to provide an analytical result. These processes, depicted in
55 Figure 6.1, may include sample preparation, dissolution, chemical separation, mounting for
56 counting, nuclear instrumentation counting, and analytical calculations. This chapter will
57 emphasize the laboratory's selection of the radioanalytical method that will be proposed in
58 response to a SOW. Each each method is assumed to address a particular analyte in a specified
59 matrix or, in some cases, a group of analytes having the same decay emission category that can
60 be identified through spectrometric means (e.g., gamma-ray spectrometry). However, it should be
61 emphasized that the project planning team should have evaluated every component of the APSs
62 for compatibility with respect to all analytes in a sample and the foreseen use of multiple
63 analytical methods by the laboratory. For example, samples containing multiple analytes must be
64 of sufficient size (volume or mass) to ensure proper analysis and to meet detection and quantifi-
65 cation requirements. Multiple analytes in a sample will require multiple analyses for which a
66 laboratory may use a sequential method that addresses multiple analytes or stand-alone individual
67 methods for each analyte. The analytical protocol must ensure that the samples are properly
MARLAP JULY 2001
DO NOT CITE OR QUOTE 6-2 DRAFT FOR PUBLIC COMMENT
-------
Selection and Application of an Analytical Method
Field Sample Preparation
and Preservation
Sample Receipt
and Tracking
Laboratory Sample
Preparation
Sample Dissolution
Chemical Separation
Sample Preparation
for Instrument Measurement
Instrument Measurement of
Radionuclides
Analytical Calculations and
Data Reduction
Data Verification, Validation
and Reporting
May be Included
These Steps Are
Typically Considered
to be "The Method"
FIGURE 6.1 — Analytical process
JULY 2001
DRAFT FOR PUBLIC COMMENT
6-3
MARLAP
DO NOT CITE OR QUOTE
-------
Selection and Application of an Analytical Method
68 preserved for each analyte and sufficient sample is collected in the field to accommodate the
69 analytical requirements.
70 Certain aspects of a method are defined in this chapter in order to facilitate the method selection
71 process. The following subsections describe the underlying basis of a performance-based
72 approach to method selection and provide a functional definition related to MARLAP.
73 Performance-Based Approach and Prescriptive Method Application
74 MARLAP uses a performance-based approach to select a method, which is based on a
75 demonstrated capability to meet defined project performance criteria (e.g., MQOs). With a
76 properly implemented quality system, a validated method should produce appropriate and
77 technically defensible results under the applicable conditions. The selection of any new method
78 usually requires additional planning and, in some cases, may result in additional method
79 development or validation. The selection of a method under the performance-based approach
80 involves numerous technical, operational, quality, and economic considerations. However, the
81 most important consideration in the selection of a method under the performance-based approach
82 is compliance with the required MQOs for the analytical data. These requirements should be
83 defined in the SOW or appropriate project plan document.
84 When developing the MQOs, the project planning team should have evaluated all processes that
85 have a potential to affect the analytical data. Those involved in the directed planning process
86 should understand and communicate the needs of the project. They should also understand how
87 the sampling (field, process, system, etc.) and analytical activities will interact and the ramifica-
88 tions that the data may have on the decisionmaking process. These interactive analysis and
89 communication techniques should be applied in all areas where analytical data are produced. As
90 new projects are implemented, it should not be assumed that the current methods are necessarily
91 the most appropriate and accurate; they should be reevaluated based on project objectives. The
92 application of a performance-based approach to method selection requires the quantitative
93 evaluation of all aspects of the analytical process. Once the MQOs for a project have been
94 determined and incorporated into the APSs, under the performance-based approach, the
95 laboratory will evaluate its existing methods and propose one or more methods that meet each
96 APS. This chapter contains guidance on how to use the APSs in the laboratory's method
97 evaluation process.
98 The objective of a performance-based approach to method selection is to facilitate the selection,
99 modification, or development of a method that will reliably produce quality analytical data as
100 defined by the MQOs. Under the performance-based approach, a laboratory, responding to a
MARLAP JULY 2001
DO NOT CITE OR QUOTE 6-4 DRAFT FOR PUBLIC COMMENT
-------
Selection and Application of an Analytical Method
101 SOW, will propose a method that best satisfies the requirements of the MQO and the laboratory's
102 operations.
103 In certain instances, the requirement to use prescribed methods may be included in the SOW. The
104 term "prescribed methods" has been associated with those methods that have been selected by
105 industry for internal use or selected by a regulatory agency, such as the U.S. Environmental
106 Protection Agency (EPA), for specific programs. The methods for analyzing radionuclides in
107 drinking water prescribed by EPA (1980 ) provides an example of applying a limited number of
108 methods to a well-defined matrix. In many companies or organizations, prescribed methods are
109 widely used. Methods that have been validated for a specific application by national standard
110 setting organizations such as the American Society for Testing and Materials (ASTM), American
111 National Standards Institute (ANSI), American Public Health Association (APHA), etc., may
112 also be used as prescribed methods by industry and government agencies.
113 Typically, the prescribed methods were selected by an organization to meet specific objectives
114 for a regulation under consideration or for a program need. In most cases, the prescribed methods
115 had undergone some degree of method validation, and the responsible organization had required
116 a quality system to demonstrate continued applicability and quality, as well as laboratory
117 proficiency. The use of any analytical method, whether prescribed or from the performance-based
118 approach, has a life cycle that can be organized into the major categories of selection, validation,
119 and continued demonstrated capability and applicability. This chapter will cover in detail only
120 the first two of these categories. A discussion on ongoing laboratory evaluations is presented in
121 Chapter 7 and Appendix C.
122 A final note should be made relative to prescribed methods and the performance-based approach
123 to method selection. The performance-based approach for method selection allows more latitude
124 in dealing with the potential diversity of matrices (such as waste-, sea-, ground- or surface water;
125 biota; air filters; waste streams; swipes; soil; sediment; or sludge) from a variety of projects, or in
126 dealing with different levels of data quality requirements or a laboratory's analytical proficiency.
127 Even though the prescribed method approach may initially appear suitable and cost effective, it
128 does not allow a laboratory to select a method from the many possible methods that will meet the
129 MQOs.
130 Many individuals have the wrong impression that prescribed methods do not need to be validated
131 by a laboratory. However, as discussed in this chapter, all methods should be validated to some
132 level of performance for a particular project by the laboratory prior to their use. In addition, the
133 laboratory should demonstrate continued proficiency in using the method through internal QC
134 and external performance evaluation (PE) programs (Chapter 18).
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 6-5 DO NOT CITE OR QUOTE
-------
Selection and Application of an Analytical Method
135
6.3 Life Cycle of Method Application
136 In responding to a SOW for a given analyte/matrix combination, a laboratory may have one or
137 more methods that may be appropriate for meeting the MQOs. The final method selected from a
138 set of methods may be influenced by many other technical, operational, or quality considerations.
139 Figure 6.2 provides an overview of the life cycle of the method application. Figure 6.3 expands
140 the life cycle into a series of flow diagrams.
Analyte / Matrix
Process Knowledge-
Project
Management
Documentation of
Method Validation
& Performance
During Project
Continued
Performance
Assessments
Analytical
Protocol
Specifications
Proposed
Method
Modification
External
PE Programs
Laboratory
Management
Method
Selection
Initial
Approval
Method
Control
(Quality System)
Method
Validation
Demonstrated
Performance)
Analyst
Selection /
Qualification
Available
Methods
Existing Method Method
Methods Development Modification
P.
Approval
Samples
Project External QC
Approved^
Method
FIGURE 6.2 — Method application life cycle
MARLAP
DO NOT CITE OR QUOTE
6-6
JULY 2001
DRAFT FOR PUBLIC COMMENT
-------
Selection and Application of an Analytical Method
Analyte / Matrix
Process Knowledge
Analytical Protocol
Specifications
Analytes - Radionuclides w/ decay products
- Health significance of nuclides
- Scaling factors (alternative related analytes) for related nuclides; decay corrected and
based on process knowledge and the uncertainty of the alternate analyte
measurements
- Chemical species of analyte; process knowledge or experiment
- Analyte stability during and following sampling
- Preservation requirements
Matrix - Description from process knowledge or field collection reports
- Chemical or radioactive interferences and inherent analyte in matrix
- Analyte Contaminant or inherent in matrix; process knowledge
- Analyte uniformly distributed within matrix
- Sample prep considerations
Data use - Define final form for analysis
- wet or dry for soil and vegetation
- analyte cone. / particle size distribution
- analyte cone. / dissolved or suspended or both
- analyte cone. / chemical & physical species
MQO - Define action level for each analyte / matrix
- Define MDC or MQC
- Define required method uncertainty at the action level
- Define MQOs for other method performance characteristics as
appropriate - method specificity, ruggedness and analyte concentration
range
Method Validation Testing Protocol
- Select Method Validation Level
- Test at several analyte concentration levels including zero analyte (blanks);
MDC or MQC requirement
- Include known chemical or radionuclide interferences at appropriate levels
- Select project specific or appropriate surrogate matrix PT samples
- Establish acceptable chemical / radiotracer yield values
- State data testing criteria
Method Development / Selection
(Section 6.5)
FIGURE 6.3 — Expanded Figure 6.2 addressing the laboratory's method evaluation process
JULY 2001
DRAFT FOR PUBLIC COMMENT
6-7
MARLAP
DO NOT CITE OR QUOTE
-------
Selection and Application of an Analytical Method
Method Development / Selection
(Section 6.5)
Method Validation
(Section 6.6)
Method Approval
Sample Prep / Sample Dissolution / Chemical Separation /
Test Source / Nuclear Counting
- Measurement quality objectives
- Analyte / radionuclide of interest
- Sample volume
- Chemical / physical species
- Preservation applied in field / lab
- Chemical / radionuclide interferences
- Matrix considerations - subsampling considerations
- Method of analyte (or alternative analyte) detection
- Method complexity
- Required turnaround and radiological holding times
- Validation status of possible methods
- Availability of qualified staff
- Hazardous waste production
- Facility / benchspace and equipment availability
- Associated costs
Review specificied method validation requirements and deterr
- Use of exisiting validated method
- Modify existing method and validate
- Develop new method and validate
Prepare method validation documentation
Laboratory proposes method and submits:
- SOP
- Method validation documentation
- Previous history of method use
Project Manager evaluates proposed method
- Technical review of proposed method
- Review of method validation documentation
- Approval signoff
Analyst
Selection / Qualifications
(Section 6.7)
Expanded Figure 6.2 addressing the laboratory's method
evaluation process
FIGURE 6.3 (continued)
MARLAP
DO NOT CITE OR QUOTE
6-8
JULY 2001
DRAFT FOR PUBLIC COMMENT
-------
Selection and Application of an Analytical Method
Analyst
Selection / Qualifications
(Section 6.7)
Method Control
(Section 6.8)
Continued Performance
Assessment
(Section 6.9)
Analyst selection consistent with level of method difficulty
- Education
- Experience & familiarity of method concepts
Documented training in lab safety, radiation safety, chemical hygiene and waste
management
Documented training on selected method
- Method review
- Supervised hands on training
Analyst completes profiency tests
- Analytical results meet quality performance requirements for MQOs
Controlled Method Manual
- Latest revision applied
- Signature signoff
Instrument calibration & radiotracers - NIST traceable standards
Instrumentation quality control
- Balances, pipettes, volumetric glassware
- Daily / prior-to-use nuclear and chemistry instrumentation QC checks
Radiotracer / gravimetric yield within specified range
Internal batch QC samples
SOPs for troubleshooting "out of control" situations
FIGURE 6.3 (continued) — Expanded Figure 6.2 addressing the laboratory's method
evaluation process
JULY 2001
DRAFT FOR PUBLIC COMMENT
6-9
MARLAP
DO NOT CITE OR QUOTE
-------
Selection and Application of an Analytical Method
Continued Performance
Assessment
(Section 6.9)
Documentation
(Section 6.10)
- Internal batch QC samples meet quality performance criteria
- External double and single blind QC / PT samples from contracting organization and / or agency
monitoring laboratory
- External single blind PT samples from national PE program - traceable to NIST
- Data verification and validation
- Internal assessments / audits / surveillances
- External assessments / audits / surveillances
- Method validation records
- Analyst training program
- Method manual control and archiving
- Software verification and validation records
- Instrument calibration & QC records
- Internal method batch QC sample results
- Internal & external assessments
- External double / single QC sample results
- Corrective action reports
- Analytical results - hard and electronic copy
141
FIGURE 6.3 (continued)
Expanded Figure 6.2 addressing the laboratory's method
evaluation process
6.4 Generic Considerations for Method Development and Selection
142 This section provides guidance on the technical, quality, and operational considerations for the
143 development of a new method or the selection of an existing radioanalytical method. Unless
144 required by a regulatory or internal policy, rarely should a method be specified in an APS or a
145 SOW. MARLAP recommends that a SOW containing the MQOs and analytical process
146 requirements be provided to the laboratory.
147 If the nature of the samples and analytes are known in advance, and variations in a sample matrix
148 and analyte concentration are within a relatively small range, the development or selection of
149 analytical methods is easier. In most situations, however, the number of samples, sample
150 matrices, analyte interferences, chemical form of analytes, and variations among and within
151 samples may influence the selection of a method for a given analyte. A number of radioanalytical
152 methods are available, but no single method provides a general solution (all have advantages and
MARLAP
DO NOT CITE OR QUOTE
6-10
JULY 2001
DRAFT FOR PUBLIC COMMENT
-------
Selection and Application of an Analytical Method
153 disadvantages). The method selection process should consider not only the classical
154 radiochemical methods involving decay emission detection (alpha, beta or gamma) but also non-
155 nuclear methods, such as mass spectrometric and kinetic phosphorescence analysis.
156 In the performance-based approach to method selection, the laboratory may select and propose a
157 gross measurement (alpha, beta, or gamma) method that can be applied to analyte concentrations
158 well below the action level for the analyte, as well as an analyte specific method for analyte
159 levels exceeding a proposed "screening level" that is a fraction of the action level. For example,
160 it may be acceptable to propose a gross measurement method when its combined standard
161 uncertainty meets the method uncertainty requirement at concentration levels much below the
162 action level. A gross measurement method may be employed initially for some projects. Such an
163 approach would have to be agreed to by the laboratory and project manager. The method
164 validation, discussed in Section 6.6, should demonstrate that the gross measurement method can
165 measure the analyte of interest (directly or indirectly) at the proposed analyte concentration and
166 meet the uncertainty requirement in the presence of other radionuclides. Appendix C provides
167 guidance on how to determine the acceptable method uncertainty at an analyte concentration
168 relative to the action level.
169 In general, the development or selection of a method follows several broad considerations. These
170 include analyte and matrix characteristics, technical complexity and practicality of methods,
171 quality requirements, availability of equipment, facility and staff resources, regulatory concerns,
172 and economic considerations. Each of the broad considerations can be detailed. The following
173 list, although not inclusive, provides insight into the selection of an appropriate method. Many of
174 these categories are discussed in subsequent MARLAP Part n chapters.
175 0 Analyte/radionuclide/isotope of interest
176 o Decay emission (particle or photon), atom detection, or chemical (photon detection)
177 o Half-life of analyte
178 o Decay products (progeny); principal detection method or interference
179 ° Chemical/physical forms (e.g., gas, volatile)
180 o Use of nondestructive or destructive sample analysis
181 0 Level of other radionuclides or chemical interference
182 ° Level of decontamination or selectivity required, e.g., a decontamination factor of 103for
183 an interfering nuclide (60Co) present with the analyte of interest (241Pu)
184 ° Resolution of measurement technique
185 o Robustness of technique for handling large fluctuations in interference levels and
186 variations in a matrix
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 6-11 DO NOT CITE OR QUOTE
-------
Selection and Application of an Analytical Method
187 ° Radionuclides inherent in background
188 0 Matrix
189 o Destructive testing
190 - Stable elemental interferences
191 - Difficulty in dissolution of a matrix
192 - Difficulty in ensuring homogeneity of aliquant
193 - Inconsistency in chemical forms and oxidation states of the analyte versus the tracer
194 ° Non-destructive testing
195 - Heterogeneity of final sample for analysis
196 - Self absorption of particle/photon emissions within a matrix
197 0 Degree of method complexity
198 o Level of technical ability required of analysts
199 ° Reproducibility of quality results between analysts
200 o Method applicability to sample batch processing
201 ° Extensive front-end chemical-processing technique (sample dissolution, analyte
202 concentration and purification/isolation, preparation for final form for radiometrics)
203 ° Nuclear instrumentation oriented technique (minimal chemical processing)
204 0 Required sample turnaround time
205 o Half-life of analyte
206 ° Sample preparation or chemical method processing time
207 o Nuclear instrumentation measurement/analysis time
208 ° Chemical or sample matrix preservation time
209 o Batch processing
210 ° Degree of automation available/possible
211 0 Status of possible methods and applications
212 o Validated for the intended application
213 ° Staff qualified and trained to use method(s)
214 o Existing QC program for method(s)
215 ° Specialized equipment, tracers, reagents, or materials available
216 0 Hazardous or Mixed waste production
217 o Older classical techniques versus new advanced chemical technologies
218 ° Availability and expense of waste disposal
219 0 Associated costs
MARLAP JULY 2001
DO NOT CITE OR QUOTE 6-12 DRAFT FOR PUBLIC COMMENT
-------
Selection and Application of an Analytical Method
220 o Labor, instrumentation usage, facilities, radiological waste costs
221 ° Method applicability to portable or mobile laboratory facilities
222 o Availability of service hookups
223 ° Need for facility environmental controls
224 o Need for regulatory permitting of mobile laboratory facility
225 6.5 Project-Specific Consideration for Method Selection
226 Certain parameters of the APSs (See Chapter 3 and the example in Figure 3.2) within the SOW
227 are important to the method selection process. These include the analytes, matrix type, matrix
228 characterization, analyte and matrix interferences, analyte speciation information gathered from
229 process knowledge, sample process specifications (such as radiological holding times and sample
230 processing turnaround times), and the MQOs. While these issues should be resolved during
231 project planning, they are presented here as guidance to the laboratory for their review and
232 evaluation of the technical adequacy of the SOW and to provide context for the method
233 evaluation and selection process. Many of the issues from the project planning point of view are
234 discussed in Section 3.3.
235 6.5.1 Matrix and Analyte Identification
The first step in selecting a method is knowing what analytes and sample matrices are involved.
The following sections discuss what important information should accompany analyte and matrix
1 /~\ £itt+i TI /"»o+i r\-r\
236
237
238 identification.
239 6.5.1.1 Matrices
240 A detailed identification and description of the sample matrix are important aspects in the
241 selection of an analytical method to meet the MQOs. The SOW should provide the necessary
242 detailed sample matrix description, including those important matrix characteristics gathered
243 from process knowledge. The laboratory should evaluate whether the existing sample preparation
244 and dissolution steps of a method (Chapters 10 and 12 through 15) will be sufficient to meet the
245 MQOs or the method validation requirements. The matrix will also determine, to a certain extent,
246 waste handling and disposal at the laboratory. If the matrix description is too vague or generic,
247 the laboratory should contact the technical representative named in the SOW and request
248 additional information.
249
250 The laboratory should ensure that the sample matrix description in the SOW reflects what is
251 considered to be the "sample" by the project manager and the description is of sufficient detail to
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 6-13 DO NOT CITE OR QUOTE
-------
Selection and Application of an Analytical Method
252 select the method preparation or analyte isolation steps that will meet the MQOs for the matrix.
253 The laboratory should not accept generic sample matrix descriptions such as liquids or solids. For
254 example, the differences between potable water and motor oil are obvious, but both may be
255 described as a "liquid sample." However, there may be only subtle differences between potable
256 surface water and groundwater but major differences between potable and process effluent
257 waters. The laboratory should consider how much method robustness is needed in order to
258 address the varied amounts of possible stable elements or compounds within a non-specified
259 water matrix. Furthermore, when water from a standing pool is received in the laboratory, it may
260 contain some insoluble matter. Now the questions arise whether the sample is the entire contents
261 of the container, what remains in the container, the insoluble material, or just the water? A clay
262 will act as an ion exchange substrate, while a sand may have entirely different retention
263 properties. Both can be described as a soil or sediment, but the properties with which they retain
264 a radionuclide are substantially different; thus, the method to properly isolate a particular
265 radionuclide will vary. The laboratory should ensure that the selected method is consistent with
266 the intended sample matrix, and the analytical results convey analyte concentration related to the
267 proper matrix (i.e., Bq/L dissolved, Bq/L suspended, or Bq/L total). For such cases, the
268 laboratory should request the project manager to clarify the "matrix" or "sample" definition.
269 Matrices genetically identified as "solid" require additional clarification or information in order
270 to select and validate a method properly. For example, sludges from a sewerage treatment facility
271 may be classified as a solid, but the suspended and aqueous portions (and possibly the dried
272 residual material) of the sample may have to be analyzed. Normally, the radioanalyte concentra-
273 tion in soils and sediments is reported in terms of becquerels per dry weight. However, certain
274 projects may require additional sample process specifications (Section 6.5.4) related to the soil or
275 sediment matrix identification that will affect the method selection process and the reporting of
276 the data. This may involve sectioning of core samples, specified drying temperature of the
277 sample, determining wet-to-dry weight ratio, removing organic material or detritus, homogeni-
278 zing and pulverizing, sieving and sizing samples, etc. In order to determine the average analyte
279 concentration of a sample of a given size containing radioactive particles, proper sample
280 preparation and subsampling coupled with the applicable analytical methods are required
281 (Chapter 12 and Appendix F). For alpha-emitting radionuclides, the method selected may only be
282 suitable to analyze a few grams of soil or sediment, depending on the organic content. The
283 laboratory should identify to the project manager the typical subsample or aliquant size that is
284 used for the proposed method. If information provided to the laboratory on process knowledge
285 indicates that there may be a possibility of radioactive particles, or selected analyte adsorption
286 onto soil or sediment particles, the laboratory should propose sample preparation and analytical
287 methods that will address these matrix characteristics. The laboratory should submit the proposed
288 methods annotated with the suspected matrix characterization issues.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 6-14 DRAFT FOR PUBLIC COMMENT
-------
Selection and Application of an Analytical Method
289 When selecting the methods for the analysis of flora (terrestrial vegetation, vegetables, aquatic
290 plants, algae, etc.) or fauna (terrestrial or aquatic animals) samples, the detailed information on
291 the matrix or the unique process specifications should be used by the laboratory to select or
292 validate the method, or both. The laboratory should ensure that the specific units for the
293 analytical results are consistent with the matrix identification and unique process specifications
294 stated in the SOW. Most flora and fauna results are typically reported in concentrations of wet
295 weight. However, for dosimetric pathway analyses, some projects may want only the edible
296 portion of the sample processed and the results to reflect this portion, e.g., fillet of sport fish,
297 meat and fluid of clams, etc. For the alpha- and beta-emitting radionuclides, aquatic vegetation
298 normally is analyzed in the dry form, but the analyte concentration is reported as wet weight. The
299 laboratory should ensure that the sample preparation method (Chapter 12) includes the
300 determination of the necessary wet and dry weights.
301 These considerations bear not only on the method selected but also on how the sample should be
302 collected and preserved during shipment. When possible, the laboratory should evaluate the
303 proposed sample collection and preservation methods, as well as timeliness of shipping, for
304 consistency with the available analytical methods. Discrepancies noted in the SOW for such
305 collateral areas should be brought to the attention of the project manager. For example, sediment
306 samples that have been cored to evaluate the radionuclide depth profile should have been
307 collected and treated in a fashion to retain the depth profile. A common method is to freeze the
308 core samples in the original plastic coring sleeves and ship the samples on ice. The SOW should
309 define the specifics on how to treat the core samples and the method of sectioning the samples
310 (e.g., cutting the cores into the desired lengths or flash heating the sleeves with subsequent
311 sectioning).
312 The SOW should have properly delineated the proper matrix specifications required for method
313 validation. In some cases, sufficient information may have been provided to define the
314 parameters necessary to prepare method validation reference material (MVRM) for method
315 validation purposes (Section 6.6). The laboratory should ensure that sufficient information and
316 clarity have been provided on the matrix to conduct a proper method validation.
317 6.5.1.2. Analytes and Potential Interferences
318 The SOW should describe the analytes of interest and the presence of any other chemical and
319 radionuclide contaminants (potential method interferences and their anticipated concentration)
320 that may be in the samples. This information should be provided in the SOW to allow the
321 laboratory's radiochemist to determine the specificity and robustness of a method that will
322 address the multiple analytes and their interferences. The delineation of other possible interfering
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 6-15 DO NOT CITE OR QUOTE
-------
Selection and Application of an Analytical Method
323 radionuclides is extremely important in the selection of a method to ensure that the necessary
324 decontamination factors and purification steps are considered.
325 The size of the sample needed by the laboratory will depend on the number of analytes and
326 whether the laboratory will select individual methods for each analyte or a possible "sequential"
327 analytical method, where several analytes can be isolated from the same sample and analyzed. If
328 a sample size is listed in the SOW, the laboratory should determine if there will be sufficient
329 sample available to analyze all analytes, the associated QC samples, and any backup sample for
330 re-analyses. Other aspects, such as the presence of short-lived analytes or analytes requiring very
331 low detection limits, may complicate the determination of a proper sample size.
332 The laboratory should ensure that the method validation requirements in the SOW are consistent
333 with the analytes and matrix. The method validation protocols defined in Section 6.6 are
334 applicable to methods for single analyte analyses or to a "sequential method" where several
335 analytes are isolated and analyzed. The laboratory should develop a well-planned protocol
336 (Section 6.6.2) for method validation that considers the method(s), analyte(s), matrix and
337 validation criteria.
338 6.5.2 Process Knowledge
339 Process knowledge typically is related to facility effluent and environmental surveillance
340 programs, facility decommissioning, and site remediation activities. Important process
341 knowledge may be found in operational history or regulatory reports associated with these
342 functions or activities. It is imperative that the laboratory review the information provided in the
343 SOW to determine whether the anticipated analyte concentration and matrix are consistent with
344 the scope of the laboratory operations. Process knowledge contained in the SOW should provide
345 sufficient detail for the laboratory to determine, quickly and decisively, whether or not to pursue
346 the work. If sufficient detail is not provided in the SOW, the laboratory should request the project
347 planning documents. Laboratories having specialized sample preparation facilities that screen the
348 samples upon arrival can make the necessary aliquanting or dilutions to permit the processing of
349 all low-level samples in the laboratories. Laboratories that have targeted certain sectors of the
350 nuclear industry or a particular nuclear facility may be very knowledgeable in the typical
351 chemical and physical forms of the analytes of a given sample matrix and may not require
352 detailed process knowledge information. However, under these circumstances, the laboratory's
353 method should be robust and rugged enough to handle the expected range of analyte concen-
354 trations, ratios of radionuclide and chemical interferences, and variations in the sample matrix.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 6-16 DRAFT FOR PUBLIC COMMENT
-------
Selection and Application of an Analytical Method
355 Process knowledge may provide valuable information on the possible major matrix constituents,
356 including major analytes, chemical/physical composition, hazardous components, radiation
357 levels, and biological growth (e.g., bacteria, algae, plankton, etc.) activities. When provided, the
358 laboratory should use this information to determine if the sample collection and preservation
359 methodologies are consistent with the proposed radioanalytical method chosen. In addition, the
360 information also should be reviewed to ensure that the proposed sample transportation or
361 shipping protocols comply with regulations governing the laboratory operation.
362 Process knowledge information in the SOW may be used by the laboratory to refine method
363 selection from possible radiometric/chemical interferences, chemical properties of the analytes or
364 matrix, and hazardous components, among others. Chapter 14 describes the various generic
365 chemical processes that may be used to ensure proper decontamination or isolation of the analyte
366 from other interferences in the sample. These include ion exchange, co-precipitation, oxidation/
367 reduction, and solvent extraction among others. The process knowledge information provided in
368 the SOW should be reviewed to determine whether substantial amounts of a radionuclide that
369 normally would be used as a radiotracer will be present in the sample. Similarly, information on
370 the levels of any stable isotope of the analyte being evaluated is equally important. Substantial
371 ambient or background amounts of either a stable isotope of the radionuclide or the radiotracer in
372 the sample may produce elevated and false chemical yield factors. In addition, substantial
373 amounts of a stable isotope of the analyte being evaluated may render certain purification
374 techniques inadequate (e.g., ion exchange or solid extractants).
375 6.5.3 Radiological Holding and Turnaround Times
376 The SOW should contain the requirements for the analyte's radiological holding and sample
377 turnaround times. MARLAP defines radiological holding time as the time differential between
378 the date of sample collection and the date of analysis. It is important that the laboratory review
379 the specifications for radionuclides that have short half-lives (less than 30 days), because the
380 method proposed by the laboratory may depend on the required radiological holding time. For
381 very short-lived radionuclides, such as 131I or 224Ra, it is very important to analyze the samples
382 within the first two half-lives in order to meet the MQOs conveniently. A laboratory may have
383 several methods for the analysis of an analyte, each having a different analyte detection and
384 quantification capability. Of the possible methods available, the method selected and proposed by
385 the laboratory most likely will be dependent on the radiological holding time requirement, half-
386 life of the analyte, and the time available after sample receipt at the laboratory. When a
387 laboratory has several methods to address variations in these constraints, it is recommended that
388 the laboratory propose more than one method with a clarification that addresses the radiological
389 holding time and MQOs. In some cases, circumstances arise which require the classification of
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 6-17 DO NOT CITE OR QUOTE
-------
Selection and Application of an Analytical Method
390 sample processing into several time-related categories (Chapter 5). For example, the determina-
391 tion of 131I in water can be achieved readily within a reasonable counting time through direct
392 gamma-ray spectrometry (no chemistry) using a Marinelli beaker counting geometry, when the
393 detection requirement is 0.4 Bq/L and the radiological holding time is short. However, when the
394 anticipated radiological holding time is in the order of weeks, then a radiochemistry method
395 using beta detection or beta-gamma coincidence counting would be more appropriate to meet the
396 detection requirement. The more sensitive method also may be used when there is insufficient
397 sample size or when the analyte has decayed to the point where the less sensitive method cannot
398 meet the required MQOs. Another example would be the analysis of 226Ra in soil, where the
399 laboratory could determine the 226Ra soil concentration through the quantification of a 226Ra
400 decay product by gamma-ray spectrometry after a certain ingrowth period, instead of direct
401 counting of the alpha particle originating from the final radiochemical product (micro-
402 precipitate) using alpha spectrometry.
403 Sample (processing) turnaround time normally means the time differential from the receipt of the
404 sample at the laboratory to the reporting of the analytical results. As such, the laboratory should
405 evaluate the SOW to ensure that the sample turnaround time, radiological holding time, data
406 reduction and reporting times, and project needs for rapid data evaluation are consistent and
407 reasonable. Method selection should take into consideration the time-related SOW requirements
408 and operational aspects. When discrepancies are found in the SOW, the laboratory should
409 communicate with the project manager and resolve any issue. Additionally, the response to the
410 SOW should include any clarifications needed for sample turnaround time and/or radiological
411 holding time issues.
412 6.5.4 Unique Process Specifications
413 Some projects may incorporate detailed sample processing parameters, specifications, or both
414 within the SOW. Specifications for parameters related to sample preparation may include the
415 degree of radionuclide heterogeneity in the final sample matrix prepared at the laboratory, the
416 length of the sections of a soil or sediment core for processing, analysis of dry versus wet weight
417 material, partitioning of meat and fluid of bivalves for analyses, and reporting of results for
418 certain media as a dry or wet weight. Specifications related to method analysis could include
419 radionuclide chemical speciation in the sample matrix. The laboratory must evaluate these
420 specifications carefully, since various parameters may affect the method proposed by the
421 laboratory. When necessary, the laboratory should request clarification of the specifications in
422 order to determine a compatible method. In addition, the laboratory should ensure that the
423 method validation process is consistent with the unique process requirements. In some cases, not
424 all special process specifications must be validated and, in other cases, site-specific materials
MARLAP JULY 2001
DO NOT CITE OR QUOTE 6-18 DRAFT FOR PUBLIC COMMENT
-------
Selection and Application of an Analytical Method
425 (also referred to as MVRM) will be required for method validation. When necessary, the
426 laboratory also should request site-specific reference materials having the matrix characteristics
427 needed for proper method validation consistent with the special process requirements. It is
428 incumbent upon the laboratory to understand clearly the intent of the special process
429 specifications and how they will be addressed.
430 6.5.5 Measurement Quality Objectives
431 The specific method performance characteristics having a measurement quality objective may
432 include:
433 • Method uncertainty at a specified analyte concentration level;
434 • Quantification capability (minimum quantifiable concentration);
435 • Detection capability (minimum detectable concentration);
436 • Applicable analyte concentration range;
437 • Method specificity; and
438 • Method ruggedness.
439 How each of these characteristics affect the method selection process will be discussed in detail
440 in the subsequent paragraphs.
441 6.5.5.1 Method Uncertainty
442 From the directed planning process, the required method uncertainty at a stated analyte
443 concentration should have been determined for each analyte/matrix combination. The method
444 uncertainty requirement may be linked to the width of the gray region (Appendix C). MARLAP
445 recommends that the SOW include the specifications for the action level and the required method
446 uncertainty for the analyte concentration at the action level for each analyte/matrix. For research
447 and baseline monitoring programs, the action level and gray region concepts may not be
448 applicable. However, for these applications, the project manager should establish a concentration
449 level of interest and a required method uncertainty at that level. The laboratory should ensure that
450 this method uncertainty requirement is clearly stated in the SOW.
451 The laboratory should select a method that will satisfy the method uncertainty requirement at the
452 action level or other required analyte level. MARLAP uses the term "method uncertainty" to
453 refer to the predicted uncertainty of a result that would be measured if a method were applied to a
454 hypothetical laboratory sample with a specified analyte concentration. The uncertainty of each
455 input quantity (method parameter) that may contribute significantly to the total uncertainty
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 6-19 DO NOT CITE OR QUOTE
-------
Selection and Application of an Analytical Method
456 should be evaluated. For some methods, the uncertainty of an input quantity may vary by analyst
457 or spectral unfolding software. Chapter 19 provides guidance on how to calculate the combined
458 standard uncertainty of the analyte concentration, and Section 19.6.12 shows how to predict the
459 uncertainty for a hypothetical measurement. For most basic methods, uncertainty values may be
460 included for the following input quantities (parameters):
461 • Poisson counting statistics (net count rate);
462 • Detector efficiency, if applicable;
463 • Chemical yield (when applicable) or tracer yield;
464 • Sample volume/weight;
465 • Decay/ingrowth factor; and
466 • Radiometric interference correction factor.
467 Typically, for low-level environmental remediation or surveillance activities, only those input
468 quantities having an uncertainty greater than one percent significantly contribute to the combined
469 standard uncertainty. Other than the radiometric interference correction factor and Poisson
470 counting uncertainties, most input quantity uncertainties normally do not vary as a function of
471 analyte concentration. At analyte levels near or below the detection limit, the Poisson counting
472 uncertainty may dominate the method's uncertainty. However, at the action level or above, the
473 Poisson counting uncertainty may not dominate.
474 When appropriate, the laboratory should determine the method uncertainty over the MQO analyte
475 concentration range (Section 6.5.5.3), including the action level or other specified analyte
476 concentration. The laboratory's method validation (Section 6.6) should demonstrate or show
477 through extrapolation or inference (e.g., from a lower or higher range of concentrations) that this
478 method uncertainty requirement can be met at the action level or specified analyte concentration
479 value. Method validation documentation should be provided in the response to the SOW.
480 6.5.5.2 Quantification Capability
481 For certain projects or programs, the project planning team may develop an MQO for the
482 quantification capability of a method. The quantification capability, expressed as the minimum
483 quantifiable concentration (MQC), is the smallest concentration of the analyte that ensures a
484 result whose relative standard deviation is not greater than a specified value, usually 10 percent.
485 Chapter 19 provides additional information on the minimum quantifiable concentration.
486 MARLAP recommends that, when required, a laboratory analyze each sample to meet the MQC
487 requirement. For example, if the MQC requirement for 89Sr is 1.0 Bq/g (with a 10 percent relative
MARLAP JULY 2001
DO NOT CITE OR QUOTE 6-20 DRAFT FOR PUBLIC COMMENT
-------
Selection and Application of an Analytical Method
488 standard deviation), the laboratory should select a method that has sufficient chemical yield
489 (Chapter 19), beta detection efficiency, low background, sample (processing) turnaround time for
490 a given sample mass, and radioactive decay to achieve a nominal measurement uncertainty of 0.1
491 Bq/g when the 89Sr concentration is 1.0 Bq/g. The same forethought that a laboratory gives to
492 estimating a method's minimum detectable concentration (MDC) for an analyte should be given
493 to the MQC requirement. The laboratory should consider the uncertainties of all input quantities
494 (detector efficiency, chemical yields, interferences, etc.), including the Poisson counting
495 uncertainty when selecting a method. This is an important consideration, because for some
496 methods, the Poisson counting uncertainty at the MQC level may contribute only 50 percent of
497 the combined standard uncertainty. Therefore, the laboratory may have to select a method that
498 will meet the MQC requirement for a variety of circumstances, including variations in matrix
499 constituents and chemical yields, radionuclide and chemical interferences, and radioactive decay.
500 In addition, sufficient sample size for processing may be critical to achieving the MQC
501 specification.
502 During the method validation process, the ability of the method to meet the required MQC
503 specification should be tested. The method validation acceptance criteria presented in Section 6.6
504 have been formulated to evaluate the MQC requirement at the proper analyte concentration level,
505 i.e., action level or other specified analyte concentration.
506 Since the laboratory is to report the analyte concentration value and its measurement uncertainty
507 for each sample, the project manager or data validator easily can evaluate the reported data to
508 determine compliance with the MQC requirement. Some projects may send performance testing
509 (PT) material spiked at the MQC level as a more in-depth verification of the compliance with this
510 requirement.
511 6.5.5.3 Detection Capability
512 For certain projects or programs, the method selected and proposed by the laboratory should be
513 capable of meeting a required MDC for the analyte/matrix combination for each sample
514 analyzed. For certain monitoring or research projects, the analyte MDC may be the important
515 MQO to be specified in the SOW. For such projects, the MDC specification may be based on the
516 analyte concentration of interest or the state-of-the-art capability of the employed technology or
517 method. No matter what premise is used to set the value by the project planning team, the
518 definition of, or the equation used to calculate, the analyte MDC should be provided in the SOW
519 (Chapter 19). Furthermore, the SOW should specify how to treat appropriate blanks or the
520 detector background when calculating the MDC. The laboratory should be aware that not all
521 agencies or organizations define or calculate the MDC in the same manner. It is important for the
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 6-21 DO NOT CITE OR QUOTE
-------
Selection and Application of an Analytical Method
522 laboratory to check that the SOW clearly defines the analyte detection requirements. In most
523 cases, it would be prudent for the laboratory to use a method that has a lower analyte MDC than
524 the SOW required MDC.
525 In some situations, a radiochemical method may not be robust or specific enough to address
526 interferences from other radionuclides in the sample. The interferences may come from the
527 incomplete isolation of the analyte of interest resulting in the detection of the decay emissions
528 from these interfering nuclides. These interferences would increase the background of the
529 measurement for the analyte of interest and, thus, increase the uncertainty of the measurement
530 background. Consequently, an aprioriMDC, since it is calculated without prior sample
531 knowledge or inclusion of the interference uncertainties, would underestimate the actual
532 detection limit for the sample under analysis. Another example of such interferences or increase
533 in an analyte's background uncertainty can be cited when using gamma-ray spectrometry to
534 determine 144Ce in the presence of 137Cs. The gamma energy usually associated with the
535 identification and quantification of 144Ce is 133.5 keV. The gamma energy for 137Cs is 661.6 keV.
536 If a high concentration of 137Cs is present in the sample, the Compton scattering from the 661.6
537 keV into the 133.5 keV region may decrease the ability to detect 144Ce by one to two orders of
538 magnitude over an a priori calculation that uses a nominal non-sample specific background
539 uncertainty. Another example can be cited for alpha-spectrometry and the determination of
540 isotopic uranium. If some interfering metal is present in unexpected quantities and carries onto
541 the final filter mount or electrodeposited plate, a substantial decrease in the peak resolution may
542 occur (resulting in an increased width of the alpha peak). Depending on the severity of the
543 problem, there may be overlapping alpha peaks resulting in additional interference terms that
544 should be incorporated into the MDC equation. In order to avoid subsequent analyte detection
545 issues, it is important for the laboratory to inquire whether or not the project manager has
546 considered all the constituents (analytes and interferences) present in the sample when specifying
547 a detection limit for an analyte.
548 The laboratory should include documentation in the response to the SOW that the method
549 proposed can meet the analyte's MDC requirements for the method parameters (e.g., sample size
550 processed, chemical yield, detector efficiency, counting times, decay/ingrowth correction factors,
551 etc.). When practicable, care should be given to ensure the blank or detector background
552 uncertainty includes contributions from possible anthropogenic and natural radionuclide
553 interferences. In addition, any proposed screening method should meet the detection limit
554 requirement in the presence of other radionuclide interferences or natural background
555 radioactivity. When appropriate or required, the laboratory should test the method's capability of
556 meeting the required MDC using MVRMs that have analytes and interferences in the expected
MARLAP JULY 2001
DO NOT CITE OR QUOTE 6-22 DRAFT FOR PUBLIC COMMENT
-------
Selection and Application of an Analytical Method
557 analyte concentration range. Upon request, the project manager should arrange to provide
558 MVRMs to the laboratory.
559 6.5.5.4 Applicable Analyte Concentration Range
560 The SOW should state the action level for the analyte and the expected analyte concentration
561 range. The proposed method should provide acceptable analytical results over the expected
562 analyte concentration range for the project. Acceptable analytical results used in this context
563 means consistent method precision (at a given analyte concentration) and without significant
564 bias. The applicable analyte concentration range may be three or four orders of magnitude.
565 However, most radioanalytical methods, with proper analyte isolation and interference-decon-
566 tamination steps, will have a linear relationship between the analytical result and the analyte
567 concentration. For certain environmental monitoring or research projects, the laboratory should
568 ensure that there are no instrument or analytical blank background problems. If the background is
569 not well-defined, there may be an inordinate number of false positive and false negative results.
570 In its response to the SOW, the laboratory should include method validation documentation that
571 demonstrates the method's capability over the expected range. The laboratory's method
572 validation (Section 6.6) should demonstrate or show through extrapolation or inference (e.g.,
573 from a different range of concentrations) that the method is capable of meeting the analyte
574 concentration range requirement.
575 6.5.5.5 Method Specificity
576 The proposed method should have the necessary specificity for the analyte/matrix combination.
577 Method specificity refers to the method's capability, through the necessary decontamination or
578 separation steps, to remove interferences or to isolate the analyte of interest from the sample over
579 the expected analyte concentration range. Method specificity is applicable to both stable and
580 radioactive constituents inherent in the sample. Certain matrices, such as soil and sediments,
581 typically require selective isolation of femtogram amounts of the analyte from milligrams to
582 gram quantities of matrix material. In these circumstances, the method requires both specificity
583 and ruggedness to handle variations in the sample constituents.
584 If other radionuclide interferences are known or expected to be present, the SOW should provide
585 a list of the radionuclides and their expected concentration ranges. This information enables the
586 laboratory to select and propose a method that has the necessary specificity to meet the MQOs.
587 As an alternative, the project manager may specify in the SOW the degree of decontamination a
588 method needs for the interferences present in the samples. If the laboratory is not provided this
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 6-23 DO NOT CITE OR QUOTE
-------
Selection and Application of an Analytical Method
589 information, method specificity cannot be addressed properly. The laboratory should ensure that
590 related information on the matrix characteristics, radiometric or chemical interferences, and
591 chemical speciation is provided to properly select a method.
592 6.5.5.6MethodRuggedness
593 Ruggedness is the ability of the method to provide accurate analytical results over a range of
594 possible sample constituents, interferences, and analyte concentrations, as well as to tolerate
595 subtle variations in the application of the method by various chemists (EPA, 1998; APHA,
596 1989). Ruggedness is somewhat qualitative (Chapter 7). Therefore, the desirable parameters of a
597 rugged method are difficult to specify quantitatively. A ruggedness test usually is conducted by
598 systematically altering the critical variables (or quantities) associated with the method and
599 observing the magnitude of the associated changes in the analytical results. ASTM El 169
600 provides generic guidance on how to conduct method ruggedness tests under short-term, high-
601 precision conditions. In many cases, a rugged method may be developed over time (typically
602 when difficulty is experienced applying an existing method to variations in the sample matrix or
603 when two analysts have difficulty achieving the same level of analytical quality or precision).
604 A laboratory may have several methods for an analyte/matrix combination. Samples from
605 different geographical locations or having different processes may have completely different
606 characteristics. Therefore, the laboratory should select a method that is rugged enough to meet
607 the APSs in the SOW. As indicated in Section 6.6, the prospective client may send site-specific
608 MVRM samples for the method validation process or for PT samples (Chapter7).
609 6.5.5.7 Bias Considerations
610 As discussed earlier, the proposed method should provide acceptable analytical results over the
611 expected analyte concentration range for the project. Acceptable results used in this context
612 means consistent method precision (at a given analyte concentration) and without significant
613 bias. According to ASTM (E177, E1488, D2777, D4855), "bias of a measurement process is a
614 generic concept related to a constant or systematic difference between a set of test results from
615 the process and an accepted reference value of the property being measured," or "the difference
616 between a population mean of the measurements or test results and the accepted reference or true
617 value." In contrast, ASTM (D2777) defines precision as "the degree of agreement of repeated
618 measurements of the same property, expressed in terms of dispersion of test results (measure-
619 ments) about the arithmetical mean result obtained by repetitive testing of a homogeneous
620 sample under specified conditions." MARLAP considers bias to be a persistent difference of the
621 measured result from the true value of the quantity being measured, which does not vary if the
MARLAP JULY 2001
DO NOT CITE OR QUOTE 6-24 DRAFT FOR PUBLIC COMMENT
-------
Selection and Application of an Analytical Method
622 measurement is repeated. Normally, bias cannot be determined from a single result or a few
623 results (unless the bias is large) because of the analytical uncertainty component in the measure-
624 ment. Bias may be expressed as the percent deviation from a "known" analyte concentration.
625 Note that the estimated bias, like any estimated value, has an uncertainty—it is not known
626 exactly.
627 If bias is detected in the method validation process or from other QA processes, the laboratory
628 should make every effort to eliminate it when practical. Implicitly, bias should be corrected
629 before using the method for routine sample processing. However, in some cases, the bias may be
630 very small and not affect the overall data quality. The project manager should review the method
631 validation documentation and results from internal QC and external PE programs obtained during
632 the laboratory review process (Chapter 7) and determine if there is a bias and its possible impact
633 on data usability.
634 6.6 Method Validation
635 For the purposes of MARLAP, method validation is the demonstration that the radioanalytical
636 method selected by the laboratory for the analysis of a particular radionuclide in a given matrix is
637 capable of providing analytical results to meet the project's MQOs and any other requirements in
638 the APS. Without reliable analytical methods, all the efforts of the project may be jeopardized.
639 Financial resources, timeliness, and public perception and confidence are at risk, should the data
640 later be called into question. Proof that the method used is applicable to the analyte and sample
641 matrix of concern is paramount for defensibility. The project manager should ensure the methods
642 used in the analyses of the material are technically sound and legally defensible.
643 The method selected and proposed by the laboratory must be based on sound scientific principles
644 and must be demonstrated to produce repeatable results under a variety of sample variations.
645 Each step of the method should have been evaluated and tested by a qualified expert (radio-
646 analytical specialist) in order to understand the limits of each step and the overall method in
647 terms of the MQOs. These steps may involve well-known and characterized sample digestion,
648 analyte purification and decontamination steps that use ion exchange, solvent extraction,
649 precipitation and/or oxidation /reduction applications. Method validation will independently test
650 the scientific basis of the method selected for a given analyte and sample matrix.
651 A method validation protocol should be a basic element in the quality system employed by a
652 laboratory. A proposed method for a specific analyte should be validated in response to the
653 requirements within a SOW. Demonstration of method performance to meet the MQOs prior to
654 processing project samples is a critical part of the MARLAP process. As a result of internal QC
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 6-25 DO NOT CITE OR QUOTE
-------
Selection and Application of an Analytical Method
655 and external PE programs, most laboratories normally have documentation on the general or
656 overall performance of a method. As discussed later, this information, depending on many
657 aspects, may be sufficient in meeting the method validation criteria.
658 Methods obtained from the literature, from recognized industry standards (ASTM, ANSI, APHA)
659 or government method manuals may have been validated for certain general applications by the
660 developing or issuing laboratory. However, other laboratories would have to validate the method
661 for specific project use.
662 6.6.1 Laboratory's Method Validation Protocol
663 During the discussion on method validation, certain terms are used. These include MVRM, QC,
664 and PT materials. QC samples and programs are related to those samples or processes that are
665 used to evaluate the quality of the analytical results for the fundamental purpose of directly
666 controlling the quality of the analytical process by initiating control mechanisms. PT materials
667 are materials prepared for use in a PE program or for validating methods. MVRM refers to site-
668 specific materials that have the same or similar chemical and physical properties as the proposed
669 project samples. Although the MVRM is the most appropriate material for testing a laboratory's
670 project-specific performance, or for validating a method for a particular project, its availability
671 may be limited depending on the project manager's ability to supply such material.
672 The laboratory's method validation protocol should include the evaluation of the method for
673 project specific MQOs for an analyte or generic quality performance criteria as well as other
674 generic parameters. With a properly designed method validation protocol, important information
675 may be ascertained from the analytical results generated by the method validation process.
676 The parameters that should be specified, evaluated, or may be ascertained from the analytical
677 results generated by the method validation process are listed below:
678 0 Defined Method Validation Level (Table 6.1)
679 <> APSs including MQOs for each analyte/matrix
680 ° Chemical or physical characteristics of analyte when appropriate
681 o Action level (if applicable)
682 ° Method uncertainty at a specific concentration
683 o MDC or MQC
684 ° Bias (if applicable)
685 o Applicable analyte concentration range including zero analyte (blanks)
686 ° Other qualitative parameters to measure the degree of method ruggedness or specificity
MARLAP JULY 2001
DO NOT CITE OR QUOTE 6-26 DRAFT FOR PUBLIC COMMENT
-------
Selection and Application of an Analytical Method
687 <> Defined matrix for testing, including chemical and physical characteristics that approximate
688 project samples
689 <> Selected project-specific or appropriate alternative matrix PT samples, including known
690 chemical or radionuclide interferences at appropriate levels
691 <> Defined sample preservation
692 0 Stated additional data testing criteria (such as acceptable chemical/radiotracer yield values)
693 In order to properly demonstrate that a method will meet project MQOs, the method should be
694 evaluated over a range of analyte concentrations. The analyte concentration range of the matrix
695 spikes (covering the testing levels) used for method validation should cover the expected analyte
696 concentration range for the project (Section 6.5.5.3), with the middle of the range set near the
697 action level. At the upper end of the range, the method validation samples should be analyzed to
698 have a Poisson counting uncertainty between 1 percent (ANSI N42.23) and 3 percent (1 sigma).
699 Keeping the Poisson uncertainty <3 percent (1 sigma) will ensure the observed precision, as
700 measured by multiple samples, is not dominated by the Poisson counting uncertainty. In addition,
701 anticipated or known chemical and radionuclide interferences should be added in the appropriate
702 "interference to analyte" activity or concentration ratio. Appropriate method blanks (also
703 containing interferences when practical) should be analyzed concurrently with the matrix spikes
704 to determine analyte interferences or biases near the detection limit.
705 The number of samples for the method validation process varies according to the method
706 validation level needed. As proposed in Table 6.1, the number of samples may vary from 6 to 21,
707 depending on the robustness of the method validation.
708 6.6.2 Tiered Approach to Validation
709 While MARLAP recommends that as each new project is implemented, the methods used in the
710 analysis of the associated samples undergo some level of validation, it is the project manager's
711 responsibility to assess the level of method validation necessary. Although the end result of
712 method validation is to ensure that the method selected meets the MQOs for an analyte/matrix,
713 the extent of the validation process depends on whether the laboratory should elect to develop a
714 new method or whether there is an existing validated method available that can be adapted or
715 validated for another specific project need. Therefore, MARLAP recommends that a tiered
716 approach be taken for method validation. The recommended protocols to be considered for
717 existing methods are provided in the next four sections, requiring from least to most effort: no
718 additional validation, modification of a method for a similar matrix, new application of a method,
719 and newly developed or adapted methods. Table 6.1 consolidates recommended validation
720 requirements from various government agencies and consensus organizations. The suggested
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 6-27 DO NOT CITE OR QUOTE
-------
Selection and Application of an Analytical Method
721 levels of validation are indicative of the modification required of the method. It should be noted
722 that the method validation requirements of Table 6.1 permit the laboratory to use internal QC, PE
723 program, or site-specific MVRM samples, or permit the project manager may provide PT, PE
724 program, or site-specific MVRM samples for the laboratory to use. Sometimes, a project
725 manager may provide PT samples as part of the qualifying process. In this case, the project
726 manager should ensure consistency with the method validation requirements of Table 6.1.
727
728
729
730
731
732
733
734
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
TABLE 6.1 — Tiered method validation approach
Validation
Level
A
Without
Additional
Validation
B
C
D
ASTM D2777
E
ASTM D2777
F
EPA
Equivalency
G
ASTMD2777
H
ASTMD2777
(Involves the
two testing
protocols stated
to the right)
Application
Existing
Validated
Method
Similar Matrix
Similar Matrix/
New Application
New
Application
New Application
New
Application/
newly Developed
or Adapted
Method
Newly
Developed
or
Adapted Method
Sample
Type
-
Internal QC
External PE
Internal QC
External PE
MVRM
Samples
MVRM
Samples
MVRM
Samples for
both
protocols
Acceptance
Criteria*
Method previously
validated
(by one of Validation
Levels B though H)
Measured value within
±3 uMRof known value
Measured value within
±3 UMR of known value
Each measured value
<30% of known
at 5 times MDC
Levels
(Concentrations) * *
-
3
3
Replicates
-
3
7
Three to five groups of two
samples with concentrations
within 20% of each other
3
7
Three to five groups of two
samples with concentrations
within 20% of each other
Three to five groups of two
samples with concentrations
within 20% of each other
Three to five groups of two
samples with concentrations
within 20% of each other
bracketing 5 times the MDC
# of Analyses
-
9
21
6-10
6-10
21
6-10
6-10
6-10
* Assumes that each sample is counted to have a Poisson counting uncertainty of < 3% (sigma) when the analyte concentration is
near the action level or MQC. This criterion is applied to each analysis in the method validation, not to the mean of the analyses.
UMR is the required method uncertainty at the action level or required concentration. UMR is an absolute value for concentrations
less than the action level and a relative (%) value for concentrations greater than the action level. In the absence of a specified
value, the default of ± 3 UMR acceptance criterion is: each measured value at the action level or other specified concentration must
be within ± 30% of known value. See references for ASTM D2777.
"Concentration levels should cover the expected analyte concentration range for a project including the action level
concentration. A set of three blanks (not considered a level) should be analyzed during the method validation process.
MARLAP
DO NOT CITE OR QUOTE
6-28
JULY 2001
DRAFT FOR PUBLIC COMMENT
-------
Selection and Application of an Analytical Method
759 The tiered approach to method validation outlined Table 6.1 was developed to give the project
760 manager flexibility in the method validation process according to the project requirements. The
761 degree of method validation increases from the lowest (Level A) to the highest (Level H). The
762 table's acceptance criteria for the validation process for a given project are based on the MQO for
763 the method uncertainty at the action level or other stated concentration. Each of the validation
764 levels evaluates the proposed method over the expected concentration range of the analytes and
765 interferences. The acceptance criterion of having each analytical result meet the ± 3 u^ of the
766 known value ensures a high degree of confidence that a method will meet the required method
767 uncertainty (MQO) at the action level or other specified concentration. (See Appendix C for the
768 definition of the method uncertainty at the action level or other stated concentration, %£.) In
769 addition to evaluating the method uncertainty, the method should be evaluated for bias.
770 During the method validation process, the laboratory should ensure that the observed precision
771 for the samples processed is consistent with the estimated individual sample measurement uncer-
772 tainty. An evaluation should be conducted for replicate sample analyses that have the same
773 approximate relative measurement uncertainties. Samples having analyte concentrations within a
774 narrow range of one another (ASTM D2777 Youden Pairs) may be considered when their
775 relative measurement uncertainties are approximately the same. If the estimated measurement
776 uncertainty of a given sample is much smaller than the observed method precision for the
777 replicate samples, then the laboratory may not have properly estimated the uncertainty of one of
778 the input quantities (parameters) or has omitted an input quantity in the measurement uncertainty
779 (combined standard uncertainty).
780 6.6.2.1 Existing Methods Requiring No Additional Validation
781 For completeness, it is necessary to discuss the possibility that a previously validated method of
782 choice requires no additional validation (Level A of Table 6.1) for a specific project use. As
783 noted in the table, the method has undergone some level (Level B through H) of previous
784 validation. It may be that the samples (matrix and analyte specific) associated with a new project
785 are sufficiently similar to past samples analyzed by the same laboratory that the project manager
786 feels additional validation is unwarranted. The decision to use Level A method validation should
787 be made with caution. While the sampling scheme may be a continuation, the analytical
788 processing capabilities at the laboratory may have changed sufficiently to merit limited method
789 validation. Without some level of method validation, the project manager has no assurance that
790 the analytical laboratory will perform to the same standards as an extension of the earlier work.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 6-29 DO NOT CITE OR QUOTE
-------
Selection and Application of an Analytical Method
791 6.6.2.2 Use of a Validated Method for Similar Matrices
792 When a previously validated method is to be used in the analysis of samples that are similar to
793 the matrix and analyte for which the method was developed, MARLAP recommends that
794 validation of the method be implemented according to Level B or C of Table 6.1. These levels
795 will provide a reasonable assurance to both the laboratory and the project manager that the
796 method will meet the required MQOs associated with the project. Level B may be used if the
797 laboratory has the capability to produce internal QC samples. When the laboratory does not have
798 the capability to produce internal QC samples, the Level C validation protocol should be used.
799 However, PE programs may not provide the necessary matrices needed for the Level C validation
800 protocol.
801 Since a method inherently includes initial sample preparation, projects that have severe
802 differences in analyte heterogeneity may require a moderate change in a radiochemical method's
803 initial sample treatment. A change in the method to address the increased heterogeneity of the
804 analyte distribution within the sample may require another method validation depending on the
805 robustness of the method and the degree of analyte heterogeneity.
806 6.6.2.3 New Application of a Validated Method
807 Methods that have been validated for one application normally require another validation for a
808 different application, such as a different sample matrix. In addition, the MQOs may change from
809 one project to another or from one sample matrix to another. The validation process for an
810 existing validated method should be reviewed to ensure applicability of the new (which can be
811 more or less restrictive) measurement quality objectives. In most cases, applying an existing
812 method for one matrix to another matrix is not recommended without another method validation.
813 MARLAP recommends, based on the extent of the modification and the difficulty of the matrix,
814 that Levels C-F of Table 6.1 be used to validate the performance of the modified method. The
815 following paragraphs and the next section provide information on whether a validated method
816 requires a slight modification or a complete revision.
817 Validation of an existing method for a different application depends on the extent of the
818 departure from the original method application, in terms of:
819 • Dissimilarity of matrices;
820 • Chemical speciation of the analyte or possible other chemical interference;
821 • Analyte, chemical or radiometric interferences;
822 • Complete solubilization of the analyte and sample matrix; and
MARLAP JULY 2001
DO NOT CITE OR QUOTE 6-30 DRAFT FOR PUBLIC COMMENT
-------
Selection and Application of an Analytical Method
823 • Degree of analyte or sample matrix heterogeneity.
824 When the chemical species of the analyte in a sample from a new project varies from the
825 chemical species for which the method was validated, then the method will have to be altered
826 and another validation performed. An example would be when a method had been developed to
827 extract iodide via ion exchange chromatography but the new application may have I2, iodate, or
828 iodide in the sample. Another example would be the initial development of a method for Pu in
829 soil generated from liquid effluents using acid dissolution and then trying to apply the same
830 method to high-fired plutonium oxide in soil. For these two examples, if the original methods
831 were to undergo the validation process for the new application, definite deficiencies and poor
832 results would become evident. Portions of the original method would have to be modified to
833 address the chemical speciation problems. The modified method requires validation to ensure
834 that the measurement quality objectives for the new application can be met.
835 When additional analyte, chemical, or sample matrix interferences are known to exist for a new
836 application compared to the old method application, the previously validated method should
837 undergo another validation, depending on the degree of interference and the problems anticipa-
838 ted. For example, applying a method used for the analysis of an analyte in an environmental
839 matrix containing few interfering radionuclides would typically be inappropriate for the analysis
840 of process waste waters containing many interfering radionuclides at high concentrations. In
841 essence, the degree of decontamination (degree of interference removal) or analyte purification
842 (isolation of the analyte from other radionuclides) necessary for one application may be
843 completely inadequate or inappropriate for another application (an indication of method
844 specificity).
845 Another example would be the use of a method for soil analysis employing 234Th as a radiotracer
846 for chemical yield for the isotopic analysis of thorium when the soil also has a high concentration
847 of uranium. 234Th is an inherent decay product of 238U and will exist in the sample as a natural
848 analyte, thus creating erroneous chemical yield factors. A third example would be the application
849 of a 90Sr method developed for freshwater to seawater samples for which the amount of chemical
850 interferences and ambient Sr levels are extensive. For these three examples, conducting the
851 validation process for the original methods for the new applications would, depending on the
852 severity of the analyte and chemical interference, illustrate method deficiencies and the inability
853 to meet measurement quality objectives.
854 Some matrices and analytes may be solubilized easily through acid dissolution or digestion. For
855 some applications, the analyte of interest may be solubilized from the sample matrix through an
856 acid extraction process. The applicability of such methods should be carefully chosen and, most
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 6-31 DO NOT CITE OR QUOTE
-------
Selection and Application of an Analytical Method
857 important, the method must be validated for each application. Definite problems and
858 misapplication can be the result of using an acid extraction process when a more robust complete
859 sample dissolution is necessary.
860 6.6.2.4 Newly Developed or Adapted Methods
861 MARLAP recommends that methods under development by the laboratory or adapted from the
862 literature that have not been previously validated for a project be validated according to Levels
863 F to H of Table 6.1. These levels provide the most comprehensive testing of method perfor-
864 mance. For low-level environmental surveillance applications, it may be advantageous to use the
865 second set of requirements of Level H (each measured value must be within ±30 percent of the
866 known value at 5 times the MDC) as part of the other validation levels as well. This requirement
867 will assess the method's ability to perform at the concentration ranges more commonly associated
868 with environmental samples. When process knowledge is available or the matrix under
869 consideration is unique or site-specific, it is best to validate the method using the matrix (e.g.,
870 MVRM) under consideration. This is extremely important for process/effluent waters versus
871 laboratory deionized water and for various heavy metal radionuclides in soils or sediments when
872 compared to spiked sand or commercial topsoil. For site-specific materials containing severe
873 chemical and radionuclides interferences, many methods have been unable to properly address
874 the magnitude of interferences.
875 6.6.4 Method Validation Documentation
876 Method validation, depending on the required level of validation, can be accomplished by the
877 project manager sending PT samples to the laboratory or by the laboratory using internal or
878 external PT/QC samples. When PT samples are sent to a laboratory to evaluate or validate the
879 laboratory's method and capabilities, the appropriate technical representative should retain all
880 records dealing with applicable method validation protocols (Section 6.6.3), PT sample
881 preparation certification, level of validation (from Table 6.1), results, and evaluations. The
882 laboratory should provide the necessary documentation to the project manager for these PT
883 samples as required by the SOW. The laboratory should request feedback from the project
884 manager as to the method performance. This information, along with the sample analytical
885 results documentation, should be retained by the laboratory for future method validation
886 documentation.
887 When the laboratory conducts its own method validation, all records, laboratory workbooks, and
888 matrix spike data used to validate an analytical method should be retained on file and retrievable
889 for a specified length of time after the method has been discontinued.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 6-32 DRAFT FOR PUBLIC COMMENT
-------
Selection and Application of an Analytical Method
890 6.7 Analyst Qualifications and Demonstrated Proficiency
891 The required level of qualification of an analyst is commensurate with the degree of difficulty
892 and sophistication of the method in use. The selection of the analyst for the method application is
893 typically determined initially on experience, education and proven proficiency in similar
894 methods. Basic guidance for the minimum education and experience for radioassay laboratory
895 technicians and analysts has been provided in Appendix E and ANSI N42.23.
896 For radiochemical methods, there may be several analysts involved. At most major laboratories,
897 different individuals may be involved in the sample preparation, radiochemistry, and radiation
898 detection aspects of the method. In these cases, the entire staff involved in the method should
899 undergo method proficiency tests to demonstrate their ability to meet quality requirements and
900 performance goals. The staff involved in the initial validation of an acceptable method would be
901 considered proficient in their particular role in the method application and the results of their
902 performance should be documented in their training records.
903 Successful proficiency is established when the performance of the analyst or staff meet
904 predefined quality requirements defined in the laboratory's quality system or a SOW, as well as
905 processing goals. Parameters involved in operational processing goals are typically turnaround
906 time, chemical yields, frequency of re-analyses (percent failure rate), and frequency of errors.
907 The continued demonstrated analyst proficiency in the method is usually measured through the
908 acceptable performance in internal QC and external PE programs associated with routine sample
909 processing.
910 6.8 Method Control
911 Method control is an inherent element of a laboratory's quality system. Simply stated, method
912 control is the ongoing process used to ensure that a validated method continues to meet the
913 expected requirements as the method is routinely used. Method control is synonymous with
914 process control in most quality systems. For a laboratory operation, method control can be
915 achieved by the application of the following:
916 • Controlled method manual (latest revision and signature sign-off);
917 • NIST traceable calibration standards and the conduct of an instrument QC program that
918 properly evaluates the variable parameters on an appropriate frequency;
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 6-33 DO NOT CITE OR QUOTE
-------
Selection and Application of an Analytical Method
919 • Radiotracers or chemical yields for each sample and the evaluation of the measured chemical
920 yield values to expected ranges;
921 • Internal QC and external PT samples to determine deviations from expected quality
922 performance ranges;
923 • Standard operating procedures for troubleshooting "out of control" situations; and
924 • Problem reporting, corrective action, and quality improvement process.
925 The above method control elements are typically addressed in the quality manual of the
926 laboratory or the project plan document for the project under consideration. Refer to Chapter 18
927 for additional information.
928 6.9 Continued Performance Assessment
929 The assessment of a laboratory's continued performance is covered in detail in Chapter 7.
930 However, it is important to briefly discuss certain aspects of evaluating a method's continued
931 performance from a laboratory's perspective.
932
933 In order to properly perform statistical analyses or compliance interpretation of the analytical data
934 produced from an analytical method, it is assumed that data quality does not vary significantly.
935 Therefore, the user of the data expects that the overall data quality will not change throughout the
936 program or project. From a laboratory management perspective, a performance indicator system
937 should be in place that assesses and provides feedback on the quality of the routine processing.
938 The most useful and cost-effective means of assessing a method's performance is through the
939 implementation of internal QC or external performance evaluation programs or both. Of course,
940 it can be argued that method assessment through a QC or PE program evaluates the combined
941 performance of the method and the analyst. However, statistical and inferential interpretation of
942 the QC/PE data can provide insight into whether the method is failing or whether an analyst is
943 underperforming. Chapters 7 and 18 and Appendix C provides guidance on quality control
944 programs and the use of the internal laboratory QC or external PE data to assess the laboratory's
945 performance in meeting performance criteria.
946 The laboratory management should use the internal QC program to detect and address
947 radioanalytical issues before the client does. Many SOWs require the use of internal QC samples
948 for every batch of project samples (Chapter 18). In effect, the client is essentially setting the level
949 of internal quality control and the frequency of method performance evaluation. It should be
MARLAP JULY 2001
DO NOT CITE OR QUOTE 6-34 DRAFT FOR PUBLIC COMMENT
-------
Selection and Application of an Analytical Method
950 recognized that an internal QC program evaluates method performance related to the initial
951 calibrations or internal "known values." An external NIST-traceable PE program will explain
952 method biases relative to the national standard or to the agency's PE program.
953 Some users of laboratory services have developed "monitoring" laboratory programs (ANSI
954 N42.23). For these programs, the user engages a recognized independent monitoring laboratory
955 to intersperse double- and single-blind external PT materials into batches of normal samples
956 submitted to a laboratory. The complexity and frequency of the monitoring laboratory PT
957 samples vary among programs, projects, and Federal and state agencies. An external double-blind
958 PE program conducted by a monitoring laboratory using site-specific matrices probably provides
959 the most realistic estimate of the method's or laboratory's true performance. When the
960 monitoring laboratory is traceable to NIST, either directly or through a NIST reference laboratory
961 (ANSI N42.23), the monitoring laboratory program will provide an estimate of any method bias
962 as related to the national standard.
963 Method performance can also be determined, although on a less frequent basis, through the
964 laboratory's participation in the various PE programs. For a laboratory providing services to
965 government agencies, the participation in such programs is typically a requirement. The PE
966 programs commonly send out non site-specific PT materials on a quarterly or semiannual basis.
967 The laboratory's performance in certain PE program is public knowledge. Such information is
968 useful to project managers in selecting a laboratory during the laboratory selection and qualifying
969 processes. Similar to the monitoring laboratory, when the laboratory conducting the PE program
970 is traceable to NIST, either directly or through a NIST reference laboratory (ANSI N42.23), the
971 PE program may provide an estimate of the bias as related to the national standard as well as the
972 precision of the method, depending on the distribution of replicate samples.
973 Some projects require that all analytical results received from a laboratory undergo a data
974 verification and validation process. Chapter 8 provides more detail on these processes. When
975 properly conducted, certain aspects and parameters of the method can be assessed during the data
976 verification and validation process.
977 Internal and external audits/assessments are also key elements in a laboratory's quality system to
978 assess the continuing performance of a method (Chapter 7). The level and frequency of the audits
979 and assessments typically vary according to the magnitude and importance of the project and on
980 the performance of the laboratory. Another quality system element that is very effective is a self-
981 assessment program. A functioning and effective self-assessment program may identify
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 6-3 5 DO NOT CITE OR QUOTE
-------
Selection and Application of an Analytical Method
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
1001
1002
1003
1004
1005
1006
1007
1008
weaknesses or performance issues more readily and timely than formal internal and external
audits.
6.10 Documentation To Be Sent to the Project Manager
The documentation related to the life cycle of a method application is essentially the information
gathered during the use of the method. A formal method documentation program is unnecessary
since the information should be part of the quality system documentation. Documented
information available from the quality system, related to a method's development, validation, and
control, include the following:
• Method validation protocol and results;
• Analyst training and proficiency tests;
• Method manual control program;
• Instrument calibration and QC results;
• Internal QC and external PT sample results;
• Internal and external assessments; and
• Corrective actions.
Data verification and validation information should be kept available and retained for those
projects requiring such processes. In addition to QA documentation, the analytical results, either
in hard copy or electronic form, should be available from the laboratory for a specified length of
time after the completion of a project.
Summary of Recommendations
MARLAP recommends the performance-based approach for method selection.
MARLAP recommends that only methods validated for a project's application be used.
MARLAP recommends that a SOW containing the MQOs and analytical process
requirements be provided to the laboratory.
MARLAP recommends that the SOW include the specifications for the action level and
the required method uncertainty for the analyte concentration at the action level for each
analyte/matrix.
MARLAP
DO NOT CITE OR QUOTE
6-36
JULY 2001
DRAFT FOR PUBLIC COMMENT
-------
Selection and Application of an Analytical Method
1009
1010
1011
MARLAP recommends that as each new project is implemented, the methods used in the
analysis of the associated samples undergo some level of validation.
MARLAP recommends that a tiered approach (Table 6.1) be taken for method validation.
1012 6.11 References
1013 American National Standards Institute (ANSI) N42.23. Measurement and Associated
1014 Instrumentation Quality Assurance for Radioassay Laboratories. 1996.
1015 American Public Health Association (APHA) 1989. Standard Methods for the Examination of
1016 Water and Waste Water. Washington, DC.
1017 American Society for Testing and Materials (ASTM) D 2777. Standard Practice for
1018 Determination of Precision and Bias of Applicable Test Methods of Committee D-19 on
1019 Water.
1020 American Society for Testing and Materials (ASTM) D 4855. Standard Practice for Comparing
1021 Methods.
1022 American Society for Testing and Materials (ASTM) E 177. Standard Practice for Use of the
1023 Terms Precision and Bias in ASTM Test Methods.
1024 American Society for Testing and Materials (ASTM) E 1169. Standard Guide for Conducting
1025 Ruggedness Tests.
1026 American Society for Testing and Materials (ASTM) E 1488. Standard Guide for Statistical
1027 Procedures to Use in Developing and Applying ASTM Test Methods.
1028 U.S. Environmental Protection Agency (EPA). 1980. Prescribed Procedures for Measurement of
1029 Radioactivity in Drinking Water. Environmental Monitoring and Support Laboratory,
1030 Cincinnati, OH. EPA 600-4-80-032, August.
1031 U.S. Environmental Protection Agency (EPA). 1998. Guidance for Quality Assurance Project
1032 Plans EPA QA/G-5. EPA 600-R-98-018, February.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 6-37 DO NOT CITE OR QUOTE
-------
7 EVALUATING METHODS AND LABORATORIES
2 7.1 Introduction
3 This chapter provides guidance for the initial and ongoing evaluation of radioanalytical labora-
4 tories and methods proposed by laboratories. Appendix E, Contracting Laboratory Services.,
5 provides additional guidance on the initial laboratory evaluation. More details about evaluating
6 and overseeing a laboratory's performance can be found in ASTM El691 and ASTM E548.
7 The performance-based approach to method selection allows a laboratory the freedom to propose
8 one or several methods for a specific analyte/matrix combination that will meet the needs of the
9 Analytical Protocol Specifications (APSs) and measurement quality objectives (MQOs)
10 delineated in the Statement of Work (SOW). However, the laboratory should demonstrate,
11 through a method validation process, that the method is capable of producing analytical results of
12 quality that meet the needs of the SOW (Chapter 5). Guidance and recommendations on the
13 selection of an analytical method based on the performance-based approach were presented in
14 Chapter 6. Section 7.2 of this chapter provides guidance on how to evaluate the methods
15 proposed by a laboratory. Section 7.3 provides guidance on the initial evaluation of a laboratory,
16 and Section 7.4 discusses the continual evaluation of the quantitative measures of quality and
17 operational aspects of the laboratory once sample processing has commenced.
18 Method applicability and performance compliance should be demonstrated prior to the initiation
19 of the sample analyses, as well as during the project period. A defined logical process for demon-
20 strating and documenting that the analytical method selected meets the project's data needs and
21 requirements may involve, for example, a review of the method validation documentation, an
22 evaluation of past performance data from other proj ects (if available), the analysis of external
23 performance evaluation (PE) program results, the analysis of matrix-specific standard reference
24 materials (or method validation reference materials) sent during the initial work period and
25 throughout the project, and the final evaluation of the protocol's performance during the data
26 verification and validation process. Chapter 8, Radiochemical Data Verification and Validation.,
27 covers the final evaluation of the protocol's performance.
28 In addition to the evaluation of the analytical methods, the capability of the laboratory to meet all
29 SOW requirements needs to be reviewed and evaluated. Supporting information, such as method
30 validation documentation, safety manuals, licenses and certificates, and quality manual are typi-
31 cally submitted with the response to the Request for Proposals (RFP). A generic evaluation of the
32 laboratory operation may be conducted during the initial laboratory audit or assessment. This
33 may be an initial onsite audit. This first evaluation covers those generic SOW requirements
34 dealing with the laboratory's capability and operation, including verification of adequate
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 7-1 DO NOT CITE OR QUOTE
-------
Evaluating Methods and Laboratories
35 facilities, instrumentation, and staffing and staff training and qualifications. Following the first
36 audit, emphasis should be on ensuring the laboratory continues to meet the APSs through a
37 continuous or ongoing evaluation effort.
38 7.2 Evaluation of Proposed Analytical Methods
39 A laboratory may submit several methods for a particular APS contained in the SOW, but each
40 method should be evaluated separately and, if appropriate, approved by the project manager or
41 designee. The method should be evaluated to be consistent with the overall analytical process
42 that includes the proposed field sampling and preservation protocols (Chapter 1). The project
43 manager may delegate the method review process to a technical evaluation committee (TEC) that
44 has a radioanalytical specialist. MARLAP recommends that a radioanalytical specialist review
45 the methods for technical adequacy. The acceptance, especially of a new method, may be the
46 most critical aspect of the performance-based approach for method selection. Acceptance of the
47 method requires the project manager to verify that the method is scientifically sound.
48 Each step of the method should be evaluated by a radioanalytical specialist in order to understand
49 how the results are derived. These steps may involve sample digestion, analyte purification and
50 decontamination steps that use ion exchange, solvent extraction, precipitation or oxidation/
51 reduction applications. Once these steps have been reviewed, and the method evaluation data
52 (e.g., from method validation documentation or various performance evaluation results) confirm
53 that the proposed method is acceptable, the project manager should have the confidence
54 necessary to endorse and verify the use of the method in the analysis of the routine samples.
55 As discussed in Chapter 6, the laboratory should provide method validation and analytical data
56 that demonstrates method performance. The data should show conclusively that the proposed
57 method meets the requirements as defined by the APSs. If method performance is questionable,
58 additional data may be required. For such cases, the project manager may decide to send per-
59 formance testing (PT) materials to the laboratory in order to evaluate or validate the method. The
60 preparation of the PT material used to evaluate the method should be based on sound scientific
61 principles and representative of the expected sample matrix (see Chapter 6 on method validation
62 options using site-specific materials). If there is sufficient reason to believe that the PT material
63 is an adequate substitute for the sample matrix and that the laboratory will follow the same
64 method, then the need to justify each step in the method may be drastically reduced.
65 7.2.1 Documentation of Required Method Performance
66 Certain documentation submitted by the laboratory with the proposed methods, as well as
MARLAP JULY 2001
DO NOT CITE OR QUOTE 7-2 DRAFT FOR PUBLIC COMMENT
-------
Evaluating Methods and Laboratories
67 available external information on the laboratory's analytical performance, should be reviewed
68 and evaluated by the radioanalytical specialist. Table 7.1 outlines where such information can be
69 typically found by the TEC. This section will discuss various information categories that may be
70 available during the method evaluation process.
71 7.2.1.1 Method Validation Documentation
72 Chapter 6 outlines the various method validation options that can be specified by the project
73 manager. In the MARLAP process, the method validation requirements will be contained in the
74 SOW. The laboratory must submit the necessary method validation documentation consistent
75 with the SOW specification. The laboratory may choose to validate a method to a higher degree
76 of validation or to submit method validation documentation for a higher degree of validation than
77 that specified by the SOW. The radioanalytical specialist or project manager should review the
78 documentation to ensure that validation criteria for the number of analyte concentration levels
79 and replicates meet or exceed the required validation criteria (Chapter 6, Table 6.1). Although
80 not specified in the method validation protocol, some laboratories may include chemical and
81 analytical interferences in their method validation plan to gain a perspective on the method's
82 specificity and ruggedness. However, it should be noted that the graded approach to method
83 validation presented in Chapter 6 does inherently increase the degree of ruggedness in terms of
84 having the method address site-specific materials which may include chemical and radionuclide
85 interferences.
86 In addition to reviewing the documentation for compliance with the method validation protocol,
87 the results of the method validation process should be evaluated to determine if the project
88 specific MQOs will be met. The method validation may or may not have been specifically
89 conducted for the project at hand. When the method has been validated (Chapter 6, Section 6.6)
90 to the SOW specifications (validation level and MQOs), then evaluation of the documentation
91 can be straight forward. If the method has been previously validated for the MQOs of other
92 projects, then the laboratory should provide a justification and calculations to show that the
93 method validation results will meet the MQOs for the new project. The TEC should verify these
94 calculations and review the assumptions and justifications for reasonableness and technical
95 correctness.
96 7.2.1.2 Internal Quality Control or External PE Program Reports
97 The documentation of internal QC and external PE program results should be reviewed relative
98 to the MQOs. Method uncertainty and internal biases can be estimated from the information
99 available in the laboratory's internal quality control reports, summaries of batch QC results that
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 7-3 DO NOT CITE OR QUOTE
-------
Evaluating Methods and Laboratories
ta
I
o
•s
s-
I
Q-i
O
8
I
O
o
M
I
5 n
O U
& •£
o — o
•^ ^ SM
5 " C/5
o b
•a*!
HH
Ji
W V)
2 4>
<3
Evaluation Element
Addressed
^f
^P
Analyte/Matrix
^f
Process Knowledge
^p
Previous Experience
^p
•
o
o
_
^F
Radiological Holding
Time
^F
•
O
O
Turnaround Time
^f
^P
Unique Process
Specifications
^f
O
t
•
^p
VI
a
3
^P
O
0
•
w
Method Uncertainty
(MQQ/MDC
^p
0
•
w
Analyte/Interference
Range
^p
•
0
O
w
Method Ruggedness
^p
•
0
O
^p
Method Specificity
^j
S
<5 c
ion shouk
ion may b
||
> >•
3 2
O3 ^
1> D
(D P
B B
.2 -2
o o
^^
jl
ui t«
-------
Evaluating Methods and Laboratories
100 may be submitted with the RFP response and external PE program reports. The TEC should
101 review these documents and, when possible, estimate the method uncertainty and bias for various
102 analyte concentration levels. However, it is imperative that no confusion exists in terms of what
103 method produced the results: the proposed method or another method available to the laboratory.
104 This is especially important when reviewing external PE program results. It should also be noted
105 that although a laboratory may meet performance acceptance criteria for an external PE program,
106 this fact may have no bearing on whether the method will meet the MQOs of the SOW.
107 Review of the internal batch QC data can provide additional information on typical sample
108 analysis times and rates of blank contamination and sample reanalysis. This information is
109 important when comparing methods (from the same or between laboratories) in terms of APS
110 characteristics. The frequency of blank contamination would be very important to national char-
111 acterization studies (groundwater or soil analyses) for the determination of ambient analyte
112 levels. Method evaluation for these projects may weight the blank contamination rate more
113 heavily than other SOW parameters. The rate of sample re-analysis would be important to
114 projects having pending operations that are conducted based on a short sample processing turn-
115 around time (TAT). In some site remediation projects, the contractor may remain onsite pending
116 analytical results. A delay in reporting data or not meeting a TAT due to sample re-analysis may
117 be costly. Projects of this nature may weight TAT and low sample re-analyses more heavily than
118 other SOW parameters.
119 7.2.1.4 Method Experience, Previous Projects, and Clients
120 When permitted by former clients, the laboratory may submit information relative to the previous
121 or ongoing clients and projects for which the proposed method has been used. The TEC should
122 verify with the laboratory's clients that the laboratory has previous experience using the method.
123 When available and allowed, the information should also include the analyte(s) and interferences
124 and their applicable concentration range, matrix type, and project size in terms of the number of
125 samples per week or other time periods. From this information, the TEC can evaluate whether or
126 not to contact the laboratory's client for further information on the operational adequacy of the
127 method. The client may offer some information on the quality of the results based on their
128 external single- or double-blind QC program, percent completion of reports, TAT, and sample re-
129 analysis frequency. The sharing of laboratory assessment reports may be advantageous when
130 reviewing the performance of the laboratory during its employment of the method.
131 7.2.1.5 Internal and External Quality Assurance Assessments
132 When available, internal and external quality assurance assessment reports should be evaluated to
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 7-5 DO NOT CITE OR QUOTE
-------
Evaluating Methods and Laboratories
133 determine the adequacy of the method performance based on previous projects. Problems with
134 the conduct of the method due to procedural and technical issues may be readily evident. These
135 issues may include an ineffective corrective action program creating delayed remedies to
136 problems, insufficient understanding of the method, inadequate training of staff, internal and
137 project-specific QC issues, and higher-than-expected failure rates for sample TATs and re-
138 analyses. Information in these reports may disclose problems with a particular method that are
139 not common to another proposed method. As such, the TEC may give one method a higher
140 weighting factor than another method.
141 7.2.2 Performance Requirements of the SOW—Analytical Protocol Specifications
142 Under the performance-based approach to method selection, a laboratory will propose one or
143 several analytical methods that can meet the stated APSs and MQOs in the SOW for a given
144 analyte and matrix combination. Chapters 3, 5, and 6 discuss the APSs and MQOs in detail in
145 terms of their basic description, their inclusion in a SOW, and as key considerations for
146 identifying existing validated methods or developing new methods. The purpose of this section is
147 to provide guidance on what available information should be evaluated in order to approve the
148 various proposed methods.
149 The following subsections cover key aspects of the SOW that should be addressed during the
150 method evaluation and approval process.
151 7.2.2.1 Matrix and Analyte Identification
152 The TEC should review the method(s) proposed by the laboratory to determine if the method
153 under evaluation is applicable for the analyte/matrix combination specified in the SOW. In some
154 cases, several methods may be proposed, including gross screening methods and specific
155 radionuclide or isotopic methods having high specificity and ruggedness (Section 6.5.1.1 has
156 additional guidance). Each method should be evaluated on its own application and merit. When
157 methods are proposed by the laboratory that use alternative nuclides (such as decay products) to
158 determine the analyte of interest, the TEC should carefully review the objective or summary of
159 the method to determine if the proposed method is truly applicable for the analyte of interest
160 given the radiological holding time and MQOs (i.e., can it properly quantify the analyte of
161 interest through decay progeny measurements?). For gross screening techniques, the TEC should
162 evaluate the analyte's decay scheme to determine the underlying gross radiation category (beta,
163 alpha, X-ray, or gamma-ray emitting) and the applicability of the proposed method's radiation
164 detection methodology.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 7-6 DRAFT FOR PUBLIC COMMENT
-------
Evaluating Methods and Laboratories
165 Each proposed method should be evaluated to determine if the method can analyze the sample
166 matrix identified in the SOW. A method validated for water cannot be applied to soil samples
167 without modification and validation (Section 6.5). The planning team should have made—
168 through historical process knowledge, previous matrix characterization studies or common
169 experience—a determination on the uniqueness of the site-specific matrices compared to typical
170 matrices and provided guidance in the SOW as to the level of method validation. In addition, if
171 the radioanalytical specialist of the project planing team is concerned that the physiochemical
172 form of the analyte or the sample matrix substrate may present special problems to the radio-
173 analytical process, a detailed description of the analyte and matrix should have been included in
174 the SOW. Chapters 12 and 13 discuss possible sample matrix problems and Section 6.5 provides
175 guidance on the need for method validation. The radioanalytical specialist should carefully
176 review the summary of the method to determine if the proposed method is applicable for the
177 sample matrix.
178 At this point, if it is determined that the proposed method(s) is not applicable and cannot meet
179 the SOW specifications, there is no need to continue the method evaluation process.
180 7.2.2.2 Process Knowledge
181 The radioanalytical specialist should review the process knowledge information and determine if
182 the proposed method is capable of addressing these issues by virtue of its specificity, ruggedness
183 and applicability. Discussions on method specificity and ruggedness may be found on in
184 subsections on pages 7-13 and 7-15, respectively.
185 As discussed in Section 6.5.2 and above, process knowledge is extremely important for identify-
186 ing potential radioanalytical problems on some projects. Historical information or process
187 knowledge may identify chemical and radionuclide interferences, expected analyte and inter-
188 fering radionuclide concentration ranges, sample analyte heterogeneity issues, and the physio-
189 chemical form of the analyte, and the sample matrix substrate. In some special cases, it may be
190 necessary to determine if the radiological holding time will be an issue if the laboratory must
191 analyze an alternative nuclide to determine supported and unsupported radionuclides (decay
192 progeny nuclides) in the matrix.
193 7.2.2.3 Radiological Holding and Turnaround Times
194 The radioanalytical specialist should review the proposed method in light of the radiological
195 holding time, analyte's half-life and typical sample delivery options and determine if the method
196 is capable of meeting the MQOs in a reasonable counting period given the typical method param-
JULY2001 MARLAP
DRAFT FOR PUBLIC COMMENT 7-7 DO NOT CITE OR QUOTE
-------
Evaluating Methods and Laboratories
197 eters (such as sample weight processed, chemical yields, radiation detection efficiency, branching
198 ratio and background, ingrowth periods for decay progeny analysis, etc.). Radiological holding
199 time is defined as the time between the sample collection and the end of the sample analysis (end
200 of final measurement), while sample processing TAT refers to the time between sample receipt at
201 the laboratory and the issuance of an analytical report. The physical (analyte's half-life) and
202 chemical (stability or preservation concerns) characteristics of the analyte, as well as biological
203 degradation for some matrices, usually will dictate the radiological holding time. Project-specific
204 schedules and practicalities related to project and laboratory processing capacities normally enter
205 into establishing TATs. If the radiological holding time appears to be a critical issue, then the
206 laboratory should submit information on the typical batch size being processed by the method.
207 This information is needed in the method evaluation and review process. Without special
208 problems (e.g., inadvertent delay of sample delivery), the laboratory should be able to meet the
209 MQOs with a good margin of error for the maj ority of the samples processed. For very short-
210 lived analytes, too large a batch size may result in the last samples in the batch having difficulty
211 in meeting the radiological holding time. For short-lived analytes, counting the sample (or final
212 processing products) longer typically is not practical because the analyte is decaying too rapidly
213 to make any gain counting the sample longer.
214 In some cases, the laboratory may want to propose two methods for a short-lived analyte: one for
215 normal delivery and processing schedules and another method for situations when lower detec-
216 tion limits are needed. An example of such a situation is the analysis of 131I in environmental
217 media. A method with adequate detection limits for reasonable radiological holding times is
218 gamma spectrometry. Another method that can be applied for lower detection limits or longer
219 radiological holding times is radiochemical separation followed by beta-gamma coincidence
220 counting.
221 Certain projects may be concerned with the chemical speciation of the analyte in the sample. For
222 these projects, the radiological holding time should have been specified to ensure that the chem-
223 ical species are not altered prior to processing. The project normally should specify chemical
224 preservation specifications applicable at the time of sample collection.
225 In the case of biological media, sample deterioration (Chapter 10) may become a problem, and
226 biological preservatives should be added to the sample to retard degradation. However, the
227 radiological holding time should be specified to limit problems with sample degradation. The
228 radioanalytical specialist should evaluate the method in light of the foregoing information and
229 determine its adequacy to meet the radiological holding time and the pertinent MQOs
230 A laboratory's sample (processing) TAT for a method typically is not related to the method's
MARLAP JULY 2001
DO NOT CITE OR QUOTE 7-8 DRAFT FOR PUBLIC COMMENT
-------
Evaluating Methods and Laboratories
231 technical basis unless the radiological holding time and the TAT are nearly equal for a short-
232 lived analyte. However, sufficient time should be available between the completion of sample
233 analysis and the delivery of the analytical report. Meeting the radiological holding time but
234 failure to meet the TAT will not affect the quality of the analytical results but may place a
235 hardship on the project to meet schedules. The TEC should review the proposed method, the
236 radiological holding time and the TAT to determine if the method can process the samples in a
237 reasonable time period to meet the TAT. The sample delivery rate, sample batch size, level of
238 data automation and the laboratory's existing sample processing capacity will affect the
239 laboratory's ability to meet the TAT requirement.
240 7.2.2.4 Unique Processing Specifications
241 The TEC should review the proposed methods for compliance or applicability to unique sample
242 processing specifications stated in the SOW. Chapter 6 provides a limited discussion on what a
243 project may identify as unique or special sample process specifications. Examples may include
244 chemical speciation, analyte depth profiles, analyte particle size distribution, analyte hetero-
245 geneity within the sample, wet-to-dry analyte concentration ratios in biologicals, and possible
246 scaling factors between radionuclides in the sample. In some cases, the proposed method(s) for
247 the analyte(s) may have to be evaluated with respect to all analytes or other sample preparation
248 specifications in order to determine method applicability and adequacy.
249 7.2.2.5 Measurement Quality Objectives
250 Method performance characteristics (Method Uncertainty, Quantification Capability, Detection
251 Capability, Applicable Analyte Concentration Range, Method Specificity, and Method
252 Ruggedness) will be discussed in the following subsections. For a particular project, MQOs
253 normally will be developed for several (but not all) of the performance characteristics discussed
254 below.
255 METHOD UNCERTAINTY
256 The SOW should specify the required method uncertainty at a stated analyte concentration (or
257 activity level) for each sample matrix and the level of method validation (Section 6.6) needed to
258 qualify the method at the stated analyte concentration.
259 MARLAP uses the term "method uncertainty" to refer to the predicted uncertainty of a result that
260 would be measured if a method were applied to a hypothetical laboratory sample with a specified
261 analyte concentration. As presented in Chapter 6 and formulated in Chapter 19, the method
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 7-9 DO NOT CITE OR QUOTE
-------
Evaluating Methods and Laboratories
262 uncertainty of the analyte concentration for a given method is determined by mathematically
263 combining the standard uncertainties of the many input quantities (parameters), involved in the
264 entire radioanalytical process. This will involve making some assumptions and normally involve
265 using typical or worst case values for a conservative estimate of the method uncertainty. Some of
266 these input quantities, and thus the method uncertainty, vary according to analyte level or concen-
267 tration in the final measured product; others do not. In some cases, the magnitude of the method
268 uncertainty for an analyte may increase in proportion to the magnitude (concentration/activity) of
269 any interfering radionuclide present in the final measurement product. Therefore, it is imperative
270 that the TEC evaluate the laboratory's submitted documentation relative to this requirement,
271 especially the information provided on method specificity, given the historical or expected inter-
272 fering nuclides and the needed decontamination factors (chemical separation factors) to render a
273 good measurement for the analyte of interest.
274
275 In evaluating the documentation relevant to meeting the method uncertainty requirement, it is
276 important to determine if the method validation requirements stated in the SOW have been met.
277 The TEC should review the submitted method validation documentation and verify that the
278 method's performance meets the requirements of Table 6.1 (Chapter 6) for the specified valida-
279 tion level. It is important that the laboratory submit definitive documentation of method
280 validation compliance for the method uncertainty requirement.
281 The method performance documentation may include documentation or data from method
282 validation, internal or external (organization sending QC samples) QC data, external PE program
283 data, and results of pre-qualifying laboratories by sample analyses. By evaluating the actual QC
284 and PE program performance data, it can be determined if the quoted measurement uncertainty
285 for a reported QC sample result (calculated by the laboratory) truly reflects the method uncer-
286 tainty under routine processing of samples. The required method uncertainty can be viewed as a
287 target value for the overall average measurement uncertainty for the samples at a specified
288 analyte concentration. It is important that the precision, as calculated from repeated measure-
289 ments, is consistent with the laboratory's stated measurement uncertainty for a given sample
290 result whose analyte concentration is near the specified concentration. If the quoted measurement
291 uncertainty of a QC or test measurement is quoted to be ± 10 percent and QC or PE program data
292 indicates a data set standard deviation of ± 20 percent, then the laboratory may not have
293 identified all possible uncertainty components or may have underestimated the magnitude of a
294 component.
295 QUANTIFICATION CAPABILITY
296 A requirement for the quantification capability of a method and the required method validation
MARLAP JULY 2001
DO NOT CITE OR QUOTE 7-10 DRAFT FOR PUBLIC COMMENT
-------
Evaluating Methods and Laboratories
297 criteria may be specified in a SOW. The quantification capability, expressed as the minimum
298 quantifiable concentration (MQC), is the smallest concentration of the analyte that ensures a
299 result whose relative standard deviation is not greater than a specified value, usually 10 percent.
300 The project manager or TEC should review available documentation on the method to determine
301 if the laboratory can meet the method quantification requirement. Method validation documen-
302 tation sent by the laboratory should demonstrate explicitly, or by extrapolation, that the method,
303 using certain input quantities and their uncertainties, can meet the quantification requirement.
304 The method validation acceptance criteria presented in Section 6.6 have been formulated to eval-
305 uate the MQC requirement at the proper analyte concentration level, i.e., action level or other
306 specified analyte concentration.
307 Some projects may send performance testing material spiked at the MQC level as a more in-
308 depth verification of the compliance with this requirement. Laboratories may also submit docu-
309 mentation for internal QC or external PE program results that cover the MQC value. The TEC
310 should evaluate the reported results to determine if the MQC requirement can be met.
311 DETECTION CAPABILITY
312 A radiochemical method's detection capability for an analyte is usually expressed in terms of
313 minimum detectable concentration (MDC) or activity (MDA). Chapter 19 provides the definition
314 and mathematical equations for the MDC1 and MDA. A MDC requirement for each analyte/
315 matrix combination may be stated in a SOW. Any proposed method should document the basis
316 and equation for calculating the MDC. The supporting documentation on the method should
317 contain the input quantity values that may be entered into the MDC equation to calculate the
318 detection capability under a variety of assumptions. The TEC should evaluate the assumptions
319 and parameter values for reasonableness and practicality. This evaluation is especially important
320 for recently validated methods that have a limited routine processing history. It is recommended
321 that the TEC perform an independent calculation of the method's MDC using laboratory-stated
322 typical or sample-specific parameters.
323 When the proposed method has been validated recently or previously used on similar projects,
324 sufficient data should exist that either are directly related to testing the method's detection capa-
325 bility or can be used to estimate the method's detection capability. Any data submitted that
326 document direct testing of the method's detection capability should be reviewed for appropri-
327 ateness or applicability, reasonableness, and accuracy. If method detection testing is performed, it
'The MDC should not be confused with the concept of the critical value (Chapter 19).
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 7-11 DO NOT CITE OR QUOTE
-------
Evaluating Methods and Laboratories
328 normally will be for one analyte concentration level or value. It should not be expected that the
329 MDC testing process included varying the magnitude of the method's many parameters over a
330 wide range.
331 The reported quantitative results of the blanks can be used to estimate the MDC to within a
332 certain degree of confidence (for most methods). At or below the MDC value, the majority of the
333 measurement uncertainty typically is due to the Poisson counting uncertainty. For well-controlled
334 methods, the uncertainties of the other method parameters (input quantities), such as sample
335 weight, detection efficiency, and chemical yield, may range up to 10 percent. Therefore, a simple
336 rule of thumb to estimate the MDC for most methods involves reviewing the measurement
337 uncertainty for the reported blank results. If the blanks were analyzed to meet the MDC
338 requirement, then the reported MDC (based on blank and sample paired observations) for most
339 methods should be between 3 and 4 times the measurement uncertainty of the blank when the
340 background counts (per measurement interval) are greater than 10. It is more complicated to
341 estimate the MDC for methods that use low background detectors (such as alpha spectrometry)
342 having background counts less than 10 per counting interval. The TEC should evaluate the blank
343 data to determine the reasonableness of the quoted MDC values. These rules of thumb can be
344 applied to actual samples when the quoted analyte concentration value is less than two times its
345 associated combined standard uncertainty value.
346 APPLICABLE ANALYTE CONCENTRATION RANGE
347 The applicable analyte concentration range can vary substantially depending on whether the
348 project deals with process waste streams, environmental remediation or monitoring, or environ-
349 mental or waste tank characterization research. The proposed method being evaluated should
350 provide accurate results over the analyte concentration range stated in the SOW. Acceptable
351 analytical results used in this context means consistent method uncertainty (at a given analyte
352 concentration) and without significant bias. The range may be over several decades, from a
353 minimum value (the MDC for some projects) to 100 times the action level or MQC.
354 Due to the effects of the Poisson counting uncertainty, most methods will provide more precise
355 results at higher analyte concentration levels compared to those concentration levels near zero. At
356 concentration levels near zero, background effects will render the results less precise. If the
357 background (instrument or ambient levels of analyte in the matrix) is not well characterized, a
358 bias may also exist. For projects or programs (environmental characterization research) that have
359 no action level requirement, the lower portion of the required concentration range or the MDC
360 requirement may be most important. For those situations, particular emphasis should be placed
361 on evaluating method and reagent blank data (i.e., net results that take into account inherent
MARLAP JULY 2001
DO NOT CITE OR QUOTE 7-12 DRAFT FOR PUBLIC COMMENT
-------
Evaluating Methods and Laboratories
362 analyte content in the reagents or tracers) to ensure that a bias does not exist. Refer to Section
363 7.2.2.6, "Bias Considerations," on page 7-15 for additional guidance.
364 Typically, radiation detection systems are linear in signal response over a very large range of
365 count rates. However, depending on the magnitude of the chemical or radionuclide interferences
366 in the sample, the method may not produce linear results over the entire application range.
367 Therefore, it is critical that when a mixture of radionuclides is present in a sample, the method
368 must provide sufficient "analyte selectivity/isolation or impurity decontamination" to ensure
369 valid results and "method linearity." In some cases, such as that for pure beta-emitting analytes,
370 the degree of needed decontamination from other interfering nuclides may be as much as six
371 orders of magnitude.
372 There are several sources of information available from the laboratory that should be reviewed
373 and possibly evaluated to ensure the method is capable of meeting this MQO. These include
374 method validation documentation, previous projects or experience using the method, PE program
375 results, internal and external QC sample results, and pre-qualifying test samples. When evalua-
376 ting the data, the TEC should evaluate the method's performance as a function of analyte concen-
377 tration with and without interferences. However, this evaluation would be most valid when the
378 samples were processed to the same MQO (especially MDC or MQC), a situation that may not
379 be realistic for different projects. If the MDC requirement results in a longer counting time from
380 one project to another, there may be an impact on the method's uncertainty for a given analyte
381 concentration due to difference in the Poisson counting uncertainty. Bias typically is not affected
382 by increasing the counting time. A graphical plot of this data would be visually helpful and may
383 be used to determine if the method uncertainty requirement would be met at the action level
384 (extrapolation may be necessary).
385 METHOD SPECIFICITY
386 Method specificity refers to the ability of the method to measure the analyte of concern in the
387 presence of other radionuclide or chemical interferences. The need for or degree of method
388 specificity depends on the degree or magnitude of the interferences and their effect on the ability
389 to measure the analyte of interest. Gross alpha, beta, and gamma-ray methods are considered to
390 be methods of low specificity and are used when individual nuclide specificity is not possible or
391 needed. Radiochemical methods involving sample digestion, purification and decontamination
392 steps followed by alpha spectrometry, such as for 239Pu in soil, are considered methods of high
393 specificity. However, the relative degree of specificity of these nuclide specific methods depends
394 on the number of analyte isolation and interference decontamination steps. High resolution
395 gamma-ray spectrometry employing a germanium (Ge) detector is considered to have better
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 7-13 DO NOT CITE OR QUOTE
-------
Evaluating Methods and Laboratories
396 specificity than the lower resolution sodium iodide (Nal) gamma-ray spectrometry.
397 The TEC should evaluate the proposed methods for adequacy to meet the specificity require-
398 ments stated in the SOW. As mentioned in Chapter 6, methods of low specificity, such as gross
399 radiation detection methods, may be proposed if the methods meet the MQOs. For example,
400 when a single analyte having a relatively elevated action level needs to be evaluated, such as 60Co
401 in soil at an action level of 26 Bq/kg (0.7 pCi/g), then a method with less specificity (gross
402 counting methods for gamma-ray or beta emitting nuclides) may be sufficient to meet the MQOs.
403 For this example, a less expensive Nal gamma-ray spectrometric analysis with a lower resolution
404 capability may be more desirable compared to a more costly high resolution germanium gamma-
405 ray spectrometric analysis. If greater method specificity for a certain analyte/matrix combination
406 has been required in the SOW, then a high resolution non-destructive sample analysis method
407 (such as high resolution gamma-ray spectrometry) or a destructive sample analysis by a detailed
408 radiochemical method would be appropriate. For proposed methods of high specificity, it is
409 important that the TEC review and evaluate the basic purification and decontamination steps of
410 the method, or the resolution of the radiation detection system, for adequacy in relation to the
411 expected mixture of analytes and interferences. For radiochemical methods, the TEC may be able
412 to estimate the needed distribution/partition coefficients, extraction and solubility factors, etc., of
413 the various purification steps and compare the values against the needed decontamination factors
414 for the interfering chemical or radionuclide interferences.
415 The adequacy of method specificity can be evaluated by the analytical results from the analysis of
416 site-specific PT materials during method validation and/or laboratory pre-qualifying tests. A
417 further discussion on the use of these materials is presented below.
418 METHOD RUGGEDNESS
419 Method ruggedness refers to the ability of the method to produce accurate results over wide
420 variations in sample matrix composition and chemical and radionuclide interferences, as well as
421 when steps (such as pH adjustments) in the method are varied slightly by the analyst. For some
422 projects, the matrix composition and level of analyte or interferences may very dramatically in a
423 given project.
424 Ruggedness studies have been defined by EPA (1998). A testing protocol for method ruggedness
425 has been outlined by the American Public Health Association (APHA). Some laboratories may
426 have developed methods according to the APHA protocol for method ruggedness or are using
427 methods contained in standards methods (APHA, 1989). Documentation on any internal
428 ruggedness study may be available from the laboratory.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 7-14 DRAFT FOR PUBLIC COMMENT
-------
Evaluating Methods and Laboratories
429 As mentioned in Chapter 5 and 6, the use of site-specific PT materials is a means of testing the
430 ruggedness of a method for a defined project. If ruggedness and method specificity are concerns
431 due to the sample matrix of a defined project, then a variety of site-specific performance testing
432 materials should be sent to the laboratory as part of the pre-qualification process or as a method
433 validation requirement. National PE programs, such as DOE's Multiple Analyte Performance
434 Evaluation Program (MAPEP) and Quality Assessment Program (QAP), use generic PT
435 materials and may not be applicable or representative of the matrices for a defined project. The
436 results of the pre-qualifying or method validation processes using site-specific PT materials
437 should be evaluated by the TEC to determine the adequacy of the method to meet this MQO
438 parameter. If the sample matrix and analytes are fairly standard, then no other evaluation of the
439 available information may be necessary.
440 7.2.2.6 Bias Considerations
441 The method proposed by the laboratory should produce analytical results that are unbiased.
442 MARLAP considers bias to be a persistent difference of the measured result from the true value
443 of the quantity being measured, which does not vary if the measurement is repeated. Normally,
444 bias cannot be determined from a single result or a few results (unless the bias is large). Bias may
445 be expressed as the percent deviation in (or deviation from) the "known" analyte concentration.
446 Since bias is estimated by repeated measurements, there will be an uncertainty in the calculated
447 value. It is incumbent upon the project manager or TEC to evaluate the proposed methods for
448 possible bias over the applicable analyte concentration range. A laboratory should eliminate all
449 known biases before using a method. However, there may be circumstances, such as the
450 processing of site-specific sample matrices, that may produce some inherent bias that is difficult
451 to assess or correct in a reasonable time or economical fashion. For the methods proposed, the
452 project manager must determine if the magnitude of the bias will significantly affect the data
453 quality.
454 A bias can be positive or negative. Methods may have a bias at all analyte concentration levels
455 due to the improper determinations of chemical yield, detector efficiency or resolution, subtrac-
456 tion of interferences, and improper assumptions for the analyte's half-life or an emission
457 branching ratio. When reporting an analyte concentration based on a decay progeny analysis,
458 improper ingrowth assumptions may lead to a bias.
459 It is recommended that the project manager or TEC evaluate the available data provided by the
460 laboratory or from performance evaluations for bias, based on multiple analyses covering the
461 applicable analyte concentration range. One means of estimating a bias is through the evaluation
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 7-15 DO NOT CITE OR QUOTE
-------
Evaluating Methods and Laboratories
462 of external PE program data.2 For proper evaluation of the PE program sample results, it is
463 essential that the PE program provider use sample preparation techniques that will produce
464 performance testing (PT) samples (or a sample distribution) having insignificant "within or
465 between" sample analyte heterogeneity and whose analyte concentrations are accurately known.
466 For the purpose of evaluating whether a laboratory method has an observable bias based on
467 multiple laboratory internal QC samples (matrix or method spikes) or external PE program
468 samples, the following equations can be used:
Dj =100
Known
Known
V
(7.1)
469 where D; is the percent deviation, Xj is an individual analytical result and Y; known is the "known"
470 value for the sample analyzed. The D; should be determined for each test sample in the data set.
471 The mean percent deviation for the method for a series of analyses in the data set can be
472 estimated by the equation:
(7.2)
N
473 Refer to various references (ASTM D2777, NBS 1963, Taylor 1990) for applicable tests that may
474 be performed to determine if there is a statistical difference at a given significance level.
475 There may be a negative or positive bias at low analyte concentrations due to the improper
476 determination of the appropriate detector background or analytical blank value. For an individual
477 blank result, the result (net activity or concentration value) would be considered to be a
478 statistically positive value if the magnitude of its value is greater than 1.65 times the quoted
479 measurement uncertainty. An older, much more conservative approach was to consider a reported
480 value as a positive value when the magnitude of a result was greater than 3 times the measure-
481 ment uncertainty.
482 Since the measurement process is statistical in nature and involves the subtraction of an
483 appropriate background or blank which also has an uncertainty, there is a 50 percent probability
2 In order to standardize against the national standard (NIST), an external performance evaluation program should
be implemented by a well-qualified provider that has standardized its reference materials to NIST or is participating
in a NIST traceability program
MARLAP JULY 2001
DO NOT CITE OR QUOTE 7-16 DRAFT FOR PUBLIC COMMENT
-------
Evaluating Methods and Laboratories
484 (half of the results) that the analytical result for a blank sample will have a negative magnitude,
485 e.g., -1.5 ± 2.0. For an individual blank measurement, the measurement may be considered to be
486 problematic when the negative magnitude is greater than 2 or 3 times the measurement
487 uncertainty.
488 For most radionuclides, other than those that are naturally occurring, the major source of a
489 positive blank is from contamination, either cross-contamination from other samples or dirty
490 glassware during sample processing or from tracer impurities. A poor estimate of the instrument
491 background or ambient analyte levels in the matrix/reagent can lead to results being too negative
492 in magnitude. A statistical test should be performed on a series of the data results to determine if
493 there is a negative bias. The relative importance of the negative bias depends on the magnitude of
494 the negative bias, magnitude of the action level and type of project.
495 7.3 Initial Evaluation of a Laboratory
496 The basic information to be considered in the initial evaluation of a laboratory has been
497 summarized according to major categories in Figure 7.1. Not all categories will be discussed in
498 detail as subsections. Some categories may be grouped and discussed under a single generic
499 subsection heading. In order to allow for flexibility, no definitive guidance or detailed acceptance
500 criteria for the parameters under discussion will be provided.
501 7.3.1 Review of Quality System Documents
502 A radiochemical laboratory providing usable analytical data should have a quality manual. A
503 review of this document by a knowledgeable evaluator can reveal a great deal about the quality
504 and acceptability of the laboratory relative to the work to be performed. A well-developed quality
505 manual contains a description of the quality system and descriptive material covering most other
506 aspects of a laboratory's operation. The standard operating procedures, method documentation,
507 list of instrumentation, and personnel resumes should be reviewed. For some projects, the project
508 manager may require the laboratory to develop a specific project quality plan, system, and
509 manual. The following items, taken from the NELAC Quality Systems (NELAC 2000), should be
510 discussed at a minimum:
511 • Organization and management
512 • Quality system establishment, audits, essential quality controls and evaluation, and data
513 verification
514 • Personnel (qualifications and resumes)
515 • Physical facilities (accommodations and environment)
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 7-17 DO NOT CITE OR QUOTE
-------
Evaluating Methods and Laboratories
INITIAL EVALUATION OF LABORATORY
GENERIC SPECIFICATIONS
FOR LABORATORY
OPERATIONS
CAPABILITY TO MEET SOW
SPECIFICATIONS
ADMINISTRATIVE / PROJECT
MANAGEMENT
ADEQUACY OF FACILITIES,
INSTRUMENTATION, AND STAFF
LEVELS
QUALITY SYSTEM
DOCUMENTATION
EXPERIENCE BASE FOR SOW
ANALYTICAL REQUIREMENTS
QUALITY MANUAL
Organization and Management
Quality Sytem - Establishment, Audits,
Essential Quality Controls and
Evaluations, and Data Verification
Personnel
Physical Facilities - Accomodations and
Environment
Equipment and Reference Materials
Measurement Traceabiltty and Calibration 9
Method and Standard Operating
Procedures
Sample Handling, Sample Acceptance
Policy, and Sample Receipt
Records
Subcontracting Analytical Samples
Outside Support Services and Supplies
Complaints
CAP ABILITY TO MEET SAMPLE
PROCESSING AND REPORTING
REQUIREMENTS
RADIOAHALYTICAL METHODS
APPLICABILITY I QUALITY
METHOD REVIEW
«. COMPARABILITY
METHOD
VALIDATION
DOCUMENTATION
DEMONSTRATED
QUALITY ON
SIMILAR PROJECTS
516
517
518
519
520
521
522
523
FIGURE 7.1 — Considerations for the initial evaluation of a laboratory
Equipment and reference materials
Measurement traceability and calibration
Test methods and standard operating procedures (methods)
Sample handling, sample acceptance policy and sample receipt
Records
Subcontracting analytical samples
Outside support services and supplies
Complaints
524 The laboratory evaluation should involve a review of the quality system documents for
525 completeness, thoroughness, and clarity.
MARLAP
DO NOT CITE OR QUOTE
7-18
JULY 2001
DRAFT FOR PUBLIC COMMENT
-------
Evaluating Methods and Laboratories
526 7.3.2 Adequacy of Facilities, Instrumentation, and Staff Levels
527 Many factors enter into a laboratory's ability to meet the analytical requirements of a SOW. The
528 resources and facilities of a laboratory may become stretched depending on the number of clients,
529 the analytical services needed, and the deadlines of the committed work activities. Some SOWs
530 may request information about the current workload of the laboratory and available facilities,
531 staff and nuclear instrumentation for the specified work scope. The resources needed will vary
532 considerably depending on the analysis and number of samples: from minimal bench space,
533 hoods, and nuclear instrumentation for fairly simple gross analyses to maximum bench space,
534 hoods, staff, and nuclear instrumentation for low-level analyses of soil. In addition, the laboratory
535 capacity also depends on the number of samples that are routinely processed in a batch. Various
536 factors may control the batch size, including the hood processing area, bench space, and
537 equipment setup, available number of radiation detectors, counting time, and half-life of
538 radionuclide, among others.
539 The adequacy of the facilities, instrumentation, and staff levels can be estimated by two general
540 mechanisms: detailed supporting information in the SOW and an initial onsite audit. Information
541 received from the prospective laboratory may provide an estimate of the laboratory's resources,
542 but an initial onsite audit goes verifies the actual existence and maintenance of the resources.
543 7.3.3 Review of Applicable Prior Work
544 If required in a SOW, a laboratory will provide a list of clients for whom radioanalytical services
545 had been performed that are considered comparable in terms of work scope, DQOs, MQOs,
546 APSs, and project type. A written or oral verification of the client list should be performed. As
547 part of the verification process, the following items related to adherence to contract or project
548 requirements should be discussed and documented:
549 • Radionuclides analyzed;
550 • Sample matrices types;
551 • Laboratory capacity (number of samples per week or another time period);
552 • MQO for method uncertainty, detection and quantification capability;
553 • Radiological holding times;
554 • Sample turnaround times;
555 • Corrective actions; and
556 • Communications related to schedule, capacity, or quality issues.
557 It should be noted that under performance-based contracting, a laboratory's prior work for an
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 7-19 DO NOT CITE OR QUOTE
-------
Evaluating Methods and Laboratories
558 agency should be considered, either as a positive or negative performance weighting factor, when
559 scoring a laboratory's performance during the technical evaluation process.
560 7.3.4 Review of Performance Indicators
561 Some laboratories compile a semiannual or annual QA report summarizing the internal QC
562 sample results for the methods used during a given time period, as well as an internal quality
563 assessment report summarizing the internal and external audit findings and corrective actions
564 taken. Although the laboratory's internal quality criteria for a given radionuclide/matrix may be
565 different from the project MQOs, the internal QC sample results can be used to gauge the
566 laboratory's performance capabilities. If these documents are available, they should be reviewed
567 for documentation of process control and pertinent quality parameters such as bias, precision,
568 unusually high number of positive blank detection, chemical recoveries, turnaround times,
569 number of recurring deficiencies or findings, and corrective action effectiveness.
570 7.3.4.1 Review of Internal QC Results
571 A quality assessment report may contain a summary of various QA-related activities, including
572 internal audits and surveillance, report of conditions adverse to quality, investigation requests,
573 corrective actions, and the results of external PE programs and internal QC samples. The content
574 and frequency of the reports normally are outlined in the laboratory's quality manual. Frequently,
575 this type of quality assessment report may be submitted with the laboratory's response to the RFP
576 without request. The TEC may want to specifically request such a report when available.
577 When the laboratory's quality system is effectively implemented, the information contained in
578 these QA reports can be used not only to gauge the quality of the analyses but also the effective-
579 ness and timeliness of such quality system activities as identifying conditions adverse to quality,
580 controlling and monitoring the radioanalytical quality using internal QC samples, and corrective
581 actions. The internal QC sample results can be used to gauge the laboratory's performance
582 capability. Results of the QC samples for a radionuclide and sample matrix should be reviewed
583 for both the batch QC samples and single- or double-blind samples submitted by the QA officer.
584 Batch QC samples typical include laboratory control samples, method blanks, matrix spikes, and
585 duplicates. Such parameters as acceptable percent deviation for spiked samples, acceptable
586 precision as measured by duplicate sample analyses, false nuclide detection, positive blanks, and
587 compliance to internal quality requirements should be reviewed, depending on the type of QC
588 sample. The single- and double-blind samples submitted independently by the QA officer are
589 considered more operationally independent than the batch QC samples.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 7-20 DRAFT FOR PUBLIC COMMENT
-------
Evaluating Methods and Laboratories
590 When quality problems are observed by the reviewer, it is important to check if the laboratory's
591 quality system also has found and reported the same problem and whether an investigation or
592 corrective action has been undertaken.
593 Additional specific guidance is provided in Chapter 18 on evaluating internal QC samples to
594 meet internal laboratory QC performance criteria. It is recommended that the project managers
595 review this chapter to gain a perspective on how to use reported internal QC results to gauge a
596 laboratory's potential to meet project MQOs.
597 7.3.4.2 External PE Program Results
598 Typically, a laboratory's performance or capability to perform high quality radioanalyses can be
599 evaluated through two external PE program mechanisms. The first mechanism, which may not be
600 available for all projects, is the submittal, as an initial laboratory evaluation process, of project-
601 specific PT samples prepared by the organization or a contracted source manufacturer. When
602 previous knowledge or experience exists, well-characterized site-specific matrix samples
603 containing the nuclides of interest can be used. This approach can use site-specific matrix
604 materials for background samples or for samples spiked with target analytes. For this evaluation
605 mechanism, and depending on the number and type of samples, the laboratory's capability to
606 meet all proposed project MQOs and quality performance specifications may be evaluated.
607 The second mechanism, available to most projects, is the laboratory's participation in
608 government or commercial PE programs for radioanalyses. Each PE program has its own
609 acceptable performance criteria related to a laboratory's bias with respect to the PE program's
610 "known" analyte concentration value. Acceptable performance criteria are established for each
611 nuclide/matrix combination. A PE program may also evaluate a laboratory based on a false
612 positive analyte detection criterion. Typically, the laboratory's performance data in government
613 PE programs are provided in reports available to the public.
614
615 The project manager should be aware that the acceptable performance criteria used by the PE
616 programs may be inconsistent with or more lenient than the MQOs of the project. The
617 laboratory's performance should be evaluated in terms of the established MQOs of the project
618 rather than a PE program's acceptable performance criteria. In some cases, the laboratories could
619 be ranked as to their level of performance in these programs.
620 7.3.4.3 Internal and External Quality Assessment Reports
621 Most laboratories undergo several external and internal QA audits per year, with resultant audit
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 7-21 DO NOT CITE OR QUOTE
-------
Evaluating Methods and Laboratories
622 reports. Typically, a summary of the findings and commitments of internal and external quality
623 audits or assessments are tracked on some type of QA database as part of the laboratory's
624 corrective action process. Access to the audit reports or database information may be limited.
625 This information is not normally requested as part of the RFP process, nor do most laboratories
626 submit such information with their response to an RFP. Therefore, obtaining previous QA audit
627 information from a laboratory outside a formal, external, onsite audit process may be limited.
628 7.3.5 Initial Audit
629 An initial assessment or audit may be performed to provide assurance that a potentially selected
630 laboratory is capable of fulfilling the project requirements in accordance with the SOW.
631 Essentially, the objectives of an initial audit are twofold. The first objective is to verify that what
632 the laboratory claims in response to the SOW or RFP, such as the various quality and safety
633 programs, are being correctly and fully implemented, and when used during the project period,
634 will ensure that stipulated requirements will be met. The second objective is to determine if the
635 laboratory has the instruments, facilities, staffing levels and other operational requirements
636 available to handle the anticipated volume of work. In other words, is the laboratory's proposal
637 realistic when compared to the actual facilities? To answer this question, auditors will be looking
638 to see whether a candidate laboratory has all the required elements to meet the project needs.
639 Detailed guidance and information on what should be evaluated in an initial audit has been
640 provided in Appendix E, Section E5.5 and Table E7. This section also contains recommendations
641 on the key items or parameters that should be reviewed during the initial audit. Depending on the
642 project, other quality or operational parameters/requirements (such as requirements related to
643 chemical speciation or subsampling at the laboratory) not covered in Appendix E should be
644 included in the initial audit plan.
645 7.4 Ongoing Evaluation of the Laboratory's Performance
646 The evaluation framework presented here is intended to be sufficiently generic to cover the
647 operations of a laboratory performing work according to a SOW as recommended in Chapter 5.
648 As described in MARLAP, MQOs are a key component of the SOW. Therefore, the sample
649 schedule, analyses to be performed, MQOs, and other analytical requirements have been defined.
650 The methods selected by the laboratory have been demonstrated to meet the MQOs and have
651 been approved by the project manager. In addition, the laboratory and its programs should have
652 undergone an initial audit to ensure that the laboratory has met or is capable of meeting project
653 requirements, including sample processing capacity, sample TATs, deliverables for analytical
654 reports, etc. This would include maintaining a satisfactory quality system that includes
MARLAP JULY 2001
DO NOT CITE OR QUOTE 7-22 DRAFT FOR PUBLIC COMMENT
-------
Evaluating Methods and Laboratories
655 monitoring and controlling the radioanalytical processes through an instrument and internal
656 sample QC program and the acceptable performance in an external PE program.
657 The ongoing evaluation of a laboratory's performance includes the evaluation of the method
658 applicability or the quality of the data produced, and assessing the laboratory's quality system
659 and operations through onsite or desk audits or assessments. The continued method performance
660 can be evaluated through the laboratory's internal sample QC program, a possible external QC
661 program maintained by the project manager, or an external PE program. It should be noted that
662 samples used to control and monitor the quality of laboratory analyses have been defined
663 according to their use. For example, batch or external QC samples are used to control as well as
664 monitor the quality of the analytical process (the process can be stopped immediately if the QC
665 sample results indicate that the process is outside appropriate SOW specifications or laboratory
666 control limits). As defined previously, PT samples are used to compare the performance of the
667 radioanalytical processing to some acceptance criteria but are not used to control the process.
668 The ongoing evaluation of the laboratory quality system and operations is accomplished through
669 a visit to the laboratory or by a desk audit (the review of records and data from the laboratory).
670 These audits or assessments are more focused on whether the laboratory is meeting project
671 specifications rather than whether the laboratory has the capability to meet project or SOW
672 requirements.
673 Once a laboratory has initiated work on a project, the laboratory's performance should be
674 evaluated for the duration of the project. The quality of the radioanalytical measurements, as well
675 as the pertinent key operational aspects of the laboratory, should be evaluated against the
676 requirements of the MQOs and SOW. Both the quantitative and qualitative measures of
677 laboratory performance should be evaluated on a continual basis. In addition, the operational
678 aspects of the laboratory germane to the effective implementation of the project requirements
679 should be evaluated/monitored on a continual basis.
680 7.4.1 Quantitative Measures of Quality
681 The laboratory's ongoing demonstrated ability to meet the MQOs and other APS requirements
682 can be evaluated through various quantitative measures using internal QC data and external PE
683 program QC data. From these data, quantitative tests, as outlined in Appendix C can be used to
684 measure and monitor the MQO parameters on a short-term basis. Also, the QC and PE program
685 data can be used to evaluate the laboratory's performance, on a long-term trending basis, in
686 meeting other quality related parameters such as bias and precision, unusually high number of
687 positive blank detection, false nuclide detection, MDC or MQC adherence, radiological holding
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 7-23 DO NOT CITE OR QUOTE
-------
Evaluating Methods and Laboratories
688 times, etc. The following subsections will discuss the use of data from these samples to evaluate
689 the laboratory's radioanalytical quality with respect to the requirements.
690 7.4. 1 . 1 MQO Compliance
691 MARLAP recommends that project specific MQOs be established and incorporated into the
692 SOW for laboratory radioanalytical services. Appendix C provides guidance on developing the
693 MQOs for method uncertainty, detection capability, and quantification capability. Establishing a
694 gray region and action level are important to the development of the MQOs. For certain research
695 programs and characterization studies, the concept of an action level may not be applicable. For
696 these studies or programs, the MDC requirement and restrictions on the frequency of false
697 positive detections may be more important. As such, the project planning team for these
698 programs should establish the basis for their own MQOs and develop tests to evaluate a
699 laboratory's performance to meet the requirements. These tests may be different from those
700 presented below.
701 MARLAP recommends that a MQO for method uncertainty be established for each analyte/
702 matrix combination. The method uncertainty is affected by laboratory sample preparation, sub-
703 sampling, and the analytical method. In the absence of other information, the required method
704 uncertainty (HMR) at the upper bound of the gray region (UBGR) may be defined as:
UMR = ~^ (7.3)
705 where u^ is the method uncertainty and A is the width of the gray region (difference between the
706 upper and lower bounds of the gray region) as defined in Appendix C. In terms of the relative
707 fraction of the upper bound of the gray region (action level), (pMft is defined:
(74)
708 The following subsections describe methods to quantitatively monitor a laboratory's performance
709 relative to meeting this principal MQO through the use of internal or external batch QC samples.
710 In some cases, the laboratory's internal quality program may have more restrictive quality control
711 limitations for method performance compared to the proposed control limits used by the project
712 manager to monitor adherence to the MQO for method uncertainty. Evaluation of the labora-
713 tory's performance in NIST-traceable external PE programs will determine the degree of bias of
714 the laboratory's method with respect to the national standard, as opposed to the determination of
MARLAP JULY 2001
DO NOT CITE OR QUOTE 7-24 DRAFT FOR PUBLIC COMMENT
-------
Evaluating Methods and Laboratories
715 the laboratory's internal bias through the use internal QC samples. The tests presented assume
716 that all known internal (related to QC values and calibrations) and external (calibration differ-
717 ences with respect to the national standard) biases have been defined and eliminated and, as such,
718 the difference between the measured result and the "expected known" value is a result of the
719 method uncertainty only.
720 USE OF INTERNAL QC SAMPLE RESULTS
721 For most projects, the SOW will specify that the laboratory incorporate internal QC samples
722 within a defined batch of samples. The QC samples may include a laboratory control sample,
723 sample duplicates, a matrix spike sample and a method or reagent blank, or both. Appendix C
724 provides examples on the use of the following quantitative tests to measure a laboratory's
725 performance in meeting the MQO for method uncertainly.
726 Quality Performance Tests and Acceptance Criteria for Quality Control Samples
111 Laboratory Control Sample (LCS). The analyte concentration of an LCS should be high enough
728 so that the resulting Poisson counting uncertainty is small and the relative uncertainty limit cp^ is
729 appropriate with respect to the action level and the spike concentration chosen. The percent
730 deviation (%D) for the LCS analysis is defined as
%D= SSR"SAxlQO% (75)
SA ( }
731 where
732 SSR is the measured result (spiked sample result) and
733 SA is the spike activity (or concentration) added.
734 It is assumed that the uncertainty of SA is negligible with respect to the uncertainty of SSR.
735 Refer to Appendix C for the basic assumption and limitation of this test. For long-term trending,
736 the %D results should be plotted graphically in terms of a quality control chart as described in
737 Chapter 18. The warning and control limits on %D are summarized below:
738
739
740
741
Laboratory Control Samples
Statistic: %D
Warning limits: (± 2(pm) x 100%
Control limits: (± 3qm) x 100%
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 7-25 DO NOT CITE OR QUOTE
-------
Evaluating Methods and Laboratories
742
743
744
745
746
747
748
749
Duplicate Analyses. The acceptance criteria for duplicate analysis results depend on the analyte
concentration of the sample, which is estimated by the average x of the two measured results xl
and x2.
~x=^-^- (7.6)
When x< UBGR, the absolute difference] xl - x2\ of the two measurements is used in the testing
protocol. For these tests, only upper warning and control limits are used, because the absolute
value]xl - x2 \ is being tested.
When x> UBGR, the acceptance criteria may be expressed in terms of the relative percent
difference (RPD) defined as
RPD
x 100%
x
(7.7)
750 The requirements for duplicate analyses are summarized below.
751
752
753
754
755
756
757
758
759
760
761
762
Duplicate Analyses
If *UBGR:
Statistic: RPD
Warning limit: 2
Control limit: 4
Method Blanks. When an
However, the measured v
control uncertainty limits
ri " xi 1
83 "MR
24 u
- v 100%
x
83 (pivjR x 100%
o/i „ y i nn°/
Z4- (pMR x lUU/o
aliquant of a blank material is analyzed, the target value is zero.
alue may be either positive or negative. The applicable warning and
for blank samples are defined as:
MARLAP
DO NOT CITE OR QUOTE
7-26
JULY 2001
DRAFT FOR PUBLIC COMMENT
-------
Evaluating Methods and Laboratories
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
783
784
785
786
Method Blanks
Statistic:
Warning limits:
Control limits:
Measured Concentration Value
±2u,
MR
±31
^
Matrix Spikes. The acceptance criteria for matrix spikes are more complicated than those
described above for the other laboratory QC samples because of the pre-existing activity that is
inherent to the unspiked sample. The pre-existing activity (or concentration) must be measured
and subtracted from the activity measured after spiking.
MARLAP recommends the "Z score," defined below, as the test for matrix spikes.
z= SSR - SR - SA
max(SR, UBGR)2
(7.8)
where:
SSR is the spiked sample result,
SR is the unspiked sample result,
SA is the spike concentration added (total activity divided by aliquant mass), and
max(SR,UBGR) denotes the maximum of SR and UBGR.
The warning and control limits for Zare set at ± 2 and ± 3, respectively. It is assumed that the
uncertainty of SA is negligible with respect to the uncertainty of SSR. For long-term trending, the
Z results should be plotted graphically in terms of a quality control chart, as described in Chapter
18.
782 The requirements for matrix spikes are summarized below.
Matrix Spikes
Statistic: Z =
Warning limits: ± 2
Control limits: ± 3
SSR-SR-SA
+ max(SR, UBGR)2
JULY 2001
DRAFT FOR PUBLIC COMMENT
7-27
MARLAP
DO NOT CITE OR QUOTE
-------
Evaluating Methods and Laboratories
787 USE OF EXTERNAL PE PROGRAM AND QC SAMPLE RESULTS
788 Information on a laboratory's performance in an external PE program or from double-blind QC
789 samples is very useful in monitoring a laboratory's ability to meet MQOs. A PE program will
790 provide a snapshot in time whereas external QC samples included with samples submitted to the
791 laboratory permit a continuous evaluation of the method's performance. When traceable to NIST,
792 the PE program will elucidate any measurement or instrument calibration biases as related to the
793 national standard. An external QC program may not have NIST traceability, and thus calibration
794 biases to the national standard would not be determined.
795 For monitoring the performance of a laboratory using external PE program and QC sample
796 results, the tests provided in the previous subsection ("Use of Internal QC Sample Results," page
797 7-25) may be used when there are sufficient data. The test equations assume that the project has
798 an MQO for method uncertainty at a specific concentration. In addition, it is assumed that the
799 Poisson counting uncertainty for the radioanalysis of these samples is minimal.
800 Results from PE Programs
801 In many SOWs, the laboratory is required to participate in a recognized PE program for the
802 nuclides and media of interest. In some cases, a certificate of participation may be needed as part
803 of response to the RFP. However, it also should be noted that although a laboratory may meet
804 performance acceptance criteria for an external PE program, this fact may have no bearing on
805 whether the method will meet the MQOs of the SOW.
806 Monitoring ongoing laboratory performance is limited due to the minimum frequency of testing
807 of the PE program, i.e., usually quarterly or semiannually. Some PE programs require multiple
808 measurements to estimate precision but most only request a single result be reported. In addition,
809 the concentration of the analyte typically never approaches an action level value and the media
810 used are not site specific. For PE program samples, when possible, the laboratory should analyze
811 a sample to reach a la Poisson counting uncertainty that is less than five percent.
812 Multiple Analyses and Results
813 When a PE program requires the analysis of multiple samples, the laboratory's measurement
814 precision and bias (to a "known value") at the analyte concentration may be estimated and
815 reported by the PE program provider. When only duplicates sample results are reported, then the
816 tests for laboratory control samples and duplicate analyses given in the previous section should
817 be used. The duplicate analysis test can be used as is, but the laboratory control sample test
MARLAP JULY 2001
DO NOT CITE OR QUOTE 7-28 DRAFT FOR PUBLIC COMMENT
-------
Evaluating Methods and Laboratories
818 should be evaluated based on the mean of the duplicate results. By using the mean of the two
819 results, the LCS test provides a better estimate of any laboratory measurement bias with respect
820 to the PE program provider. As discussed in Appendix C, the measurement (combined standard)
821 uncertainty of each measured result value should be smaller than the required u^ or cp^.
822 Results from External QC Samples
823 The project manager may elect to establish an external QC program wherein QC samples are
824 submitted to the laboratory with each batch of routine samples for the purpose of "controlling,"
825 rather than monitoring, the quality of the analytical processes. The types of QC samples may
826 include matrix spikes, blanks, and possibly duplicates if prepared under controlled and exacting
827 protocols. An agency may use a qualified reference or monitoring laboratory (ANSI N42.23) to
828 prepare the performance testing materials. When available, these QC samples may be prepared
829 from site-specific materials.
830 When acceptance criteria are not met, the organization may issue a stop-work order and request
831 corrective actions and reanalysis before routine processing can resume. In order to do this, the
832 SOW must define the performance acceptance criteria and stipulate that the agency or
833 organization has the right to stop laboratory processing when the performance requirements are
834 not met. This application is not widespread but may have merit for certain project types. For
835 example, research or national monitoring programs may monitor groundwater for specific
836 naturally occurring radionuclides at state-of-art detection levels. For these programs, frequent
837 false positive results, due to the application of incorrect instrument background or an analytical
838 blank to the analytical result, would be unacceptable. Rather than permit a high rate of false
839 positive results to continue, the agency can use the external batch QC samples to detect problems
840 early and have the laboratory discontinue sample processing until a root cause is discovered and a
841 corrective action undertaken. Non-conformance of a single analysis to performance criteria
842 would not warrant the issuance of a stop work order unless a severe blunder has occurred.
843 Typically, a certain amount of statistical trending of the data is in order to truly elucidate
844 deficiencies.
845 Since the number of QC samples is similar to the recommendations for the laboratory's internal
846 batch QC samples, there should be sufficient data for trending. The statistical tests provided in
847 the section on "Use of Internal QC Sample Results," beginning on page 7-25, may be applied to
848 these QC samples.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 7-29 DO NOT CITE OR QUOTE
-------
Evaluating Methods and Laboratories
849 7.4.1.2 Other Parameters
850 The laboratory's performance in meeting the requirements for the other APSs that are listed in
851 the SOW should be evaluated quantitatively when possible. In some cases, the information
852 needed to perform the evaluations may be found in the final analytical results data package. For
853 certain types of evaluations, a follow-up onsite or desk audit may be needed to complete the
854 evaluation, e.g., a review of logbooks on unique processes or software algorithms and the
855 analytical data base for proper spectral resolution.
856 RADIOLOGICAL HOLDING AND TURNAROUND TIMES
857 The data packages or analytical results report should contain the sample collection (reference),
858 sample analysis, and reporting dates. From this information, the radiological holding and sample
859 processing TATs can be calculated and compared against requirements. When a method uses a
860 decay progeny to measure the analyte of interest (222Rn to measure 226Ra), the decay of the parent
861 nuclide and ingrowth of the decay progeny are important parameters for evaluation. Unless
862 requested in the SOW, most laboratories do not report the ingrowth factor as a standard output.
863 Therefore, the information on the sample specific ingrowth factor may be available in the data
864 reports or during audits. When required, these time related requirements will be evaluated for
865 compliance during data verification and validation.
866 CHEMICAL YIELD
867 When appropriate, the SOW may specify limits on the chemical yield for each analyte. For
868 radionuclides, this requirement typically is related to the provision of robust or rugged methods
869 so that extreme yields become flags indicating potential problems. Wide swings in the chemical
870 yield may be indicative of method's difficulty handling matrix or radionuclide interferences. The
871 data packages or analytical results report should contain the chemical yield for each analyte
872 listed. This reported value can be compared to the SOW yield limit. When required, these
873 requirements will be evaluated for compliance during data verification and validation.
874 SPECTRAL RESOLUTION
875 Problems with spectral resolution of gamma-ray and alpha spectra cannot be evaluated through a
876 review of the analytical results report. If spectral resolution limits have been stated in the SOW,
877 the evaluator should review and evaluate each sample spectrum against the SOW limit. Spectral
878 information may be available in data packages when required or may be obtained during audits.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 7-30 DRAFT FOR PUBLIC COMMENT
-------
Evaluating Methods and Laboratories
879 During an initial audit, a preliminary evaluation of the method's SOP and review of past
880 performance data for spectral resolution should be undertaken. The TEC may want to determine
881 the baseline or typical spectral resolution for the radiation detection systems that will be used in
882 the analysis of project samples. Trends of the spectral resolution of each detection system during
883 the conduct of the project may be used to determine compliance with a spectral resolution
884 specification.
885 7.4.2 Operational Aspects
886 Once a laboratory begins providing radioanalytical services, certain operational aspects need to
887 be reviewed and evaluated periodically to determine if the laboratory is maintaining project
888 requirements or if new problems have occurred. It is also important to ensure that the laboratory
889 has been properly maintained and is operated and managed in a manner that will not create a
890 liability to any client. Many of the operational areas that were discussed in Sections 7.3.1 and
891 7.3.2 for the initial evaluation of a laboratory also should be evaluated periodically to ensure
892 commitments are being met. The audit frequency varies according to the organization and the
893 extent of the project or contract. Desk audits can be conducted more frequently than onsite audits
894 because they require fewer resources. However, not all operational aspects may be reviewed
895 during desk audits. The operational aspects that may be considered during desk and onsite audits
896 are presented below.
897 7.4.2.1 Desk Audits
898 A desk audit is conducted as an off-site activity, usually by a technical representative of the
899 project manager. A radioanalytical specialist should review all technical aspects of the desk
900 audit, including method and calculation (data reduction) changes, method performance,
901 instrument recalibrations, corrective actions, and case narratives. The desk audit is most useful
902 when performed periodically to monitor certain activities or programs following an extensive
903 onsite laboratory audit. However, for some smaller projects, the desk audit may be the only
904 assessment mechanism used to monitor the laboratory's operations. The desk audit may be used
905 to review or monitor the following operational aspects or items:
906 <> Organization and Management
907 o Changes in key personnel
908 o Reassignments
909 Quality System
910 o Internal and external audits conducted, including laboratory certification audits
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 7-31 DO NOT CITE OR QUOTE
-------
Evaluating Methods and Laboratories
911 o Corrective action implementations
912 o Quality control and performance evaluations
913 - Instrument and batch sample QC results
914 - External PE program results
915 o Laboratory data verification (narrative status reports)
916 o Additional method validation studies
917 0 Certificates, licenses, equipment, and reference materials
918 o Standard and tracer certificates
919 o New and updates to instrument calibrations
920 o Instrument repairs and new instruments put into service
921 o NRC/State radioactive materials licence updates
922 o State or EPA drinking water certification status changes
923 <> Personnel
924 o Updates to staff qualification/proficiency for methods
925 o Updates to staff training files
926 - Radiation and chemical safety
927 - Quality assurance
928 - Technical principles
929 - Hands-on training records
930 <> Radioanalytical Methods and Standard Operating Procedures
931 o Updates to methods and SOPs
932 o Technical basis for updates
933 o Detection limits or method uncertainty studies
934 0 Sample Receipt, Handling and Disposal
935 o Sample receipt acknowledgment
936 o Chain-of-custody
937 o Sample- and waste-disposal tracking logs and manifests
938 Desk audits may also be used to review the data packages provided by the laboratory and,
939 periodically, to verify certain method results by hand calculations. In addition, verification of
940 compliance to radiological holding and turnaround times may be performed during the desk
941 audit. In the absence of a full data verification and validation program (Chapter 8), the desk audit
942 may be used to periodically evaluate the detailed instrument and data reduction reports of the
943 data packages for method adherence, technical correctness and valid application.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 7-32 DRAFT FOR PUBLIC COMMENT
-------
Evaluating Methods and Laboratories
944 7.4.2.2 Onsite Audits
945 The onsite laboratory audit is more comprehensive and resource intensive than a desk audit. An
946 onsite audit typically is conducted to assess, periodically and in depth, a laboratory's capability to
947 meet project requirements. Section E.5.5 of Appendix E provides guidance on the conduct of an
948 initial onsite audit during a contract award process. EPA (1997) provides limited guidance on the
949 conduct of an audit for a radiological laboratory. NELAC (2000) provides some generic guidance
950 on laboratory assessments, although not specifically for a radiological laboratory.
951 Onsite audits usually cover the operational aspects delineated in Section 7.4.2.1 and also provide
952 an opportunity to evaluate the physical conditions at the laboratory, in terms of adequacy and
953 upkeep of the facilities, and the full application or conduct of programs and resources. Informa-
954 tion sent in data packages or submitted for desk audits can be confirmed or verified during an
955 onsite audit. Furthermore, an onsite audit permits the tracking of a sample from receipt through
956 processing to sample storage and disposition and can verify the related instrument and batch QC
957 samples specific to the sample being tracked. During an onsite audit, the auditors may have
958 interviews with the staff to gauge their technical proficiency and familiarity with methods.
959 For large projects, onsite audits may be formal in nature and have a predefined audit plan, which
960 has been developed by a designated audit team, for a specific project or program. The audit team
961 typically is comprised of qualified QA representatives and technical experts. MARLAP
962 recommends that the audit team include a radioanalytical specialist familiar with the project's or
963 program's technical aspects and requirements.
964 In addition to the items in Section 7.4.2.1 ("Desk Audits"), the following items and programs
965 should be assessed during an onsite laboratory audit:
966 0 Organization and Management
967 o Qualifications of assigned laboratory project manager
968 o Implementation of management's policy on quality
969 o Timeliness of addressing client complaints
970 o Timeliness of implementing corrective actions
971 0 Physical Facilities
972 o Adequacy of facilities (sample receipt, processing, instrumentation and storage areas,
973 waste processing and storage, offices, etc.)
974 o Physical conditions of facilities including laboratories, hoods, bench tops, floors, offices,
975 etc.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 7-33 DO NOT CITE OR QUOTE
-------
Evaluating Methods and Laboratories
976 o Environmental controls, such as climate control (heating, ventilation, air conditioning)
977 and electrical power regulation
978 o Sample processing capacity
979 o Sample storage conditions including chain-of-custody lockup areas and cross
980 contamination control (separation of samples by project and from radioactive sources or
981 wastes)
982 0 Instrumentation and Equipment
983 o Age of nuclear instrumentation and equipment
984 o Functionality of nuclear instrumentation and equipment
985 o Calibrations and QC logs
986 o Maintenance and repair logs
987 o Sample throughput capacity
988 o Contamination control for radiation detectors
989 o Background spectra of radiation detectors
990 <> Methods and Standard Operating Procedures
991 o Use of latest revisions of methods and SOPs (spot check method manuals used by
992 technical staff)
993 o Conformance to method application (surveillance of method implementation)
994 o Effectiveness of administering the controlled method manual
995 <> Certifications, Licenses and Certificates of Traceability
996 o Ensure existence and applicability of, and conformance to, certifications and licenses
997 o Noted citations during audits related to certifications and licenses
998 o Ensure use of NIST-traceable materials (calibration standards)/review of vendors' report
999 of NIST traceability
1000 <> Waste Management Practices
1001 o Adherence to waste management SOPs
1002 o Proper packaging, labeling, manifests, etc.
1003 o Sample storage and records
1004 o Training and qualification records
1005 <> Radiological Controls
1006 o Adherence to radiological safety SOPs
1007 o Contamination control effectiveness (spill control, survey requirements and adherence,
1008 posted or restricted areas, proper ventilation, cleaning policies, etc.)
MARLAP JULY 2001
DO NOT CITE OR QUOTE 7-34 DRAFT FOR PUBLIC COMMENT
-------
Evaluating Methods and Laboratories
1009 o Badging and survey adherence
1010 <> Personnel
1011 o Number and technical depth of processing staff
1012 o Training files
1013 o Testing/qualifications
1014 o Personal interviews to determine familiarity of methods and safety SOPs
1015 Quality Systems
1016 o Performance indicator program (feedback from program)Quality assurance reports (QC
1017 and audits) for all laboratory processing
1018 o Ongoing method evaluations and validations
1019 o Corrective action program (effectiveness and outstanding issues for all processing; spot
1020 check for implementation of corrective actions)
1021 o Records/reports related to audits of vendors used by laboratory
1022 o Reagent control program (spot check conformance for effectiveness)
1023 o Audits of laboratories that are subcontracted
1024 o Laboratory's data verification and validation processes
1025 0 Software Verification and Validation
1026 o Spot review of key method calculation and data reduction programs that include MDC,
1027 MQC, and measurement uncertainty; spectral unfolding routines or crosstalk factors;
1028 application of instrument background and analytical blanks; etc.
1029 o Spot verification of consistency between electronic data deliverable and data packages
1030 0 Radiological Holding and Sample Turnaround Times
1031 o Verification of compliance to radiological holding and sample TAT specifications (spot
1032 check samples and confirm paperwork)
1033
1034
1035
1036
1037
1038
Summary of Recommendations
MARLAP recommends that a radioanalytical specialist review the methods for technical
adequacy.
MARLAP recommends that project specific MQOs be established and incorporated into
the SOW for laboratory radioanalytical services.
MARLAP recommends that a MQO for method uncertainty be established for each
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 7-3 5 DO NOT CITE OR QUOTE
-------
Evaluating Methods and Laboratories
1039
1040
1041
analyte/matrix combination.
MARLAP recommends that an audit team include a radioanalytical specialist familiar with
the project's or program's technical aspects and requirements.
1042 7.5 References
1043 American National Standards Institute (ANSI) N42.23. Measurement and Associated
1044 Instrumentation Quality Assurance for Radioassay Laboratories.
1045 American Public Health Association (APHA) 1989. Standard Methods for the Examination of
1046 Water and Waste Water. Washington, DC.
1047 American Society for Testing and Materials (ASTM) D2777. Standard Practice for
1048 Determination of Precision and Bias of Applicable Test Methods of Committee D-19 on
1049 Water.
1050 American Society for Testing and Materials (ASTM) El 77. Standard Practice for Use of the
1051 Terms Precision and Bias in ASTM Test Methods.
1052 American Society for Testing and Materials (ASTM) E548. Standard Guide for General Criteria
1053 Used for Evaluating Laboratory Competence.
1054 American Society for Testing and Materials (ASTM) E1580. Standard Guide for Surveillance of
1055 Accredited Laboratories.
1056 American Society for Testing and Materials (ASTM) El 691. Standard Guide for Evaluation and
1057 Assessment of Analytical Chemistry Laboratories.
1058 U.S. Environmental Protection Agency (EPA). 1997. Manual for the Certification of
1059 Laboratories Analyzing Drinking Water. EPA 815-B-97-001.
1060 U.S. Environmental Protection Agency (EPA). 1998. Guidance for Quality Assurance Project
1061 Plans EPA QA/G-5. EPA 600-R-98-018, February.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 7-36 DRAFT FOR PUBLIC COMMENT
-------
Evaluating Methods and Laboratories
1062 International Organization for Standardization (ISO) 17025. General Requirements for the
1063 Competence of Testing and Calibration Laboratories. International Organization for
1064 Standardization, Geneva, Switzerland. 1999.
1065 National Bureau of Standards (NBS). 1963. Experimental Statistics. NBS Handbook 91, National
1066 Bureau of Standards, Gaithersburg, MD.
1067 National Environmental Laboratory Accreditation Conference (NELAC) 2000. Chapter 5,
1068 Quality Systems. July. Available at: http://www.epa.gov/ttn/nelac/.
1069 Taylor, John Keenan. 1990. Statistical Techniques for Data Analysis. Lewis Publishers, Chelsea,
1070 MI.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 7-37 DO NOT CITE OR QUOTE
-------
i 8 RADIOCHEMICAL DATA VERIFICATION AND
2 VALIDATION
3 8.1 Introduction
4 The goal of the data collection process is to produce credible and cost-effective data to meet the
5 needs of a particular project. The process can be divided into several stages, as illustrated in the
6 data life cycle (Chapter 1). This chapter is the first of two chapters that address the assessment
7 phase of the project. Because the efficiency and success of these assessment activities are heavily
8 dependent on the completion of the preceding steps in the data collection process, especially the
9 initial planning activity (Chapter 2), the integration of planning and assessment is discussed in
10 Section 8.2 prior to presenting material on data verification and validation.
11 Data verification compares the material delivered by the laboratory to the requirements in the
12 statement of work (SOW) and identifies problems, if present, that should be investigated during
13 data validation. Data validation compares the data produced with the measurement quality
14 objectives (MQOs) and any other analytical process requirements contained in the analytical
15 protocol specifications (APSs) developed in the planning process. It may not be necessary in all
16 instances to validate all project data. This chapter outlines a validation plan that specifies the data
17 deliverables and data qualifiers to be assigned that will facilitate the data quality assessment. The
18 project-specific data validation plan should establish a protocol that prioritizes the data to be
19 validated. This is to eliminate unnecessarily strict requirements that commit scarce resources to
20 the in-depth evaluation of data points with high levels of acceptable uncertainty. For example,
21 results very much above or below an action level may not require rigorous validation, since
22 relatively large measurement uncertainty would not affect the ultimate decision or action.
23 Planners should also identify those samples or data sets that have less rigorous standards for data
24 quality and defensibility.
25 This chapter presents suggested criteria to evaluate data and addresses the appropriate function
26 and limits of radiochemical techniques and measurements. Since calibration is more efficiently
27 evaluated as part of an audit, this chapter does not recommend that the complete calibration-
28 support documentation be included as part of the data package. MARLAP recommends that
29 calibration be addressed in a Quality System and through an audit (Chapter 18), although
30 demonstration of calibration may be required as part of a project's deliverables. Detector
31 calibration, self absorption curves and efficiencies should be addressed as part of the evaluation
32 of laboratories during the procurement process and continued during subsequent assessments
33 (Chapter 7). Availability and retention of calibration records are decisions that are project-
34 specific, but should be clearly identified for contract clarity and to assure project completeness
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 8-1 DO NOT CITE OR QUOTE
-------
Radiochemical Data Verification and Validation
35 (i.e., customer needs met). External sources of information, such as performance evaluation
36 sample results and internal laboratory control samples, provide useful interim information on
37 calibration status and accuracy.
38 8.2 Data Assessment Process
39 Figure 1.1 of Chapter 1 graphically depicts the three phases—planning, implementation, and
40 assessment—of the data life cycle, and the associated activities and products of each phase.
41 While these activities are addressed in separate chapters in MARLAP, it should be emphasized
42 that integration of planning, sampling, and analysis with subsequent data verification, data
43 validation, and data quality assessment (DQA) is essential.
44 This section reviews the data life cycle from the perspective of the assessment phase and focuses
45 on those issues that have the potential to impact the quality and usability of the data. Section
46 8.2.1 addresses the development of the assessment procedures during project planning. Section
47 8.2.2 considers assessment needs for documentation and a quality system during implemen-
48 tation. Section 8.2.3 focuses on the assessment phase and addresses the interrelationship of the
49 three assessment processes. This introduction to the data life cycle process emphasizes the
50 importance of linkages among planning, implementation, and assessment.
51 8.2.1 Planning Phase of the Data Life Cycle
52 Directed project planning and the development of the associated DQOs, MQOs, and other
53 specifications for the project were reviewed in Chapters 2 and 3. These chapters emphasize the
54 need for planners to thoroughly define the assessment processes (i.e, verification, validation and
55 data quality assessment) in sufficient detail that success or failure in meeting goals can be
56 determined upon project completion. MARLAP recommends that the assessment phase of a
57 project (verification, validation, and DQA processes) be designed during the directed planning
58 process and documented in the respective plans as part of the project plan documents. This
59 requires the project planning team to develop detailed procedures for data verification, data
60 validation, and data quality assessment, as well as identify the actual personnel who will perform
61 assessment or the required qualifications and expertise of the assessors.
62 The development of these procedures during the directed planning process will increase the
63 likelihood that the appropriate documentation will be available for assessment, and that those
64 generating and assessing data will be aware of how the data will be assessed. A secondary
65 advantage, which assessment plans have, is that prior to their completion, they often result in the
66 detection of design flaws (e.g., lack of proper quality control [QC] samples, lack of a field audit)
MARLAP JULY 2001
DO NOT CITE OR QUOTE 8-2 DRAFT FOR PUBLIC COMMENT
-------
Radiochemical Data Verification and Validation
67 that upon correction will result in the complete information necessary for the proper assessment
68 of data usability.
69 The culmination of the planning process is documentation of the outputs of the directed planning
70 process in the project plan documents. The project plan documents should capture the DQOs,
71 MQOs, and the optimized data collection design (i.e., Analytical Protocol Specifications,
72 sampling and analysis plans, and SOPs). The project plans should also include the assessment
73 plans as discussed above, and describe the field, lab, safety, and QA activities in sufficient detail
74 that the project can be implemented as designed. Chapter 4 discusses guidance for the authoring
75 and content of project plan documents.
76 If the directed planning process, its outputs (DQOs, MQOs, optimized sampling and analysis
77 designs), and associated assumptions are not documented well in project plan documents, the
78 assessment phase will have difficulties evaluating the resulting data in terms of the project's
79 objectives.
80 8.2.2 Implementation Phase of the Data Life Cycle
81 The project plans are executed during the implementation phase. Ideally, the plans would be
82 implemented as designed, but due to errors, misunderstandings, the uncontrolled environments
83 under which sampling is implemented, and matrix-specific issues that complicate sample
84 handling and analysis, most project plans are not implemented without some deviation.
85 Understanding the realities of implementation, the assessment process, in particular the DQA
86 process, will evaluate the project's implementation by considering: (a) if the plans were adequate
87 to meet the project's DQOs, (b) if the plans were implemented as designed, and (c) if the plans as
88 implemented were adequate to meet the project DQOs. MARLAP recommends that project
89 objectives, implementation activities and QA/QC data be well documented in project plans,
90 reports, and records, since the success of the assessment phase is highly dependent upon the
91 availability of such information.
92 Documentation and record keeping during the planning and implementation phase of the data life
93 cycle are essential to subsequent data verification, data validation, and data quality assessment.
94 Thorough documentation will allow for a determination of data quality and data usability.
95 Missing documentation can result in uncertainty, and a lack of critical documentation (e.g.,
96 critical quality control results) can result in unusable data. The quality and usability of data can
97 not be assessed if the supporting documentation is not available.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 8-3 DO NOT CITE OR QUOTE
-------
Radiochemical Data Verification and Validation
98 8.2.2.1 Project Objectives
99 The DQOs, MQOs, and other specifications, requirements, and assumptions developed during
100 the planning phase will influence the outcomes during the subsequent implementation and
101 assessment phases of the data life cycle. It is important that these objectives, specifications,
102 requirements, and assumptions are well documented and available to those implementing the
103 program so they can make informed decisions. This documentation is reviewed during the DQA
104 process (see discussion of the review of DQOs in Section 9.6.1.1, sampling plan in
105 Section 9.6.2.1, and analysis plan in Section 9.6.3.1).
106 8.2.2.2 Documenting Project Activities
107 The assessment of data in terms of sampling and analytical MQOs requires an accurate record of
108 QC sample data and compliance with specifications and requirements. If these records are
109 missing or inadequate, then compliance with APSs, including the MQOs that were identified
110 during the planning phase, will not be ascertainable and will raise questions regarding quality.
ill Additional documentation is required to assess compliance with plans and contracts, and to
112 assess field and lab activities (e.g., compliance with SOPs) and the associated organizational
113 systems (e.g., laboratory Quality Manual). This information is gleaned from the review of field
114 and laboratory notebooks, deviation reports, chain-of-custody forms, verification reports, audit
115 reports, surveillance reports, performance evaluation sample analyses, corrective action reports
116 and reports to management that may identify deviations, contingencies, and quality problems.
117 Assessment of these types of contemporaneous records allow for the assessment of data in the
118 context of pertinent issues that may have arisen during proj ect implementation.
119 Project records should be maintained for an agreed upon period of time, which should be
120 specified in project plan documents. Record maintenance should comply with all regulatory
121 requirements and parallel the useful life of the data for purposes of re-assessment as questions
122 arise or for purposes of secondary data uses that were not originally anticipated.
123 8.2.2.3 QA/QC
124 To ensure that the data collection activity generates data of known quality, it is essential that the
125 project plan documents specify the requirements for an appropriate quality system that is capable
126 of implementing the quality controls and the quality assurance necessary for success.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 8-4 DRAFT FOR PUBLIC COMMENT
-------
Radiochemical Data Verification and Validation
127 The quality system will oversee the implementation of QC samples, documentation of QC
128 sample compliance or non-compliance with MQOs, audits, surveillances, performance evaluation
129 sample analyses, corrective actions, quality improvement and reports to management. The
130 documentation generated by these quality assurance activities and their outputs during project
131 implementation will be a key basis for subsequent assessments and data usability decisions.
132 8.2.3 Assessment Phase of the Data Life Cycle
133 Assessment of environmental data currently consists of three separate and identifiable phases:
134 data verification, data validation, and DQA. Verification and validation pertain to evaluation of
135 analytical data. Verification and validation are considered as two separate processes, but as the
136 MARLAP recommended planning process is implemented, they may be combined—with the
137 verification activities constituting the bulk of the review. DQA considers all sampling, analytical,
138 and data handling details, external QA assessments, and other historical project data to determine
139 the usability of data for decision-making.
140 Figure 8.1 is a graphical depiction of the assessment phase. Although, it portrays a linear
141 progression through the various steps, and from verification and validation to data quality
142 assessment, this linear advancement is not entirely necessary. It is possible for parallel progress
143 within an assessment process (e.g., existing documents are verified while waiting for the
144 production of others) and between assessment processes (e.g., analysis of the DQOs for data
145 quality assessment while data validation is being completed). Typically, the focus of verification
146 and validation is on the analytical process and on a data point by data point review, while data
147 quality assessment considers the entire data collection process and the entire data set as it
148 assesses data quality.
149 Analytical data verification assures laboratory conditions and operations were compliant with the
150 SOW based on project plan documents. The updated project plan documents specify the
151 analytical protocols the laboratory should use to produce data of acceptable quality and the
152 content of the analytical data package (see MARLAP Process in Chapter 1). Verification
153 compares the analytical data package delivered by the laboratory to these requirements
154 (compliance), and checks for consistency and comparability of the data throughout the data
155 package, correctness of basic calculations, data for basic calculations, and completeness of the
156 results to ensure all necessary documentation is available. Verification can be accomplished
157 through use of a plan or simply a check list. The verification process produces a report
158 identifying which requirements are not met (i.e., exceptions qualified with an "E" to alert the
159 validator). The verification report is used to confirm laboratory compliance with the SOW and to
160 identify problems that should be investigated during data validation. Verification works
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 8-5 DO NOT CITE OR QUOTE
-------
Radiochemical Data Verification and Validation
C^Start of Assessment^)
1
f DATA VERIFICATION 1
1
| Identification of missing 1
1 documentation I
1
a^d contract requirements
1
Identification of noncompliant
procedures
1
Identification of noncompliance
with SOW and MQOs
1
Identification of Exceptions
1
Verification Report
V Y—
COMPLIANCE
(Measurable Factors)
I
f DATA VALIDATION j
1
C Review exceptions identified in
[ Verification Report
1
Determine if analytical system
was in control
(Compliance with MQOs)
1
Determine if analytical system was
applicable to sample matrix
i
and uncertainty
1
Apply Qualifiers
1
Validation Report
J ^
1
f DATA QUALITY ASSESSMENT 1
1
Review DQOs and Project Plans
1
1
Determine if data are
accurate
1
^__L_^^
C^End of Assessment^)
(^JVIAKE DECISIoi\P^)
.
V ~
USABILITY
(Measurable & Nonmeasurable Factors)
^ - '
Focus is typically on the analytical process
& individual datum
y-^^
Focus on the entire data collection process
& the entire dataset
FIGURE 8.1 — The Assessment Process
161 iteratively and interactively with the generator (i.e., laboratory) to assure receipt of all necessary
162 data. Although the verification process identifies specific problems, the primary function should
163 be to apply appropriate feedback to the lab resulting in corrective action improving the analytical
164 services before the project is completed.
165 Validation addresses the reliability of the data. The validation process begins with a review of the
166 verification report and laboratory data package to identify its areas of strength and weakness.
167 This process involves the application of qualifiers that reflect the impact of not meeting the
168 MQOs. Validation then evaluates the data to determine the presence or absence of an analyte,
169 and the uncertainty of the measurement process. During validation, the technical reliability and
170 the degree of confidence in reported analytical data are considered. The data validator should be
171 a scientist with radiochemistry experience.
MARLAP
DO NOT CITE OR QUOTE
JULY 2001
DRAFT FOR PUBLIC COMMENT
-------
Radiochemical Data Verification and Validation
172 Validation flags (i.e., qualifiers) are applied to data that do not meet the performance acceptance
173 criteria established in the SOW and the project plan documents. The products of the validation
174 process are validated data and a validation report stating which data are acceptable, which data
175 are sufficiently inconsistent with the validation acceptance criteria in the expert opinion of the
176 validator, and a summary of the QC sample performance. The appropriate data validation tests
177 should be established during the project planning phase. The point of validation is to perform a
178 systematic check on a set of data being used to meet the project MQOs and any other analytical
179 process requirements. Documenting that such a check cannot be done is an appropriate and
180 essential validation activity. (For example, applying numerical tests to data already determined to
181 be unreliable data are of no value.)
182 Data Quality Assessment'^ the last phase of the data collection process, and consists of a
183 scientific and statistical evaluation of project-wide knowledge to assess the usability of data sets.
184 To assess and document overall data quality and usability, the data quality assessor integrates the
185 data validation report, field information, assessment reports, and historical project data, and
186 compares the findings to the original project DQOs. The DQA process uses the combined
187 findings of these multi-disciplinary assessments to determine data usability for the intended
188 decisions, and to generate a report documenting that usability and the causes of any deficiencies.
189 It may be useful for a validator to work with the assessor to assure the value of the validation
190 process (e.g., appropriateness of rejection decision), and to make the process more efficient.
191 DQA will be covered in Chapter 9.
192 8.3 Validation Plan
193 The validation plan should integrate the contributions and requirements of all stakeholders and
194 present this information in a clear, concise format. To achieve this goal, validation planning
195 should be part of initial planning (e.g., directed planning process) to assure that the data will be
196 validated efficiently to determine its reliability and technical defensibility in an appropriate
197 context and to an appropriate degree.
198 The validation plan is an integral part of the project plan documents (Chapter 4), and should be
199 included as either a section within the plan or as a stand-alone document attached as an appendix.
200 The validation plan should be approved by an authorized representative of the project, the
201 validation group performing the validation, and any other stakeholder whose agreement is
202 needed.
203 The information and documentation identified in the validation plans should be communicated to
204 the laboratory as part of the SOW. Integration of validation plan specifications, contractual
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 8-7 DO NOT CITE OR QUOTE
-------
Radiochemical Data Verification and Validation
205 requirements, and validator instructions/contracts is essential to ensure data collection process
206 efficiency. Implementation of the data validation plan will ensure that proper laboratory
207 procedures are followed and data are reported in a format useful for validation and assessment,
208 and will improve cost-effectiveness of the data collection process.
209 The data validation plan should contain the following information:
210 • Summarize the project that provides sufficient detail about the project technical and quality
211 objectives in terms of sample and analyte lists, required measurement uncertainty, and
212 required detection limit and action level on a sample/analyte-specific basis. Specify the scope
213 of validation, e.g., whether all the raw data will be reviewed and in what detail (see Section
214 8.3.1).
215 • Specify the necessary validation criteria, as derived from the MQOs, and performance
216 objectives deemed appropriate for achieving project objectives (see Section 8.3.2).
217 • Direction to the validator on what qualifiers are to be used and how final qualifiers are
218 assigned (see Section 8.3.3).
219 • Direction to the validator on the content of the validation report (see Section 8.3.4).
220 8.3.1 Technical and Quality Objectives of the Project
221 The identity of key analytes and how the sample results drive project decisions should be
222 specified in the validation plan. In addition, the plan should define the association of required
223 quality control samples with project environmental samples.
224 This section of the validation plan should specify the following:
225 • Quality control (QC) acceptance criteria;
226 • Level of measurement uncertainty considered unusually high and unacceptable (tests of
227 unusual uncertainty and rejection); and
228 • Action level and MQOs for detection and quantification capability (e.g., required detection
229 and quantification limit) (tests of detection).
MARLAP JULY 2001
DO NOT CITE OR QUOTE 8-8 DRAFT FOR PUBLIC COMMENT
-------
Radiochemical Data Verification and Validation
230 The quality control acceptance criteria serve two purposes: (1) to establish if the analytical
231 process was in control; and (2) to determine if project requirements were met. If the analytical
232 process is in control, the assumption was that the analysis was performing within established
233 limits and indicates a reasonable match among matrix/analyte/method. Generally this means that
234 routine data quality expectations are appropriate. The tests of unusual (i.e., analysis not in
235 control) uncertainty should verify the data meet the statistical confidence limits for uncertainty
236 associated with the planning process. During validation, the uncertainty associated with sampling
237 cannot be estimated. The tests of detection determine the presence or absence of analytes.
238 8.3.2 Validation Tests
239 Validating data requires three specific decisions that will allow the validator to qualify the data.
240 The project planning team should determine:
241 • Which QC samples should be employed and how do they relate to the environmental
242 samples?
243 • Which validation tests are appropriate?
244 • What validation limits should be used for the specific tests?
245 The answers to these questions are driven by the need to know whether the data meets the MQOs
246 for the project, and the allocation of resources between planning and implementation (i.e.,
247 conservative review may be more costly than real or perceived value in the decision). This
248 section of the validation plan should address the following:
249 • QC sample validation criteria;
250 • Specific validation tests to be used; and
251 • Statistical confidence intervals or fixed limit intervals applied to each of the validation tests
252 and criteria based on the MQOs for the project (Appendix C).
253 8.3.3 Data Qualifiers
254 Data qualifiers are codes placed on an analytical result that alert data users to the validator's or
255 verifier's concern about the result. This section of the validation plan should outline:
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 8-9 DO NOT CITE OR QUOTE
-------
Radiochemical Data Verification and Validation
256 • The basis for rejection or qualification of data; and
257 • The qualification codes that will be assigned.
258 These issues are discussed in detail in Section 8.5, which provides guidance for assigning data
259 qualifiers.
260 The verification process uses a qualifier (E) to alert the validator to non-compliance, including
261 missing documentation, contract compliance, etc. This qualifier may be removed or replaced
262 during validation, based on the validator's interpretation of the effect of the non-compliance on
263 the data's integrity.
264 E A notice to the validator that something was noncompliant.
265 The validation process uses the qualifiers listed below to identify data points that do not meet the
266 project MQOs or other analytical process requirements listed in the SOW or appropriate project
267 plan document. The assignment of the J and R qualifiers relies heavily on the judgement and
268 expertise of the reviewer and therefore, these qualifiers should be assigned as appropriate at the
269 end of data validation.
270 U A normal, not detected (< critical value) result.
271
272 Q A reported combined standard uncertainty, which exceeds the project's required method
273 uncertainty.
274 J An unusually uncertain or estimated result.
275 R A rejected result: the problems (quantitative and/or qualitative) are so severe that the data
276 can not be used.
277 The data validator should be aware that a data qualifier or a set of qualifiers does not apply to all
278 similar data. The data validator should incorporate the project MQOs into the testing and
279 qualifying decision-making process. During the data validation process the data validator may
280 use additional qualifiers based on QC sample results and acceptance criteria. These qualifiers
281 may be summarized as U, J, R or Q in the final validation report. The final validation reports
282 should also include a summary of QC sample performance for use by the data assessor.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 8-10 DRAFT FOR PUBLIC COMMENT
-------
Radiochemical Data Verification and Validation
283 S A result with a related spike result (laboratory control sample (LCS), matrix spike(MS) or
284 matrix spike duplicate MSD), which is outside the control limit for recovery (%R), S+ or
285 S- used to indicate high or low recovery.
286 P A result with an associated replicate result that exceeds the control limit.
287 B A result with associated blank result, which is outside the control limit, B+ or B-.
288 8.3.4 Reporting and Documentation
289 The purpose of this section is to define the format and program needs for validation reports and
290 supporting documentation. This section should include:
291 • Documentation and records that should be included in a validation report;
292 • Disposition requirements for records and documents from the project;
293 • Report format, i.e., a summary table with results, uncertainties and qualifiers; and
294 • Procedures for non-conformance reporting, which detail the means by which the laboratory
295 communicates non-conformances against the validation plan. The procedures should include
296 all instances where the analytical data requirements and validation requirements established
297 by the planning process and validation plan, respectively, cannot be met due to sample matrix
298 problems and/or unanticipated laboratory issues (loss of critical personnel or equipment).
299 Detailed information about the Validation Report is presented in Section 8.6.
300 8.4 Other Essential Elements
301 Effective data validation is dependent on:
302 • A SOW and project plan documents that clearly define the data needs and the data quality
303 requirements (i.e., MQOs); and
304 • A data package that has been verified for completeness, consistency, compliance, and
305 correctness.
306 8.4.1 Statement of Work
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 8-11 DO NOT CITE OR QUOTE
-------
Radiochemical Data Verification and Validation
307 The analytical services procurement options should be considered during the planning process.
308 The SOW should specify the QC requirements that will be evaluated by the validator (see
309 Chapter 5). The elements that should be specified include, but are not limited to:
310 • External performance evaluation (PE) participation and acceptance criteria;
311 • Replicate sample frequency and acceptance criteria;
312 • LCS and acceptance criteria;
313 • Blank requirements and acceptance criteria;
314 • MS and MSD samples and acceptance criteria;
315 • Uncertainty calculations; and
316 • Sample result equations and calculations including corrections for yield, percent moisture,
317 efficiencies and blank, if applied.
318 Section 8.5.2 provides guidance on evaluating QC sample results based on the project's MQO for
319 measurement uncertainty.
320 8.4.2 Verified Data Deliverables
321 Verification compares the sample receipt information and the sample report delivered by the
322 laboratory against the SOW and produces a report that identifies those requirements that were not
323 met (called exceptions). Verification can be accomplished using a plan or checklist, which
324 doesn't necessarily need to be project-specific. Verification exceptions normally identify:
325 • Required steps not carried out by the laboratory (i.e., correction for yield, proper signatures);
326 • Method QC not conducted at the required frequency (i.e., blanks, duplicates); and
327 • Method QC not meeting pre-set acceptance criteria (i.e., non-compliant laboratory control
328 sample analysis).
329 The verifier checks the data package (paper or electronic) for completeness, consistency,
330 correctness, and compliance. Completeness means all required information is present.
331 Consistency means values are the same when reported redundantly on different reports, or
332 transcribed from one report to another. Correctness means the reported results are based on
333 properly documented and correctly applied algorithms. Compliance means the data pass
334 numerical QC tests based on parameters or limits derived from the MQOs specified in the SOW.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 8-12 DRAFT FOR PUBLIC COMMENT
-------
Radiochemical Data Verification and Validation
335 The verifier should provide, within the verification package, checklists for contract or SOW
336 specifications, noted deficiencies related to contract compliance, noted discrepancies or obvious
337 quality related problems, and pertinent external QC results. The verification package notes the
338 deficiencies, discrepancies, and quality-related problems that could not be resolved with the
339 laboratory. The validator should take this information into consideration during the data
340 validation process.
341 8.5 Data Verification and Validation Process
342 In its most basic form, data validation focuses on the reliability of each data point. After each
343 point is evaluated, summary conclusions concerning the validity of groups of data (sets) are
344 drawn and finally, after the reliability of all data sets has been established, an overall conclusion
345 about the quality and defensibility of a project's analytical database is reached (DQA).
346 The first step in establishing the reliability of an analytical measurement is to determine that the
347 measurement analytical process used in making the measurement is in control. That is, the
348 sample handling and analysis system is performing within an accepted operating range
349 (established by instrument manufacturer, method, or contract specifications and/or long-term
350 historical laboratory performance). After it has been determined that the measurement analytical
351 process is in control, it is necessary to demonstrate that the sample is responding as expected
352 when introduced into the measurement system.
353 The measurement process includes devices such as detectors for measuring radioactive decay and
354 balances for determining the mass of materials. The measurement process also includes the
355 software that takes the output from the measurement device and calculates the result as a quantity
356 of target radionuclide (activity/mass activity/volume). The measurement process performance
357 normally is specified by the SOW and appropriate project plan documents, and monitored by
358 routine laboratory quality control procedures. Laboratory performance against these requirements
359 is determined by the verification process uses these requirements to determine laboratory
360 performance.
361 When an environmental sample is analyzed, new sources of variability are encountered in
362 addition to those associated with the measurement process. These sources include laboratory
363 subsampling, sample preparation (e.g., digestion, leaching, etc.), sample matrix effects, and data
364 transcription, to list a few. These processes, taken together with the previously discussed
365 measurement process, comprise the analytical process.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 8-13 DO NOT CITE OR QUOTE
-------
Radiochemical Data Verification and Validation
366 The performance of the analysis can be predicted based on previous experience with similar
367 materials. Analysis performance is monitored by laboratory quality control procedures specified
368 in the SOW and appropriate project plan documents. Unlike the analytical process performance,
369 the overall performance of the analysis is not amenable to assessment by the data verification
370 process. Since each sample matrix, analyte, and method set is unique, the evaluation of overall
371 analysis performance and resulting data is the role of a knowledgeable validator.
372 Using the validation plan, which specifies QC samples, validation tests, and validation limits,
373 validation occurs in four stages:
374 • Determine whether the sample handling and analysis system is in control (Section 8.5.1);
375 • Determine whether quality control sample analyses meet specified MQOs (Section 8.5.2);
376 • Apply validation tests of detection and unusual uncertainty (Section 8.5.3); and
377 • Determine final data qualifiers and document the results (Section 8.5.4).
378 For other chemistry methods, identification of the analyte is also a primary decision. Except for
379 gamma spectroscopy, this is rarely an issue in radiochemistry. For radiochemistry, the
380 laboratory's ability to reliably identify analytes do reliable identifications is best checked by
381 auditors and verified by checking the calibration check samples.
382 8.5.1 The Sample Handling and Analysis System
383 As described in earlier sections of this guidance, it is necessary to know the extent to which the
384 data delivered for validation meet the requirements of the SOW and appropriate project plan
385 documents. These documents normally specify the minimum acceptable performance of the
386 analytical process. These specifications are the basis of the tests of quality control (QC tests) that
387 establish that the sample handling and analysis system is in control at the time the analyses were
388 performed. It is also necessary to know that all reporting requirements are complete. Normally,
389 this evaluation against the requirements is made during the data verification process. If the data
390 do not conform to the requirements, notification should be provided in the verification report.
391 The review of the verification package (and data package) by the validator determines if
392 sufficient information is provided to proceed with data validation. The outcome of the
393 verification process is the designation of exceptions to the quality control tests. These exceptions
394 should be flagged with a qualifier (re-evaluated by the validator), which is appended to a data or
395 report requirement that does not meet specifications to alert the validator of potential problems.
396 The validator should then determine if sufficient reliable data are available to proceed with
397 validation. The validator should use the data requirements and criteria developed in the
MARLAP JULY 2001
DO NOT CITE OR QUOTE 8-14 DRAFT FOR PUBLIC COMMENT
-------
Radiochemical Data Verification and Validation
398 validation plan to determine if the quality control exceptions have an adverse impact on one or
399 more of the data points being validated.
400 Rarely, if ever, should quality control exceptions result in the decision to reject a complete data
401 set. Those types of situations should have been detected by the laboratory during the analytical
402 process and the samples reanalyzed. The validator should not reject (assign an "R" code) single
403 data points based on a single QC test exception. Normally, only numerous QC exceptions and
404 failures in one or more of the tests of detection and uncertainty are sufficient reason to reject
405 data. The validation report should fully explain the assignment of all qualifiers as previously
406 discussed.
407 The following paragraphs discuss some of the more important evaluations that should be applied
408 to the sample handling and analysis system. Limited guidance is provided on how the QC test
409 may impact data quality and defensibility.
410 8.5.1.1 Sample Descriptors
411 Sample descriptors include sample identification number, analytical method, analyte, and matrix,
412 among others.
413 Criteria. Each sample should have a unique identifier code that can be cross-referenced to a
414 unique field sample or an internally generated laboratory sample. This unique identifier and
415 associated sample descriptors should be included in all analytical reports to properly document
416 the sample and requested analysis (Chapters 10 and 11).
417 The matrix and other characteristics of the sample that affect method selection and performance
418 should be clearly identified. The method(s) used in sample preparation and analysis should be
419 identified.
420 If laboratory replicate analyses are reported for a sample, they should be distinguishable by a
421 laboratory-assigned code.
422 Verification. Each of the criteria related to describing the sample should be checked for and
423 found in the analytical data package. If any of the criteria are missing, they should be flagged
424 with an "E" code.
425 Validation. Missing information will increase the uncertainty on any result reported on a
426 sample(s) and justify the assignment of a " J" code. Missing information may be inferred from
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 8-15 DO NOT CITE OR QUOTE
-------
Radiochemical Data Verification and Validation
427 other information in the data package and eliminate the added uncertainty. For example, if the
428 sample matrix is not provided, it may be inferred from:
429 • The aliquant units are expressed in units of mass or volume;
430 • The sample preparation method is specific for soils;
431 • The final results are expressed in units of mass; and
432 • The sampling report describes sampling soil.
433 The majority of related information should support the decision that the exception does not
434 increase the uncertainty of the result. If the supporting information is incomplete or conflicting,
435 the assignment of a "J" code to data points is warranted. If documentation is inadequate to
436 support the reporting of a data point, the data point should be qualified with an "R" code.
437 8.5.1.2 Aliquant Size
438 Criteria. The aliquant or sample size used for analysis should be documented so that it can be
439 checked when reviewing calculations, examining dilution factors or analyzing any data that
440 requires aliquant as an input. It is also imperative that the appropriate unit (liter, kilogram, etc.) is
441 assigned to the aliquant.
442 Verification. The criteria related to describing the sample aliquant should be checked for and
443 found in the analytical data package. If the aliquant size is missing, it should be flagged with an
444 "E" code.
445 Validation. The missing information will increase the uncertainty on any result reported on a
446 sample(s) and justify the assignment of a " J" code.
447 8.5.1.3 Dates of Sample Collection, Preparation, and Analysis
448 Criteria. The analytical data package should report date of sampling, preparation, and analysis.
449 These data are used to calculate radiological holding times, some of which may be specified in
450 the sampling and analysis plan.
451 There are few circumstances where radiological holding times are significant for radionuclides.
452 The best approach to minimize the impact of holding time on analysis is to analyze the samples
453 as quickly as possible. Holding times may be applied to environmental samples that contain
454 radionuclides with short half lives. Holding times would apply to these radionuclides to prevent
MARLAP JULY 2001
DO NOT CITE OR QUOTE 8-16 DRAFT FOR PUBLIC COMMENT
-------
Radiochemical Data Verification and Validation
455 reporting of high measurement uncertainties and MDCs, and to detect the radionuclide, if present
456 at low concentration, before it decays to undetectable levels.
457 Verification. Each of the criteria related to sample holding time should be checked for and found
458 in the analytical data package. If any of the objectives are missing, they should be flagged with
459 an "E" code.
460 If a holding time is specified in the project plan documents or validation plan, the reported values
461 should be compared to this specification. If the holding time is exceeded, the affected criteria
462 (holding time) should be flagged with an "E" code.
463 Validation. The data points impacted by the missed holding time should be flagged with a "J"
464 code by the validator or the justification for discounting the holding time impact described in the
465 narrative section of the validation report.
466 8.5.1.4 Preservation
467 Criteria. Appropriate preservation is dependent upon analyte and matrix, and should be defined
468 in sampling and analysis documentation. Generally, preservation is applied to samples being
469 analyzed for radionuclides to prevent precipitation, adsorption to container walls, etc. The criteria
470 (required presence or absence) for this QC process should be provided in the sampling and
471 analysis plan (see Chapter 10).
472 Verification. The criteria related to preservation should be checked for and found in the
473 analytical data package. If any of the criteria are missing, they should be flagged with an "E"
474 code.
475 Validation. If exceptions to the preservation criteria are noted, the validator should decide if a
476 "J" code should be assigned to data points because the improper preservation increased the
477 overall uncertainty in the data point(s). In some cases where improper preservation severely
478 impacts data quality or defensibility (e.g., the use of acid preservation in water samples being
479 analyzed for 14C), the validator should assign an "R" qualifier. The assessor may elect to use the
480 data, but they have the responsibility of addressing the data quality and defensibility in the
481 assessment report.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 8-17 DO NOT CITE OR QUOTE
-------
Radiochemical Data Verification and Validation
482 8.5.1.5 Tracking
483 Criteria. Each analytical result should be traceable to the instrument or detector on which it was
484 counted. The requirement for this traceability normally is found in the project plan documents.
485 The analytical sequence log (or some other suitable record) should be available in the data
486 package submitted by the laboratory.
487 Verification. If any of the analytical data are not traceable to the instrument or detector, it should
488 be flagged with an "E" code.
489 Validation. The validator may factor the absence of the traceability into their evaluation of data
490 quality and usability. At most, this should result in increasing the uncertainty of the
491 determination and the possible assignment of a " J" code to the data. This would not occur
492 normally unless one or more of the detectors used in analyzing the samples was shown to be
493 unreliable. Then, the inability to trace a reliable detector to a sample increases the uncertainty of
494 the data point(s).
495 8.5.1.6 Traceability
496 Criteria. The traceability of standards and reference materials to be used during the analysis
497 should be specified in the sampling and analysis plan.
498 Verification. The source of the reference materials and standards should be checked for and
499 found or referenced in the analytical data package. If any of the sources are missing they should
500 be flagged with an "E" code.
501 Validation. The validator may factor the absence of the traceability into their evaluation of data
502 quality and usability. At most, this should result in increasing the uncertainty of the
503 determination and the possible assignment of a " J" code to the data. This would not occur
504 normally unless one or more of the standards used in analyzing the samples was shown to be
505 unreliable. Then, the inability to trace a reliable standard to a sample increases the uncertainty of
506 the data point(s).
507 8.5.1.7 QC Types and Linkages
508 Criteria. The type and quantity of QC samples should be identified and listed in the SOW, and
509 the results provided by the laboratory in a summary report. Replicates and matrix spike results
510 should be linked to the original sample results. The approximate level of matrix spike
MARLAP JULY 2001
DO NOT CITE OR QUOTE 8-18 DRAFT FOR PUBLIC COMMENT
-------
Radiochemical Data Verification and Validation
511 concentrations should be specified in the SOW, but the actual levels should be reported by the
512 laboratory. The QC analyses should be traceable to the original field sample.
513 Verification. Each of the criteria related to the QC samples should be checked for and found in
514 the analytical data package. If any of the objectives are missing, they should be flagged with
515 an "E" code.
516 Validation. The validator should compare any QC sample exceptions to similar ones that
517 precede and follow the non-conforming QC sample. If these are in control, the validator can
518 discount the impact of the single QC sample exception on the data results (i.e., analytical
519 blunder). If a trend of failing values is found, the validator should consider if they affected a
520 group of data points to the extent that the level of uncertainty was increased. This may warrant
521 the assignment of a "J" code to the data.
522 8.5.1.8 Chemical Separation (Yield)
523 Criteria. Yield assesses the effects of the sample matrix and the chemical separation steps on the
524 analytical result and estimates the analyte loss throughout the total analytical process. Yield is
525 typically measured gravimetrically (via a carrier) or radiometrically (via a tracer). All the
526 components in the calculation of the yield should be identified in a defined sequence. These
527 specifications are found in the project plan documents.
528 Criteria for both analytical process and sample analysis may be given in the project plan
529 documents. The criteria should be based on historical data for the method and matrix. In that
530 case, yield is determined on both quality control samples and actual field samples.
531 The most important yield-related question is whether the yield has been determined accurately.
532 Typically, a yield estimate that is much greater than 100 percent cannot be accurate, but the
533 estimate may also be questionable if the yield is far outside its historical range. Extremely low
534 yields also tend to have large measurement uncertainties, which increase the uncertainties of the
535 results. The uncertainties of factors such as the yield, counting efficiency, and aliquant volume,
536 which affect the sensitivity of the measurement, should be kept relatively small.
537 Verification. Each of the yield-related criteria pertaining to the sample should be checked for
538 and found in the analytical data package. If missing, the data should be returned to the lab to
539 correct for yi el d.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 8-19 DO NOT CITE OR QUOTE
-------
Radiochemical Data Verification and Validation
540 Validation. The experimentally determined yield is used to normalize the observed sample
541 results to 100% yield. Exceptions to the yield value outside the range specified in the project plan
542 documents may result in the validator assigning a "J" qualifier to otherwise acceptable data.
543 8.5.1.9 Self-Absorption (Residue)
544 Criteria. For some radiochemical analytical methods, the SOW may specify the generation of a
545 self-absorption curve, which correlates mass of sample deposited in a known geometry to
546 efficiency.
547 Verification. Each self-absorption curve called for in the SOW should be checked for and found
548 in the analytical data package. If missing, they should be flagged with an "E" code.
549 Validation. If required self-absorption curves are missing, the validator may select to qualify
550 affected data with a "J" qualifier to signify an increased level of uncertainty in the measurement
551 because of the inability to correct the measured value for self-absorption.
552 8.5.1.10 Efficiency, Calibration Curves, and Instrument Background
553 Criteria. For some methods based on decay emission counting, efficiency is reported as count
554 rate divided by disintegration rate. Methods employing radiotracers determine a sample-specific
555 effective efficiency factor that is a product of the chemical yield and the detector efficiency. This
556 criteria may be specified in the SOW. Instrument background count rate is determined for each
557 detector for each region of interest and subtracted from the sample count rate.
558 Verification. Each efficiency determination, efficiency calibration curve, and instrument
559 background called for in the project plan documents should be checked for and found in the
560 analytical data package. If missing, they should be flagged with an "E" code.
561 Validation. If required factors are missing, the validator may select to qualify affected data with
562 a "J" qualifier to signify an increased level of uncertainty in the measurement because of the
563 inability to correct the measured value for efficiency.
564 8.5.1.11 Spectrometry Resolution
565 Criteria. The measured resolution of alpha, gamma-ray, and liquid scintillation spectrometers, in
566 terms of the full width of a peak at half maximum (FWHM), can be used to assess the adequacy
567 of instrument setup, detector selectivity, and chemical separation technique that may affect the
MARLAP JULY 2001
DO NOT CITE OR QUOTE 8-20 DRAFT FOR PUBLIC COMMENT
-------
Radiochemical Data Verification and Validation
568 identification and quantification of the analyte. When sufficient peak definition (i.e., sufficient
569 number of counts to provide an adequate Gaussian peak shape) has been reached for a sample,
570 the resolution of the analyte peak should be evaluated to determine if proper peak identification
571 and separation or deconvolution was made. Spectral information should be provided in the data
572 packages to accomplish this evaluation.
573 Verification. There are no established acceptance criteria, but should be provided in the package
574 or available in the audit.
575 Validation. If required calculations are missing, the validator may select to qualify affected data
576 with a "J" qualifier to signify an increased level of uncertainty in the measurement because of the
577 inability to evaluate instrument setup and separation technique. An "R" code may be applied if
578 there is no separation.
579 8.5.1.12 Dilution and Correction Factors
580 Criteria. Samples for radiochemistry are usually not diluted, but a larger sample may be
581 digested, taking an aliquant for analysis to obtain a more representative subsample. The dilution
582 factors are normally used for tracers and carriers. Dilutions of the stock standards are prepared
583 and added to the samples. This dilution normally affects yield calculations, laboratory control
584 samples, and matrix spikes. This data should be provided in the data package so that the final
585 calculations of all data affected by dilution factors can be recalculated and confirmed, if required.
586 Other correction factors that may be applied to the data are dry weight correction, ashed weight
587 correction, and correction for a two-phased sample analyzed as separate phases.
588 Verification. Each dilution and correction factor affecting the sample should be checked for and
589 found in the analytical data package. If any of the factors are missing, they should be flagged
590 with an "E" code.
591 Validation. Those results impacted by missing dilution factors should be flagged with a "J" or
592 "R" qualifier, reflecting increased uncertainty in the data point(s). "R" may be warranted if the
593 calculation cannot be confirmed due to missing data.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 8-21 DO NOT CITE OR QUOTE
-------
Radiochemical Data Verification and Validation
594 8.5.1.13 Counts and Count Time (Duration)
595 Criteria. The count time for each sample, QC analysis, and instrument background should be
596 recorded in the data package. The ability to detect radionuclide disintegrations is directly related
597 to the count time. The longer the count time, the lower the detection limit. The project plan
598 documents should specify the MQOs, which will drive the count time for each analyte.
599 Verification. Each count time relating to the sample analysis should be checked for and found in
600 the analytical data package. If any of the objectives are missing, they should be flagged with
601 an "E" code.
602 Validation. The validator should estimate the impact of the actual count times on the ability to
603 detect the target analyte and the impact on the uncertainty of the measurement. If the MQOs are
604 met, the sample should not be qualified for count time. It should be noted that preset count
605 determination, rather than preset count time, will result in the same uncertainly for all the
606 samples. The qualifiers should be adjusted accordingly and the justification provided in the
607 validation report.
608 8.5.1.14 Result of Measurement, Uncertainty, Minimum Detectable Concentration, and Units
609 Criteria. MARLAP recommends that the result of each measurement, its expanded measurement
610 uncertainty, and the estimated sample- or analyte-specific MDC be reported for each sample in
611 the appropriate units. These values, when compared with each other, provide information about
612 programmatic problems with the calculations, interference of other substances, and bias. The
613 report should state the coverage factor used if calculating expanded measurement uncertainties,
614 and the Type I and Type n error probabilities used to calculate MDCs.
615 Verification. The linkage between the result, measurement uncertainties, MDC, and the sample
616 identification should be checked. If linkage is not evident, data should be flagged with an "E"
617 code.
618 Validation. The validator should assign data qualifiers to those data points for which they feel
619 sufficient justification exists. Each qualifier should be discussed in the validation report.
620 8.5.2 Quality Control Samples
621 Historically, data validation has placed a strong emphasis on review of QC sample data
622 (laboratory control samples, duplicates, etc). The assumption is that if the analytical process was
MARLAP JULY 2001
DO NOT CITE OR QUOTE 8-22 DRAFT FOR PUBLIC COMMENT
-------
Radiochemical Data Verification and Validation
623 in control and the QC samples responded properly, then the environmental samples (field
624 samples plus the preparation sequences used to prepare the sample for analysis) would respond
625 properly. It is possible to have excellent performance on simple matrices (e.g., quality control
626 samples), but unacceptable performance on complex matrices (i.e., environmental) reported in
627 the same batch as the QC samples. Directly evaluating the environmental sample performance is
628 essential to determine measurement uncertainty and the likelihood of false positive and negative
629 detect!on of the target analyte.
630 Method blanks and laboratory control samples relate to the analytical batch (a series of similar
631 samples prepared and analyzed together as a group) quality control function. They are required
632 by most analytical service contracts, sampling and analysis plans, and project plan documents.
633 They serve a useful function as monitoring tools that track the continuing analytical process
634 during extended analytical sequences. They are the most ideal samples analyzed as part of a
635 project. Normally, their performance is compared to fixed limits derived from historical
636 performance or additionally project specific limits derived from the MQOs.
637 Laboratory duplicates and matrix spikes are quality control samples that directly monitor sample
638 system performance. The laboratory duplicates (two equal-sized samples of the material being
639 analyzed, prepared, and analyzed separately as part of the same batch) measure the overall
640 precision of the sample measurement process beginning with laboratory sub-sampling of the field
641 sample. Matrix spikes (a known amount of target analyte added to the environmental sample)
642 provide a direct measure of how the target analyte responds when the environmental sample is
643 prepared and measured, thereby estimating the bias introduced by the sample matrix.
644 Other QC tests can be applied to determine how the analytical process performs during the
645 analysis of environmental samples. These are yield/recovery, efficiency, self-absorption,
646 resolution, and drift. They are the same QC tests that were applied to routine QC samples (blanks
647 and laboratory control samples) in the previous discussion of the analytical process, but now are
648 applied to environmental samples. The difference lies in how performance is measured. Fixed
649 limits based on historical performance and/or statistics are usually the basis for evaluating the
650 results of routine QC samples.
651 The following paragraphs discuss how QC tests should be used to determine if the results for QC
652 samples meet the project MQOs. Guidance is provided on how to relate QC sample and
653 environmental sample performance to determine environmental sample data quality and
654 defensibility. Direction is also given about how to assign data qualifiers to environmental sample
655 data based on the tests of quality control. Appendix C provides guidance on developing criteria
656 for evaluating QC sample results. Specifically, Appendix C contains equations that allow for the
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 8-23 DO NOT CITE OR QUOTE
-------
Radiochemical Data Verification and Validation
657 determination of warning and control limits for QC sample results based on the project's MQO
658 for measurement uncertainty.
659 8.5.2.1 Method Blank
660 The method blank (Section 18.4.1) is generated by carrying all reagents and added materials
661 normally used to prepare an environmental sample through the same preparation process. It
662 establishes how much, if any, of the measured analyte is contributed by the reagents and
663 equipment used in the preparation process. For an ideal system, there will be no detected
664 concentration or activity.
665 Since measured results are usually corrected for instrument and reagent background levels, it is
666 possible to obtain final results that are less than zero. A method blank result that is much less
667 than zero may indicate that the correction term is too large and therefore analyte concentrations
668 in actual samples may be underestimated.
669 Criteria. The requirement for a method blank is usually established in the SOW and appropriate
670 plan documents. The objective is to establish the target analyte concentration or activity
671 introduced by the sample preparation sequence. Method blanks are normally analyzed once per
672 analytical batch.
673 Other types of blanks, such as field blanks and trip blanks, are used to evaluate aspects of the
674 data collection effort and laboratory operations that are not directly related to the validation of
675 environmental analytical data quality or technical defensibility. They can be important to the
676 overall data assessment effort, but are beyond the scope of this guidance (Chapter 10).
677 See Appendix C for guidance on developing criteria for evaluating blanks based on the project's
678 MQO for method uncertainty.
679 Verification. If a method blank was required but not performed, or if the required data is
680 missing, the verifier flags the missing information with an "E" code.
681 Validation. If a blank result does not comply with the established criteria, the associated samples
682 are flagged "B+" to indicate that the blank result is greater than the upper limit, or "B-" to
683 indicate that the blank result is less than the lower limit.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 8-24 DRAFT FOR PUBLIC COMMENT
-------
Radiochemical Data Verification and Validation
684 8.5.2.2 Laboratory Control Samples
685 The laboratory control sample (LCS) is a QC sample of known composition or an artificial
686 sample (created by spiking a clean material similar in nature to the environmental sample), which
687 is prepared and analyzed in the sample manner as the environmental sample. In an ideal situation,
688 the LCS would give 100 percent of the concentration or activity known to be present in the
689 fortified sample or standard material. Acceptance criteria for the LCS sample are based on the
690 complexity of the matrix and the historical capability of the lab and method to recover the
691 activity. The result normally is expressed as percent recovery. The LCS recovery differs from the
692 recovery of a matrix spike in that the matrix spike is added directly to the environmental sample
693 and the percent recovery is determined by comparing the difference between the original and
694 spiked samples.
695 Criteria. The objective of the LCS is to measure the response of the analytical process to a QC
696 sample with a matrix similar to the environmental sample. This will allow inferences to be drawn
697 about the reliability of the analytical process.
698 See Appendix C for guidance on developing control limits for LCS results based on the project's
699 MQO for method uncertainty.
700 Verification. If a required LCS is not analyzed, or if required information is missing, the verifier
701 flags the missing information with an "E" code.
702 Validation. When the measured result for the LCS is outside the control limits, the associated
703 samples are flagged with the "S" qualifier (S+ or S-).
704 8.5.2.3 Laboratory Replicates
705 Replicates are used to determine the precision of laboratory preparation and analytical
706 procedures. Laboratory replicates are two aliquants selected from the laboratory sample and
707 carried through preparation and analysis as part of the same batch.
708 The discussion of field replicates is beyond the scope of this chapter.
709 Criteria. The objective of replicate analyses is to measure laboratory precision based on each
710 sample matrix. The variability of the samples due to field sample heterogeneity is also reflected
711 in the replicate result. The laboratory may not be in control of the precision. Therefore, replicate
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 8-25 DO NOT CITE OR QUOTE
-------
Radiochemical Data Verification and Validation
712 results are used to evaluate reproducibility of the complete laboratory process that includes
713 subsampling, preparation, and analytical process.
714 See Appendix C for guidance on developing control limits for replicate results based on the
715 project's MQO for method uncertainty.
716 Verification. If replicate analyses are required but not performed, or if the required data is not
717 present in the report, the verifier flags the missing information with an "E" code.
718 Validation. When the replicate analysis is outside the control limit, the associated samples are
719 flagged with the "P" qualifier.
720 8.5.2.4 Matrix Spikes and Matrix Spike Duplicates
721 The matrix spike is an aliquant of a sample, fortified (spiked) with known quantities of target
722 analytes and subjected to the entire analytical procedure to establish if the method or procedure is
723 appropriate for the analysis of the particular matrix.
724 Criteria. Matrix spike samples provide information about the effect of each sample matrix on
725 the preparation and measurement methodology. The test uncovers the possible existence of
726 recovery problems, based on either a statistical test or a specified fixed control limit.
727 See Appendix C for guidance on developing criteria for evaluating matrix spikes based on the
728 proj ecf s MQO for method uncertainty.
729 Verification. If a required matrix spike analysis was not performed, or if the required
730 information is missing, the missing information should be flagged with an "E" code.
731 Validation. If the results of the matrix spike analysis do not meet the established criteria, the
732 samples should be qualified with an "S+" or "S-" indicating unacceptable spike recoveries.
733 8.5.3 Tests of Detection and Unusual Uncertainty
734 8.5.3.1 Detection
735 The purpose of a test of detection is to decide if each result for a regular sample is significantly
736 different from zero. Since most radiochemistry methods always produce a result, even if a very
737 uncertain or negative one, some notion of a non-detected but measured result may be needed for
MARLAP JULY 2001
DO NOT CITE OR QUOTE 8-26 DRAFT FOR PUBLIC COMMENT
-------
Radiochemical Data Verification and Validation
738 some projects. A non-detected result is generally as valid as any other measured result, but it is
739 too small relative to its measurement uncertainty to give high confidence that a positive amount
740 of analyte was actually present in the sample. Ordinarily, if the material being analyzed is
741 actually analyte-free, most results should be "non-detected."
742 For some projects, detection may not be an important issue. For example, it may be known that
743 all the samples contain a particular analyte, and the only question to be answered is whether the
744 mean concentration is less than an action level. However, all laboratories should be able to
745 perform a test of detection routinely for each analyte in each sample.
746 Criteria. An analyte is considered detected when the measured analyte concentration exceeds its
747 critical value (see Chapter 19). Both values are calculated by the laboratory performing the
748 measurement; so, the detection decision can be made at the laboratory and indicated in its report.
749 If there is no evidence of additional unquantified uncertainty in the result (e.g., lack of statistical
750 control or blank contamination), the laboratory's decision may be taken to be final.
751 Verification. Typically, the role of the verifier is limited to checking that required information,
752 such as the critical value, is present in the report. If information is missing, the result should be
753 flagged with an "E" code.
754 Validation. The validator examines the result of the measurement, its critical value, and other
755 information associated with the sample and the batch in which it was analyzed, including method
756 blank results in particular, to make a final determination of whether the analyte has been detected
757 with confidence. If the data indicates the analyte has been detected in both the sample and the
758 method blank, its presence in the sample may be questionable. A quantitative comparison of the
759 total amounts of analyte in the sample and method blank, which takes into account the associated
760 measurement uncertainties, may be needed to resolve the question.
761 8.5.3.2 Detection Capability
762 Criteria. If the project requires a certain detection capability, the requirement should be
763 expressed as a required minimum detectable concentration (RMDC). The data report should
764 indicate the RMDC and the sample-specific estimate of the actual minimum detectable
765 concentration (MDC) for each analyte in each sample.
766 In some situations, it may not be necessary or even possible for a laboratory to meet the MDC
767 requirement for all analytes in all samples. In particular, if the analyte is present and quantifiable
768 at a concentration much greater than the action level, a failure to meet a contract-required
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 8-27 DO NOT CITE OR QUOTE
-------
Radiochemical Data Verification and Validation
769 detection limit is usually not a cause for concern. A failure to meet the RMDC is more often an
770 important issue when the analyte is not detected.
771 Verification. The RMDC specified in the contract is compared to the sample-specific MDC
772 achieved by the method. The analytes that do not meet the RMDC are flagged with an "E" code.
773 Validation. If the sample-specific MDC estimate exceeds the RMDC, the data user may be
774 unable to make a decision about the sample with the required degree of certainty. A "UJ"
775 qualifier is warranted if the estimated MDC exceeds the RMDC and the analyte was not detected
776 by the analysis. A final decision about the usability of the data should be made during the data
777 assessment phase of the data collection process.
778 An assignment of "R" to the data points affected by this type of exception may be appropriate in
779 some cases, but the narrative report may classify the data as acceptable (no qualifier), "U," or " J,"
780 based on the results of the tests of detection and uncertainty. This allows the assessor to make an
781 informed judgement about the usability of the data point(s) and allows them the opportunity to
782 provide a rationale of why the data can be used in the decision process.
783 8.5.3.3 Large or Unusual Uncertainty
784 When project planners follow MARLAP's recommendations for developing MQOs, they
785 determine a required method uncertainty at a specified analyte concentration. The required
786 method uncertainty is normally expressed in concentration units, but it may be expressed as a
787 relative method uncertainty (percent based on the upper bound of the gray region, which is
788 normally the action level). It is reasonable to expect the laboratory's combined standard
789 uncertainty at concentrations lower than the action level to be no greater than the required
790 method uncertainty (expressed in concentration units) and to expect the laboratory's relative
791 combined standard uncertainty at concentrations above the action level to be no greater than the
792 required relative method uncertainty (expressed as a percent). Each measured result should be
793 checked against these expectations (see Appendix C).
794 Criteria. The reported combined standard uncertainty is compared to the maximum allowable
795 standard uncertainty. Either absolute (in concentration units) or relative uncertainties (expressed
796 as a percent) are used in the comparison, depending on the reported concentration. The result is
797 qualified with a "Q" if the reported uncertainty is larger than the requirement allows.
798 Verification. The test for large uncertainty is straightforward enough to be performed during
799 either verification or validation. If there is a contractual requirement for measurement
MARLAP JULY 2001
DO NOT CITE OR QUOTE 8-28 DRAFT FOR PUBLIC COMMENT
-------
Radiochemical Data Verification and Validation
800 uncertainty, the verifier should perform the test and assign the "E" qualifier to results that do not
801 meet the requirement. Note that it may sometimes happen that circumstances beyond the control
802 of the laboratory make it impossible to meet the requirement.
803 Validation. If a "Q" qualifier is assigned, the validator may consider any special circumstances
804 that tend to explain it, such as interferences, small sample sizes, or long decay times, which were
805 beyond the control of the laboratory. He or she may choose to remove the qualifier, particularly if
806 it is apparent that the original uncertainty requirement was too restrictive.
807 8.5.4 Final Qualification and Reporting
808 The final step of the validation process is to assign and report final qualifiers for all regular
809 sample results. The basis for assignment of final qualifiers is qualifiers and reasons from all
810 previous tests, patterns of problems in batches of samples, and validator judgement.
811 The difficult issue during final qualifier assignment is rejecting data. What follows summarizes
812 some of the issues to consider when thinking about rejecting data.
813 Rejecting a result is an unconditional statement that it is not useable for the intended purpose. A
814 result should only be rejected when the risks of using it are significant relative to the benefits of
815 using whatever information it carries. If the DQA team or users feel data is being rejected for
816 reasons that don't affect usability, they may disregard all validation conclusions. Rejected results
817 should be discarded and not used in the DQA phase of the data life cycle.
818 There are three bases on which to reject data:
819 1. Insufficient or only incorrect data are available to make fundamental decisions about data
820 quality. For example, if correctly computed uncertainty estimates are not available, it is
821 not possible to do most of the suggested tests. If the intended use depends on a consistent,
822 high level of validation, it may be proper to reject such data.
823 The missing data should be fundamental. For example, missing certificates for standards
824 are unlikely to be fundamental if lab performance on spiked samples is acceptable. In
825 contrast, if no spiked sample data is available, it may be impossible to determine if a
826 method gives even roughly correct results, and rejection may be appropriate.
827 2. Available data indicate that the assumptions underlying the method are not true. For
828 example, QC samples may demonstrate that the lab's processes are out of control.
829 Method performance data may indicate that the method simply does not work for
830 particular samples. These problems should be so severe that is not possible to make
831 quantitative estimates of their effects.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 8-29 DO NOT CITE OR QUOTE
-------
Radiochemical Data Verification and Validation
832 3. A result is "very unusually uncertain." It is difficult to say what degree of uncertainty
833 makes a result unusable. Whenever possible, uncertain data should be rejected based on
834 multiple problems with one result, patterns in related data, and the validator's judgement,
835 not the outcome of a single test. This requires radiochemistry expertise and knowledge of
836 the intended use.
837 Based on an evaluation of the tentative qualifiers, final qualifiers are assigned to each regular
838 sample result.
839 After all necessary validation tests have been completed and a series of qualifiers assigned to
840 each data point based on the results of the tests, a final judgment to determine which, if any, final
841 qualifiers will be attached to the data should be made. The individual sample data from the
842 laboratory should retain all the qualifiers. The basic decision making process for each result is
843 always subj ect to validator judgement:
844 • As appropriate, assign a final "R";
845 • IP'S", "P", or "B" were assigned, determine whether the qualifiers warrant the assignment of
846 an "R";
847 • If "R" is not assigned, but some test assigned a tentative S, P, B, Q, or J, or a pattern exists
848 that makes it appropriate, assign a final S, P, B, Q, or J and summarize QC sample
849 performance;
850 • If a final S, B, or J was assigned, + or -, but not both, was tentatively assigned, and the
851 potential bias is not outweighed by other sources of uncertainty, make the + or - final; and
852 • For non-R results, if any test assigned a tentative "U," make it final.
853 The final validation decision should address the fact that the broader purpose of validation is to
854 contribute to the total data collection process, i.e., effectively translate and interpret analytical
855 results for efficient use by an assessor. This means the validator should examine the full range of
856 data available to search for and utilize relationships among the data elements to support the
857 acceptance and use of data that falls outside method or contract specifications and data validation
858 plan guidance.
859 8.6 Validation Report
860 The final product of validation is a package that summaries the validation process and its
861 conclusions in an orderly fashion. This package should include:
MARLAP JULY 2001
DO NOT CITE OR QUOTE 8-30 DRAFT FOR PUBLIC COMMENT
-------
Radiochemical Data Verification and Validation
862 • A narrative or summary table written by the validator that summarizes exceptional
863 circumstances: In particular, it should document anything that prevented executing the
864 planned validation tests. Further, the narrative should include an explicit statement explaining
865 why data has been rejected or qualified based on the findings of the validation tests and the
866 validator's judgment.
867 • A list of validated samples that provides a cross-reference of laboratory and client sample
868 identifiers: This report should also include other identifiers useful in the context of the
869 project, such as reporting batch, chain of custody, or other sample management system
870 sample information.
871 • A summary of all validated results with associated uncertainty for each regular sample with
872 final qualifiers: Unless specified in the sampling and analysis plan, non-detects are reported
873 as measured, not replaced by a detection limit or other "less than" value.
874 • A summary of QC sample performance and the potential effect on the data both qualified and
875 not qualified.
876 Assuming the client wants additional information, the following, more detailed reports can be
877 included in the validation package. Otherwise, they are simply part of the validation process and
878 the verification contract compliance:
879 • A detailed report of all tentative qualifiers and associated reasons for their assignment;
880 • QC sample reports that document analytical process problems; and
881 • Reports that summarize performance by method—these should support looking across related
882 analyses at values such as yields and result ratios.
883 The data in the summary reports should be available in a computer-readable format. If no result
884 was obtained for a particular analyte, the result field should be left blank. The validation report
885 should package analytical results as effectively as possible for application and use by the
886 individual assembling and assessing all project data.
887 The validation report should contain a discussion describing the problem(s) found during the
888 validation process. For the validation codes, the discussion summarizes the performance criteria
889 established in the validation plan. If the validation test performance criteria were changed (e.g.,
890 increased or decreased level of unusual uncertainty) because the nature of the sample matrix or
891 analyte was different than expected, the new criteria should be explained in the report and the
892 qualifiers applied using the new criteria. The approval of the project manager should be obtained
893 (and documented) before the new criteria are applied. The project manager should communicate
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 8-31 DO NOT CITE OR QUOTE
-------
Radiochemical Data Verification and Validation
894 the changes to the project planning team to maintain the consensus reached and documented
895 during validation planning.
896 Well-planned and executed analytical activities can be expected to meet reasonable expectations
897 for data reliability. This means that for most data points or data sets, the results of the tests of
898 quality control, detection, and unusual uncertainty will show that the data are of sufficient quality
899 and defensibility to be forwarded to the assessor with little or no qualification for final
900 assessment. A small number of points will be rejected because random errors in the analytical
901 process or unanticipated matrix problems resulted in massive failure of several key validation
902 tests.
903 A smaller number of data points will show conflicting results from the validation tests and
904 present the greatest challenge to the validator. The more important the decision and/or the lower
905 the required detection limit, the more common this conflict will become, and the more critical it
906 is that the data validation plan provide guidance to the validator about how to balance the
907 conflicting results. Is the ability to detect the analyte more important than the associated
908 statistical unusual uncertainly, or is the presence of the analyte relatively definite but the unusual
909 uncertainty around the project decision point critical to major decisions? The necessary guidance
910 should be developed during the planning phase to guide the final judgment of the validator.
9ii 8.7 Other Sources of Information
912 American National Standards Institute (ANSI) N13.30. 1996. Performance Criteria for
913 Radiobioassay.
914 U.S. Environmental Protection Agency (EPA). 1994. Contract Laboratory Program National
915 Functional Guidelines for Inorganic Data Review. EPA-540/R-94-013 (PB94-963502).
916 February. Available from http://www.epa.gov/oerrpage/superfund/programs/clp/download/
917 fginorg.pdf.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 8-32 DRAFT FOR PUBLIC COMMENT
-------
9 DATA QUALITY ASSESSMENT
2 9.1 Introduction
3 This chapter provides an overview of the data quality assessment (DQA) process, the third and
4 final process of the overall data assessment phase of a project. Assessment is the last phase in the
5 data life cycle and precedes the use of data. Assessment—in particular DQA—is intended to
6 evaluate the suitability of project data to answer the underlying project questions or the suitability
7 of project data to support the project decisions. The output of this final assessment process is a
8 determination as to whether a decision can or cannot be made within the project-specified data
9 quality objectives (DQOs).
10 The discussions in this chapter assume that prior to the DQA process, the individual data
11 elements have been subjected to the first two assessment processes, "data verification" and "data
12 validation" (see Chapter 8, Radiochemical Data Verification and Validation). The line between
13 these three processes has been blurred for some time and varies from guidance to guidance and
14 practitioner to practitioner. Although the content of the various processes is the most critical
15 issue, a common terminology is necessary to minimize confusion and to improve communication
16 among planning team members, those who will implement the plans, and those responsible for
17 assessment. MARLAP defines these terms in Section 1.4 and discusses assessment in Section 8.2
18 This chapter is not intended to address the detailed and specific technical issues needed to assess
19 the data from a specific project but rather to impart a general understanding of the DQA process
20 and its relationship to the other assessment processes, as well as of the planning and implemen-
21 tation phases of the project's data life cycle. The target audience for this chapter is the project
22 planner, project manager, or other member of the planning team who wants to acquire a general
23 understanding of the DQA process; not the statistician, engineer, or radiochemist who is seeking
24 detailed guidance for the planning or implementation of the assessment phase. Guidance on
25 specific technical issues is available (EPA, 2000; MARSSEVI, 2000; NRC, 1998).
26 This chapter emphasizes that assessment, although represented as the last phase of the project's
27 data life cycle, should be planned for during the directed planning process, and the needed
28 documentation should be provided during the implementation phase of the project.
29 Section 9.2 reviews the role of DQA in the assessment phase. Section 9.3 discusses the graded
30 approach to DQA. The role of the DQA team is discussed in Section 9.4. Section 9.5 describes
31 the content of DQA plans. Section 9.6 details the activities that are involved in the DQA process.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 9-1 DO NOT CITE OR QUOTE
-------
Data Quality Assessment
32 9.2 Assessment Phase
33 The assessment phase was discussed in Section 8.2. This subsection provides a brief overview of
34 the individual assessment processes, their distinctions, and how they interrelate.
35 "Data verification" generally evaluates compliance of the analytical process with project-plan
36 and other project-requirement documents, and the statement of work (SOW), and documents
37 compliance and noncompliance in a data verification report. Data verification is a separate
38 activity in addition to the checks and review done by field and laboratory personnel during
39 implementation.
40 Documentation generated during the implementation phase will be used to determine if the
41 proper procedures were employed and to determine compliance with project plan documents
42 (e.g., QAPP), contract-specified requirements, and measurement quality objectives (MQOs). Any
43 data associated with noncompliance will be identified as an "exception," which should elicit
44 further investigation during data validation.
45 Compliance, exceptions, missing documentation, and the resulting inability to verify compliance
46 should be recorded in the data verification report. Validation and DQA employ the verification
47 report as they address the usability of data in terms of the project DQOs.
48 "Data validation" qualifies the usability of each datum after interpreting the impacts of
49 exceptions identified during verification. The validation process should be well defined in a
50 validation plan that was completed during the planning phase. The validation plan, as with the
51 verification plan or checklist, can range from sections of a project plan to large and detailed
52 stand-alone documents. Regardless of its size or format, the validation plan should address the
53 issues presented in Section 8.3. Data validation begins with a review of project objectives and
54 requirements, the data verification report, and the identified exceptions. The data validator
55 determines if the analytical process was in statistical control (Section 8.5.1) at the time of sample
56 analysis, and whether the analytical process as implemented was appropriate for the sample
57 matrix and analytes of interest (Section 8.5.2). If the system being validated is found to be under
58 control and applicable to the analyte and matrix, then the individual data points can be evaluated
59 in terms of detection (Section 8.5.3.1), detection capability (Section 8.5.3.2), and unusual
60 uncertainty (Section 8.5.3.3). Following these determinations, the data are assigned qualifiers
61 (Section 8.5.4) and a data validation report is completed (Section 8.6). Validated data are rejected
62 only when the impact of an exception is so significant that the datum is unreliable.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 9-2 DRAFT FOR PUBLIC COMMENT
-------
Data Quality Assessment
63 While both data validation and DQA processes address usability, the processes address usability
64 from different perspectives. "Data validation" attempts to interpret the impacts of exceptions
65 identified during verification and the impact of project activities on the usability of an individual
66 datum. In contrast, "data quality assessment" considers the results of data validation while
67 evaluating the usability of the entire data set.
68 During data validation, the guidance in Chapter 8 strongly advises against the rejection of data
69 unless there is a significant argument to do so. As opposed to rejecting data, it is generally
70 preferable that data are qualified and that the data validator details the concerns in the data
71 validation report. However, there are times when data should be rejected, and the rational for the
72 rejection should be explained in the data validation report. There are times when the data
73 validator may have believed data should be rejected based on a viable concern, yet during DQA,
74 a decision could be made to employ the rejected data.
75 In summary, data validation is a transition from the compliance testing of data verification to
76 usability determinations. The results of data validation, as captured in the qualified data and
77 validation reports, will greatly influence the decisions made during the final assessment process,
78 data quality assessment, which is discussed in Section 9.6.
79 9.3 Graded Approach to Assessment
80 The sophistication of the assessment phase—and in particular DQA and the resources applied—
81 should be appropriate for the project (i.e., a "graded approach"). Directed planning for small or
82 less complex projects usually requires fewer resources and typically involves fewer people and
83 proceeds faster. This graded approach to plan design is also applied to the assessment phase.
84 Generally, the greater the importance of a project, the more complex a project, or the greater the
85 ramifications of an incorrect decision, the more resources will be expended on assessment in
86 general and DQA in particular.
87 It is important to note that the depth and thoroughness of a DQA will be affected by the
88 thoroughness of the preceding verification and validation processes. Quality control or statement
89 of work (SOW) compliance issues that are not identified as an "exception" during verification, or
90 qualified during validation, will result in potential error sources not being reviewed and their
91 potential impact on data quality will not be evaluated. Thus, while the graded approach to
92 assessment is a valid and necessary management tool, it is necessary to consider all assessment
93 phase processes (data verification, data validation, and data quality assessment) when assigning
94 resources to assessment.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 9-3 DO NOT CITE OR QUOTE
-------
Data Quality Assessment
95 9.4 The Data Quality Assessment Team
96 The project planning team is responsible for ensuring that its decisions are scientifically sound
97 and comply with the tolerable decision-error rates established during planning. MARLAP
98 recommends the involvement of the data assessment specialist(s) on the project planning team
99 during the directed planning process. This should result in a more efficient assessment plan and
100 should increase the likelihood that flaws in the design of the assessment processes will be
101 detected and corrected during planning. Chapter 2.4 noted that it is important to have an
102 integrated team of operational and technical experts. The data assessment specialist(s) who
103 participated as members of the planning team need not be the final assessors. However, using the
104 same assessors who participated in the directed planning process is advantageous, since they will
105 be aware of the complexities of the project's goals and activities.
106 The actual personnel who will perform data quality assessment, or their requisite qualifications
107 and expertise, should be specified in the project plan documents. The project planning team
108 should choose a qualified data assessor (or team of data assessors) who is technically competent
109 to evaluate the project's activities and the impact of these activities on the quality and usability of
110 data. Multi-disciplinary projects may require a team of assessors (e.g., radiochemist, engineer,
ill statistician) to address the diverse types of expertise needed to assess properly the representa-
112 tiveness of samples, the accuracy of data, and whether decisions can be made within the specified
113 levels of confidence. Throughout this manual, the term "assessment team" will be used to refer to
114 the assessor expertise needed.
ii5 9.5 Data Quality Assessment Plan
116 To implement the assessment phase as designed and ensure that the usability of data are assessed
117 in terms of the project objectives, a detailed DQA plan should be completed during the planning
118 phase of the data life cycle. This section focuses on the development of the DQA plan and its
119 relation to DQOs and MQOs.
120 The DQA plan should address the concerns and requirements of all stakeholders and present this
121 information in a clear, concise format. Documentation of these DQA specifications,
122 requirements, instructions, and procedures are essential to assure process efficiency and that
123 proper procedures are followed. Since the success of a DQA depends upon the prior two
124 processes of the assessment phase, it is key that the verification and validation processes also be
125 designed and documented in respective plans during the planning phase. Chapter 8 lists the types
126 of guidance and information that should be included in data verification and validation plans.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 9-4 DRAFT FOR PUBLIC COMMENT
-------
Data Quality Assessment
127 MARLAP recommends that the DQA process should be designed during the directed planning
128 process and documented in a DQA plan. The DQA plan is an integral part of the project plan
129 documents and can be included as either a section or appendix to the project plan or as a cited
130 stand-alone document. If a stand-alone DQA plan is employed, it should be referenced by the
131 project plan and subjected to a similar approval process.
132 The DQA plan should contain the following information:
133 • A short summary and citation to the project documentation that provides sufficient detail
134 about the project objectives (DQOs), sample and analyte lists, required detection limit, action
135 level, and level of acceptable uncertainty on a sample- or analyte-specific basis;
136 • Specification of the necessary sampling and analytical assessment criteria (typically
137 expressed as MQOs for selected parameters such as method uncertainty) that are appropriate
138 for measuring the achievement of project objectives and constitute a basis for usability
139 decisions;
140 • Identification of the actual assessors or the required qualifications and expertise that are
141 required for the assessment team performing the DQA (Section 9.4);
142 • A description of the steps and procedures (including statistical tests) that will constitute the
143 DQA, from reviewing plans and implementation to authoring a DQA report;
144 • Specification of the documentation and information to be collected during the project's
145 implementation;
146 • A description for any project-specific notification or procedures for documenting the usability
147 or non-usability of data for the project's decision making;
148 • A description of the content of the DQA report;
149 • A list of recipients for the DQA report; and
150 • Disposition and record maintenance requirements.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 9-5 DO NOT CITE OR QUOTE
-------
Data Quality Assessment
151 9.6 Data Quality Assessment Process
152 MARLAP's guidance on the DQA process has the same content as other DQA guidance (ASTM
153 D6233; EPA, 2000; MARSSIM, 2000; NRC, 1998; USAGE, 1998), however, MARLAP
154 presents these issues in an order that parallels project implementation more closely. The
155 MARLAP guidance on the DQA process can be summarized as an assessment process that—
156 following the review of pertinent documents (Section 9.6.1)—answers the following questions:
157
158
159
Are the samples representative? (Section 9.6.2)
Are the analytical data accurate? (Section 9.6.3)
Can a decision be made? (Section 9.6.4)
160 Each of these questions is answered first by reviewing the plan and then evaluating the
161 implementation. The process concludes with the documentation of the evaluation of the data
162 usability in a DQA Report (Section 9.7).
163 The DQA Process is more global in its purview than the previous verification and validation
164 processes. The DQA process should consider the combined impact of all project activities in
165 making a data usability determination. The DQA process, in addition to reviewing the issues
166 raised during verification and validation, may be the first opportunity to review other issues, such
167 as field activities and their impact on data quality and usability. A summary of the DQA steps
168 and their respective output is presented in Table 9.1.
169
170
171
172
TABLE 9.1 — Summary of the DQA process
DQA PROCESS
Input
Output for DQA Report
1. Review Project
Plan Document
The project plan document (or a cited
stand-alone document) that addresses:
(a) Directed Planning Process Report,
including DQOs, MQOs and
optimized Sampling and Analysis
Plan.
(b) Revisions to documents in (a) and
problems or deficiency reports.
(c) DQA Plan.
Identification of project documents.
Clear understanding by the assessment team of
project's DQOs and MQOs.
Clear understanding of assumptions made
during the planning process.
If a clear description of the DQOs does not
exist, the assessment team should record the
DQOs (as they were established for
assessment).
173
174
2.
Are the Samples
Representative?
The project plan document (or a cited
stand-alone document) that addresses:
(a) The sampling portion of the
Sampling and Analysis Plan.
Documentation of all assumptions as potential
limitations and, if possible, a description of
their associated ramifications.
The determination of whether the design
MARLAP
DO NOT CITE OR QUOTE
9-6
JULY 2001
DRAFT FOR PUBLIC COMMENT
-------
Data Quality Assessment
DQA PROCESS
Input
Output for DQA Report
(b) SOPs for sampling.
(c) Sample handing and preservation
requirements of the APS
resulted in a representative sampling of the
population of interest.
The determination of whether the sampling
locations introduced bias.
The determination of whether the sampling
equipment and their use as described in the
sampling procedures were capable of extracting
a representative set of samples from the material
of interest.
An evaluation of the necessary deviations
(documented), as well as those deviations
resulting from misunderstanding or error, and a
determination of their impact on the representa-
tiveness of the affected samples.
175
176
3. Are the Data The project plan documents (or a
Accurate? cited stand-alone document) which
address:
(a) The analysis portion of the
Sampling and Analysis Plan.
(b) The MQOs.
(c) SOPs for analysis.
(d) Analytical Protocol Specifications,
including quality control
requirements and MQOs.
(e) SOW.
(f) The selected analytical protocols.
(g) Ongoing evaluations of
performance.
(h) Data Verification and Validation
plans and reports.
A determination of whether the selected method
was appropriate for the intended application.
The identification of any potential sources of
inaccuracy.
An assessment of whether the sample analyses
were implemented according to the analysis
plan.
An evaluation of the impact of any deviations
from the analysis plan on the usability of the
data set.
177
178
4.
Can a Decision
be Made?
The project plan document (or a cited stand-alone
document) that addresses:
(a) The DQA plan, including the statistical tests to
be used.
(b) The DQOs and the tolerable decision error
rates.
• Results of the statistical tests. If new tests were
selected, the rationale for their selection and the
reason for the inappropriateness of the
statistical tests selected in the DQA plan.
• Graphical representations of the data set and
parameters) of interest.
• Determination of whether the DQOs and
JULY 2001
DRAFT FOR PUBLIC COMMENT
9-7
MARLAP
DO NOT CITE OR QUOTE
-------
Data Quality Assessment
DQA PROCESS Input Output for DQA Report
tolerable decision error rates were met.
A final determination as to whether the data are
suitable for decision-making, estimation, or
answering questions within the levels of
certainty specified during planning.
179 9.6.1 Review of Project Documents
180 The first step of the DQA process is for the team to identify and become familiar with the DQOs
181 of the project and the DQA plan. Like the planning process, the steps of the DQA process are
182 iterative, but they are presented in this text in a step-wise fashion for discussion purposes.
183 Members of the assessment team may focus on different portions of the project plan documents
184 and different elements of the planning process. Some may do an in-depth review of the directed
185 planning process during this step; others will perform this task during a later step. The
186 assessment team should receive revisions to the project planning documents and should review
187 deficiency reports associated with the project. The subsections below discuss the key and
188 minimum project documents that should be reviewed.
189
190 9.6.1.1 The Project DQOs and MQOs
191 Since the usability of data is measured in terms of the project DQOs, the first step in the DQA
192 process is to acquire a thorough understanding of the DQOs. If the DQA will be performed by
193 more than one assessor, it is essential that the assessment team shares a common understanding
194 of the project DQOs and tolerable decision error rates. The assessment team will refer to these
195 DQOs continually as they make determinations about data usability. The results of the directed
196 planning process should have been documented in the project plan documents. The project plan
197 documents, at a minimum, should describe the DQOs and MQOs clearly and in enough detail
198 that they are not subject to misinterpretation or debate at this last phase of the project.
199 If the DQOs and MQOs are not described properly in the project plan documents or do not
200 appear to support the project decision, or if questions arise, it may be necessary to review other
201 planning documents (such as memoranda) or to consult the project planning team or the core
202 group (Section 2.4). If a clear description of the DQOs does not exist, the assessment team
203 should record any clarifications the assessment team made to the DQO statement as part of the
204 DQA report.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 9-8 DRAFT FOR PUBLIC COMMENT
-------
Data Quality Assessment
205 9.6.1.2 The DQA Plan
206 If the assessment team was not part of the directed planning process, the team should familiarize
207 itself with the DQA plan and become clear on the procedures and criteria that are to be used for
208 the DQA Process. If the assessment team was part of the planning process, but sufficient time has
209 elapsed since the conclusion of planning, the assessment team should review the DQA plan. If
210 the process is not clearly described in a DQA plan or does not appear to support the project
211 decision, or if questions arise, it may be necessary to consult the project planning team or the
212 core group. If necessary, the DQA plan should be revised. If it cannot be, any deviations from it
213 should be recorded in the DQA report.
214 During DQA, it is important for the team, including the assessors and statistician, to be able to
215 communicate accurately. Unfortunately, this communication can be complicated by the different
216 meanings assigned to common words (e.g., samples, homogeneity). The assessment team should
217 be alert to these differences during their deliberations. The assessment team will need to
218 determine the usage intended by the planning team.
219 It is important to use a directed planning process to ensure that good communications exist from
220 planning through data use. If the statistician and other experts are involved through the data life
221 cycle and commonly understood terms are employed, chances for success are increased.
222 9.6.1.3 Summary of the DQA Review
223 The review of project documents should result in:
224 • An identification and understanding of project plan documents, including any changes made
225 to them and any problems encountered with them;
226 • A clear understanding of the DQOs for the project. If a clear description of the DQOs does not
227 exist, the assessment team should reach consensus on the DQOs prior to commencing the
228 DQA and record the DQOs (as they were established for assessment) as part of the DQA
229 report; and
230 • A clear understanding of the terminology, procedures, and criteria for the DQA process.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 9-9 DO NOT CITE OR QUOTE
-------
Data Quality Assessment
231 9.6.2 Sample Representativeness
232 MARLAP does not provide guidance on developing sampling designs or a sampling plan. The
233 following discussion of sampling issues during a review of the DQA process is included for
234 purposes of completeness.
235 "Sampling" is the process of obtaining a portion of a population (i.e., the material of interest as
236 defined during the planning process) that can be used to characterize populations that are too
237 large or complex to be evaluated in their entirety. The information gathered from the samples is
238 used to make inferences whose validity reflects how closely the samples represent the properties
239 and analyte concentrations of the population. "Representativeness" is the term employed for the
240 degree to which samples properly reflect their parent populations. A "representative sample," as
241 defined in ASTM D6044, is "a sample collected in such a manner that it reflects one or more
242 characteristics of interest (as defined by the project objectives) of a population from which it was
243 collected" (Figure 9.1). Samples collected in the field as a group and subsamples generated as a
244 group in the laboratory (Appendix F) should reflect the population physically and chemically. A
245 flaw in any portion of the sample collection or sample analysis design or their implementation
246 can impact the representativeness of the data and the correctness of associated decisions.
247 Representativeness is a complex issue related to analyte of interest, geographic and temporal
248 units of concern, and project objectives.
249 The remainder of this subsection discusses the issues that should be considered in assessing the
250 representativeness of the samples: the sampling plan (Section 9.6.2.1) and its implementation
251 (Section 9.6.2.2). MARLAP recommends that all sampling design and statistical assumptions be
252 identified clearly in project plan documents along with the rationale for their use.
253 9.6.2.1 Review of the Sampling Plan
254 The sampling plan and its ability to generate representative samples are assessed in terms of the
255 project DQOs. The assessors review the project plan with a focus on the approach to sample
256 collection, including sample preservation, shipping and subsampling in the field and laboratory,
257 and sampling standard operating procedures (SOPs). Ideally the assessors would have been
258 involved in the planning process and would be familiar with the DQOs and MQOs and the
259 decisions made during the selection of the sampling and analysis design. If the assessors were
260 part of the project planning team, this review to become familiar with the project plan will go
261 quickly, and the team can focus on deviations from the plan that will introduce unanticipated
262 imprecision or bias (Section 9.6.2.2).
MARLAP JULY 2001
DO NOT CITE OR QUOTE 9-10 DRAFT FOR PUBLIC COMMENT
-------
Data Quality Assessment
ANALYTICAL
SUBSAMPLES
n n n
FIELD SAMPLES
Collectively Subsamples
represent population
DATABASE
Collectively Samples
represent population
Database accurately represents
the measured population
characteristic
POPULATION
263
FIGURE 9.1 — Using physical samples to measure a characteristic of the population representatively.
APPROACH TO SAMPLE COLLECTION
264 Project plan documents (e.g., QAPP, SAP, Field Sampling Plan) should provide details about the
265 approach to sample collection and the logic that was employed in its development. At this stage,
266 the assessment team should evaluate whether the approach, as implemented, resulted in
267 representative samples. For example, if the approach was probabilistic, the assessment team
268 should determine if it was appropriate to assume that spatial or temporal correlation is not a
269 factor, and if all portions of the population had an equal chance of being sampled. If an
270 "authoritative" sample collection approach was employed (i.e., a person uses his knowledge to
271 choose sample locations and times), the assessment team—perhaps in consultation with the
272 appropriate experts (e.g., an engineer familiar with the waste generation process)—should
273 determine if the chosen sampling conditions do or do not result in a "worst case" or "best case."
274 The assessment team should evaluate whether the chosen sampling locations resulted in a
275 negative or positive bias, and whether the frequency and location of sample collection accounted
276 for the population heterogeneity.
JULY 2001
DRAFT FOR PUBLIC COMMENT
9-11
MARLAP
DO NOT CITE OR QUOTE
-------
Data Quality Assessment
277 Optimizing the data collection activity (Section 2.5.4 and Appendix B3.8) involved a number of
278 assumptions. These assumptions are generally employed to manage a logistical, budgetary, or
279 other type of constraint, and are used instead of additional sampling or investigations. The
280 assessment team needs to understand these assumptions in order to fulfill its responsibility to
281 review and evaluate whether their continued validity based on the project's implementation. The
282 assessment team should review the bases for the assumptions made by the planning team because
283 they can result in biased samples and incorrect conclusions. For example, if samples are collected
284 from the perimeter of a lagoon to characterize the contents of the lagoon, the planning team's
285 assumption was that the waste at the lagoon perimeter has the same composition as that waste
286 located in the less-accessible center of the lagoon. In this example, there should be information to
287 support the assumption, such as historical data, indicating that the waste is relatively homogen-
288 ous and well-mixed. Some assumptions will be stated clearly in project plan documents. Others
289 may only come to light after a detailed review. The assessment team should review assumptions
290 for their scientific soundness and potential impact on the representativeness of the samples.
291 Ideally, assumptions would be identified clearly in project plan documents, along with the
292 rationale for their use. Unfortunately, this is uncommon, and in some cases, the planners may be
293 unaware of some of the implied assumptions associated with a design choice. The assessment
294 team should document any such assumptions in the DQA report as potential limitations and, if
295 possible, describe their associated ramifications. The assessment team may also suggest
296 additional investigations to verify the validity of assumptions which are questionable or key to
297 the project.
298 SAMPLING SOPs
299 Standard operating procedures for sampling should be assessed for their appropriateness and
300 scientific soundness. The assessment team should assess whether the sampling equipment and
301 their use, as described in the sampling procedures, were capable of extracting a representative set
302 of samples from the material of interest. The team also should assess whether the equipment's
303 composition was compatible with the analyte of interest. At this stage, the assessment team
304 assumes the sampling device was employed according to the appropriate SOP. Section 9.6.2.2
305 discusses implementation and deviations from the protocols.
306 In summary, the assessment team should investigate whether:
307 • The sampling device was compatible with the material being sampled and with the analytes of
308 interest;
MARLAP JULY 2001
DO NOT CITE OR QUOTE 9-12 DRAFT FOR PUBLIC COMMENT
-------
Data Quality Assessment
309 • The sampling device accommodated all particle sizes and did not discriminate against
310 portions of the material being sampled;
311 • The sampling device resulted in contamination or loss of sample components;
312 • The sampling device allowed access to all portions of the material of interest;
313 • The sample handling, preparation, and preservation procedures maintained sample integrity;
314 and
315 • The field and laboratory subsampling procedures resulted in a subsample that accurately
316 represents the contents of the original sample.
317 These findings should be detailed in the DQA report.
318 9.6.2.2 Sampling Plan Implementation
319 The products of the planning phase are integrated project plan documents that define how the
320 planners intend the data collection process to be implemented. At this point in the DQA process,
321 the assessment team determines whether sample collection was done according to the plan,
322 reviews any noted deviations from the protocols, identifies any additional deviations, and
323 evaluates the impact of these deviations on sample representativeness and the usability of the
324 data. The success of this review will be a function of the documentation requirements specified
325 during the planning process, and how thoroughly these requirements were met during sample
326 collection.
327 The determination as to whether the plans were implemented as written typically will be based
328 on a review of documentation generated during the implementation phase, through on-site
329 assessments, and during verification, if sampling activities (e.g., sample preservation) were
330 subjected to verification. In some instances, assessment team members may have firsthand
331 knowledge from an audit that they performed, but in general the assessment team will have to
332 rely upon documentation generated by others. The assessment team will review field notes,
333 sample forms, chain-of-custody forms, verification reports, audit reports, deviation reports,
334 corrective action documentation, QA reports, and reports to management. The assessment team
335 also may choose to interview field personnel to clarify issues or to account for missing
336 documentation, .
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 9-13 DO NOT CITE OR QUOTE
-------
Data Quality Assessment
337 Due to the uncontrolled environments from which most samples are collected, the assessment
338 team expects to find some deviations even from the best-prepared plans. Those not documented
339 in the project deficiency and deviation reports should be detailed in the DQA report. The
340 assessment team should evaluate these necessary deviations, as well as those deviations resulting
341 from misunderstanding or error, and determine their impact on representativeness of the affected
342 samples. These findings also should be detailed in the DQA report.
343 In summary, the assessment team will develop findings and determinations regarding any
344 deviations from the original plan, the rationale for the deviations, and if the deviations raise
345 question of representativeness.
346 9.6.2.3 Data Considerations
347 Sample representativeness also can be evaluated in light of the resulting data. Favorable
348 comparisons of the data to existing data sets (especially those data sets collected by different
349 organizations and by different methods) offer encouraging evidence of representativeness, but
350 not absolute confirmation of sample representativeness, since both data sets could suffer from the
351 same bias and imprecision. The project plan documents should have referenced any credible and
352 applicable existing data sets identified by the planning team. Comparisons to existing data sets
353 may offer mutual support for the accuracy of each other, and when differences result they tend to
354 raise questions about both data sets. Quite often, the DQA assessors are looking for confirmatory
355 or conflicting information. How existing data sets are used during the DQA will be determined
356 by how much confidence the assessors place in them. If they are very confident in the accuracy of
357 existing data sets, then they may classify the new data as unusable if it differs from the existing
358 data. If there is little confidence in the existing data set, then the assessors may just mention in
359 the DQA report that the new data set was in agreement or not in agreement. However, if the
360 planning team has determined that additional data were needed, they probably will not have
361 sufficient confidence in the existing data set for purposes of decision-making.
362 Data comparison is an issue that could be addressed during validation to some degree, depending
363 on the validation plan. However, at this point in the DQA, comparable data sets serve a different
364 purpose. For example, the MDCs, concentration units, and the analytical methods may be the
365 same and allow for data comparison in validation. However, the assessors during DQA would
366 look for similarities and dissimilarities in reported concentrations for different areas of the
367 populations, and whether any differences might be an indication of a bias or imprecision that
368 makes the samples less representative. Temporal and spatial plots of the data also may be helpful
369 in identifying portions of the sampled population that were over- or under-represented by the data
370 collection activity.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 9-14 DRAFT FOR PUBLIC COMMENT
-------
Data Quality Assessment
371 The planning process and development of probabilistic sampling plans typically require
372 assumptions regarding average concentrations and variances. If the actual average concentrations
373 and variances are different than anticipated, it is important for the assessment team to evaluate
374 the ramifications of these differences on sample representativeness. As reported values approach
375 an action level, the greater the need for the sample collection activities to accurately represent the
376 population characteristics of interest.
377 During the evaluation of sample representativeness, as discussed in the previous subsections, the
378 assessment team has the advantage of hindsight, since they review the sample collection design
379 in light of project outcomes and can determine if the sample collection design could have been
380 optimized differently to better achieve project objectives. Findings regarding the representative-
381 ness of samples and how sampling can be optimized should be expeditiously passed to project
382 managers if additional sampling will be performed.
383 In summary, results of the evaluation of the sample representativeness are:
384 • An identification of any assumptions that present limitations and, if possible, a description of
385 their associated ramifications;
386 • A determination of whether the design resulted in a representative sampling of the population
387 of interest;
388 • A determination of whether the specified sampling locations, or alternate locations as
389 reported, introduced bias;
390 • A determination of whether the sampling equipment and their use, as described in the
391 sampling procedures or as implemented, were capable of extracting a representative set of
392 samples from the material of interest; and
393 • An evaluation of the necessary deviations from the plan, as well as those deviations resulting
394 from misunderstanding or error, and a determination of their impact on the representativeness
395 of the affected samples.
396 The product of this step is a set of findings regarding the impact of representativeness—or the
397 lack thereof—that affects data usability. Findings and determinations regarding representative-
398 ness will impact the usability of the resulting data to varying degrees. Some findings may be so
399 significant (e.g., the wrong waste stream was sampled) that the samples can be determined to be
400 non-representative and the associated data cannot be used; as a result, the DQA need not progress
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 9-15 DO NOT CITE OR QUOTE
-------
Data Quality Assessment
401 any further. Typically, findings will be subject to interpretation, and the impacts on representa-
402 tiveness will have to be evaluated in light of other DQA findings to determine the usability of
403 data.
404 9.6.3 Data Accuracy
405 The next step in the DQA process is the evaluation of the analysis process and accuracy of the
406 resulting data. The term "accuracy" describes the closeness of the result of a measurement to the
407 true value of the quantity being measured. The accuracy of results may be affected by both
408 imprecision and bias in the measurement process, and by blunders and loss of statistical control
409 (see Chapter 19, Measurement Statistics, for a discussion on measurement error).
410 Since MARLAP uses "accuracy" only as a qualitative concept, in accordance with the
411 International Vocabulary of Basic and General Terms in Metrology (ISO, 1993), the agreement
412 between measured results and true values is evaluated quantitatively in terms of the
413 "imprecision" and "bias" of the measurement process. "Imprecision" usually is expressed as a
414 standard deviation, which measures the dispersion of results about their mean. "Bias" is a
415 persistent distortion of results from the true value.
416 During the directed planning process, the project planning team should have made an attempt to
417 identify and control sources of imprecision and bias (Appendix B3.8). During DQA, the
418 assessment team should evaluate the degree of imprecision and bias and determine its impact on
419 data usability. Quality control samples are analyzed for the purpose of assessing imprecision and
420 bias. Spiked samples and method blanks typically are used to assess bias, and duplicates are used
421 to assess imprecision. Since a single measurement of a spike or blank principle cannot
422 distinguish between imprecision and bias, a reliable estimate of bias requires a data set that
423 includes many such measurements. Control charts of quality control (QC) data, such as field
424 duplicates, matrix spikes, and laboratory control samples are graphical representations and
425 primary tools for monitoring the control of sampling and analytical methods and identifying
426 precision and bias trends (Chapter 18, Quality Control).
427 Measurable types of bias are identified and controlled through the application of quantitative
428 MQOs to QC samples, such as blanks, standard reference materials, performance evaluation
429 samples, calibration check standards, and spikes samples. Non-measurable forms of bias (e.g., a
430 method being implemented incorrectly, such as reagents being added in the incorrect order) are
431 usually identified and controlled by well-designed plans that specify quality assurance systems
432 that detail needed training, use of appropriate SOPs, deficiency reporting systems, assessments,
433 and quality improvement processes.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 9-16 DRAFT FOR PUBLIC COMMENT
-------
Data Quality Assessment
434 Bias in a data set may be produced by measurement errors that occur in steps of the measurement
435 process that are not repeated. Imprecision may be produced by errors that occur in steps that are
436 repeated many times. The distinction between bias and imprecision is complicated by the fact
437 that some steps, such as instrument calibration and tracer preparation and standardization, are
438 repeated at varying frequencies. For this reason, the same source of measurement error may
439 produce an apparent bias in a small data set and apparent imprecision in a larger data set. During
440 data assessment, an operational definition of bias is needed. A bias may exist if results for
441 analytical spikes (i.e., laboratory control samples, matrix spike, matrix spike duplicate),
442 calibration checks, and performance evaluation samples associated with the data set are mostly
443 low or mostly high, if the results of method blank analyses tend to be positive, or if audits
444 uncover certain types of biased implementation of the SOPs. At times, the imprecision of small
445 data sets can incorrectly indicate a bias, while at other times, the presence of bias may be masked
446 by imprecision. Statistical methods can be applied to imprecise data sets and used to determine if
447 there are statistically significant differences between data sets or between a data set and an
448 established value. If the true value or reference value (e.g., verified concentration for a standard
449 reference material) is known, then statistics can be used to determine whether there is a bias.
450 Figure 9.2 employs targets to depict the impacts of imprecision and bias on measurement data.
451 The true value is portrayed by the bulls-eye and is 100 units (e.g., dpm, Bq, pCi/g). Ideally, all
452 measurements with the same true value would be centered on the target, and after analyzing a
453 number of samples with the same true value, the reported data would be 100 units for each and
454 every sample. This ideal condition of precise and unbiased data is pictured in Figure 9.2(a). If the
455 analytical process is very precise but suffers from a bias, the situation could be as pictured in
456 Figure 9.2(b) in which the data are very reproducible but express a significant 70 percent
457 departure from the true value—a significant bias. The opposite situation is depicted in Figure
458 9.2(c), where the data are not precise and every sample yields a different concentration. However,
459 as more samples are analyzed, the effects of imprecision tend to average out, and lacking any
460 bias, the average measurement reflects the true concentration. Figure 9.2(d) depicts a situation
461 where the analytical process suffers from both imprecision and bias, and even if innumerable
462 samples with the same true value are collected and analyzed to control the impact of imprecision,
463 the bias would result in the reporting of an incorrect average concentration.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 9-17 DO NOT CITE OR QUOTE
-------
Data Quality Assessment
Precise
Precise
100ppm = true
concentration
Ave. = 100 = True Value
i
I
Biased
Ave. = 170
True Value
(100 ppm)
Ave. = 170
Imprecise
Imprecise
100ppm = true
concentration
(c) Unbiased
Ave. = 100
Ave. = 100 = True Value
BiaFed
Ave. = 150
True Value
(100 ppm) ]
i Ave. = 150
FIGURE 9.2 — Types of sampling and analytical errors.
464 Each target in Figure 9.2 has an associated frequency distribution curve. Frequency curves are
465 made by plotting a concentration value versus the frequency of occurrence for that concentration.
466 Statisticians employ frequency plots to display the imprecision of a sampling and analytical
467 event, and to identify the type of distribution. The curves show that as imprecision increases the
468 curves flatten-out and there is a greater frequency of measurements that are distant from the
MARLAP
DO NOT CITE OR QUOTE
9-18
JULY 2001
DRAFT FOR PUBLIC COMMENT
-------
Data Quality Assessment
469 average value (Figures 9.2c and d). More precise measurements result in sharper curves (Figures
470 9.2a and b), with the majority of measurements relatively closer to the average value. The greater
471 the bias (Figures 9.2b and d), the further the average of the measurements is shifted from the true
472 value. The smaller the bias (Figures 9.2a and c), the closer the average of the measurements is to
473 the true value.
474 The remainder of this subsection focuses on the review of analytical plans (Section 9.6.3.1) and
475 their implementation (Section 9.6.3.2) as a mechanism to assess the accuracy of analytical data
476 and their suitability for supporting project decisions.
477 9.6.3.1 Revi ew of the Analyti cal PI an
478 The analytical plan is that portion of the project plan documentation (e.g., in QAPP or SAP) that
479 addresses the optimized analytical design and other analytical issues (e.g., analytical protocol
480 specifications, SOPs). Its ability to generate accurate data is assessed in terms of the project
481 DQOs. The assessment team will refer to the DQOs and the associated MQOs as they review the
482 analytical protocol specifications to understand how the planning team selected methods and
483 developed the analytical plan. If the assessors were part of the project planning team, this review
484 process will go quickly and the team can focus on deviations from the plan that will introduce
485 unanticipated imprecision or bias. (The term "analytical plan" is not meant to indicate a separate
486 document.)
487 REVIEW OF THE MQOs, ANALYTICAL PROTOCOL SPECIFICATIONS, AND OPTIMIZED ANALYTICAL
488 DESIGN
489 The assessment team's review of the analytical plan first should focus on the analytical protocol
490 specifications, including the MQOs, which were established by the project planning team
491 (Chapter 3). The team should understand how the analytical protocol specifications were used to
492 develop the SOW (Chapter 5) and select the method (Chapter 6). If the project and contractual
493 documentation are silent or inadequate on how they address these key issues, the assessment
494 team may be forced to review the analytical results in terms of the project DQOs and determine if
495 the MQOs achieved were sufficient to meet the project's objectives.
496 As with the approach to sample collection, optimizing the analytical activity involved a number
497 of assumptions. Assumptions were made when analytical issues were resolved during planning
498 and the decisions were documented in the analytical protocol specifications (Chapter 3). It is
499 important for the assessment team to be aware of these assumptions because they can result in
500 biases and incorrect conclusions. Some assumptions will be clearly stated in the project plan
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 9-19 DO NOT CITE OR QUOTE
-------
Data Quality Assessment
501 documents. Others may only come to light after a detailed review. The assessment team should
502 review assumptions for their scientific soundness and potential impact on the data results.
503 Ideally, assumptions would be identified clearly in project plan documents, along with the
504 rationale for their use. Unfortunately, this is uncommon, and in some cases, the planners may be
505 unaware of some of the implied assumptions associated with a design choice. The assessment
506 team should document any such assumptions in the DQA report as potential limitations and, if
507 possible, describe their associated ramifications. The assessment team may also suggest
508 additional investigations to verify the validity of assumptions which are questionable or key to
509 the project.
510 REVIEW OF THE ANALYTICAL PROTOCOLS
511 The analytical plan and the associated analytical protocols will be reviewed and assessed for their
512 scientific soundness, applicability to the sample matrix and the ability to generate precise and
513 unbiased data. The analytical protocols review should consider the entire analytical process, from
514 sample preparation through dissolution and separations, counting, data reduction, and reporting.
515 MARLAP, whose focus is on the analytical process, defines "analytical process" as including
516 sample handling in the field (e.g., filtration, sample preservation) to ensure that all activities that
517 could impact analyses would be considered. The assessment team should consider both sampling
518 and analytical processes in assessing data quality—and such field activities as sample preserva-
519 tion—along with other issues that can affect representativeness (Section 9.6.2). The assessment
520 team also should review the contract evaluation (under the performance-based approach) for the
521 selection of the analytical protocols to assure that the documentation showed that the protocol
522 could meet the analytical protocol specifications (which defines the MQOs).
523 Since the review of the analytical protocols will be performed with the advantage of hindsight
524 gained from the data verification and data validation reports, the assessment team also should
525 attempt to identify any flaws in the analytical protocols that may have resulted in noncompliance
526 with MQOs. The identification of these flaws is essential if future analyses will be required.
527 REVIEW OF VERIFICATION AND VALIDATION PLANS
528 To understand how the verification and validations processes were implemented and the degree
529 to which the assessors can rely upon their findings, the assessors should familiarize themselves
530 with the verification and validation plans that were developed during the planning phase. A
531 review of these plans will indicate whether the issues deemed of importance to the assessors were
532 evaluated and the thoroughness of the evaluations.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 9-20 DRAFT FOR PUBLIC COMMENT
-------
Data Quality Assessment
533 9.6.3.2 Analytical Plan Implementation
534 After reviewing the analytical plan, the assessment team should assess whether sample analyses
535 were implemented according to the analysis plan. Typically, the first two steps of the assessment
536 phase—data verification and data validation—have laid most of the groundwork for this
537 determination. However, the issue of whether the plan was implemented as designed needs to be
538 reviewed one final time during the DQA process. This final review is needed since new and
539 pertinent information may have been uncovered during the first steps of the DQA process.
540 The goal of this assessment of the analytical protocols and associated MQOs is to confirm that
541 the selected method was appropriate for the intended application and to identify any potential
542 sources of inaccuracy, such as:
543 • Laboratory subsampling procedures that resulted in the subsample that may not accurately
544 represent the content of the original sample;
545 • Sample dissolution methods that may not have dissolved sample components quantitatively;
546 • Separation methods whose partitioning coefficients were not applicable to the sample matrix;
547 • Unanticipated self-absorption that biased counting data;
548 • Non-selective detection systems that did not resolve interferences; or
549 • Data reduction routines that lacked needed resolution or appropriate interference corrections
550
551 The success of the assessment of the analytical plan implementation will be a function of the
552 documentation requirements specified during the planning process, and how thoroughly these
553 requirements were met during sample analysis. In some instances, assessment team members
554 may have firsthand knowledge from an audit that they performed, but in general the assessment
555 team will have to rely upon documentation generated by others.
556 In addition to verification and validation reports, the assessment team will review pertinent
557 documents such as: laboratory notebooks, instrument logs, quality control charts, internal
558 sample-tracking documentation, audit reports, deviation reports, corrective action documentation,
559 performance evaluation sample reports, QA reports, and reports to management provided for
560 verification and validation. To clarify issues or to account for missing documentation, the
561 assessment team may choose to interview laboratory personnel.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 9-21 DO NOT CITE OR QUOTE
-------
Data Quality Assessment
562 Verification and validation reports will be used to identify nonconformance, deviations, and
563 problems that occurred during the implementation of the analytical plan. The challenge during
564 DQA is to evaluate the impact of nonconformance, deviations, problems, and qualified data on
565 the usability of the overall data set and the ability of the data set to support the decision.
566 Deviations from the plan will be encountered commonly and the assessment team will evaluate
567 the impact of these deviations upon the accuracy of the analytical data. The deviations and the
568 assessment team's related findings should be detailed in the data quality assessment report.
569 The prior verification and validation processes and the prior DQA steps involving the evaluation
570 of sampling are all an attempt to define the quality of data by (1) discovering sources of bias,
571 quantifying their impact, and correcting the reported data; and (2) identifying and quantifying
572 data imprecision. The products of this step are a set of findings regarding the analytical process
573 and their impact on data usability. Some findings may be so significant (e.g., the wrong analytical
574 method was employed) that the associated data cannot be used, and as a result, the DQA need not
575 progress any further. Typically, findings will be subject to interpretation and a final
576 determination as to the impacts will have to wait until the data has been subjected to evaluations
577 described in Section 9.6.4.
578 After reviewing the verification and validation reports, the outputs of the analytical data
579 evaluation are:
580 • A determination of whether the selected analytical protocols and analytical performance
581 specifications were appropriate for the intended application;
582 • An identification of any potential sources of inaccuracy; and
583 • A determination of whether sample analyses were implemented according to the analysis plan
584 and the overall impact of any deviations on the usability of the data set.
585 9.6.4 Decisions and Tolerable Error Rates
586 A goal of DQA is to avoid making a decision based on inaccurate data generated by analytical
587 protocols found to be out of control or on data generated from samples found to be nonrepresen-
588 tative, and to avoid making decisions based on data of unknown quality. Preferably, a decision
589 should be made with data of known quality (i.e., with data of known accuracy from samples of
590 known representativeness) and within the degree of confidence specified during the planning
591 phase.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 9-22 DRAFT FOR PUBLIC COMMENT
-------
Data Quality Assessment
592 This section focuses on the final determination by the assessment team, who uses the information
593 taken from the previous assessment processes and statistics to make a final determination of
594 whether the data are suitable for decision-making, estimating, or answering questions within the
595 levels of certainty specified during planning.
596 9.6.4.1 Statistical Evaluation of Data
597 Statistics are used for the collection, presentation, analysis, and interpretation of data. The two
598 major branches of statistics, "descriptive statistics" and "inferential statistics," are applicable to
599 data collection activities. "Descriptive statistics" are those methods that describe populations of
600 data. For example, descriptive statistics include the mean, mode, median, variance, and
601 correlations between variables, tables, and graphs to describe a set of data. "Inferential statistics"
602 use data taken from population samples to make estimates about the whole population
603 ("inferential estimations") and to make decisions ("hypothesis testing"). Descriptive statistics is
604 an important tool for managing and investigating data in order that their implications and
605 significance to the project goals can be understood.
606 Sampling and inferential statistics have identical goals—to use samples to make inferences about
607 a population of interest and to use sample data to make defensible decisions. This similarity is
608 the reason why planning processes, such as those described in Chapter 2, couple sample
609 collection activities with statistical techniques to maximize the representativeness of samples, the
610 accuracy of data, and the certainty of decisions.
611 Due to the complexity of some population distributions (Appendix 19A) and the complex
612 mathematics needed to treat these distributions and associated data, it is often best to consult
613 with someone familiar with statistics to ensure that statistical issues have been addressed
614 properly. However, it is critical for the non-statistician to realize that statistics has its limitations.
615 The following statistical limitations should be considered when assessment teams and the project
616 planning team are planning the assessment phase and making decisions:
617 • Statistics are used to measure precision and, when true or reference values are known,
618 statistics can be applied to imprecise data to determine if a bias exists. Statistics do not
619 address all types of sampling or measurement bias directly.
620 • If the characteristic of interest in a sample is more similar to that of samples adjacent to it than
621 to samples that are further removed, the samples are deemed to be "correlated" and are not
622 independent of each other (i.e., there is a serial correlation such that samples collected close in
623 time or space have more similar concentrations than those samples further removed).
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 9-23 DO NOT CITE OR QUOTE
-------
Data Quality Assessment
624 Conventional parametric and non-parametric statistics require that samples be independent
625 and are not applicable to populations that have significantly correlated concentrations.
626 The statistical tests typically are chosen during the directed planning process and are documented
627 in the project plan documents (e.g., DQA plan, QAPP). However, there are occasions when the
628 conditions encountered during the implementation phase are different than anticipated (e.g., data
629 were collected without thorough planning, or data are being subjected to an unanticipated
630 secondary data use). Under these latter conditions, the statistical tests will be chosen following
631 data collection.
632 The statistical analysis of data consists of a number of steps. The following outline of these steps
633 is typical of the analyses that a statistician would implement in support of a data quality
634 assessment.
635 CALCULATE THE BASIC STATISTICAL PARAMETERS
636 Chapter 19 has a detailed discussion of statistical issues, so a few concepts key to understanding
637 are summarized here. Statistical "parameters" are fundamental quantities that are used to describe
638 the central tendency or dispersion of the data being assessed. The mean, median, and mode are
639 examples of statistical parameters that are used to describe the central tendency, while range,
640 variance, standard deviation, coefficient of variation, and percentiles are statistical parameters
641 used to describe the dispersion of the data. These basic parameters are used because they offer a
642 means of understanding the data, facilitating communication and data evaluation, and generally
643 are necessary for subsequent statistical tests.
644 GRAPHICAL REPRESENTATIONS
645 Graphical representations of the data are similar to basic statistical parameters in that they are a
646 means of describing and evaluating data sets. Graphical representations of QC-sample results
647 used to evaluate project-specific control limits and warning limits derived from the MQO criteria
648 are discussed in Appendix C. Graphical representations of field data over space or time have the
649 additional ability of offering insights, such as identifying temporal and spatial patterns, trends,
650 and correlations. Graphical depictions are also an excellent means of communicating and
651 archiving information.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 9-24 DRAFT FOR PUBLIC COMMENT
-------
Data Quality Assessment
652 REVIEW AND VERIFY TEST ASSUMPTIONS
653 Statistical tests are the mathematical structure that will be employed to evaluate the project's data
654 in terms of the project decision, question, or parameter estimate. Statistical tests are not
655 universally applicable, and their choice and suitability are based on certain assumptions. For
656 example:
657 • Some tests are suitable for "normal" distributions, while others are designed for other types of
658 distributions.
659 • Some tests assume that the data are random and independent of each other.
660 • Assumptions that underlie tests for "outliers" should be understood to ensure that hot spots or
661 the high concentrations symptomatic of skewed distributions (e.g., lognormal) are not
662 incorrectly censored.
663 • Assumptions are made regarding the types of population distributions whenever data are
664 transformed before being subjected to a test.
665 • Assumptions of test robustness need to be reviewed in light of the analyte. For example,
666 radiological data require statistical tests that can accommodate positive and negative numbers.
667 It is important that a knowledgeable person identify all assumptions that underlie the chosen
668 statistical tests, and that the data are tested to ensure that the assumptions are met. If any of the
669 assumptions made during planning proved to be not true, the assessment team should evaluate
670 the appropriateness of the selected statistical tests. Any decision to change statistical tests should
671 be documented in the DQA report.
672 APPLYING STATISTICAL TESTS
673 The chosen statistical tests will be a function of the data properties, statistical parameter of
674 interest, and the specifics of the decision or question. For example, choice of the appropriate tests
675 will vary according to whether the data are continuous or discrete; whether the tests will be
676 single-tailed or double-tailed, whether a population is being compared to a standard or to a
677 second population, or whether stratified sampling or simple random sampling was employed.
678 Once the statistical tests are deemed appropriate, they should be applied to the data by an
679 assessor who is familiar with statistics. The outputs from applying the statistical tests and
680 comparisons to project DQOs are discussed in the following section.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 9-25 DO NOT CITE OR QUOTE
-------
Data Quality Assessment
681 9.6.4.2 Evaluation of Decision Error Rates
682 The heterogeneity of the material being sampled and the imprecision of the sampling and
683 analytical processes generate uncertainty in the reported data and in the associated decisions and
684 answers. The project planning team, having acknowledging this decision uncertainty, will have
685 chosen "tolerable decision errors rates" during the planning process, which balanced resource
686 costs against the risk of making a wrong decision or arriving at a wrong answer. During this final
687 step of DQA process, the assessment team will use the project's tolerable levels of decision error
688 rates as a metric of success.
689 The DQA process typically corrects data for known biases and then subjects the data to the
690 appropriate statistical tests to make a decision, answer a question, or supply an estimate of a
691 parameter. The assessment team will compare statistical parameters—such as the sample mean
692 and sample variance estimates employed during the planning process—to those that were
693 actually obtained from sampling. If the distribution was different, if the mean is closer to the
694 action level, or if the variance is greater or less than estimated, one or all of these factors could
695 have an impact on the certainty of the decision. The assessment team also will review the results
696 of the statistical tests in light of missing data, outliers, and rejected data. The results of the
697 statistical tests are then evaluated in terms of the project's acceptable decision error rates. The
698 assessment team determines whether a decision could or could not be made, or why the decision
699 could not be made, within the project specified decision error rates.
700 In summary, outputs from this step are:
701 • Generated statistical parameters;
702 • Graphical representations of the data set and parameters of interest;
703 • If new tests were selected, the rationale for selection and the reason for the inappropriateness
704 of the statistical tests selected in the DQA plan;
705 • Results of application of the statistical tests; and
706 • A final determination as to whether the data are suitable for decision making, estimating, or
707 answering questions within the levels of certainty specified during planning.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 9-26 DRAFT FOR PUBLIC COMMENT
-------
Data Quality Assessment
708 9.7 Data Quality Assessment Report
709 The DQA process concludes with the assessment team documenting the output of the statistical
710 tests and the rationale for why a decision could or could not be made, or why the decision could
711 not be made within the project specified decision error rates. The DQA report will document
712 findings and recommendations and include or reference the supporting data and information. The
713 DQA report will summarize the use of the data verification and data validation reports for data
714 sets of concern, especially if rejected for usability in the project's decision making. The report
715 also will document the answers to the three DQA questions:
716 • Are the samples representative?
717 • Are the data accurate?
718 • Can a decision be made?
719 Although there is little available guidance on the format for a DQA report, the report should
720 contain, at a minimum:
721 • An executive summary that briefly answers the three DQA questions and highlights major
722 issues, recommendations, deviations, and needed corrective actions;
723 • A summary of the project DQOs used to assess data usability, as well as pertinent
724 documentation such as the project plan document, contracts, and SOW;
725 • A listing of those people who performed the DQA;
726 • A summary description of the DQA process, as employed, with a discussion of any deviations
727 from the DQA plan designed during the planning process (the DQA plan should be appended
728 to the report);
729 • A summary of the data verification and data validation reports that highlights significant
730 findings and a discussion of their impact on data usability (the data verification and data
731 validation reports should be appended to the DQA report);
732 • A discussion of any missing documentation or information and the impact of their absence on
733 the DQA process and the usability of the data;
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 9-27 DO NOT CITE OR QUOTE
-------
Data Quality Assessment
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
• A thorough discussion of the three DQA questions addressing the details considered in
Sections 9.6.2 through 9.6.4 (possible outputs to be incorporated in the report are listed at the
conclusion of each these section);
• A discussion of deviations, sampling, analytical and data management problems, concerns,
action items, and suggested corrective actions (the contents of this section should be
highlighted in the executive summary if the project is ongoing and corrections or changes are
needed to improve the quality and usability of future data); and
• A recommendation or decision on the usability of the data set for the project's decision
making.
Upon completion, the DQA report should be distributed to the appropriate personnel as specified
in the DQA plan and archived along with supporting information for the period of time specified
in the project plan document. Completion of the DQA report concludes the assessment phase and
brings the data life cycle to closure.
Summary of Recommendations
MARLAP recommends that the assessment phase of a project (verification, validation, and
DQA processes) be designed during the directed planning process and documented in the
respective plans as part of the project plan documents.
MARLAP recommends that project objectives, implementation activities, and QA/QC data
be well documented in project plans, reports, and records, since the success of the
assessment phase is highly dependent upon the availability of such information.
MARLAP recommends the involvement of the data assessment specialist(s) on the project
planning team during the directed planning process.
MARLAP recommends that the DQA process should be designed during the directed
planning process and documented in a DQA plan.
MARLAP recommends that all sampling design and statistical assumptions be clearly
identified in project plan documents along with the rationale for their use.
MARLAP
DO NOT CITE OR QUOTE
9-28
JULY 2001
DRAFT FOR PUBLIC COMMENT
-------
Data Quality Assessment
760 9.8 References
761 9.8.1 Cited Sources
762 American Society for Testing and Materials (ASTM) D6044. Guide for Representative Sampling
763 and Management of Waste and Contaminated Media. 1996.
764 American Society for Testing and Materials (ASTM) D6233. Standard Guide for Data
765 Assessment for Environmental Waste Management Activities. 1998.
766 International Organization for Standardization (ISO) 1993. International Vocabulary of Basic
767 and General Terms in Metrology.
768 MARSSIM. 2000. Multi-Agency Radiation Survey and Site Investigation Manual, Revision 1.
769 NUREG-1575 Rev 1, EPA 402-R-97-016 Revl, DOE/EH-0624 Revl. August. Available
770 from http://www.epa.gov/radiation/marssim/filesfm.htm.
771 U.S. Army Corps of Engineers (USAGE). 1998. Technical Project Planning (TPP) Process.
Ill Engineer Manual EM-200-1-2.
773 U.S. Environmental Protection Agency (EPA). 2000. Guidance for the Data Quality Objective
774 Process (EPA QA/G-4). EPA/600/R-96/055, Washington, DC. available from www.epa.gov/
775 qualityl/qa_docs.html.
776 U.S. Nuclear Regulatory Commission (NRC). 1998. A Nonparametric Statistical Methodology
111 for the Design and Analysis of Final Status Decommissioning Surveys. NUREG 1505, Rev. 1.
778 9.8.2 Other Sources
779 American Society for Testing and Materials (ASTM). 1997. Standards on Environmental
780 Sampling, 2nd Edition, ASTM PCN 03-418097-38. West Conshohocken, PA.
781 American Society for Testing and Materials (ASTM) D5956. Standard Guide for Sampling
782 Strategies for Heterogeneous Wastes.
783 American Society for Testing and Materials (ASTM) D6051. Guide for Composite Sampling and
784 Field Subsampling for Environmental Waste Management A ctivities. 1996.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 9-29 DO NOT CITE OR QUOTE
-------
Data Quality Assessment
785 American Society for Testing and Materials (ASTM) D6311. Standard Guide for Generation of
786 Environmental Data Related to Waste Management Activities: Selection and Optimization of
787 Sampling Design. 1998
788 American Society for Testing and Materials (ASTM) D6323. Standard Guide for Laboratory
789 Subsampling of Media Related to Waste Management Activities. 1998.
790 Taylor, John Keenan. 1987. Quality Assurance of Chemical Measurements. Lewis Publishers,
791 Chelsea MI, ISBN 0-87371 -097-5.
792 U. S. Environmental Protection Agency (EPA). 1994. Guidance for the Data Quality Objective
793 Process, EPA QA/G-4, EPA/600/R-96/055, September.
794 U. S. Environmental Protection Agency (EPA). 1997. Guidance for Quality Assurance Project
795 Plans, EPA QA/G-5, August.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 9-30 DRAFT FOR PUBLIC COMMENT
-------
i 10 FIELD AND SAMPLING ISSUES THAT AFFECT
2 LABORATORY MEASUREMENTS
3 Part I: Generic Issues
4 10.1 Introduction
5 The primary purpose of this chapter is to provide guidance on issues that affect laboratory
6 measurements to project planners and managers tasked with developing a field sampling plan.
7 Specifically, this chapter provides guidance on activities conducted primarily after the proper
8 collection of the sample. Sampling design and collection are beyond the scope of MARLAP. A
9 field sampling plan should be a comprehensive document that provides detailed guidance for
10 collecting, preparing, preserving, shipping, tracking field samples, and recording field data. The
11 principal objective of a well-designed sampling plan is to provide representative samples of the
12 proper size for analysis. Critical to the sampling plan are outputs of the systematic planning
13 process, which commonly define the Analytical Protocol Specifications (APS) and the
14 Measurement Quality Objectives (MQO) that must be met. While a comprehensive discussion
15 that extends to field sampling strategies is beyond the scope of this chapter, specific aspects of
16 sample collection methods and physical preparation and preservation of samples warrant further
17 discussion because they impact the analytical process and the data quality.
18 This chapter is divided into two main parts. Part I identifies general elements of a field sampling
19 plan and provides project planners with general guidance. Part II provides more detailed informa-
20 tion. Matrix-specific guidance and technical data are presented for liquid, solid, airborne, and
21 surface contaminants requiring field sampling. This information will assist project planners
22 further in the development of standard operating procedures (SOPs) and training for field
23 personnel engaged in preparation and preservation of field samples.
24 The need to specify sample collection methods, and preparation and preservation of field
25 samples, is commonly dictated by one or more of the following:
26 • The systematic planning process that identifies the type, quality, and quantity of data needed
27 to satisfy a decision process;
28 • The potential alteration of field samples by physical, chemical, and biological processes
29 during the time between collection and analysis;
30 • Requirements specified by the analytical laboratory pertaining to sample analysis;
31 • Requirements of analytical methods; and
32 • Requirements of regulators (e.g., Department of Transportation).
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 10-1 DO NOT CITE OR QUOTE
-------
Field and Sampling Issues That Affect Laboratory Measurements
33 10.1.1 The Need for Establishing Channels of Communication
34 Of critical importance to the effective design of a sampling plan are the input and recommen-
35 dations of members representing: (1) the field sampling team; (2) the health physics professional
36 staff; (3) the analytical laboratory; (4) statistical and data analyses; (5) quality assurance
37 personnel, and (6) end-users of data.
38 Beyond the initial input that assist the project planners in the design of the sampling plan, it is
39 equally important to maintain open channels of communication among key members of the
40 project team throughout the process. For example, the analytical laboratory should be provided
41 with contacts from the field sampling team to ensure that modifications discrepancies and
42 changes are addressed and the timely resolution of potential problems.
43 Communication among project staff, field personnel, and the laboratory offer a means to
44 coordinate activities, schedules, and sample receipt. Project planning documents generated from
45 the systematic planning process, such as APS and statements of work (SOWs), should be
46 consulted, but they cannot address all details. Additional communication likely will be necessary.
47 Communication conveys information about the number and type of samples the laboratory can
48 expect at a certain time. Documentation with special instructions regarding the samples should be
49 received before the samples arrive. This information notifies the laboratory of any health and
50 safety concerns so that laboratory personnel can implement proper contamination management
51 practices. Health and safety concerns may affect analytical procedures, sample disposition, etc.
52 The analytical laboratory should have an initial understanding about the relative number of
53 samples that will be received and the types of analyses that are expected for specific samples.
54 Furthermore, advance communications allow laboratory staff to adjust to modifications,
55 discrepancies, and changes.
56 10.1.2 Developing Field Documentation
57 The field organization must conduct its operations in such a manner as to provide reliable
58 information that meets the data quality objectives (DQOs). To achieve this goal, all relevant
59 procedures pertaining to sample collection and processing should be based on documented
60 standard operating procedures that include the following activities:
61 • Developing a technical basis for defining the size of individual samples;
62 • Selecting field equipment and instrumentation;
63 • Using proper sample containers and preservatives;
64 • Using consistent container labels and sample identification codes;
65 • Documenting field sample conditions and exceptions;
66 • Documenting sample location;
67 • Tracking, accountability and custody, and shipment forms;
MARLAP JULY 2001
DO NOT CITE OR QUOTE 10-2 DRAFT FOR PUBLIC COMMENT
-------
Field and Sampling Issues That Affect Laboratory Measurements
68 • Legal accountability, such as chain-of-custody record, when required;
69 • Selecting samples for field QC program;
70 • Decontaminating equipment and avoiding sample cross-contamination;
71 • Sample packaging, shipping, and tracking; and
72 • Health and safety plan.
73 10.2 Field Sampling Plan: Non Matrix Specific Issues
74 10.2.1 Determination of Analytical Sample Size
75 When collecting environmental samples for radioanalysis, an important parameter for field
76 personnel is the mass, volume, or weight of an individual sample that must be collected. The
77 required minimum sample size is best determined through the collective input of project
78 planners, field technicians, and laboratory personnel who must consider the likely range of the
79 contaminant concentrations, the type of radiation emitted by constituents or analytes (a, p, y),
80 field logistics, and the radioanalytical methods that are to be employed. For samples to yield
81 useful data, it is important to have a quantitative understanding of the relationship between
82 sample size and project specific requirements.
83 10.2.2 Field Equipment and Supply Needs
84 Before starting field sampling activities, all necessary equipment and supplies should be
85 identified, checked for proper operation and availability, and—when appropriate—pre-
86 assembled. Instrumentation and equipment needs will depend not only on the medium to be
87 sampled, but also on the accessibility of the medium and the physical and chemical properties of
88 radionuclide contaminants under investigation.
89 Independent of specialized field equipment and instrumentation, field sampling supplies
90 commonly include the following:
91 • Sampling devices (e.g., trowel, hand auger, soil core sampler, submersible water pump, high
92 volume air filter, etc.);
93 • Sampling preparation equipment (e.g., weighing scales, volume measuring devices, soil
94 screening sieves, water filtering equipment, etc.);
95 • Sample preservation equipment and agents (e.g., refrigeration, ice, formaldehyde or acid
96 additives);
97 • Personnel protective gear (e.g., respiratory protective devices, protective clothing such as
98 gloves and booties, life-preservers, etc.);
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 10-3 DO NOT CITE OR QUOTE
-------
Field and Sampling Issues That Affect Laboratory Measurements
99 • Proper writing utensils (e.g., permanent pens and markers);
100 • Field logbooks and field tracking forms;
101 • Maps, distance measuring equipment, global positioning systems, or other location-
102 determining equipment;
103 • Field sampling flags or paint;
104 • Chain-of-custody (COC) forms;
105 • Sample tags, labels, documents;
106 • Appropriately labeled sample containers;
107 • Shipment containers and packing materials that meet DOT regulations;
108 • Shipment forms;
109 • Analysis request form identifying the type of radioanalysis to be performed; and
110 • Health and Safety Plan requirements (medical kit, etc.).
ill 10.2.3 Selection of Sample Containers
112 There are several physical and chemical characteristics that must be considered when selecting a
113 suitable container for shipping and storing samples. Important characteristics include the
114 container material and its size, configuration, and method for ensuring a proper seal.
115 10.2.3.1 Container Material
116 Sample containers must provide reasonable assurance of maintaining physical integrity (i.e.,
117 against breakage, rupture, or leakage) during handling, transport, and potentially long periods of
118 storage. The most important factor to consider in container selection is the chemical
119 compatibility between container material and sample. Containers may include ordinary bottle
120 glass, borosilicate glass (such as Pyrex or Corex), plastics (e.g., high density polyethylene—
121 FIDPE), low density polyethylene, polycarbonate, polyvinyl chloride (PVC), fluorinated ethylene
122 propylene (Reflon), or polymethelpentene. For select samples, the choice of containers may
123 require metal construction or be limited to paper envelopes.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 10-4 DRAFT FOR PUBLIC COMMENT
-------
Field and Sampling Issues That Affect Laboratory Measurements
124 10.2.3.2 Container Opening and Closure
125 Selection of a suitable container also must consider the ease with which the sample is introduced
126 into the container. For example, a wide-mouthed container will provide easier access for the
127 introduction and withdrawal of sample material and eliminate spills or the need for additional
128 tools or equipment (e.g., funnel) that may become a source of cross contamination among
129 samples.
130 Equally important is the container closure or seal. As a rule, snap-on caps should not be
131 considered for liquid samples because they do not ensure a proper seal. Even when screw caps
132 are used, it is frequently prudent to protect against vibration by securing the cap with electrical or
133 duct tape. A proper seal is important for air samples, such as radon samples. The container cap
134 material, if different from the container material, must be equally inert with regard to sample
135 constituents.
136 10.2.3.3 Sealing Containers
137 Tamper-proof seals offer an additional measure to ensure sample integrity. A simple example
138 includes placing a narrow strip of paper over a bottle cover and then affixing this to the container
139 with a wide strip of clear tape (EPA, 1987, Exhibit 5-6, example of custody seals). The paper
140 strip can be initialed and dated in the field to indicate the staff member who sealed the sample
141 and the date of the seal. Individually sealing each sample with a custody seal with the collector's
142 initials and the date the sample was sealed may be required by the project. The seal ensures legal
143 defensibility and integrity of the sample at collection. Tamper-proof seals should only be applied
144 once field processing and preservation steps are completed. Reopening this type of sealed
145 container in the field might warrant using a new container or collecting another sample.
146 10.2.3.4 Precleaned and Extra Containers
147 The reuse of sample containers is discouraged because traces of radionuclides might persist from
148 initial container use to subsequent use. The use of new containers for each collection removes
149 doubts concerning radionuclides from previous sampling. New containers might also require
150 cleaning (ASTM D5245) to remove plasticizer used in container production or to pretreat glass
151 surfaces. Retaining extra empty containers from a new lot or a special batch of precleaned and
152 treated containers offers the laboratory container blanks for use as part of quality control. Extra
153 containers are also useful for taking additional samples as needed during field collection and to
154 replace broken or leaking containers.
155 10.2.4 Container Label and Sample Identification Code
156 Each sample can only be identified over the life of a study if a form of permanent identification
157 is provided with or affixed to the container or available in sample log. The most useful form of
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 10-5 DO NOT CITE OR QUOTE
-------
Field and Sampling Issues That Affect Laboratory Measurements
158 identification utilizes a unique identifier for each sample. Such unique identification codes
159 ensure the project's ability to track individual samples. The standard operating procedure (SOP)
160 that addresses sample identification should describe the method to be used to assure that samples
161 are properly identified and controlled in a consistent manner. Containers sometimes may be pre-
162 labeled with identification numbers already in place.
163 Any identification recorded on a container or a label affixed to the container should remain with
164 the container throughout sample processing and storage. The identification information should be
165 written with a permanent marker—especially if the labels are exposed to liquids. Information can
166 be recorded directly on the container or on plastic or paper tags securely fixed to the container.
167 However, tags are more likely to become separated from containers than are properly secured
168 labels.
169 Labels, tags, and bar codes should be rugged enough so no information is lost or compromised
170 during field work, sample transport, or laboratory processing. Transparent tape can be used to
171 cover the label once it is completed. The tape protects the label, adds moisture resistance,
172 prevents tampering with the sample information, and helps secure the label to the container.
173 The project manager needs to determine if a sample number scheme may introduce bias into the
174 analysis process. That is, the lab may be aware of trends or locations from the sample
175 identification and this could influence their judgment as to the anticipated result and thereby
176 introduce actions on the part of lab personnel that they would not otherwise take. The project
177 manager needs to determine the applicability of electronic field data recorders and the issue of
178 electronic signatures for the project.
179 A unique identifier can include a code for a site, the sample location at the site, and a series of
180 digits identifying the year and day of year (e.g., "1997-127" uses the Julian date, and "062296"
181 describes a month, day, and year). Alternatively, a series of digits can be assigned sequentially by
182 site, date, and laboratory destination. The use of compass headings and grid locations also
183 provides additional unique information (e.g., "NW fence, sampled at grid points: Al through
184 C25, 072196, soil"). With this approach, samples arriving at a laboratory are then unique in two
185 ways. First, each sample can be discriminated from materials collected at other sites. Second, if
186 repeat samples are made at a single site, then subsequent samples from the same location are
187 unique only by date. Labeling of samples sequentially might not be appropriate for all studies.
188 Bar coding may reduce transcription errors and should be evaluated for a specific project.
189 10.2.5 Field Data Documentation
190 All information pertinent to field sampling is documented in a log book or on a data form. The
191 log book should be bound and the pages numbered consecutively and forms should be page-
192 numbered and dated. Where the same information is requested routinely, preprinted log books or
193 data sheets will minimize the effort and will standardize the presentation of data. Even when
MARLAP JULY 2001
DO NOT CITE OR QUOTE 10-6 DRAFT FOR PUBLIC COMMENT
-------
Field and Sampling Issues That Affect Laboratory Measurements
194 standardized preprinted forms are used, all information recorded should be in indelible ink, with
195 all entry errors crossed out with a single line and initialed. The color of ink used should be
196 compatible with the need to copy that information. All entries should be dated and signed on the
197 date of entry. Initials should be legible and traceable, so that it is clear who made the entry.
198 Whenever appropriate, log or data form entries should contain—but are not limited to—the
199 following:
200 • Identification of Project Plan or Sampling Plan;
201 • Location of sampling (e.g., reference to grid location, maps, photographs, location in a
202 room);
203 • Date and time of sample collection;
204 • Sample medium (e.g., surface water, soil, sediment, sludge, etc.);
205 • Suspected radionuclide constituents;
206 • Sample-specific ID number;
207 • Sample volume, weight, depth;
208 • Sample type (e.g., grab, composite);
209 • Sample preparation used (e.g., removal of extraneous matter);
210 • Sample preservation used;
211 • Requested analyses to be performed (e.g., gross beta/gamma, gamma spectroscopy for a
212 specific radionuclide, radiochemical analysis);
213 • Sample destination including name and address of analytical laboratory;
214 • Names of field persons responsible for collecting sample;
215 • Physical and meteorological conditions at time of sample collection;
216 • Special handling or safety precautions;
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 10-7 DO NOT CITE OR QUOTE
-------
Field and Sampling Issues That Affect Laboratory Measurements
217 • Recommendations regarding time to date of analysis that reflect (1) the loss of radioactivity
218 due to natural decay, (2) the ingrowth and secular equilibrium of short-lived progeny, or (3)
219 the potential loss of radioactivity due to evaporation or volatility; and
220 • Signatures or initials of appropriate field personnel. When using initials, ensure that they can
221 be uniquely identified with an individual.
222 Labels affixed to individual sample containers should contain key information that is an abstract
223 of log book data sheets. When this is not practical, a copy of individual sample data sheets may
224 be included along with the appropriately ID-labeled sample.
225 10.2.6 Field Tracking, Custody, and Shipment Forms
226 A sample tracking procedure must be in place for all projects in order that the proper location and
227 identification of samples is maintained throughout the process from collection through handling,
228 preservation, storage, transfer to laboratory, and disposal. The term "tracking," when used here,
229 connotes a tracking and accountability process that meets generally acceptable laboratory
230 practices as described by accrediting bodies, but is less stringent than a formal chain-of-custody
231 process. Tracking also develops a record of all individuals responsible for the custody and
232 transfer of the samples. Chapter 4 {Project Plan Documents) discusses the process of tracking
233 and accountability. Also, Chapter 11 (Sample Receipt, Inspection, and Tracking) discusses the
234 laboratory process of tracking.
235 When transferring the possession of samples, the individuals relinquishing and the individuals
236 receiving the samples should sign, date, and note the time on the form. A standardized form
237 should be designed for recording tracking or formal chain-of-custody information related to
238 tracking sample possession. If samples are to be split and distributed to more than one analytical
239 laboratory, multiple forms will be needed to accompany sample sets. The sample collector is
240 responsible for initiating the sample tracking record. The following information is considered
241 minimal for sample tracking:
242 • Name of project;
243 • Sampler's signature;
244 • Sample ID;
245 • Sample location
246 • Date and time sampled;
247 • Sample type;
248 • Preservatives;
249 • Number of containers;
250 • Analysis required;
251 • Signatures of persons relinquishing, receiving, and transporting the samples;
252 • Signature for laboratory receipt;
MARLAP JULY 2001
DO NOT CITE OR QUOTE 10-8 DRAFT FOR PUBLIC COMMENT
-------
Field and Sampling Issues That Affect Laboratory Measurements
253 • Method of shipment or carrier and air bill when shipped or shipping manifest identification
254 upon receipt; and
255 • Comments regarding the integrity of shipping container and individual samples.
256 10.2.7 Chain of Custody
257 The legal portion of the tracking and handling process that ensures legal defensibility from
258 sample collection to data reporting has become relatively standardized and is referred to as the
259 chain-of-custody (COC) process (APHA, 1996). Guidance is provided in "Standard Practice for
260 Sampling Chain-of-Custody Procedures" (ASTM D4840) and NIOSH (1983). The level of
261 security required to maintain an adequate chain of custody is that necessary to establish a
262 "reasonable probability" that the sample has not been tampered with. For court proceedings, the
263 requirements are established in law. COC procedures are important in demonstrating sample
264 control when litigation is involved. In many cases, Federal, State or local agencies may require
265 that COC be maintained for specific projects. COC is usually not required for samples that are
266 generated and immediately tested within a facility or continuous (rather than discrete or
267 integrated) samples that are subject to real- or near-real-time analysis (e.g., continuous
268 screening).
269 When COC is required, the custody information is recorded on a COC form. Chain-of-custody
270 documents vary by organization. Communication between field and laboratory personnel is
271 critical to the successful use of COC. Any error made on a custody form is crossed out with a
272 single line and dated and initialed. Use of correction ink or obliteration of data is not acceptable.
273 Inform the laboratory when COC is required before the samples are received (see Section 11.2
274 for further information). The COC documents are signed by personnel who collect the samples.
275 A chain-of-custody record accompanies the shipment and one or more copies are distributed to
276 the project coordinator or other office(s) where field and laboratory records are maintained. An
277 example of a COC form is shown in Figure 10.1. Additional information and examples of
278 custody forms are illustrated by EPA (1987) and EPA (1994).
279 10.2.8 Field Quality Control
280
281 A project plan should have been developed to ensure that all data are accurate and that decisions
282 based on these data are technically sound and defensible. The implementation of a project plan
283 requires quality control (QC) procedures. QC procedures, therefore, represent specific tools for
284 measuring the degree to which quality assurance objectives are met. Field quality control
285 measures are comprehensively discussed in ASTM D5283.
286 While some types of quality control (QC) samples are used to assess analytical process, field
287 quality control samples are used to assess the actual sampling process. The type and frequency of
288 these field QC samples must be specified by the project planning process along with being
289 included in the project planning documents and identified in the sampling plan. Definitions for
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 10-9 DO NOT CITE OR QUOTE
-------
Field and Sampling Issues That Affect Laboratory Measurements
CHAIN-OF-CUSTODY RECORD
FIELD
IDENTIFI-
CATION
NUMBER
FIELD LOCATION
DATE
TIME
Relinquished by: (Signature)
Relinquished by: (Signature)
Relinquished by: (Signature)
Received by: (Signature)
Dispatched by: (Signature)
Date
SAMPLERS (Signature)
SAMPLE MATRIX
Water
Soil
Air
SEQ.
No.
No. of
Containers
Received by: (Signature)
Relinquished by: (Signature)
Received by: (Signature)
Received by Laboratory for field analysis:
(Signature)
Time
Received for Laboratory by:
Analysis
Required
Date/Time
/
Date/Time
/
Date/Time
/
Date/Time
/
Date/Time
/
Method of Shipment:
Distribution: Orig. - Accompany Shipment
1 Copy - Survey Coordinator Field Files
FIGURE 10.1—Example of chain-of-custody record.
290 certain types of field QC samples can be found in ASTM D5283 and MARS SIM (2000).
291 10.2.9 Decontamination of Field Equipment
292 Sampling SOPs must describe the recommended procedure for cleaning field equipment before
293 and during the sample collection process, as well as any pretreatment of sample containers. The
294 SOPs should include the cleaning materials and solvents used, the purity of rinsing solution or
295 water, the order of washing and rinsing, associated personnel safety precautions, and the disposal
296 of cleaning agents.
MARLAP
DO NOT CITE OR QUOTE
10-10
JULY 2001
DRAFT FOR PUBLIC COMMENT
-------
Field and Sampling Issues That Affect Laboratory Measurements
297 Detailed step-by-step procedures for the decontamination of field equipment used in the
298 sampling of low-activity soils, soil gas, sludges, surface water, and ground water are given in
299 ASTMD5608.
300 10.2.10 Packing and Shipping
301 The final responsibility of field sampling personnel is to properly prepare and package samples
302 for transport or shipment by a commercial carrier. All applicable State and Federal shipping
303 requirements, as discussed later in this section, must be followed. Samples transported over
304 shorter distances by the sampling or testing agency by way of automobile, van, or truck will
305 require less stringent packing requirements. In most instances, placing sealed sample containers
306 within cardboard boxes (or similar containers) in which individual samples are sufficiently
307 cushioned to guard against bumping, rolling, or dropping, is adequate.
308 When samples must be shipped by way of a commercial carrier or the U.S. Postal Service,
309 containers must be designed to protect samples against crushing forces, impacts, and severe
310 temperature fluctuations. Within each shipping container, the cushioning material (sawdust,
311 rubber, polystyrene, urethane foam, or material with similar resiliency) should encase each
312 sample completely. The cushioning between the samples and walls of the shipping containers
313 should have a minimum thickness of one inch. A minimum thickness of two inches should be
314 provided on the container floor.
315 Consideration must also be given to protect samples against potentially adverse impacts of
316 temperature fluctuations. When appropriate, sample protection against freezing, thawing,
317 sublimation, evaporation, or extreme temperature variation may require that the entire interior
318 surface of the shipping container be lined with an adequate layer of insulation. In many instances,
319 the insulating material may also serve as the cushioning material.
320 When metal containers are used, the requirements for container security, cushioning, and insula-
321 tion apply equally. For smaller volume and low-weight samples, properly lined containers
322 constructed with laminated fiberboard, plastic, or reinforced cardboard outer walls also may be
323 used.
324 When samples are shipped as liquids in glass or other breakable sample containers, additional
325 packaging precautions may have to be taken. Additional protection is obtained when sample
326 containers are shipped in nested containers, in which several smaller containers (i.e., inner
327 containers) are packed inside a second larger container (i.e., the outer pack or overpack). To
328 contain any spills of sample material within the shipping container, it is advisable either to wrap
329 individual samples or to line the shipping container with absorbent material, such as asbestos-
330 free vermiculite or pearlite.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 10-11 DO NOT CITE OR QUOTE
-------
Field and Sampling Issues That Affect Laboratory Measurements
331 For proper packaging of liquid samples, additional guidance has been given by EPA (1987) and
332 includes the following:
333 • All sample bottles are taped closed;
334 • Each sample bottle is placed in a plastic bag and the bag is sealed;
335 • Each sample bottle may be placed in a separate metal can filled with vermiculite or other
336 packing material, then the lid may be fixed to the can with tape;
337 • The cans are placed upright in a cooler that has its drain plug taped closed, inside and out,
338 and lined with a plastic bag; and
339 • The cooler is filled with packing material—"bubble wrap" or cardboard separators may be
340 used—and closed with sealing tape.
341 Field screening measurements are made for compliance with Department of Transportation
342 regulations, 49 CFR Parts 170 through 189, as well as compliance with the laboratory's U.S.
343 NRC (10 CFR Part 71) and Agreement State license. International requirements may also apply.
344 See International Air Transport Association (IATA) Dangerous Goods Regulations for additional
345 guidance. These regulations not only set contamination and dose limits for shipping containers,
346 but also describe the types of containers and associated materials that are to be used based on the
347 total activity and quantity of materials shipped. When the samples are screened in the field with
348 survey instrumentation, the results should be provided to the laboratory. This information should
349 also state the distance used from the probe to the packing container wall. Measurements normally
350 are made in contact or at one meter. The readings in contact are most appropriate for laboratory
351 use. The screening measurements in the field are mainly for compliance with transportation
352 requirements and are usually in units of exposure. Laboratory license requirements are usually by
353 isotope and activity. Project planning and communication are essential to ensure that a specific
354 set of samples can be transported, received, and analyzed safely while complying with applicable
355 rules and regulations.
356
357 The external surface of each shipping container must be labeled clearly, contain information
358 regarding the sender and receiver, and should include the respective name and telephone number
359 of a contact. When required, proper handling instructions and precautions should be clearly
360 marked on shipping containers. Copies of instructions, shipping manifest or container inventory,
361 chain of custody, and any other paperwork that is enclosed within a shipping container should be
362 safeguarded by placing documents within a sealed protected envelope.
363 10.2.11 Worker Health and Safety Plan
364 In some cases, field samples will be collected where hazardous agents or site conditions might
365 pose health and safety considerations for field personnel. These can include chemical, biological,
366 and radiological agents, as well as common industrial hazards associated with machinery, noise
367 levels, and heat stress. The health and safety plan established in the planning process should be
MARLAP JULY 2001
DO NOT CITE OR QUOTE 10-12 DRAFT FOR PUBLIC COMMENT
-------
Field and Sampling Issues That Affect Laboratory Measurements
368 followed. For the Department of Defense (DOD), these plans may include imminent threats to
369 life, such as unexploded ordnance, land mines, hostile forces, chemical agents, etc.
370 10.2.11.1 Physical Hazards
371 MECHANICAL EQUIPMENT
372 Personnel working with hand-held tools (e.g., sledge hammers used for near-surface coring) or
373 power tools and equipment are subject to a variety of hazards. For example, personnel drilling
374 monitoring wells are exposed to a variety of potential mechanical hazards, including moving
375 machinery, high-pressure lines (e.g., hydraulic lines), falling objects, drilling through under-
376 ground utilities, flying machinery parts, and unsafe walking and working surfaces. The
377 consequences of accidents involving these physical hazards can range from minor to fatal injury.
378 At a minimum, workers should be required to wear protective clothing, which includes hard hats,
379 gloves, safety glasses, coveralls (as an option) and steel-toed safety shoes. Workers required to
380 climb (e.g., ladders, drilling masts) must be required to wear harnesses and lanyards and be tied
381 off throughout the process.
382 For sampling operations that require drilling, open boreholes and wells must be covered or
383 secured when unattended, including during crew breaks.
384 ELECTRICAL HAZARDS
385 Electric power often is supplied by gasoline or diesel engine generators. Working conditions may
386 be wet, and electrical shock with possibly fatal consequences may occur. In addition, it is
387 possible that drilling operations may encounter overhead or buried electrical utilities, potentially
388 resulting in exposure to very high voltages, which could be fatal or initiate fires.
389 All electrical systems used during field operations should be checked for proper grounding
390 during the initial installation. Temporary electrical power provided to the drill site shall be
391 protected by ground fault circuit interrupters.
392 NOISE HAZARDS
393 Power equipment is capable of producing sound levels in excess of 85dB(A), the eight-hour
394 threshold limit value recommended by the American Conference of Governmental Industrial
395 Hygienists (ACGIH). Exposure to noise levels in excess of 85dB(A) for long periods of time can
396 cause irreversible hearing loss. If noise levels
397 exceed 85dB(A), a controlled area must be
398 maintained at this distance with a posting at
399 each entrance to the controlled area to read:
CAUTION
NOISE HAZARD
Hearing Protection Required Beyond This Point
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 10-13 DO NOT CITE OR QUOTE
-------
Field and Sampling Issues That Affect Laboratory Measurements
400 HEAT STRESS
401 The use of protective clothing during summer months significantly increases the potential for
402 personnel to experience heat stress. Adverse effects from heat stress include heat cramps,
403 dehydration, skin rash, heat edema, heat exhaustion, heat stroke or death. When heat stress
404 conditions exist, the following ought to be available:
405 • A cool and shaded rest area;
406 • Regular rest breaks;
407 • An adequate supply of drinking water; and
408 • Cotton coveralls rather than impermeable Tyvek coveralls.
409 CHEMICAL AND RADIOLOGICAL HAZARDS
410 The health and safety plan should contain information about a site's potential radionuclides and
411 hazards that might be encountered during implementation of field sampling and survey
412 procedures. All field personnel should read the health and safety plan and acknowledge an
413 understanding of the radiological hazards associated with a site. Site specific training must be
414 provided that addresses the chemical and radiological hazards likely to be associated with a site.
415 Field procedures should include either information relating to these hazards or should reference
416 appropriate sections of the Health and Safety Plan. References related to the use of protective
417 clothing are given in EPA (1987), DOE (1987, Appendix J), and in 29 CFR 1910, Subpart I.
418 When procuring environmental solid and liquid samples, unusual characteristics such as color,
419 suspended material, or number of phases and unusual odors should be noted and a description
420 should be provided to the on-site safety officer as well as the analytical laboratory. Additional
421 information concerning field methods for rapid screening of hazardous materials is presented in
422 EPA (1987). This source primarily addresses the appearance and presence of organic compounds
423 that might be present on occasions when one is collecting materials to detect radioactivity.
424 Checking samples for chemical or radiological hazards can be as simple as visual inspection or
425 using a hand-held radiation meter to detect radiation levels. Adjustments to laboratory
426 procedures, particularly those involving sample handling and preparation, can only be made
427 when pertinent field information is recorded and relayed to the project planner and to the
428 laboratory. In some cases, a laboratory might not have clearance to receive certain types of
429 samples (such as explosives or chemical agents) because of their content, and it will be necessary
430 to divert these samples to an alternate laboratory. It might be necessary to reduce the volume
431 sampled in order to meet shipping regulations if high concentrations of radioactivity are present
432 in the samples. In some cases, the activity of one radionuclide might be much higher than others
433 in the same sample. Adjustments made on the basis of the radionuclide of higher activity might
434 result in collection of too little of another radionuclide to provide adequate detection and thus
435 prevent identification of these radionuclides because of their relatively low minimum detectable
MARLAP JULY 2001
DO NOT CITE OR QUOTE 10-14 DRAFT FOR PUBLIC COMMENT
-------
Field and Sampling Issues That Affect Laboratory Measurements
436 concentrations. These situations should be considered during planning and documented in the
437 appropriate sampling plan document.
438 10.2.11.2 Biohazards
439 Precautions should be taken when handling unknown samples in the field. Some examples are
440 wearing gloves, coveralls or disposable garments, plastic booties, dust masks or other respiratory
441 protection. Some biohazards may be snakes, ticks, spiders, and rodents (Hanta virus). Prevention
442 of potential exposure is the goal of a safety program. The type of protective equipment in the
443 field should be discussed in the planning process and specified in the appropriate plan document.
444 Since there are many specifics that are site dependent, it is difficult to create a comprehensive
445 list. But the information is discussed to provide an awareness and starting point for additional
446 discussion.
447 PERSONNEL TRAINING AND QUALIFICATION
448 All field operations that could lead to injury for sample collectors should be performed by
449 personnel trained to documented procedures. When sampling is conducted in radiologically
450 controlled areas (RCAs) as defined in regulatory standards (i.e., 10 CFR 20, 10 CFR 835).
451 Formal training and qualification of field personnel may be required.
452 Training may require both classroom and practical applications in order to familiarize personnel
453 with the basic theory of radiation and radioactivity and the basic rules for minimizing external
454 exposures through time, distance, shielding, and avoidance of internal exposure (by complying
455 with rules regarding smoking, drinking, eating, and washing of hands). Other topics to cover
456 include common routes of exposure (e.g., inhalation, ingestion, skin contact); proper use of
457 equipment and the safe handling of samples; proper use of safety equipment such as protective
458 clothing, respirators, portable shielding, etc.
459 Guidance for the training and qualification of workers handling radioactive material has been
460 issued by the Nuclear Regulatory Commission (see appropriate NRC NUREGs and Regulatory
461 Guides on training of radiation workers), Department of Energy (1994), and the Institute of
462 Nuclear Power Operations (INPO 88-010). These and other documents should be consulted for
463 the purpose of training and qualifying field personnel.
464 PERSONNEL MONITORING AND BIOASSAY SAMPLING
465 When conditions dictate the need for personnel monitoring, various methods are commonly
466 employed to assess external and internal exposure that might have resulted from the inhalation or
467 ingestion of a radionuclide.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 10-15 DO NOT CITE OR QUOTE
-------
Field and Sampling Issues That Affect Laboratory Measurements
468 To monitor for external exposures to the whole body or extremities, thermoluminescent
469 dosimeters (TLDs) or film badges may be used to document a worker's exposure. For internal
470 exposures, assessment of dose may be based on: (1) air monitoring of the work area or the
471 worker's breathing zone; (2) in vivo bioassay (whole-body counting); or (3) in vitro bioassays
472 that normally involve urinalysis but may also include fecal analysis and nasal smears. For in vitro
473 bioassays (i.e., urine or fecal), the standard method involves a 24-hour sample collection in a
474 sealable container. Samples may be kept under refrigeration until laboratory analysis can be
475 performed to retard bacterial action. (Bioassay sample collection is normally not performed in the
476 "field.")
477 The following guidance documents may be used for personnel monitoring and the collection and
478 preservation of bioassay samples:
479 • ANSI/ANSHPSN13.30 (1996), Performance Criteria for Radiobioassay;
480 • ANSI/ANS HPS N13.14 (1994), Internal Dosimetry Programs for Tritium Exposure—
481 Minimum Requirements;
482 • ANSI/ANS HPS 13.22 (1995), Bioassay Programs for Uranium;
483 • ANSI/ANS HPS 13.42 (1997), Internal Dosimetry for Mixed Fission Activation Products;
484 • DOE Implementation Guide, Internal Dosimetry Program, G-10 CFR 835/C1—Rev. 1 Dec.
485 1994a;
486 • DOE Implementation Guide, External Dosimetry Program, G-10 CFR 835/C2—Rev. 1 Dec.
487 1994b;
488 • DOE Implementation Guide, Workplace Air Monitoring, G-10 CFR 835/E2-Rev. 1 Dec.
489 1994c;
490 • DOE Radiological Control Manual, DOE/EH-0256T, Rev. 1, 1994d;
491 • NRC Regulatory Guide 8.9, Acceptable Concepts, Models, Equations, and Assumptions for a
492 Bioassay Program;
493 • NRC Regulatory Guide 8.11, Applications of Bioassay for Uranium;
494 • NRC Regulatory Guide 8.20, Applications of Bioassay for 125I and 131I;
495 • NRC Regulatory Guide 8.22, Bioassays at Uranium Mills;
496 • NRC Regulatory Guide 8.26, Applications of Bioassay for Fission and Activation Products;
497 • NRC Regulatory Guide 8.32, Criteria for Establishing a Tritium Bioassay Program;
498 • NCRP (1987), Use of Bioassay Procedures for Assessment of Internal Radionuclides
499 Deposition; and
500 • INPO (1988), Guidelines for Radiological Protection at Nuclear Power Stations.
501 Part II: Matrix-Specific Issues That Impact Field Sample Collection,
502 Processing, and Preservation
503 Field processing should be planned in advance so that all necessary materials are available during
504 field work. Preparing checklists of processing equipment, instruments, and expendable
505 materials—as exemplified in part by lists accompanying sampling procedures described by EPA
MARLAP JULY 2001
DO NOT CITE OR QUOTE 10-16 DRAFT FOR PUBLIC COMMENT
-------
Field and Sampling Issues That Affect Laboratory Measurements
506 1994—helps this planning effort and serves to organize field methods. Field personnel who
507 communicate problems should prevent loss of time, effort, and improper sample collection, as
508 well as documents exactly what equipment, instruments, etc. were used.
509 The initial steps taken in the field frequently are critical to the laboratory analysis performed
510 hours, days, or even weeks after a sample is obtained. Various sample preparation steps may be
511 required before samples are packaged and shipped for laboratory analysis. The need for sample
512 processing and preservation is commonly determined by the sample matrix, the data quality
513 objectives of the analysis, the nature of the radionuclide, and the analytical method.
514 The goal of sample preservation is to maintain the integrity of the sample between the time the
515 sample is collected and the time it is analyzed, thus assuring that the analysis is performed on a
516 sample representative of the media collected. In general, the aim of sample preservation is to
517 limit biological and chemical actions that might alter the concentration or physical state of the
518 radionuclide constituents or analytes. For example, cations at very low concentrations can be lost
519 from solution (e.g., cesium can exchange with potassium in the glass container, and radio-
520 nuclides can be absorbed by algae or slime growths in sample lines or containers that remain in
521 the field for extended periods). Requirements for sample preservation should be determined
522 during project planning when analytical protocols are selected. Sample preservation in the field
523 typically follows or accompanies processing activities.
524 This section provides matrix-specific guidance that focuses on the preparation and processing of
525 field samples. In order to assist project planners in developing a sampling plan, a limited
526 discussion is also provided that describes matrix-specific methods commonly employed for the
527 collection of field samples. Guidance is presented for only the most common materials or
528 environmental media, which are generically classified as liquids, solids, and air. In some
529 instances, a solid material to be analyzed involves particulate matter suspended in a liquid or air
530 that is commonly obtained by filtration. Because filter media can affect analytical protocols, a
531 separate discussion is provided that addresses sample materials contained on filter materials,
532 including surface contamination associated with wipe samples.
533 10.3 Liquid Samples
534 Liquid samples are typically classified as aqueous, non-aqueous, and as mixtures. Aqueous
535 samples requiring analysis are likely to represent surface water, ground water, drinking water,
536 precipitation, tanks and lagoons, and runoff. Non-aqueous liquids may include a variety of
537 solvents, oils and other organic liquids. Mixtures of liquids represent a combination of aqueous
538 and non-aqueous liquids or a solid suspended in either aqueous and non-aqueous liquids.
539 Standardized water sampling procedures are described in numerous documents (APHA, 1996;
540 EPA, 1985; EPA, 1987; DOE, 1997; ASTM D3370). Important decisions include the choice of
541 instrument or tool used to obtain the sample, the sample container material, the need for sample
542 filtration, and the use of sample preservatives.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 10-17 DO NOT CITE OR QUOTE
-------
Field and Sampling Issues That Affect Laboratory Measurements
543 10.3.1 Liquid Sampling Methods
544 The effect of the sample collection process on the sample integrity needs to be understood and
545 managed. Two examples are dissolved gases and cross contamination. It may be necessary to
546 minimize dissolved oxygen and carbon dioxide which may cause some dissolved metals to
547 undergo reaction or precipitation.
548 Sampling is discussed in Navy Environmental Compliance Sampling and Field Testing
549 Procedures Manual, NAVSEA T0300-AZ-PRO-010. USAGE discusses sampling in Technical
550 Project Planning Guidance for Hazardous, Toxic and Radioactive Waste (HTRW) Data Quality
551 Design, Engineer ManuaJEM-200-1-2, Appendix H, Sampling Methods, July 1995. This
552 reference has been superseded but the revision does not include sampling. The sampling
553 references listed in Appendix H are:
554 • U.S. Environmental Protection Agency (EPA). 1984. Characterization of Hazardous Waste
555 Sites—A Method Manual, Vol. U, Available Sampling Methods, Second Edition, EPA 600-
556 4-84-076.
557 • U.S. Environmental Protection Agency (EPA). 1982. Handbook for Sampling and Sample
558 Preservation of Water and Wastewater, EPA 600-4-82-029.
559 • U.S. Environmental Protection Agency (EPA). 1986. Compendium of Methods for
560 Determination of Superfund Field Operation Methods, EPA 600-4-87/006.
561 • U.S. Environmental Protection Agency (EPA). 1987. A Compendium of Methods for
562 Determination of Superfund Field Operation Methods, EPA 540-P-87-001a, OSWER
563 Directive 9355.0-14.
564 • U.S. Department of the Interior. 1980. National Handbook of Recommended Methods for
565 Water for Water-Data Acquisition, Volume I and II.
566 10.3.2 Liquid Sample Preparation: Filtration
567 Filtration of a water sample may be a key analytical planning issue and is discussed in Chapter 3,
568 Section 3.3.2. A decision needs to be made during project planning whether or not to filter the
569 sample in the field. Filtration of water or other liquids may be required to determine contaminant
570 concentrations in solubilized form, suspended particulates, or sediment. The method of filtration
571 will depend on the required sample volume, the amount and size of suspended parti culates, and
572 the availability of portable equipment and resources (e.g., electricity).
573
574 The potential need to filter a water sample principally depends on the source of water and the
575 objectives of the project investigation. If, for example, the source of water is drinking water "at-
MARLAP JULY 2001
DO NOT CITE OR QUOTE 10-18 DRAFT FOR PUBLIC COMMENT
-------
Field and Sampling Issues That Affect Laboratory Measurements
576 the-spigot" and the intent is to assess human internal exposure from ingestion, unfiltered tap
577 water samples are likely to be required. Conversely, filtration may be required for water taken
578 from an unlined field monitor well that is likely to contain significant amounts of particulate
579 matter. These solids are of little relevance but may interfere with radioanalytical protocols (e.g.,
580 sample absorption may occur during gross alpha or beta counting where the analytical procedure
581 involve s the simple evaporation of a water aliquant on a planchet).
582 For remote sampling sites, sample processing may be restricted to gravity filtration that requires a
583 minimum of equipment and resources. Drawing samples through filters by pressure or suction
584 that is created by syringe, vacuum pump, or aspiration are alternative options. If filter papers or
585 membranes capture materials that will be retained for analysis, they should be handled with clean
586 rubber or plastic gloves, forceps, or other instruments to prevent sample contamination.
587 Each Federal Agency may have unique guidance to determine the need and process for filtering
588 samples. One performance-based example is that of EPA, discussed in the next section. This
589 guidance applies to either the field or laboratory filtration.
590 10.3.2.1 EPA Guidance for Samples/Filtration
591 The Special Topics Subcommittee of EPA's Science Advisory Board's Environmental
592 Engineering Committee met to examine the question of whether or not to filter ground-water
593 samples when analyzing for metals in the context of a review of the Office of Emergency and
594 Remedial Response's (OERR) proposed guidance on field filtration of ground-water samples
595 taken from monitoring wells for metals analysis as part of a Superfund site assessment (EPA,
596 1997). The key findings of the Subcommittee were:
597 • Several factors could introduce errors in the sampling and analysis of ground water for metals
598 or metallic radionuclides. Well construction, development, sampling, and field filtering are
599 among the steps that could influence the metals measured in the ground-water samples. Field
600 filtering is often a smaller source of variability and bias compared to these other factors.
601 Therefore, the Agency should emphasize in its guidance the importance of proper well
602 construction, development, purging, and water pumping rates so that the field filtering
603 decisions can also be made accurately.
604 • Under ideal conditions, field-filtered ground-water samples should yield identical metals
605 concentrations when compared to unfiltered samples. However, under non-ideal conditions,
606 the sampling process may introduce geological materials into the sample and would require
607 field filtration. Under such conditions, filtering to remove the geological artifacts has the
608 potential of removing colloids (small particles that may have migrated as suspended materials
609 that are mobile in the aquifer). Available scientific evidence indicates that when wells have
610 been properly constructed, developed, and purged, and when the sample has been collected
611 without stirring or agitating the aquifer materials (turbidity less than 5 nephelometric turbidy
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 10-19 DO NOT CITE OR QUOTE
-------
Field and Sampling Issues That Affect Laboratory Measurements
612 units, NTU), then field filtering should not be necessary. For Superfund site assessments, the
613 low-flow sampling technique without filtration is the preferred sampling approach for
614 subsequent metal analysis when well construction, well maintenance, and hydrogeological
615 conditions such as flow rate allow. Under such conditions, the collected samples should be
616 representative of the dissolved and particulate metals that are mobile in ground-water
617 systems. The Agency's proposal to rely on low flow sampling and unfiltered samples is a
618 conservative approach that favors false positives over false negatives.
619 • When the turbidity of the sample is high, the situation is different. In-line filtering provides
620 samples that retain their chemical integrity. Therefore, field filtering of properly collected
621 ground-water samples should be done when turbidity in the samples is higher than 5 NTU,
622 even after slow pumping has been utilized to obtain the sample.
623 They acknowledged, however, that differences in the way wells are installed, their packing
624 materials, and the techniques used to collect ground-water samples can lead to variability in
625 analytical results between wells and between individual samples. Filtering a sample can be seen
626 as a way to remove suspended particles and some colloids that contain metals that would not
627 normally be in the ground water if the material were not disturbed during sampling. Here a
628 colloid is defined as a particle that ranges in size from 0.003 to 10 jim (Puls et al., 1990) or
629 particles having diameters of less than lOjim (Puls and Powell, 1992). The literature indicates
630 that colloids as large as 2 jim can be mobile in porous media (Puls and Powell., 1992), and that
631 colloid concentration can be as high as 1,000 times higher in fractured granitic systems
632 (McCarthy and Deguelde, 1993). Saar (1997) presents a review of the industry practice of
633 filtration of ground-water samples. For some sites with low hydraulic conductivity the presence
634 of an excess of colloids presents numerous monitoring challenges and field filtration might be
635 necessary.
636 The desire to disturb the aquifer as little as possible has led to the use of low-flow sampling of
637 wells—low-flow purging and sampling occurs typically at 0.1 to 0.3 L/min (Saar, 1997). The
638 low-flow technique maximizes representativeness by (EPA, 1997):
639 • Minimizing disturbances that might suspend geochemical materials that are not usually
640 mobile;
641 • Minimizing disturbances that might expose new reactive sites that could result in leaching or
642 adsorption of inorganic constituents of ground water;
643 • Minimizing exposure of the ground water to the atmosphere or negative pressures, ensuring
644 that the rate of purging and sampling does not remove ground water from the well at a rate
645 much greater than the natural ground-water influx; and
MARLAP JULY 2001
DO NOT CITE OR QUOTE 10-20 DRAFT FOR PUBLIC COMMENT
-------
Field and Sampling Issues That Affect Laboratory Measurements
646 • Monitoring indicator parameters to identify when stagnant waters have been purged and the
647 optimum time for sample collection.
648 In summary, based on the ability of the low-flow sampling technique to collect representative
649 samples, EPA suggests that filtering of ground-water samples prior to metals analysis is usually
650 not required (EPA, 1997).
651 10.3.2.2 Filters
652 When filtration is required, it should be done in the field or as soon as practicable. The
653 advantages of filtering in the field are that acid preservatives can be added shortly thereafter
654 which minimizes both the adsorption of soluble contaminants and avoids the dissolution of
655 particulate matter, volume reduction, and waste reduction. Unless specific requirements dictate
656 otherwise, the removal of suspended particles is commonly achieved by filtration that removes
657 particles larger than 0.45 |im (ASTM D3977).
658 In other instances, the investigative objectives may not be restricted to water-solubilized
659 contaminants but include analysis of contaminated suspended particulate matter. To detect the
660 presence of radionuclides that are highly insoluble, such as isotopes of uranium, thorium, and
661 plutonium, analysis of particulate matter is considered more sensitive than the filtered water
662 (EPA, 1994).
663 The fact that small particles pass through membrane filters has been recognized for some time
664 (Kennedy et al., 1974). The arbitrary cutoff of 0.45 jim between dissolved and suspended matter
665 has gained such wide use that it is the filter size that is commonly recommended by laboratory
666 protocols. Filtering through a 0.45 jim filter may take considerable time and may require suction
667 or pressure to accomplish in a reasonable time.
668
669 It should be noted, however, that manufacturers of filters usually specify only what will not pass
670 through the filter; they make no claims concerning what actually does pass through the filter.
671 Laxen and Chandler (1982) present a comprehensive discussion of some effects of different filter
672 types. They refer to thin (5 to 10 jim) polycarbonate filters as screen types, and thick (100 to
673 150 |im) cellulose nitrate and acetate filters as depth types. The polycarbonate-screen type clogs
674 much more rapidly. Once the filtration rate drops, particles that would normally pass through the
675 filter are trapped in the material already retained. Hence, the use of so-called polycarbonate-
676 screen filters, because of their increased propensity to clog, is generally not recommended.
677 In addition to the difficulty of contending with clogging, Silva and Yee (1982) report adsorption
678 of dissolved radionuclides on membrane filters. Although these drawbacks cannot be completely
679 overcome, they are still less than the potential difficulties that arise from not filtering.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 10-21 DO NOT CITE OR QUOTE
-------
Field and Sampling Issues That Affect Laboratory Measurements
680 Finally, good laboratory practices must be used for field sampling. The most likely sources of
681 contamination for the filters are improperly cleaned tubing and filter holders and handling the
682 filters with contaminated fingers. Tubing and holders should be thoroughly cleaned and rinsed
683 between samples and the entire system should be rinsed several times with the water to be
684 sampled. Filters should be handled with clean rubber gloves.
685 10.3.3 Field Preservation of Liquid Samples
686 Sample degradation may occur between the time of collection and analysis due to microbial
687 contaminants or chemical interactions. Although sample degradation cannot destroy or alter the
688 radiological properties of a contaminant, it can alter the radionuclide's chemical properties and
689 its potential distribution within a sample. For example, microbial processes are known to affect
690 both the chemical state and the distribution of radioelements due to oxidation-reduction
691 reactions, complexation and solubilization by metabolic compounds, bioaccumulation,
692 biomylation, and production of gaseous substances such as CO2, H2, CH4, and H2S (Francis,
693 1985; Pignolet et al., 1989).
694 10.3.3.1 Sample Acidification
695 Acidification is the method of choice for preserving most types of water samples. The principal
696 benefit of acidification is that it keeps many radionuclides in solution and minimizes their
697 potential for removal by chemical and physical adsorption or by ion exchange. The mode by
698 which a radionuclide is potentially removed from solution is strongly affected by the radionuclide
699 and the container material. For example, studies conducted by Bernabee et al. (1980) and Milkey
700 (1954) demonstrated that the removal of metal ions from solution is dominated by physical (i.e.,
701 van der waals) adsorption. Their conclusion is based on: (1) their observation that the loss of
702 uranium, lead, and thorium ions from solution was significantly greater for containers made of
703 polyethylene when compared to borosilicate glass; and (2) the fact that while adsorption by glass
704 may potentially involve all three adsorption processes; with polyethylene plastic, there are no
705 valence-type attractive forces or ions to exchange and only physical van der waals adsorption is
706 possible.
707 Similar observations were reported by: (1) Dyck (1968), who compared long-term adsorption of
708 silver ions by molded plastic to glass containers; (2) Jackson (1962), who showed that
709 polyethylene containers absorbed about five times as much 90Sr as glass containers at pH of about
710 seven; and (3) Martin and Hylko (1987a; 1987b), who reported that greater than 50 percent of
711 "Tc was adsorbed by polyethylene containers from non-acidified samples.
712 For sample acidification, either nitric or hydrochloric acid is commonly added until a pH of less
713 than two. Table 7010:1 in Standard Methods for the Examination of Water and Wastewater
714 (APHA, 1995) and Method 900.0 in Prescribed Procedures for Measurement of Radioactivity in
MARLAP JULY 2001
DO NOT CITE OR QUOTE 10-22 DRAFT FOR PUBLIC COMMENT
-------
Field and Sampling Issues That Affect Laboratory Measurements
715 Drinking M/ater(EPA, 1980) provide additional guidance. Guidance for sample preservation by
716 acidification has been issued by Federal Agencies and others as summarized below.
717 In instances of very low activity samples where container adsorption poses a significant concern,
718 but where acidification of the sample interferes with the radioanalytical method, the choice of
719 sample container may be limited to glass or require alternative methods. For example, the use of
720 acids as a preservative is not recommended for the analysis of tritium (3H), carbon-14 (14C), or
721 radon in water, and precautions must be taken for the following reasons:
722 • For radon, sample preservation offers no benefit and is therefore not required for analytical
723 accuracy.
724 • The addition of acid to a sample containing 14C may result in the production of 14CO 2 and the
725 loss of radioactivity from the sample.
726 • The adverse impact of acid on tritiated water is due to the fact that water dissociates and
727 recombines continuously (i.e., H2O _ FT + OH" or HO _ T+ + OH"). The tritium ion that was
728 part of the water molecule may, therefore, be exchanged for the hydrogen ion from the acid.
729 The impact of this exchange is realized as a result of distillation, which is a common method
730 for purifying water in preparation for liquid scintillation counting. When the sample is heated
731 and the distillate is collected on a cold finger, distilled tritiated water, in the presence of acid,
732 would have a reduced specific activity over the original sample.
733 Although acidification has been shown to effectively reduce the adsorption of technetium by
734 polyethylene, technetium in the TcO 4"4 state has been observed to volatilize in strong acid
735 solutions during evaporation while preparing water samples for gross beta analysis (NAS, 1960).
736 To hasten evaporation, the planchet is commonly flamed. This dilemma can be resolved by either
737 precoating planchets with a film of detergent prior to the addition of the acidified water sample
738 or by passive evaporation of the acidified water sample that avoids the higher temperature
739 associated with flaming (Blanchard et al., 1993).
740 10.3.3.2 Non-Acid Preservation Techniques
741 If a sample contains significant organics, or if contaminants under investigation react with acids
742 that interfere with the radioanalytical methods, other methods of sample preparation should be
743 considered.
744 REFRIGERATION AND FREEZING
745 The effect of refrigeration or freezing temperatures to arrest microbial activity is a fundamental
746 concept. Temperatures near the freezing mark or below not only retard or block bacterial growth
747 but arrest essentially all other metabolic activity. It should, however, be noted that most bacteria
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 10-23 DO NOT CITE OR QUOTE
-------
Field and Sampling Issues That Affect Laboratory Measurements
748 can survive even in extreme temperatures. (Indeed, if a suspension of bacterial cells is frozen
749 rapidly with no appreciable formation of ice crystals, it can be kept at temperatures as low as
750 -194° C for indefinite periods of time with little loss of viability.)
751 The choice between refrigeration and freezing is dictated by the potential impacts of ice
752 formation on sample constituents. Besides physical changes of organic constituents, the initial
753 formation of ice crystals and the exclusion of any solutes may concentrate the solutes to the point
754 of precipitation. Quick freezing methods that minimize ice crystal formation are beneficial for
755 preserving some organic constituents. Quick freezing is commonly done by packing sealed
756 samples in liquid nitrogen or dry ice. Care must be taken, however, to avoid container breakage
757 due to sample volume expansion. An air space of a least 10 percent and a container made of
758 plastic provide reasonable assurance for container integrity.
759 When refrigeration is employed, attempts should be made to avoid temperatures that could result
760 in slow freezing and the formation of ice crystals. Optimum refrigeration temperatures for sample
761 preservation at 4 ± 2° C can be achieved by packing samples in ice or freeze packs within a
762 thermally insulated leak-proof container (ASTM D3856; ASTM D3370).
763 PAPER PULP
764 Adsorption and loss of radionuclides over time to the container wall can be avoided with the
765 addition of paper pulp. Due to its adsorptive property and large surface, paper pulp has been
766 shown to remove more than 95 percent of radionuclides from solution (Bernabee et al., 1980).
767 About two grams of finely ground paper pulp are added per liter of acidified sample at time of
768 collection. The pH should be adjusted to one or less and vigorously shaken. The sample may be
769 stored in this condition for an extended period of time. To prepare for analysis, the pulp is
770 removed from solution by filtration and subjected to wet ashing using strong acids (Chapter 12).
771 This ashed solution is commonly added to the original filtrate to make a reconstituted sample
772 solution.
773
774 The use of paper pulp and the need for wet ashing, however, pose problems for certain
775 radioanalytical laboratory protocols and must therefore be thoroughly evaluated.
776 SULFITE
777 To prevent the loss of radioiodine from solution, sodium bisulfite (NaHSO3) or sodium
778 metabisulfite (Na2S2O5) may be used. These compounds act as strong reducing agents and
779 prevent the volatilization of iodine. If acid is also employed to preserve samples for analysis of
780 other radionuclides, it is important to note that acid will negate the reductant's effectiveness in
781 behalf of iodine. For this reason, samples collected for iodine analyses typically are collected and
782 preserved in a separate container. It should also be noted that the reducing environment produced
783 by the sulfite preservative may render iron, uranium, and other easily reduced elements or their
MARLAP JULY 2001
DO NOT CITE OR QUOTE 10-24 DRAFT FOR PUBLIC COMMENT
-------
Field and Sampling Issues That Affect Laboratory Measurements
784 compounds to an insoluble state. The loss of reduced insoluble radionuclides from solution will
785 have an obvious adverse impact on radioanalytic measurements that require chemical separation.
786 Chapter 14.9 has additional information on carriers and tracers.
787
OTHERS
788 Other methods that have been used to preserve liquid samples containing organics and biological
789 materials include chemical preservatives (e.g., formaldehyde and methanol) or quick freezing by
790 means of liquid nitrogen. Table 10.1 summarizes the advantages and disadvantages of these and
791 previously described preservation methods.
792
793
794
TABLE 10.1—Summary of sample preservation techniques.
Preservation Technique
Addition of HNO3
Addition of Hcl
Addition of Sulfite
Addition of
Formaldehyde
Cooling
(Ice at approximately 0°
C)
Freezing
(Dry Ice at approximately
-78° C)
Addition of Paper Pulp
Advantages
Reduces pH and inhibits plating of
metals on container walls.
Reduces pH and inhibits plating of
metals on container walls.
Chloride forms strong anionic
complexes with Iron and Uranium.
Forms a reducing environment to
prevent the volatilization of iodine.
Preserves organic samples.
Prevents further biological activity.
Preserves organic samples (i.e.,
water, foods).
Reduces dehydration and retains
moisture.
Reduces biological activity.
Preserves organic samples (i.e.,
water, plant, animal).
Suspends biological activity.
Provides large surface area for
adsorption of metals, thus
minimizing adsorption on container
walls.
Disadvantages
Strong oxidizer that might react with organic
compounds.
Tritium might be separated preferentially as acid
hydrogen;
1/fC might be lost as 14CO2.
Tritium will be preferentially separated as acid
hydrogen;
14C might be lost as 14CO2
Might cause corrosion of stainless steel
planchets on gross analyses.
Might produce insoluble compounds from
reduced forms of iron or uranium.
May create disposal problems.
Ice melts, requiring replacement over time.
Dry ice sublimates and requires replacement.
Requires pH to be one or less.
Requires filtration and wet ashing of paper pulp
and combining liquids to make a new solution.
795
796
797
798
799
800
801
802
803
804
805
JULY 2001
DRAFT FOR PUBLIC COMMENT
10-25
MARLAP
DO NOT CITE OR QUOTE
-------
Field and Sampling Issues That Affect Laboratory Measurements
806 10.3.4 Liquid Samples: Special Cases
807
808 In some cases, liquid samples require special handling in order to preserve or retain a volatile or
809 gaseous radionuclide. The following are examples of specific methods used to recover or
810 preserve such samples of interest.
811 10.3.4.1 Radon-222 in Water
812 Waterborne radon is analyzed most commonly by liquid scintillation methods, although gamma
813 spectroscopy and other methods have been employed or proposed. Liquid scintillation has the
814 obvious advantage of being designed for automated sample processing and is, therefore, less
815 labor intensive or costly. A key to consistency in analytical results is the zero headspace sampling
816 protocol such as the one described below.
817 Since radon is inert and nonpolar, it diffuses through plastic more rapidly than glass. The use of
818 plastic scintillation vials, therefore, leads to significant loss of radon in water (Whittaker, 1989;
819 Hess and Beasley, 1990). For this reason, it is recommended that the water sample is collected in
820 a 23 mL glass scintillation vial, capped with a Teflon or foil-lined cap.
821 Samples are collected from a non-aerated faucet or spigot, which has been allowed to flow for
822 sufficient time so that the sample is representative of the water in the distribution system or well.
823 The time will vary depending on the source. The following zero headspace procedure will
824 minimize the loss of radon from the sample during collection:
825 • Place sample vial in a 300-600 mL beaker or other suitable container and attach the universal
826 adapter and fill-line to spigot, and start the flow.
827 • Fill the vial to prevent it from floating. Then fill the beaker until the vial is submerged.
828 • Place the tip of the fill line about two thirds of the way into the vial and fill until
829 approximately two or more vial (50-100 mL) volumes have been displaced.
830 • Carefully remove the vial with a pair of 10-inch tweezers and cap the vial with a Teflon or
831 foil-lined cap. Invert the sample and check for air bubbles. If any bubbles are present, discard
832 the sample and repeat the sampling procedure. Record date and time the sample was collected
833 and store the sample in a cooler to prevent temperature excursions. Transport the samples to
834 the laboratory in a cooler or other suitable insulated package.
835 10.3.4.1 Milk
836 Milk commonly is viewed as the food product of greatest potential dose significance for airborne
837 releases of radionuclides. Due to the metabolic discrimination, however, only a few radionuclides
MARLAP JULY 2001
DO NOT CITE OR QUOTE 10-26 DRAFT FOR PUBLIC COMMENT
-------
Field and Sampling Issues That Affect Laboratory Measurements
838 have a significant dose impact via the milk pathway, notably 90Sr, 131I, and 137Cs. Raw milk
839 should be obtained from the closest cows or goats downwind from a source.
840 To prevent milk from souring or curdling, samples should be refrigerated. Preservation of milk
841 may also be achieved through the addition of formaldehyde or methanol (DOE, 1987),
842 merthiolate, or Thimerosal (EPA, 1994). Analytical procedures for select radionuclides in milk
843 are well established and should be considered when deciding on a sample preservation method.
844 Owing to the volatility and potential loss of 131I, a known amount of Nal dissolved in water
845 should be added to the milk sample at time of collection. The Nal not only serves as a carrier for
846 the chemical separation of radioiodine, but also provides a quantitative tool for determining any
847 loss prior to analysis (DOE, 1990).
848 10.3.5 Non-aqueous Liquids and Mixtures
849 Non-aqueous liquids and mixtures include a wide range of organic fluids or solvents, organic
850 materials dissolved in water, oils, lubricants, etc. These liquids are not likely to represent
851 contaminated environmental media or matrices, but most likely represent waste streams that must
852 be sampled. Non-aqueous waste streams are generated as part of normal operations by nuclear
853 utilities, medical facilities, academic and research facilities, State and Federal Agencies, radio-
854 pharmaceutical manufacturers, DOE weapons complexes, mining and fuel fabrication facilities,
855 etc. Examples of these non-aqueous liquids and mixtures include waste oils and other lubricants
856 that are generated routinely from maintenance of various types equipment associated with
857 nuclear power plant operations or the production of nuclear fuel and nuclear weapon
858 components; and organic and inorganic solvents, acids, and bases that are used in a variety of
859 medical, research, and industrial applications.
860 In addition to the production of non-aqueous liquid wastes from routine operations by these
861 facilities, large quantities of non-aqueous liquids containing radionuclide contaminants are also
862 generated by routine facility decontamination efforts and final decontamination associated with
863 facility decommissioning. For decontamination and decommissioning activities, a wide range of
864 processes have been developed that employ halogenated organic compounds, such as Freon,
865 chloroform, or trichloroethane. Other aggressive chemical decontamination processes involve
866 dissolution and removal of metal and oxide layers from surfaces using acid solutions (e.g.,
867 sulfuric acid, nitric acid, phosphoric acids, and oxalic acid). Chemical decontamination also may
868 use chelating agents in concentrated processes (5 to 25 percent wt. chemical in solution) and
869 dilute processes (one percent wt. or less chemicals in solution). Examples of chemical processes
870 that can be used in both concentrated and dilute forms include the low oxidation-state transition-
871 metal ion (LOMI) and LOMI-nitric permanganate, developed by Dow Chemical Company and
872 AP/Citron. The reagents used in both the concentrated and dilute processes include chelating and
873 complexing agents such as ethylene diamine tetra acetic acid (EDTA), diethylene triamine penta-
874 acetic acid (DTPA), citric acid, oxalic acid, picolinic acid, and formic acid. Chelating agents and
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 10-27 DO NOT CITE OR QUOTE
-------
Field and Sampling Issues That Affect Laboratory Measurements _
875 organic acids are used in decontamination formulas because they form strong complexes with
876 actinides, lanthanides, heavy metals, and transition metals and assist in keeping these elements in
877 solution.
878 Generally, these chemical decontamination solutions, once used, are treated with ion-exchange
879 resins to extract the soluble activity. The ion-exchange decontamination solutions must,
880 nevertheless, be sampled to assess the amount of residual radioactivity.
881 The radionuclides that may be encountered with non-aqueous liquids and mixtures depend on
882 both the nature of the liquid and its usage. The following listing of radionuclides and liquids are
883 based on published data collected by NRC (1992) and the State of Illinois (Klebe 1998; IDNS
884 1993-1997):
885 • Toluene/xylene/scintillation fluids used by research and clinical institutions: 3H, 14C, 32P, 35S,
886 45Ca, 63Ni, "Tc, 90Sr, 125I, 147Pm, 226/228Ra, 228/230/232Th, 234/235/238U, 238/239/2«Pu, 241Am.
887 • Waste oils and lubricants from operation of motors, pumps, and other equipment: 3H, 54Mn,
888 65Zn, 60Co, 134/137Cs, 228/230/232Th.
889 • Halogenated organic and solvents from refrigeration, degreasing, and decontamination: 3H,
890 14C, 32P, 35S, 54Mn, 58/60Co, 63Ni, 90Sr, 125/129I, 134/137Cs, 226/228Ra, 22*™32Th, 232/234/238U,
891 23
892 • Other organic solvents from laboratory and industrial operations and cleaning: 3H, 32P, 35S,
893 45Ca, 125I, U-nat.
894 • Inorganic and organic acids and bases from extraction processes and decontamination: 3H,
895 14C, 32P, 35S, 54Mn, 67Ga, 1251, 60Co, 137Cs, 201/202Th, and U-nat.
896 Due to the large number of potential non-aqueous liquids and the complex mixtures of
897 radionuclide contaminants that may require radiochemical analysis, a comprehensive discussion
898 of sample preparation and preservation is beyond the scope of this discussion. In most instances,
899 however, these samples are not likely to require refrigeration or chemical preservatives that
900 protect against sample degradation.
901 Some organic solvents and highly acidic or basic liquids may react with plastic containers,
902 causing brittleness or breakage. In selecting sample containers for these non-aqueous samples, it
903 is important to assess the manufacturers product specifications, which typically provide
904 information regarding the container's resistance to chemical and physical agents. When non-
905 aqueous samples are stored for long periods of time, containers should be checked routinely.
906 10.4 Solids
MARLAP JULY 2001
DO NOT CITE OR QUOTE 1 0-28 DRAFT FOR PUBLIC COMMENT
-------
Field and Sampling Issues That Affect Laboratory Measurements
907
908 Solid samples consist of a wide variety of materials that include soil and sediment, plant and
909 animal tissue, metal, concrete, asphalt, trash, etc. In general, most solid samples do not require
910 preservation, but require specific processing in the field before transporting to the laboratory for
911 analysis. For example, soil sample field processing may require sieving in order to establish
912 sample homogeneity. These and other specific handling requirements are described below in the
913 section on each type of solid sample.
914 The most critical aspect is the collection of a sufficient amount of a representative sample. One
915 purpose of soil processing is to bring back only that sample needed for the laboratory. Unless
916 instructed otherwise, samples received by the laboratory are typically analyzed exactly as they are
917 received. This means that extraneous material should be removed at the time of sample
918 collection, if indicated in the appropriated plan document.
919 In many instances, sample moisture content at the time of collection is an important factor. Thus,
920 the weights of solid samples should be recorded at the time a sample is collected. This allows one
921 to track changes in wet weight from field to laboratory. Dry and ash weights generally are
922 determined at the laboratory.
923 Unlike liquid samples that may be introduced or removed from a container by simple pouring,
924 solid samples may require a container that is designed for easy sample placement and removal.
925 For this reason, large-mouth plastic containers with screw caps or individual boxes with sealable
926 plastic liners are commonly used. The containers also minimize the risk for breakage and sample
927 cross-contamination.
928 10.4.1 Soils
929
930 ASTM D653 (Standard Terminology Relating to Soil, Rock, and Contained Fluids) defines soil
931 as: "Sediments or other unconsolidated accumulations of solid particles produced by the physical
932 and chemical degradation of rocks, and that might or might not contain organic matter." ASTM
933 C999 provides generic guidance for soil sample preparation for the determination of
934 radionuclides. The American Society for Testing and Materials provides additional information
935 on soil and rock in the following standards:
936 • ASTM D 4914, Section 4, Construction, Volume 4.08 Soil and Rock (I).
937 • ASTM D 4943, Section 4, Construction, Volume 4.09 Soil and Rock (II): Geosynthetics.
938 The distribution of radionuclides in soil should be assumed to be heterogeneous. The degree of
939 heterogeneity is dictated by the radionuclide's mode of entry into the environment and soil, the
940 chemical characteristics of the radionuclide contaminant, soil composition, meteorological and
941 environmental conditions, and land use. For example, soil contamination from an airborne
942 release of a radionuclide with strong affinity for clay or other mineral constituents of soil (i.e.,
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 10-29 DO NOT CITE OR QUOTE
-------
Field and Sampling Issues That Affect Laboratory Measurements
943 high kd value) will likely exhibit a gradient with rapidly diminishing concentrations as a function
944 of soil depth. Moreover, contamination may be differentially distributed among soil particles of
945 different sizes. In most cases, because the contaminant is adsorbed at the surface of soil particles
946 and since the surface-to-volume ratio favors smaller particles, smaller soil particles will exhibit a
947 higher specific activity when compared to larger particles. If land areas include areas of farming,
948 tilling of soil will clearly impact the distribution of surface contamination.
949 10.4.1.1 Soil Sample Preparation
950 Extraneous material should be removed at the time of sample collection, if indicated in the
951 appropriate plan document. The material may have to be saved and analyzed separately,
952 depending on the project requirements and MQOs. If rocks, debris, and roots are removed from a
953 soil sample after it arrives at the laboratory, there might not be sufficient material to complete all
954 the requested analyses. A sufficient amount of sample should be collected to provide the net
955 quantity necessary for the analysis. Subsequent drying at the laboratory may remove a large
956 percentage of the sample weight that is available for analysis. Field-portable balances or scales
957 may be used to weigh samples as they are collected, further ensuring sufficient sample weights
958 are obtained. For certain types of samples, the project DQOs may require maintaining the
959 configuration of the sample, such as core samples where concentration verses depth will be
960 analyzed.
961 The project plan should address the impact of heterogeneity of radionuclide distribution in soil.
962 Some factors to consider that may impact radionuclide distribution are: determining sampling
963 depth, the need for removal of vegetative matter, rocks, and debris, and the homogenation of soil
964 particulates. For example, soil sampling depths of the top 5 cm is recommended for soils
965 contaminated by recent airborne releases (ASTM C998); soil depth to 15 cm may be appropriate
966 when exposure involves the need to monitor the root zone of food crops (MARSSEVI, 2000;
967 NRC, 1990). The need for sample field QC, such as field splitting, should be evaluated. Some
968 types of field QC can be used to evaluate the extent of radionuclide homogeneity. In general, no
969 special preservation measures are required for soil samples; however, preliminary soil sample
970 preparation involving drying, sieving, homogenizing, and splitting may be performed by a field
971 laboratory prior to sample shipment to the analytical laboratory.
972 If volatile elements are among other non-volatile contaminants, samples must be fractionated
973 before drying to avoid loss of the contaminant of interest. Dried samples are homogenized by
974 mortar and pestle, jaw crusher, ball mill, parallel plate grinder, blender, or a combination of these
975 techniques and sieved to obtain a uniform sample. Sieve sizes from 35 to 200 mesh generally are
976 recommended for wet chemistry procedures. ASTM C999 correlates various mesh sizes with
977 alternative designations, inclusive of physical dimensions expressed in inches or in the metric
978 system. In addition, samples for chemical separations are usually ashed in a muffle furnace to
979 remove any remaining organic materials that may interfere with the procedures.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 10-30 DRAFT FOR PUBLIC COMMENT
-------
Field and Sampling Issues That Affect Laboratory Measurements
980 10.4.1.2 Sample Ashing
981 Soil samples that require chemical separation for radionuclide analysis may also be ashed by the
982 field laboratory. The use of the term "field laboratory" can cause confusion, since no one
983 definition is possible. It is used here to define a lab that is close to the point of sample collection.
984 In no way does it imply that there is a distinction in requirements or specifications that impact
985 quality. For soil samples, ashing is performed in a muffle furnace to remove any organic
986 materials that may interfere with radiochemical procedures.
987 10.4.2 Sediments
988
989 Sediments of lakes, reservoirs, cooling ponds, settling basins, and flowing bodies of surface
990 water may become contaminated as a result of direct liquid discharges, wet surface deposition, or
991 from runoffs associated with contaminated soils. Because of various chemically and physically
992 binding interactions with radionuclides, sediments serve as integrating media that are important
993 to environmental monitoring. An understanding of the behavior of radionuclides in the aquatic
994 environment is critical to designing a sampling plan, because their behavior dictates their
995 distribution and sampling locations. Sediment cores may be sampled, frozen, and then sectioned.
996 The fate of radionuclides entering surface waters and their subsequent interaction with sediment
997 is complex due to numerous mechanisms and processes that affect the initial mixing and
998 dispersion of radionuclides, their distribution in water, sediment, plants and animals, and their
999 long-term retention within these compartments. Several factors must be considered to establish
1000 appropriate sediment sampling locations and depths and are discussed briefly below.
1001 10.4.2.1 Initial Mixing and Transport Dispersion of Radionuclides Discharged to Water
1002 The rapid initial mixing phase in the nearfield is dominated by the characteristics of the effluent
1003 and the outfall structure. The extent of nearfield mixing and dilution is strongly affected by the
1004 quantity of effluent relative to the receiving body of water, the level of turbulence produced by
1005 means of the discharge momentum (jet action), the discharge buoyancy (plume action), the
1006 outfall configuration, and the depth and current flow rate in the vicinity of outfall.
1007 Predictive models have been proposed for surface and submerged discharges; single point and
1008 multi-point outfalls; deep and shallow, stagnant and flowing water; and buoyant (positive and
1009 negative) and non-buoyant effects. An understanding of the basic hydrodynamic variables that
1010 define each of these conditions will aid in the selection of sampling locations.
1011 In the case of small and medium bodies of surface waters, where vertical thermal stratification is
1012 the primary factor that determines inflow and outflow dynamics, a simple one- or two-
1013 dimensional model may be appropriate as discussed in Regulatory Guide 1.113 (NRC 1977). For
1014 large bodies of surface water where neither horizontal nor vertical homogeneity can be assumed,
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 10-31 DO NOT CITE OR QUOTE
-------
Field and Sampling Issues That Affect Laboratory Measurements
1015 more complex three-dimensional dispersion models must be applied to properly assess
1016 hydrodynamics and the distribution of radionuclides in sediment. A review of numerical
1017 hydrodynamic models for large bodies of surface waters has been presented by Johnson (1980).
1018 10.4.2.2 Sediment Effect
1019 Following initial mixing in the nearfield (i.e., outfall), subsequent transport and distribution of a
1020 dissolved radionuclide is greatly impacted if the radionuclide is absorbed strongly from solution
1021 onto sediments by processes that include ion exchange, precipitation-mineral formation,
1022 complexation-hydrolysis, and oxidation-reduction. Both suspended and less-mobile bed
1023 sediments may absorb radionuclides, but suspended sediments usually absorb more efficiently
1024 per unit weight than bed sediments (Friend et al., 1965; Parker et al., 1965).
1025 The impacts of sediment absorption in a flowing body of water are obvious: the required time for
1026 sediment absorption allows the dissolved radionuclide to move considerable distances
1027 downstream before being absorbed, and sediment absorption steadily reduces the concentration
1028 of dissolved radionuclides with the result that an activity gradient is established in downstream
1029 water, sediment, and aquatic biota. Concentration gradients are further complicated by the high
1030 mobility of suspended sediments, the slow but steady erosion of bed sediments, the mobility and
1031 transfer of the radionuclide contaminant that has entered the aquatic food web, and the various
1032 mechanisms that modify sediment adsorption and desorption.
1033 10.4.2.3 Sample Preparation/Preservation
1034 In most cases, sediment is separated from water by simple decanting, but samples also may be
1035 obtained by filtering a slurry or through passive evaporation. As noted previously, care must be
1036 taken to avoid cross contamination from sampling by decontaminating or replacing tools and also
1037 from avoiding contact between successive samples. Suitable sample containers include glass or
1038 plastic jars with screw caps. The presence of volatile or semi-volatile organic and micro-
1039 organisms may impact the radionuclide concentration, therefore, samples should be kept on ice
1040 while in the field and refrigerated while awaiting radioanalysis.
1041 10.4.3 Other Solids
1042
1043 10.4.3.1 Structural Materials
1044 In some cases, a project plan requires sample analysis of structural materials such as concrete or
1045 steel. Concrete from floors, walls, sidewalks or road surfaces is typically collected by scabbling,
1046 coring, drilling, or chiseling. Depending on the radionuclides of interest and detection methods,
1047 these sample preparations may require crushing, pulverization, and sieving.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 10-32 DRAFT FOR PUBLIC COMMENT
-------
Field and Sampling Issues That Affect Laboratory Measurements
1048 Metal associated with structures (e.g., I-beams, rebar) or machines may be contaminated on
1049 exterior or interior surfaces or through activation may become volumetrically contaminated.
1050 Surface contamination may be assessed by swipe samples that provide a measure of removable
1051 contamination (Section 10.7) or by scraping, sandblasting, or other abrasive techniques.
1052 Volumetric contamination is frequently assessed by non-destructive field measurements that rely
1053 on gamma-emitting activation products. However, drill-shavings or pieces cut by means of a
1054 plasma arc torch may be collected for further analysis in a laboratory where they can be analyzed
1055 in a low-background environment. In general, these materials require no preservation but, based
1056 on activity/dose rate levels and sample size and weight, may require proper shielding, engineered
1057 packaging, and shipping by a licensed carrier.
1058 10.4.3.2 Biota: Samples of Plant and Animal Products
1059 The release of radionuclides to the environment from normal facility operations or as the result of
1060 an accident requires the sampling of a wide variety of terrestrial and aquatic biota. Guidance
1061 provided below is directed principally to those responsible for designing a sampling plan, who
1062 must make decisions pertaining to the type of samples that should be collected, where and how to
1063 collect the samples, and the preferred methods for sample preparation. For most biota, sample
1064 preservation usually is achieved by icing samples in the field and refrigeration until receipt by the
1065 analytical laboratory.
1066 The specific media that fall under this general category include food, domestic animals (meat and
1067 poultry), animal products, game animals, game birds, etc. The field sampling plan should
1068 describe the type of processing and preservation required.
1069 Samples of food and certain terrestrial animals are of greatest importance in environmental
1070 surveillance because they provide the most direct basis for assessing the radiation dose to man.
1071 The principal pathways for radionuclide contamination of food and plants are atmospheric
1072 deposition from airborne releases and crop irrigation from rivers, ponds or lakes receiving liquid
1073 effluents. Care should also be taken not to select a sampling site that has been fertilized or has
1074 been contaminated by runoffs from fertilized soil due to enhanced natural radioactivity content of
1075 many fertilizers (ASTM C998).
1076 To determine the dose to a population, pathway analysis may require sampling of food and biota.
1077 One example is the analysis of meat from domestic or game animals. Samples from food and
1078 biota also may be used to determine radionuclide accumulation in the environment. For example,
1079 the analysis of growth rings from trees may indicate when a radionuclide was released into the
1080 environment.
1081 Animal feeds also provide important data for determining radionuclide concentrations in the food
1082 chain. Foods may be categorized according to the U.S. Department of Agriculture scheme as
1083 leafy vegetables, grains, tree-grown fruits, etc., and representative samples from each group may
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 10-33 DO NOT CITE OR QUOTE
-------
Field and Sampling Issues That Affect Laboratory Measurements
1084 be selected for analysis. Guidance for procuring or preparing terrestrial samples is provided
1085 below.
1086 MEAT, PRODUCE, AND DAIRY PRODUCTS
1087 Meat, poultry, eggs, fresh produce, and other food should be procured from local farmers most
1088 likely to have been affected by a singular event. The choice of sample is dependent on the
1089 pathway. Meat samples also may be collected at a slaughter house if the origin of the animals can
1090 be documented. Local health departments may be able to assist in getting samples. Samples
1091 should be placed in sealed plastic bags and appropriately labeled and preserved by means of ice
1092 in the field and refrigeration during interim storage prior to delivery to the analytical laboratory.
1093 All food samples may be reduced to edible portions (depending on study objective) for analysis
1094 in a manner similar to that for human consumption (i.e., remove cores, bones, seeds, other
1095 nonedible parts) and weighed as received from the field (i.e., wet weight) within 24 hours. Wet
1096 weights are desired, since consumption data are generally on this basis.
1097 For sampling fresh produce, fruits, meats, and other domestic animal products, a local land-use
1098 study may be necessary to determine what crops and animals are important in the local diet and
1099 where they are produced with respect to the site. Fruit and vegetable samples should be collected
1100 near the point of maximum predicted annual ground concentration from airborne releases and
1101 from areas that may be contaminated by water into which liquid plant wastes have been
1102 discharged (e.g., irrigated crops). Local land usage should be reviewed periodically, as well as
1103 current farming and stock-feeding practices at sampling locations.
1104 ANIMAL FEED AND VEGETATION
1105 Crops raised for animal feed and vegetation consumed by grazing farm animals may be sampled.
1106 Depending upon radionuclides under investigation and their analytical sensitivities, kilogram
1107 quantities of vegetative matter may be needed. The choice of species and sample type must be
1108 guided by factors such as exposure pathways, species availability, seasonal growth patterns, soil
1109 types, and farming practices.
1110 As in all terrestrial samples, naturally occurring 40K and the uranium and thorium series
llll contribute to the radiation observed. Deposition of such cosmic-ray-produced nuclides as 7Be and
1112 fallout from nuclear tests also may be present. Properly selected processed items from commer-
1113 cial sources may be helpful in providing natural and anthropogenic background data.
1114 WILDLIFE
1115 Wild animals that are hunted and eaten may be of interest for potential dose estimates and
1116 therefore may require sampling. However, the data from small numbers of samples of wild
1117 animals or game birds should be viewed with caution because of their great variation in mobility,
MARLAP JULY 2001
DO NOT CITE OR QUOTE 10-34 DRAFT FOR PUBLIC COMMENT
-------
Field and Sampling Issues That Affect Laboratory Measurements
1118 age, and diet. Examples of wildlife that have been used are rabbits and rodents that may feed on
1119 and live in a contaminated site.
1120 Wildlife samples can be trapped, acquired from hunters, collected after accidental road kills, or
1121 obtained by request to the appropriate state game agency. Wildlife that is relatively rare locally
1122 should not be taken as environmental samples. Since the choice of species samples may be
1123 crucial to the usefulness of the results, local ecologists and biologists should be consulted to
1124 ensure consideration of factors that affect animal radionuclide uptake and retention, such as size,
1125 age, sex, feeding locus, and food consumption. An estimate of the radionuclide intake of the
1126 animal just before its death may be provided by analyzing the stomach content, especially the
1127 rumen in deer. However, the sample must be collected within a brief period (two to four hours)
1128 after death.
1129 AQUATIC ENVIRONMENTAL SAMPLES
1130 In addition to natural radionuclides and natural radionuclides enhanced by human activity, there
1131 are numerous man-made radionuclides that have the potential for contaminating surface and
1132 ground water. The most common of these are fission and activation products associated with
1133 reactor operation and fuel cycle facilities. Radioanalysis of aquatic samples may therefore
1134 include 54Mn, 58Co, 60Co, 65Zn, 95Zr, 90Sr, 134Cs, 137Cs, and transuranics, such as 239Pu.
1135 When surface and ground waters are contaminated, radionuclides may be transferred through a
1136 complex food web consisting of aquatic plants and animals. Aquatic plants and animals, as
1137 discussed here, are any species which derive all or substantial portions of their nourishment from
1138 the aquatic ecosystem, are part of the human food chain, and show significant accumulation of a
1139 radionuclide relative to its concentration in water. Although fish, aquatic mammals, and
1140 waterfowl provide a direct link to human exposure, lower members of the food chain also may be
1141 sampled.
1142 FLORA
1143 Aquatic biota such as algae, seaweed, and benthic organisms are indicators and concentrators of
1144 radionuclides—especially 59Fe, 60Co, 65Zn, 90Sr, and 137Cs—and can be vectors in the water-fish-
1145 human food chain. As such, they may be sampled upstream and downstream at locations similar
1146 to those described for sediment. Because of their high water content, several kilograms (wet
1147 weight) should be collected per sample. The wet weight of the sample should be recorded.
1148 Enough of the wet sample should be processed so that sufficient sample remains following the
1149 drying process. Both algae (obtained by filtering water or by scraping submerged substrates) and
1150 rooted aquatic plants should be sampled.
ii5i FISH AND SHELLFISH
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 10-3 5 DO NOT CITE OR QUOTE
-------
Field and Sampling Issues That Affect Laboratory Measurements
1152 For practical reasons, fish and shellfish may be purchased from local sources if the origin can be
1153 determined. Samples also can be obtained by pole fishing, netting, or electric shock devices. The
1154 sampling plan will describe the processing needed. Samples should include each of the principal
1155 edible types in local catches. Several kilograms of each fish sample are usually required; this may
1156 be one large fish, but preferably a composite of a number of small ones. Analysis of the edible
1157 portions of food fish as prepared for human consumption is of major interest. Fish may be de-
1158 boned, if specified in the sampling plan. The whole fish is analyzed if it is used for the
1159 preparation of a fish meal for consumption or if only trend indication is required. In a program
1160 where fish are the critical pathway, fish are analyzed by species; if less detail is required, several
1161 species with similar feeding habits (such as bottom feeders, insectivores, or predators) may be
1162 collected and the data grouped.
1163 In large bodies of water, samples from several locations are desirable because of the difficulty in
1164 knowing whether a fish caught at a given location had lived there for an extended period. Thus,
1165 the presence or absence of a radionuclide in a specific fish does not permit any definite
1166 conclusion concerning the presence of the radionuclide in water at that location. For some fish,
1167 more specific information concerning their usual location may be available; for example, dams,
1168 salinity gradients, and temperature gradients can be effective barriers to their movement.
1169 Information on fish age, feeding habits, and the quality of the aquatic environment are desirable
1170 to evaluate the significance of any findings.
1171 Shellfish, such as clams, oysters, and crabs, are collected for the same reasons as fish, but have
1172 the advantage as indicators of being relatively stationary. Their restricted mobility contributes
1173 substantially to the interpretation and application of analytical results to environmental
1174 surveillance. Edible and inedible portions of these organisms can be prepared separately.
1175 WATERFOWL
1176 Waterfowl, such as ducks and geese, may also concentrate radionuclides from their food sources
1177 in the aquatic environment and serve as important food sources to humans. The migratory
1178 patterns and feeding habits of waterfowl vary widely. Some species are bottom feeders and, as
1179 such, tend to concentrate those radionuclides associated with sediments such as 60Co, 65Zn, and
1180 137Cs. Others feed predominantly on surface plants, insects, or fish.
1181 Whenever practical, and if time permits, waterfowl should be obtained by hunting, but a trapping
1182 procedure may also be used. An important consideration in obtaining a sample from waterfowl is
1183 that their exterior surfaces, especially feathers, may be contaminated. It is important to avoid
1184 contaminating the "flesh" sample during handling. As with other biota samples, analyses may be
1185 limited to the edible portions and should be reported on a wet weight basis. Local game officials
1186 or aquatic ecologists may provide valuable information for choosing the proper species.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 10-36 DRAFT FOR PUBLIC COMMENT
-------
Field and Sampling Issues That Affect Laboratory Measurements
1187 Caution is advised in the selection of background or control locations for all biota (terrestrial and
1188 aquatic) sampled, at least for those species whose mobility and feeding habits may significantly
1189 affect the results obtained. Since this mobility makes it difficult to establish upstream/
1190 downstream sampling locations for biota in a manner analogous to those for air, water, or plants,
1191 a sound sampling strategy may require the expert advice and direction of local ecologists, and
1192 fish and game personnel. Samples from the background locations should be from an ecosystem
1193 identical to that of those collected near the site, but unaffected by site effluents.
ii94 10.5 Air Sampling
1195 The measurement of airborne radionuclides as gases or particulates provides a means of
1196 evaluating internal exposure through the inhalation pathways. The types of airborne radioactivity
1197 that may require air sampling are normally categorized as: (1) airborne particulates; (2) noble
1198 gases; (3) volatilized halogens (principally radioiodines); and (4) tritiated water. Depending upon
1199 the source term and the objectives of the investigation, air sampling may be conducted outdoors
1200 as well as indoors on behalf of a variety of human receptors. For example, routine outdoor air
1201 samples may be taken for large population groups living within a specified radius of a nuclear
1202 facility. On the other end of the spectrum, air samples may be taken for a single person or small
1203 group of persons exposed occupationally to a highly localized source of airborne radioactivity.
1204 The purpose of the samples being collected must, therefore, be well defined in terms of sampling
1205 location, field sampling equipment, and required sample volumes. Due to the wide range of
1206 conditions that may mandate air sampling, and the limited scope of this section, only generic
1207 topics of air sampling will be discussed.
1208 10.5.1 Sampler Components
1209 Common components of air sampling equipment include a sample collector (i.e., filter), a sample
1210 collector holder, an air mover, and a flow-rate measuring device.
1211 The sample holder should provide adequate structural support while not damaging the filter,
1212 should prevent sampled air from bypassing the filter, should facilitate changing the filter, and
1213 should facilitate decontamination. A backup support that produces negligible pressure drop
1214 should be used behind the filter to prevent filter distortion or deterioration.
1215 If rubber gaskets are used to seal the filter to the backing plate, the gasket should be in contact
1216 with the filter along the entire circumference to ensure a good fit.
1217 Air movers or vacuum systems should provide the required flow through the filter and to
1218 minimize air flow reduction due to filter loading. Consideration should be given to the use of air
1219 movers that compensate for pressure drop. Other factors to consider should include size, power
1220 consumption, noise, durability, and maintenance requirements.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 10-37 DO NOT CITE OR QUOTE
-------
Field and Sampling Issues That Affect Laboratory Measurements
1221 Each air sampler should be equipped with a reliable calibrated air flow measuring device with
1222 specified accuracy. To calculate the concentrations of any radionuclide in air collected, it is
1223 necessary to accurately determine the total volume of air sampled. The planning documents
1224 should state who is responsible for making volume corrections. Also, the information needed for
1225 half-life corrections for short-lived radionuclides needs to be recorded.
1226 Generally, a parameter of the air mover can be related to flow. If the mean flow during a
1227 collection period can be determined, the total volume of air sampled can be readily calculated.
1228 Accurate flow measurements and the total integrated sample volume of air can be obtained using
1229 a mass flow meter and a totalizer. This direct technique of air flow measurement becomes
1230 impractical at remote field locations, due to cost and exposure of the flow meter to harsh
1231 environments. Other procedures for the measurement of air flow in sampling systems are
1232 reviewed by Lippmann (1989a). The equipment readings (flow rate, volume, etc.) should be
1233 recorded by the sample collector.
1234 The collection medium or filter used depends on the physical and chemical properties of the
1235 materials to be collected and counted. A variety of particulate filters (cellulose, cellulose-
1236 asbestos, glass fiber, membrane, polypropylene, etc.) is available. The type of filter is selected
1237 according to needs, such as high collection efficiency, particle-size selectivity, retention of alpha
1238 emitters on the filter surface, and the compatibility with radiochemical analysis. The criteria for
1239 filter selection are good collection efficiency for submicron particles at the range efface
1240 velocities used, high particle and mass loading capacity, low-flow resistance, low cost, high
1241 mechanical strength, low-background activity, compressibility, low-ash content, solubility in
1242 organic solvents, non-hygroscopicity, temperature stability, and availability in a variety of sizes
1243 and in large quantities. The manufacturer's specifications and literature should provide a source
1244 for filter collection efficiency. In the selection of a filter material, a compromise must be made
1245 among the above-cited criteria that best satisfies the sampling requirements. An excellent review
1246 of air filter material used to monitor radioactivity was published by Lockhart and Anderson
1247 (1964). Lippmann (1989b) also provides information on the selection of filter materials for
1248 sampling aerosols by filtration. See ANSI (1999), Annex D and Table D.I, for criteria for the
1249 selection of filters for sampling airborne radioactive particles.
1250 In order to select a filter medium with adequate collection efficiency, it may be necessary to first
1251 determine the distribution of size of airborne particulates. Several methods, including impactors
1252 (e.g., multistage cascade impactor) and electrostatic precipitators, can be used to classify particle
1253 size. Waite and Nees (1973) and Kotrappa et al. (1974) discuss techniques for particle sizing
1254 based on the flow discharge perturbation method and the HASL cyclone, respectively. These
1255 techniques are not recommended for routine environmental surveillance of airborne parti culates,
1256 although their use for special studies or for the evaluation of effluent releases should not be
1257 overlooked. Specific data on various filter materials, especially retention efficiencies, have been
1258 reported by several authors (Lockhart and Anderson, 1964; Denham, 1972; Stafford, 1973;
1259 ASTM STP555) and additional information is available from manufacturers.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 10-3 8 DRAFT FOR PUBLIC COMMENT
-------
Field and Sampling Issues That Affect Laboratory Measurements
1260 10.5.2 Filter Selection Based on Destructive Versus Non-destructive Analysis
1261 Pure cellulose papers are useful for samples to be dissolved and analyzed radiochemically, but
1262 the analytical filter papers used to filter solutions are inefficient collectors for aerosols and clog
1263 easily. Cellulose-asbestos filter papers combine fairly high efficiency, high flow rates, high
1264 mechanical strength, and low pressure drops when loaded. They are very useful for collecting
1265 large samples but present difficulties in dissolution, and their manufacture is diminishing because
1266 of the asbestos. Fiberglass filters can function efficiently at high flow rates, but require fluoride
1267 treatment for dissolution and generally contain sufficient radioactive nuclides to complicate low-
1268 activity analysis. Polystyrene filters are efficient and capable of sustaining high air flow rates
1269 without clogging. They are readily destroyed for analysis by ignition (300° C) or by wet washing
1270 with oxidizing agents, and also are soluble in many organic liquids. They have the disadvantage
1271 of low mechanical and tensile strength, and they must be handled carefully. Membrane filters are
1272 excellent for surface collection efficiency and can be used for direct alpha spectrometry on the
1273 filter. However, they are fragile and suffer from environmental dust loading. An alternative
1274 choice for radionuclides in the environment is the polypropylene fiber filter, Dynaweb Grade
1275 DW7301L. Filters come in two sizes: a 20.32 cm circle and a 20.32 cm x 25.40 cm rectangle.
1276 The filter is composed of a 100 percent polypropylene web that is 100 percent binderless. Three
1277 layers of this web are collated and sandwiched between two sheets of a protective DuPont Reeme
1278 (100 percent polyester) scrim.
1279 10.5.3 Sample Preservation and Storage
1280 Since paniculate air samples are generally dry samples that are chemically and physically stable,
1281 they require no preservation. However, care must be exercised to avoid loss of sample from the
1282 filter medium and the cross contamination among individual samples. A common method is to
1283 fold filters symmetrically so that the two halves of the collection surface are in contact. Filters
1284 should be stored in individual envelopes that have been properly labeled. Filters may also be
1285 stored in special holders that attach on the filter's edge outside of the collection surface.
1286 When background levels of 222Ra and 220Ra progeny interfere with evaluation of alpha air
1287 samples, a holdup time of several hours may be required before samples are counted. Corrections
1288 or determinations can also be made for the contribution of radon or thoron progeny present on a
1289 filter (Setter and Coats, 1961).
1290 10.5.4 Special Cases: Collection of Gaseous and Volatile Air Contaminants
1291 Prominent radionuclides that may exist in gaseous states include noble gases, 14C as carbon
1292 dioxide or methane, 3H as water vapor, and volatilized radioiodines. (Radon is discussed in
1293 Section 10.5.5.)
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 10-39 DO NOT CITE OR QUOTE
-------
Field and Sampling Issues That Affect Laboratory Measurements
1294 10.5.4.1 Radioiodines
1295 The monitoring of airborne iodine, such as 129I and 131I, may be complicated by the probable
1296 existence of several species, including paniculate iodine or iodine bound to foreign particles,
1297 gaseous elemental iodine, and gaseous non-elemental compounds of iodine. A well-designed
1298 sampling program should be capable of distinguishing all possible iodine forms. While it may
1299 not always be necessary to differentiate between the various species, care should be taken so that
1300 no bias can result by missing one or more of the possible species. See ANSI (1999) Annex C.3,
1301 for information on collection media for radioiodine.
1302 In addition to the problems noted above, charcoal cartridges (canisters) for the collection of
1303 radioiodine in air are subject to channeling. Hence, they should be carefully checked before
1304 operation in the field (analogous to DOP testing of high efficiency particulate air (HEP A) filters
1305 in situ ) or several should be mounted in series to prevent loss of iodine. Too high a sampling rate
1306 reduces both the collection efficiency and retention time of charcoal filters, especially for the
1307 non-elemental forms of iodine (Keller et al., 1973; Bellamy, 1974). The retention of iodine in
1308 charcoal is dependent not only on charcoal volume, but also the length of the charcoal bed.
1309 Typical air flow rates for particulate sampling of 30 to 90 L/min (1 to 3 ft3/min) are normally
1310 acceptable for environmental concentrations of radioiodine. The method proposed by the
1311 Intersociety Committee (APHA, 1972) for 131I concentrations in the atmosphere involves
1312 collecting iodine in its solid and gaseous states with an "absolute" particulate filter in series with
1313 an activated charcoal cartridge followed by gamma spectrometric analysis of the filter and
1314 cartridge. The Intersociety-recommended charcoal cartridges are 5/8 in. diameter by 1.5 in. deep
1315 containing 3 g of 12 to 30 mesh Kl-activated charcoal. The minimum detectable level using the
1316 Intersociety method is 3.7 x 10"3 Bq/m3 (0.1 pCi/m3). Larger cartridges will improve retention,
1317 permitting longer sampling periods. A more sensitive system has been described by Baratta et al.
1318 (1968), in which concentrations as low as 0.037 Bq/m3 (0.01 pCi/mL) of air are attainable.
1319 For the short-lived radioiodines (mass numbers 132, 133, 135), environmental sampling is
1320 complicated by the need to obtain a sufficient volume for analysis, while at the same time,
1321 retrieving the sample soon enough to minimize decay (with half-lives ranging from two hours to
1322 31 hours). Short period (grab) sampling with charcoal cartridges is possible, with direct counting
1323 of the charcoal as soon as possible for gamma emissions, but radon and thoron will affect
1324 detect on 1 evel s.
1325 Because of the extremely long half-life and normally low environmental concentrations, 129I
1326 determinations must usually be performed by neutron activation or mass spectrometry analysis
1327 after chemical isolation of the iodine. For concentrations about 3 x 10"10 |iCi/mL, liquid
1328 scintillation counting can be used after solvent extraction (Gabay et al., 1974).
MARLAP JULY 2001
DO NOT CITE OR QUOTE 10-40 DRAFT FOR PUBLIC COMMENT
-------
Field and Sampling Issues That Affect Laboratory Measurements
1329 10.5.4.2 Gases
1330 Sampling for radioactive gases is either done by grab sample that employs an evacuated chamber
1331 or by airflow through a medium such as charcoal, water, or a variety of chemical absorbers. For
1332 example, radioactive CO2 is most commonly extracted by passing a known volume of air through
1333 columns filled with 3 M NaOH solution. After the NaOH is neutralized with sulfuric acid, the
1334 CO2 is precipitated in the form of BaCO3, which then can be analyzed in a liquid scintillation
1335 counter (NCRP, 1985).
1336 Because noble gases have no metabolic significance, and concern is principally limited to
1337 external exposure, surveillance for noble gases is commonly performed by ambient dose rate
1338 measurements. However, the noble gases xenon and krypton may be extracted from air by
1339 adsorption on activated charcoal (Scarpitta and Harley, 1990). However, depending upon the
1340 analytical method and instrumentation employed, significant interference may result from the
1341 presence of naturally occurring radioactive gases of 222Rn and 220Rn.
1342 10.5.4.3 Tritium Air Sampling
1343 In air, tritium occurs primarily in two forms: as water vapor (HTO) and as hydrogen gas (HT).
1344 Tritiated organic compounds in the vapor phase or attached to paniculate matter occur only
1345 occasionally. To measure tritium as HT or in tritiated organic, the gas phase can be oxidized,
1346 converting the tritium to HTO before desiccation and counting. For dosimetric purposes, the
1347 fraction present as HT can usually be neglected, since the relative dose for a given activity
1348 concentration of HTO is 400 times that for HT (NCRP, 1978). However, if HT analysis is
1349 required, it can be removed from the atmosphere by oxidation to water (HTO) using CuO/MnO2
1350 at 600° C (Pelto et al., 1975), or with air passed over platinum alumina catalyst (Bixel and
1351 Kershner 1974). These methods also oxidize volatile tritiated organic compounds to yield
1352 tritiated water (ANSI, 1999, Annex H).
1353 A basic system for sampling HTO consists of a pump, a sample collector, and a flow-measuring
1354 or flow-recording device. Air is drawn through the collector for a measured time period at a
1355 monitored flow rate to determine the total volume of air sampled. The total amount of HTO
1356 recovered from the collector is divided by the total volume of air sampled to determine the
1357 average HTO-in-air concentration of the air sampled. In some sampler types, the specific activity
1358 of the water collected is measured and the air concentration is determined from the known or
1359 measured humidity. Some common collectors are cold traps, tritium-free water, and solid
1360 desiccants, such as silica gel, DRIERITE™, or molecular sieve.
1361 Cold traps are usually made of glass and consist of cooled collection traps through which sample
1362 air flows. The trap is cooled well below the freezing point of water, usually with liquid nitrogen.
1363 The water vapor collected is then prepared for analysis, usually by liquid scintillation counting.
1364 Phillips and Easterly (1982) have shown that more than 95 percent HTO collection efficiency can
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 10-41 DO NOT CITE OR QUOTE
-------
Field and Sampling Issues That Affect Laboratory Measurements
1365 be obtained using a single cold trap. Often a pair of cold traps is used in series, resulting in a
1366 collection efficiency in excess of 99 percent.
1367 Gas-washing bottles (i.e., "bubblers") filled with an appropriate collecting liquid (usually tritium-
1368 free water) are used quite extensively for collecting HTO from air. HTO in the sample gas stream
1369 "dissolves" in the collecting liquid. For the effective collection rate to remain the same as the
1370 sample flow rate, the specific activity of the bubbler water must be negligible with respect to the
1371 specific activity of the water vapor. Thus, the volume of air that can be sampled is ultimately
1372 limited by the volume of water in the bubbler. However, except when sampling under conditions
1373 of very high humidity, sample loss (dryout) from the bubbler usually limits collection time rather
1374 than the attainment of specific activity equilibrium. Osborne (1973) carried out a thorough
1375 theoretical and experimental evaluation of the HTO collection efficiency of water bubblers over a
1376 wide range of conditions.
1377 The use of silica gel as a desiccant to remove moisture from air is a common technique for
1378 extracting HTO. The advantage of using silica gel is that lower HTO-in-air concentrations can be
1379 measured, since the sample to be analyzed is not significantly diluted by an initial water volume,
1380 which occurs when a liquid-sampling sink is used. Correcting for dilution is discussed in Rosson
1381 et al. (2000).
1382 10.5.5 Radon
1383 There are three isotopes of radon in nature: 222Rn is a member of the 238U decay chain; 220Rn is a
1384 member of the 232Th decay chain; and 219Rn is a member of the 235U decay chain. Because of the
1385 small relative abundance of the parent nuclides and the short half-lives of 220Rn (55 seconds) and
1386 219Rn (4 seconds), the term "radon" generally refers to the isotope 222Rn. Owing to its ubiquitous
1387 presence in soils, uranium mill tailings, underground mines, etc., and the health risks to large
1388 populations and occupational groups, radon is perhaps the most studied radionuclide.
1389 Consequently, many reports and articles have been published in the scientific literature dealing
1390 with the detection methods and health risks from radon exposures. Many of them appear in
1391 publications issued by the EPA, DOE, NCRP, NAS, and in radiation-related journals, such as the
1392 journals Health Physics and Radiation Research. Given the voluminous amount of existing
1393 information, only a brief overview of the sampling method can be presented here.
1394 10.5.5.1 Radon Sampling Methods
1395 Quantitative measurements of radon gas and its short-lived decay products can be obtained by
1396 several techniques that are broadly categorized as grab sampling, continuous radon monitoring,
1397 and integrative sampling. Each method imposes unique requirements that should be followed
1398 carefully. The U.S. EPA Radon Measurement Proficiency (RMP) Program should be consulted
1399 for current guidance for sample collection (EPA, 1992; EPA, 1993). Information is available on
MARLAP JULY 2001
DO NOT CITE OR QUOTE 10-42 DRAFT FOR PUBLIC COMMENT
-------
Field and Sampling Issues That Affect Laboratory Measurements
1400 the RMP home page at www.epa.gov/radonpro/index.htm. Working with the Radon Proficiency
1401 Program (RPP) is described in a separate handbook (EPA, 1996). A description of additional
1402 sampling methods and materials is also presented in EPA (1994) and Cohen (1989).
1403 In general, EPA's protocols specify that radon sampling and measurements be made under
1404 standardized conditions when radon and its progeny are likely to be at their highest concentra-
1405 tions and maximum equilibrium. For indoor radon measurement, this implies minimum building
1406 ventilation through restrictions on doors, windows, HVAC systems, etc. Also sampling should
1407 not take place during radical changes in weather conditions. Both high winds and rapid changes
1408 in barometric pressure can dramatically alter a building's natural ventilation rate. Although
1409 recommended measurements are likely to generate higher than actual average concentrations, the
1410 benefit of a standardized sampling condition is that it is reproducible, least variable, and
1411 moderately conservative. Brief descriptions of the basic techniques used to sample air for radon
1412 and its progeny are provided below.
1413 GRAB SAMPLING
1414 The term "grab sampling" refers to very short-term sampling. This method consists of evaluating
1415 a small volume of indoor air for either radon or radon decay product concentration. In the radon
1416 grab sampling method, a sample of air is drawn into and subsequently sealed in a flask or cell
1417 that has a zinc sulfide phosphor coating on its interior surfaces. One surface of the cell is fitted
1418 with a clear window that is put in contact with a photomultiplier tube to count light pulses
1419 (scintillations) caused by alpha disintegrations from the sample interacting with the zinc sulfide
1420 coating. The number of pulses is proportional to the radon concentration in the cell. The cell is
1421 counted about four hours after filling to allow the short-lived radon decay products to reach
1422 equilibrium with the radon. The results are corrected to compensate for decay during the time
1423 between collection and counting, and for decay during counting.
1424 Several methods for performing such measurements have been developed. However, two
1425 procedures that have been most widely used with good results are the Kusnetz procedure and the
1426 modified Tsivogiou procedure. In brief, the Kusnetz procedure (Kusnetz, 1956; ANSI, 1973)
1427 may be used to obtain results in working levels (WL) when the concentration of individual decay
1428 products is not important. Decay products in up to 100 liters of air are collected on a filter in a
1429 five-minute sampling period. The total alpha activity on the filter is counted any time between 40
1430 and 90 minutes after sampling is completed. Counting can be done using a scintillation-type
1431 counter to obtain gross alpha counts for a selected counting time. Counts from the filter are
1432 converted to disintegrations using the appropriate counter efficiency. The disintegrations from
1433 the decay products may be converted into working levels using the appropriate "Kusnetz factor"
1434 for the counting time used.
1435 The Tsivogiou procedure may be used to determine both WL and the concentration of the
1436 individual radon decay products. Sampling is the same as in the Kusnetz procedure. However,
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 10-43 DO NOT CITE OR QUOTE
-------
Field and Sampling Issues That Affect Laboratory Measurements
1437 the filter is counted three separate times following collection. The filter is counted between 2 and
1438 5 minutes, 6 and 20 minutes, and 21 and 30 minutes after sampling is complete. Count results are
1439 interpreted by a series of equations that calculate concentrations of the three radon decay
1440 products and WL.
1441 The advantages of grab sampling are that the analysis time is relatively short, results are available
1442 within a short time, and conditions during the measurement are known to the sampler. In
1443 addition, grab sampling does not provide a long-term average and house conditions must be
1444 controlled for 12 hours prior to measurement.
1445 CONTINUOUS RADON MONITOR
1446 A continuous radon monitor (CRM) samples the ambient air by pumping air into a scintillation
1447 cell after passing it through a paniculate filter that removes dust and radon decay products. As
1448 the radon in the air decays, the ionized radon decay products plate out on the interior surface of
1449 the scintillation cell. As the radon decays, the alpha particles strike the coating on the inside of
1450 the cell, causing scintillations. The scintillations are detected by the photomultiplier tube in the
1451 detector, which generates electrical signals. The signals are processed and the results are either
1452 stored in the memory of the CRM or printed on paper tape by the printer. The CRM must be
1453 calibrated in a known environment to obtain the conversion factor used to convert count to radon
1454 concentration.
1455 The CRM may be a flowthrough-cell type or a periodic-fill type. In the flowthrough-cell type, air
1456 flows continuously into and through the scintillation cell. The periodic-fill type fills the cell once
1457 during each preselected time interval, counts the scintillations, then begins the cycle again.
1458 An analogous device to the continuous radon monitor is the Continuous Working Level Monitor
1459 (CWLM). This device filters air at a low flow rate of about 0.2 to one liter per minute and
1460 measures the amount of radon decay products on the filter medium. An alpha detector, such as a
1461 diffused-junction or surface-barrier detector, counts the alpha particles produced by the radon
1462 decay products as they decay on the filter. The detector is normally set to detect alpha particles
1463 with energies between 2 and 8 meV. The alpha particles emitted from the radon decay products
1464 218Po and 214Po are the significant contributors to the events that are measured by the detector.
1465 The event count is directly proportional to the number of alpha particles emitted by the radon
1466 decay products on the filter. The unit typically contains a microprocessor that stores the number
1467 of counts and elapsed time. The unit can be set to record the total counts registered over specified
1468 time periods. The unit must be calibrated in a calibration facility to convert count rate to working
1469 level (WL) values. This may be done initially by the manufacturer and should be done
1470 periodically thereafter by the operator.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 10-44 DRAFT FOR PUBLIC COMMENT
-------
Field and Sampling Issues That Affect Laboratory Measurements
1471 INTEGRATING SAMPLING DEVICES
1472 By far, the most common technique for measuring radon is by means of integrating devices.
1473 Integrating devices, like the charcoal canister and the Electret-Passive Environmental Radon
1474 Monitor, are commonly employed as short-term integrating devices (two to seven days), while
1475 alpha track detectors are commonly used to provide measurements of average radon levels over
1476 periods of weeks to months.
1477 CHARCOAL CANISTERS
1478 Charcoal canisters (CC) are passive devices requiring no power to function. The passive nature
1479 of the activated charcoal allows continual adsorption and desorption of radon. During the
1480 measurement period, the adsorbed radon undergoes radioactive decay. Therefore, the technique
1481 does not uniformly integrate radon concentrations during the exposure period. As with all
1482 devices that store radon, the average concentration calculated using the mid-exposure time is
1483 subject to error if the ambient radon concentration adsorbed during the first half of the sampling
1484 period is substantially higher or lower than the average over the period. For a 2 to 7 day exposure
1485 period, the minimum detectable concentration (MDC) should be 18.5 Bq/m3 (0.5 pCi/L) or less
1486 (EPA, 1989). This detection level can normally be achieved with a counting time of up to 30
1487 minutes. This MDC should be calculated using the results of charcoal background
1488 determinations. The coefficient of variation should not exceed 10 percent (1 sigma) at radon
1489 concentrations of 148 Bq/m3 (4 pCi/L) or greater (EPA, 1989). This precision should be
1490 monitored using the results of duplicate canister analyses. CCs can achieve an average coefficient
1491 of variation of less than five percent at concentrations of 148 Bq/m3 (4 pCi/L) or greater.
1492 ELECTRET-PASSIVE ENVIRONMENTAL RADON MONITORS
1493 Electret-passive environmental radon monitors (E-perms) require no power and function as true
1494 integrating detectors that measure the average concentration during the exposure period. E-
1495 PERMS contain a permanently charged Electret (an electrostatically charged disk of Teflon) that
1496 collects ions formed in the chamber by radiation emitted from radon decay products. When the
1497 device is exposed, radon diffuses into the chamber through filtered openings. Ions that are
1498 generated continuously by the decay of radon and radon decay products are drawn to the surface
1499 of the electret and reduce its surface voltage. The amount of voltage reduction is related directly
1500 to the average radon concentration present during the exposure period. There are both short-term
1501 (2 to 7 days) and long-term (1 to 12 months) E-PERMS that are marketed currently. The
1502 thickness of the electret affects the usable measurement period. For a 7-day exposure period
1503 using a short-term E-PERM, as well as for a long-term E-PERM, the MDC is about 11.1 Bq/m3
1504 (0.3 pCi/L) (EPA, 1989). The coefficient of variation should not exceed 10 percent (1 sigma) at
1505 radon concentrations of 148 Bq/m3 (4 pCi/L) or greater. This precision should be verified by
1506 using results of duplicate detector analysis.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 10-45 DO NOT CITE OR QUOTE
-------
Field and Sampling Issues That Affect Laboratory Measurements
1507 ALPHA TRACK DETECTORS
1508 An alpha track detector (ATD) consists of a small piece of plastic or film enclosed in a container
1509 with a filter-covered opening. Radon diffuses through the filter into the container and alpha
1510 particles emitted by radon and its decay products strike the detector and produce submicroscopic
1511 damage tracks. At the end of the measurement period, the detectors are returned to a laboratory.
1512 Plastic detectors are placed in a caustic solution that accentuates the damage tracks so they can be
1513 counted using a microscope or an automated counting system. The number of tracks per unit area
1514 is correlated to the radon concentration in air, using a conversion factor derived from data
1515 generated at a calibration facility. The number of tracks produced per unit time is proportional to
1516 the radon concentration, so an ATD functions as a true integrating detector and measures the
1517 average concentration over the measurement period. The MDC and precision of an ATD system
1518 is dependent upon the tracks counted and, therefore, the area of the detector that is analyzed.
1519 With present ATDs, routine counting achieves a MDC of 6,660 Bq/m3-days (180 pCi/L-days).
1520 The coefficient of variation (precision) should be monitored using the results of duplicate
1521 detectors. The coefficient of variation should not exceed 20 percent (1 sigma) at radon
1522 concentrations of 148 Bq/m3 (4 pCi/L) or greater (EPA, 1989).
1523 10.5.5.2 Selecting a Radon Sampling Method Based on Data Quality Objectives
1524 The choice from among the sampling methods described above depends on whether the measure-
1525 ment is intended as a quick screening measurement or as a measurement that determines average
1526 exposure. In practice, the choice of a measurement system often is dictated by availability. If
1527 alternative systems are available, the cost or duration of the measurement may become the
1528 deciding factor. Each system has its own advantages and disadvantages, and the investigator
1529 must exercise some judgment in selecting the system best suited to the DQOs of the
1530 investigation.
1531 There are, however, some general guidelines concerning standardized measurement conditions
1532 and quality assurance objectives which apply to all measurement techniques. The following
1533 elements of quality assurance should be included in any measurement program: detector
1534 calibrations, replicate measurements, background measurements, and routine sensitivity checks.
1535 Detector calibrations are measurements made in a known radon environment, such as a
1536 calibration chamber. Detectors requiring laboratory readout, such as charcoal canisters and alpha-
1537 track detectors, should be exposed in the calibration chamber and then analyzed. Instruments
1538 providing immediate results, such as continuous working-level monitors and continuous radon
1539 monitors, should be operated in a chamber to establish calibration.
1540 There are two types of calibration measurements that should be made for alpha-track detectors
1541 and charcoal canisters. The first measurements determine and verify the conversion factors used
1542 to derive the concentration results. These measurements, commonly called spiked samples, are
MARLAP JULY 2001
DO NOT CITE OR QUOTE 10-46 DRAFT FOR PUBLIC COMMENT
-------
Field and Sampling Issues That Affect Laboratory Measurements
1543 done at the beginning of the measurement program and periodically thereafter. The second
1544 calibration measurements monitor the accuracy of the system. These are called blind calibration
1545 measurements and consist of detectors that have been exposed in a radon calibration chamber.
1546 The detectors are not labeled as such when sent to a processing laboratory.
1547 Background measurements, or blanks, should also be conducted. Such measurements should be
1548 made using unexposed passive detectors, or should be instrument measurements conducted in
1549 very low (outdoor) radon concentration environments and separated from the operating program.
1550 Generally, these should be equivalent in frequency to the spiked samples and should also not be
1551 identified as blanks when submitted for analysis to external laboratories. In addition to these
1552 background measurements, the organization performing the measurements should calculate the
1553 minimum detectable concentration MDC for the measurement system. This MDC is based on the
1554 system's background and can restrict the ability of some measurement systems to measure low
1555 concentrations.
1556 Duplicate measurements provide an estimate of the precision of the measurement results.
1557 Duplicate measurements should be included in at least 10 percent of the samples. If enough
1558 measurements are made, the number of duplicates may be reduced, as long as enough are used to
1559 analyze the precision of the method.
1560 A quality assurance program should include a written plan for satisfying the preceding
1561 objectives. A system for monitoring the results of the four types of quality assurance
1562 measurements should also be maintained.
1563 Calibrated radon detection devices and on-site measurements can also be obtained under contract
1564 from commercial vendors who have demonstrated their proficiency in measuring radon and
1565 radon decay products, and who have had their quality assurance programs assessed by the EPA or
1566 state agencies.
1567 10.6 Wipe Sampling for Assessing Surface Contamination
1568 Surface contamination falls into two categories: fixed and loose. The wipe test (also referred to
1569 as "swipes" or "smears") is the universally accepted technique for detecting removable
1570 radioactive contamination on surfaces (Section 12.5). It is often a stipulation of radioactive
1571 materials licenses and is widely used by laboratory personnel to monitor their work areas,
1572 especially for low-energy radionuclides that are otherwise difficult to detect with hand-held
1573 survey instruments. A comprehensive history of "Use of Smears for Assessing Removable
1574 Contamination" is presented by Frame and Abelquist (1999).
1575 The U.S. Nuclear Regulatory Commission (NRC, 1981) suggests that 100 cm2 areas be wiped
1576 and lists acceptable levels for surface contamination. However, NRC neither recommends the
1577 collection device nor the manner in which to conduct such surveys, relying instead on
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 10-47 DO NOT CITE OR QUOTE
-------
Field and Sampling Issues That Affect Laboratory Measurements
1578 suggestions by the National Committee on Radiation Protection (1964) and the National Council
1579 on Radiation Protection and Measurements (1978).
1580 10.6.1 Sample Collection Methods
1581 10.6.1.1 Dry Wipes
1582 Smears for removable surface activity are obtained by wiping an area of approximately 100 cm2
1583 using a dry filter paper, such as Whatman 50 or equivalent, while applying moderate pressure. A
1584 47 mm diameter filter is typically used, although for surveys for low-energy beta emitters,
1585 smaller sizes may be more appropriate because they can be placed directly into a liquid
1586 scintillation vial for counting. Small pieces of wipes occasionally are used for smears for tritium
1587 (Slobodine and Grandlund, 1974). A smear for removable contamination is obtained at each
1588 location of direct surface activity measurement.
1589 For surveys of small penetrations, such as cracks or anchor-bolt holes, cotton swabs are used to
1590 wipe the area of concern. Samples (smears or swabs) are placed into envelopes or other
1591 individual containers to prevent cross-contamination while awaiting analysis. Smears for alpha
1592 and medium- or high-energy beta activity can be evaluated in the field by counting them on an
1593 integrating sealer unit with appropriate detectors; the same detectors utilized for direct
1594 measurements may be used for this purpose. However, the more common practice is to return the
1595 smears to the laboratory, where analysis can be conducted using more sensitive techniques. The
1596 most common method for analyzing wipe samples is to use a proportional counter. For very low-
1597 energy beta emissions, wipe samples are commonly analyzed by liquid scintillation counting.
1598 10.6.1.2 Wet Wipes
1599 Although dry wipes are more convenient to handle, and there are fewer chances of cross
1600 contamination, a general limitation of dry wipes is their low recovery of surface contamination.
1601 The low recovery using dry wipes is due to the higher affinity for the surface by the contaminant
1602 than for the filter paper. Several studies have shown that for maximum sensitivity, a wipe
1603 material moistened with a suitable solvent may be indicated. For example, Ho and Shearer
1604 (1992) found that alcohol-saturated swabs were 100 times more efficient at removing
1605 radioactivity than dry swabs.
1606 In another study, Kline et al. (1992) assessed the collection efficiency of wipes from various
1607 surfaces that included vinyl floor tile, plate glass, and lead foil. Two different collection devices,
1608 cotton swabs and 2.5 cm diameter glass fiber filter disks, were evaluated under various collection
1609 conditions. Dry wipes were compared to collections made with the devices dampened with
1610 different amounts of either distilled H2O, 70 percent ethanol, or a working-strength solution of a
1611 multipurpose laboratory detergent known to be effective for removing contaminants from
1612 laboratory glassware (Manske et al., 1990).
MARLAP JULY 2001
DO NOT CITE OR QUOTE 10-48 DRAFT FOR PUBLIC COMMENT
-------
Field and Sampling Issues That Affect Laboratory Measurements
1613 The entire area of each square was manually wiped in a circular, inwardly-moving motion with
1614 consistent force. The collection capacity of each device was estimated by wiping progressively
1615 larger areas (multiple grids) and comparing the measured amounts of radioactivity with the
1616 amounts placed on the grids.
1617 Collection efficiency varied with both the wipe method and the surface wipe. Contamination was
1618 removed most readily from unwaxed floor tile and glass; lead foil released only about one-half
1619 the radioactivity. Stainless steel, another common laboratory surface, has contamination retention
1620 properties similar to those of glass.
1621 In most cases, collection was enhanced by at least a factor of two after dampening either the
1622 swabs or filter disks with water. Dampening with ethanol or the detergent produced removals that
1623 were statistically indistinguishable from samples dampened with an equal amount of water.
1624 The filter disks had a higher collection capacity for removable contaminants than cotton swabs,
1625 nearly doubling the radioactivity removed for each doubling of surface area wiped. Variability
1626 within all methods was high, with coefficients of variation ranging from 2 to 30 percent.
1627 For the moistened wipes, wipe efficiency depended on three factors, including the polarity of the
1628 solvent, the polarity of the contaminant being measured, and the affinity of the compound for the
1629 contaminated surface. For a solvent to readily dissolve a compound (i.e., remove it from the
1630 surface), the solvent and the compound must have similar polarities. Nonpolar solvents include
1631 ethyl acetate and petroleum ether; for polar solvents, water or methanol may be used (Cambell et
1632 al., 1993). There are other factors that influence the affinity of a compound for a surface,
1633 including porosity of the surface and available binding sites on the surface. One important factor
1634 which influences binding capacity is the type of treatment that a surface has received. When
1635 working with a surface treated with a nonpolar wax, such as that used on floor tile, a nonpolar
1636 compound will be adsorbed to the surface, which further limits recovery. In contrast, recovery
1637 from absorbent surfaces, such as lab bench paper or untreated wood, may give poor recoveries
1638 due to the porous nature of the surface.
1639 10.6.2 Sample Handling
1640 Filter paper or other materials used for wipe tests in the field should be placed in separate
1641 containers that prevent cross contamination during transport and allow for labeling of each
1642 sample. Plastic bags, paper or glassine envelopes, and disposable plastic petri dishes are
1643 containers typically used to store and transport wipe samples. Field workers can use plastic or
1644 rubber gloves and forceps when applying the wipe material to a surface and during handling as
1645 each wipe is placed into a container. Protection of the sample wipe surface is the main concern
1646 when a wipe must be placed in a container for transport. If a scintillation vial or planchet will be
1647 used in the lab, then a field worker may put wipes directly into them. Planchets containing loose
1648 or self-sticking wipes can also be put into self-sealing plastic bags to separate and protect the
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 10-49 DO NOT CITE OR QUOTE
-------
Field and Sampling Issues That Affect Laboratory Measurements
1649 integrity of the sample's surface. Excessive dust and dirt can cause self adsorption or quenching,
1650 and therefore should be minimized.
1651 To maintain constant geometry in an automatic proportional counter, it is important that the wipe
1652 remain flat during counting. Additionally, material that will curl can jam the automatic counter
1653 and cause cross contamination or even destroy the instrument window. When it is necessary to do
1654 destructive analysis on the wipe, it is critical that the wipe can easily be destroyed during the
1655 sample preparation step, and that the residue not cause interference problems.
1656 When wipes are put directly into liquid scintillation cocktail, it is important that the wipe not add
1657 color or react with the cocktail. For maximum counting efficiency, as well as reproducibility, the
1658 wipe should either dissolve or become translucent in the cocktail.
1659 10.7 References
1660 American National Standards Institute/American Nuclear Society (ANSI/ANS). 1994. Internal
1661 Dosimetry Programs for Tritium Exposure - Minimum Requirements, HP S N13.14.
1662 American National Standards Institute/American Nuclear Society (ANSI/ANS). 1995. Bioassay
1663 Programs for Uranium, HPS N13.22.
1664 American National Standards Institute/American Nuclear Society (ANSI/ANS). 1996.
1665 Performance Criteria for Radiobioassay, HPS N13.30.
1666 American National Standards Institute/American Nuclear Society (ANSI/ANS). 1997. Internal
1667 Dosimetry for Mixed Fission Activation Products, HPS N13.42.
1668 American National Standards Institute (ANSI). 1973. American National Standard for Radiation
1669 Protection in Uranium Mines. N13.8-1973.
1670 American National Standards Institute (ANSI). 1999. Sampling and Monitoring Releases of
1671 Airborne Radioactive Substances from the Stacks and Ducts of Nuclear Facilities. HPS
1672 N13.1.
1673 American Public Health Association (APHA). 1972. Intersociety Committee for a Manual of
1674 Methods for Ambient Air Sampling and Analysis, Methods of Air Sampling and Analysis.
1675 APHA, Washington, DC.
1676 American Public Health Association (APHA). 1995. Standard Methods for the Examination of
1677 Water and Wastewater. 19th Edition, APHA, Washington, DC.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 10-50 DRAFT FOR PUBLIC COMMENT
-------
Field and Sampling Issues That Affect Laboratory Measurements
1678 American Public Health Association (APHA). 1996. Standard Methods for the Examination of
1679 Water and Wastewater. 19th Edition Supplement, APHA, Washington, DC.
1680 American Society for Testing and Materials (ASTM) STP 555. Instrumentation for Monitoring
1681 Air Quality, 1974. West Conshohocken, Pennsylvania.
1682 American Society for Testing and Materials (ASTM) C998. Sampling Surface Soil for
1683 Radionuclides, 1995. West Conshohocken, Pennsylvania.
1684 American Society for Testing and Materials (ASTM) C999. Soil Sample Preparation for the
1685 Determination of Radionuclides, 1995. West Conshohocken, Pennsylvania.
1686 American Society for Testing and Materials (ASTM) D420. Site Characterization for
1687 Engineering, Design, and Construction Purposes., 1998. West Conshohocken, Pennsylvania.
1688 American Society for Testing and Materials (ASTM) D653. Terminology Relating to Soil, Rock,
1689 and Contained Fluids, 1991. West Conshohocken, Pennsylvania.
1690 American Society for Testing and Materials (ASTM) D3370, Standard Practices for Sampling
1691 Water from Closed Conduits. ASTM, West Conshohocken, Pennsylvania.
1692 American Society for Testing and Materials (ASTM) D3856. Good Laboratory Practices in
1693 Laboratories Engaged in Sampling and Analysis of Water, 1995. West Conshohocken,
1694 Pennsylvania.
1695 American Society for Testing and Materials (ASTM) D3977. Determining Sediment
1696 Concentration in Water Samples, 1991. West Conshohocken, Pennsylvania.
1697 American Society for Testing and Materials (ASTM) D4840. Sampling Chain-of-Custody
1698 Procedures, 1999. West Conshohocken, Pennsylvania.
1699 American Society for Testing and Materials (ASTM) D4914. Density of Soil and Rock in Place
1700 by the Sand Replacement Method in a Test Pit, 1999. West Conshohocken, Pennsylvania.
1701 American Society for Testing and Materials (ASTM) D4943. Shrinkage Factors of Soils by the
1702 Wax Method, 1995. West Conshohocken, Pennsylvania.
1703 American Society for Testing and Materials (ASTM) D5245. Cleaning Laboratory Glassware,
1704 Plasticware, and Equipment Used in Microbiological Analyses, 1998. West Conshohocken,
1705 Pennsylvania.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 10-51 DO NOT CITE OR QUOTE
-------
Field and Sampling Issues That Affect Laboratory Measurements
1706 American Society for Testing and Materials (ASTM) D5283. Generation of Environmental Data
1707 Related to Waste Management Activities Quality Assurance and Quality Control Planning
1708 and Implementation, 1991. West Conshohocken, Pennsylvania.
1709 American Society for Testing and Materials (ASTM) D5608. Decontamination of Field
1710 Equipment Used at Low Level Radioactive Waste Sites, 1994. West Conshohocken,
1711 Pennsylvania.
1712 Army Corps of Engineers (USAGE). 1995. Technical Project Planning Guidance for Hazardous,
1713 Toxic and Radioactive Waste (HTRW) Data Quality Design. Engineer Manual EM-200-1-2,
1714 Appendix H, Sampling Methods.
1715 Baratta, E.J., G.E. Chabot, and RJ. Donlen. 1968. Collection and Determination of Iodine-131 in
1716 the Air. Amk. Ind. Hyg. Assoc. J., 29:159.
1717 Bellamy, R.R. 1974. Elemental Iodine and Methyl Iodide Adsorption on Activated Charcoal at
1718 Low Concentrations. Nuclear Safety Volume 15, U.S. Atomic Energy Commission Technical
1719 Information Center, Oak Ridge, Tennessee.
1720 Bernabee, R. P., D. R. Percival, and D. B. Martin. 1980. Fractionation of Radionuclides in Liquid
1721 Samples from Nuclear Power Facilities. Health Physics 39:57-67.
1722 Bixel, J.C. and CJ. Kershner. 1974. A Study of Catalytic Oxidation and Oxide Adsorption for
1723 Removal of Tritium from Air. Proceedings of the 2nd AEC Environmental Protection
1724 Conference page 261, April 16-19, Report No. CONF-740406, WASH-1332 (74).
1725 Blanchard, R.L., R. Leiberman, W.S. Richardson in, and C.L. Wakamo. 1993. Considerations of
1726 Acidifying Water Samples for Tc-99 Analysis. Health Physics 65(2):214-215.
1727 Cambell, J.L., C.R. Santerre, P.C. Farina, and L.A. Muse. 1993. Wipe Testing for Surface
1728 Contamination by Tritiated Compounds. Health Phys. 64:540-544.
1729 Cohen, B. S. 1989. Sampling Airborne Radioactivity. Air Sampling Instruments for Evaluation of
1730 Atmospheric Contaminants, 7th edition, American Conference of Governmental Industrial
1731 Hygienists, Cincinnati, Ohio.
1732 Dehnam, D.H. 1972. Effectiveness of Filter Media for Surface Collection of Airborne
1733 Radioactive Particulates. Health Physics Operational Monitoring Vol. 2, Gordon and Breach,
1734 New York.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 10-52 DRAFT FOR PUBLIC COMMENT
-------
Field and Sampling Issues That Affect Laboratory Measurements
1735 Department of Energy (DOE). 1987. The Environmental Survey Manual, Appendices E, F, G, H,
1736 I, J, andK. DOE/EH-0053, Vol. 4 of 4, DOE, Office of Environmental Audit, Washington,
1737 DC.
1738 Department of Energy (DOE). 1990. EML Procedures Manual (HASL-300-Ed.27). G. de
1739 Planque Editor, Environmental Measurements Laboratory.
1740 Department of Energy (DOE). 1994a. Implementation Guide, Internal Dosimetry Program. G-10
1741 CFR835/Cl-Rev. 1.
1742 Department of Energy (DOE). 1994b. Implementation Guide, External Dosimetry Program. G-
1743 10CFR835/C2-Rev. 1.
1744 Department of Energy (DOE). 1994c. Implementation Guide, Workplace Air Monitoring. G-10
1745 CFR835/E2-Rev. 1.
1746 Department of Energy (DOE). 1994d. Radiological Control Manual. DOE/EH-0256T, Rev. 1.
1747 Department of Energy (DOE). 1997. EML Procedures Manual. HASL-300, 28th Edition,
1748 Environmental Measurements Laboratory.
1749 Department of the Interior (DOT). 1980. National Handbook of Recommended Methods for Water
1750 for Water-Data Acquisition, Volume I and II.
1751 Dyck, W. 1968. Adsorption of Silver on Borosilicate Glass. Anal. Chem. 40:454-455.
1752 Environmental Protection Agency. (1980). Prescribed Procedures for Measurement of
1753 Radioactivity in Drinking Water. EPA-600/4-80-032, EPA, Environmental Monitoring and
1754 Support Laboratory, Cincinnati, Ohio.
1755 Environmental Protection Agency (EPA). 1982. Handbook for Sampling and Sample
1756 Preservation of Water and Wastewater. EPA-600/4-82-029, EPA, Washington, DC. (PB83-
1757 124503)
1758 Environmental Protection Agency (EPA). 1984. Characterization of Hazardous Waste Sites- A
1759 Method Manual, Vol. II, Available Sampling Methods. EPA-600/4-84-076, Second Edition.
1760 Environmental Protection Agency (EPA). 1985. Sediment Sampling Quality Assurance User's
1761 Guide. EPA/600/4-85/048, Environmental Monitoring Systems Laboratory, Las Vegas, NV.
1762 (PB85-233542).
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 10-53 DO NOT CITE OR QUOTE
-------
Field and Sampling Issues That Affect Laboratory Measurements
1763 Environmental Protection Agency (EPA). 1986. Compendium of Methods for Determination of
1764 Superfund Field Operation Methods, EPA 600-4-87/006. Office of Emergency and Remedial
1765 Response, Washington, DC.
1766 Environmental Protection Agency (EPA). 1987. A Compendium of Superfund Field Operations
1767 Methods. EPA/540/P-87/001. Office of Emergency and Remedial Response, Washington,
1768 DC. (PB88-181557).
1769 Environmental Protection Agency (EPA). 1989. Indoor Radon and Radon Decay Product
1770 Measurement Protocols. Radon Division, Washington, DC.
1771 Environmental Protection Agency (EPA). 1992. Indoor Radon and Radon Decay Product
1772 Measurement Device Protocols. EPA 402-R-92-004, EPA, Office of Air and Radiation,
1773 Washington, DC.
1774 Environmental Protection Agency (EPA). 1993. Protocols for Radon and Radon Decay Product
1775 Measurements in Homes. EPA 402-R-92-003, EPA, Office of Air and Radiation,
1776 Washington, DC.
1777 Environmental Protection Agency (EPA). 1994. Routine Environmental Sampling Procedures
1778 Manual For Radionuclides. EPA, Office of Radiation and Indoor Air and National Air and
1779 Radiation Environmental Laboratory, Montgomery, AL.
1780 Environmental Protection Agency (EPA). 1996. Radon Proficiency Program - Handbook. EPA
1781 402-R-95-013, EPA, Office of Radiation and Indoor Air, Washington, DC.
1782 Environmental Protection Agency (EPA). 1997. To Filter or Not to Filter, That is the Question.
1783 EPA Science Advisory Board (SAB), July 11, 1997.
1784 Frame, P.W. and E.W. Ablequist. May 1999. Use of Smears for Assessing Removable
1785 Contamination. Operational Radiation Safety 67 (5):S57-S66.
1786 Francis, AJ. 1985. Low-Level Radioactive Wastes in Subsurface Soils. Soil reclamation
1787 Processes: Microbiological Analyses and Applications, NY.
1788 Friend, A.G., A.H Story, C.R. Henderson, and K.A. Busch. 1965. Behavior of Certain
1789 Radionuclides Released into Fresh- Water Environments. U. S. Public Health Service
1790 Publication999-RH-13.
1791 Gabay, J.J., CJ.Paperiello, S. Goodyear, J.C. Daly, and J.M. Matuszek. 1974. A Method for
1792 Determining Iodine-129 in Milk and Water. Health Physics 26:89.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 10-54 DRAFT FOR PUBLIC COMMENT
-------
Field and Sampling Issues That Affect Laboratory Measurements
1793 Hess, C.T. and S.M. Beasley. 1990. Setting Up a Laboratory for Radon in Water Measurments.
1794 Radon, Radium and Uranium in Drinking Water, Lewis Publishers, Chelsea, MI.
1795 Ho, S.Y. and D.R. Shearer. 1992. Radioactive Contamination in Hospitals from Nuclear
1796 Medicine Patients. Health Physics 62:462-466.
1797 Illinois Department of Nuclear Safety (IDNS). 1993. 1992 Annual Survey Report. Spingfield,
1798 Illinois.
1799 Illinois Department of Nuclear Safety (IDNS). 1994. 1993 Annual Survey Report. Spingfield,
1800 Illinois.
1801 Illinois Department of Nuclear Safety (IDNS). 1995. 1994 Annual Survey Report. Spingfield,
1802 Illinois.
1803 Illinois Department of Nuclear Safety (IDNS). 1996. 1995 Annual Survey Report. Spingfield,
1804 Illinois.
1805 Illinois Department of Nuclear Safety (IDNS). 1997. 1996 Annual Survey Report. Spingfield,
1806 Illinois.
1807 Institute of Nuclear Power Operations (INPO). 1988. Guidelines for Radiological Protection at
1808 Nuclear Power Stations. INPO 88-010, Atlanta, Georgia.
1809 Jackson, E.W. 1962. Prevention of Uptake of Strontium Ions on Glass. Nature 194: 672.
1810 Johnson, B.H. 1980. A Review of Numerical reservoir Hydrodynamic Modeling. U.S. Army
1811 Corps of Engineers, Waterways Experiment Station, Vicksburg, Mississippi.
1812 Keller, J.H., T.R. Thomas, D.T. Pence, and W.J. Maeck. 1973. An Evaluation of Materials and
1813 Techniques Used for Monitoring Airborne Radioiodine Species. Proceedings of the 12th AEC
1814 Air Cleaning Conference. U.S. Atomic Energy Commission, Washington, DC.
1815 Kennedy, V.C., G.W. Zellweger, and B.F. Jones. 1974. Filter Pore Size Effects on the Analysis
1816 of Al, Fe, Mn, and Ti in Water. Water Resources Research 10(4):785-790.
1817 Klebe, M. 1998. Illinois Department of Nuclear Safety. Correspondence of June 12, 1998 to Mr.
1818 J.C. Dehmel, SC&A, Inc., with copies of Tables 4 and 5 from survey questionnaires for the
1819 years of 1994 to 1997.
1820 Kline, R.C, I. Linins, E.L. Gershey. 1992. Detecting Removable Surface Contamination. Health
1821 Phys. 62:186-189.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 10-55 DO NOT CITE OR QUOTE
-------
Field and Sampling Issues That Affect Laboratory Measurements
1822 Kotrappa, P., S.K. Dua, D.P. Bhanti, and P.P. Joshi. 1974. HASL Cyclone as an Instrument for
1823 Measuring Aerosol Parameters for New Lung Model. Proceedings of the 3d International
1824 Congressional Radiation Protection Association, September 9-14, 1973.
1825 Kusnetz, H.L. 1956. Radon Daughers in Mine Atmospheres - A Field Method for Determing
1826 Concentrations. Am. Ind. Hyg. Assoc. QuarterlyVol. 17.
1827 Laxen, D.P.H. and I.M. Chandler. 1982. Comparison of Filtration Techniques for Size
1828 Distribution in Freshwaters. Analytical Chemistry 54(8): 1350-1355.
1829 Lippmann, M. 1989a. Calibration of Air Sampling Instruments. Air Sampling Instruments. 7th
1830 Edition, American Conference of Governmental Industrial Hygienists, Cincinnati, OH, pp.
1831 73-100.
1832 Lippmann, M. 1989b. Sampling Aerosols by Filtration. Air Sampling Instruments. 7th Edition,
1833 American Conference of Governmental Industrial Hygienists, Cincinnati, OH, pp. 305-336.
1834 Lockhart, L., R. Patterson and W. Anderson. 1964. Characteristics of Air Filter Media Used for
1835 Monitoring Airborne Radioactivity. Naval Research Laboratory Report NRL-6054,
1836 Washington, DC.
1837 Manske, P., T. Stimpfel, and E.L. Gershey. 1990. A Less Hazardous Chromic Acid Substitute for
1838 Cleaning Glassware. J. Chew. Educ. 67:A280-A282.
1839 MARSSEVI. 2000. Multi-Agency Radiation Survey and Site Investigation Manual, Revision 1.
1840 NUREG-1575 Rev 1, EPA 402-R-97-016 Revl, DOE/EH-0624 Revl. August. Available
1841 from http://www.epa.gov/radiation/marssim/filesfm.htm.
1842 Martin, I.E. and J.M. Hylko. 1987a. "Formation of Tc-99 in Low-Level Radioactive Waste
1843 Samples from Nuclear Plants." Radiation Protection Management, 4:6, p 67-71.
1844 Martin, I.E. and J.M. Hylko. 1987b. "Measurement of 99Tc in Low-Level Radioactive Waste
1845 from Reactors Using 99Tc as a Tracer." Applied Radiation and Isotopes, 38:6, p 447-450.
1846 McCarthy, J. F. and C. Deguelde. 1993. "Sampling and Characterization of Colloids and
1847 Particles in Ground Water for Studying Their Role in Contaminant Transport."
1848 Environmental Analytical and Chemistry Series. Environmental Particles, Vol. 2, Lewis
1849 Publishers, Chapter 6.
1850 Milkey, R.G. 1954. Stability of Dilute Solutions of Uranium, Lead, and Thorium Ions. Anal.
1851 Chem. 26(11): 1800-1803.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 10-56 DRAFT FOR PUBLIC COMMENT
-------
Field and Sampling Issues That Affect Laboratory Measurements
1852 National Academy of Sciences (NAS). 1960. The Radiochemistry of Technetium. Office of
1853 Technical Services, Washington, DC.
1854 National Committee on Radiation Protection. 1964. Safe Handling of Radioactive Materials.
1855 NCRP Report 30, Washington, DC.
1856 National Council on Radiation Protection and Measurements (NCRP). 1978. Instrumentation
1857 and Monitoring Methods for Radiation Protection. NCRP Report 57.
1858 National Council on Radiation Protection and Measurements (NCRP). 1985. A Handbook of
1859 Radioactivity Measurements Procedures. NCRP Report 81.
1860 National Council on Radiation Protection and Measurements (NCRP). 1987. Use ofBioassay
1861 Procedures for Assessment of Internal Radionuclides Deposition. NCRP Report No. 87.
1862 National Institue for Occupational Safety and Health (NIOSH). 1983. Industrial Hygiene
1863 Laboratory Quality Control-587. NIOSH, Cincinnati, Ohio.
1864 Navy Environmental Compliance Sampling and Field Testing Procedures Manual (NAVSEA).
1865 T0300-AZ-PRO-010.
1866 Nuclear Regulatory Commission (NRC). Acceptable Concepts, Models, Equations, and
1867 Assumptions for a Bioassay Program. NRC Regulatory Guide 8.9.
1868 Nuclear Regulatory Commission (NRC). Applications of Bioassay for Uranium. NRC
1869 Regulatory Guide 8.11.
1870 Nuclear Regulatory Commission (NRC). Applications of Bioassay for 1-125 and 1-131. NRC
1871 Regulatory Guide 8.20.
1872 Nuclear Regulatory Commission (NRC). Bioassays at Uranium Mills. NRC Regulatory Guide
1873 8.22.
1874 Nuclear Regulatory Commission (NRC). Applications of Bioassay for Fission and Activation
1875 Products. NRC Regulatory Guide 8.26.
1876 Nuclear Regulatory Commission (NRC). Criteria for Establishing a Tritium Bioassay Program.
1877 NRC Regulatory Guide 8.32.
1878 Nuclear Regulatory Commission (NRC). 1977. Estimating Aquatic dispersion of Effluents from
1879 A ccidental and Routine Reactor Releases for the Purpose of Implementing Appendix I. NRC
1880 Regulatory Guide 1.113.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 10-57 DO NOT CITE OR QUOTE
-------
Field and Sampling Issues That Affect Laboratory Measurements
1881 Nuclear Regulatory Commission (NRC). 1981. Radiation Safety Surveys at Medical Institutions.
1882 NRC Regulatory Guide 8.23.
1883 Nuclear Regulatory Commission (NRC). 1990. Model Feasibility Study of Radioactive Pathways
1884 From Atmosphere to Surface Water. NUREG/CR-5475, Washington, DC.
1885 Nuclear Regulatory Commission (NRC). 1992. National Profile on Commercially generated
1886 Low-Level Radioactive Mixed Waste. NUREG/CR-593 8, Washington, DC.
1887 Osborne, R.V. 1973. Sampling for Tritiated Water Vapor. Proceedings of the 3d International
1888 Congress. International Radiation Protection Association, CONF-730907-P2, 1973:1428-
1889 1433.
1890 Parker, H.M., R.F. Foster, I.L. Ophel, F.L. Parker, and W.C. Reinig. 1965. North American
1891 Experience in the Release of Low-Level Waste to Rivers and Lakes. Proceedings of the Third
1892 United National International Symposium on the Peaceful Uses of Atomic Energy Vol.
1893 14:62-71.
1894 Pelto, R.H., CJ. Wierdak, and V.A. Maroni. 1975. Tritium Trapping Kinetics in Inert Gas
1895 Streams. Liquid Metals Chemistry and Tritium Control Technology Annual Report ANL-7 5-
1896 50, p. 35 (Argonne National Laboratory, Lemont, IL).
1897 Phillips, I.E. and C.E. Easterly. 1982. Cold Trapping Efficiencies for Collecting Tritiated Water
1898 Entrained in a Gaseous Stream. Rev. Sci. Instrum. 53(1).
1899 Pignolet, L., F. Auvray, K. Fonsny, F. Capot, and Z. Moureau. 1989. Role of Various
1900 Microorganisms on Tc Behavior in Sediments. Health Physics 57(5):791-800.
1901 Puls, R.W., J.H. Eychaner, and R.M. Powell. 1990. Colloidal-Facilitated Transport of Inorganic
1902 Contaminants in Ground Water: Part I. Sampling Considerations. EPA/600/M-90/023.
1903 (NTISPB 91-168419).
1904 Puls, R. W., and R. M. Powell. 1992. Transport of Inorganic Colloids Through Natural Aquifer
1905 Material: Implications for Contaminant Transport. Environmental Science & Technology
1906 26(3):614-621.
1907 R. Rosson, R. Jakiel, S. Klima, B. Kahn, P.O. Fledderman. 2000. "Correcting Tritium
1908 Concentrations in Water Vapor with Silica Gel." Health Physics, 78:1, p 68-73. Also
1909 available at http://www.srs.gov/general/sci-tech/fulltext/ms9800901/ms9800901 .html.
1910 Saar, R. A. 1997. Filtration of Ground Water Samples: A Review of Industry Practice. GWMR,
1911 Winter, 1997, 56-62.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 10-58 DRAFT FOR PUBLIC COMMENT
-------
Field and Sampling Issues That Affect Laboratory Measurements
1912 Scarpitta, S.C. and N.H. Harley. 1990. Adsorption and Desorption of Noble Gases on Activated
1913 Charcoal. I. 133X Studies in a Monolayer and Packed Bed. Health Physics 59(4):383-392.
1914 Setter, L.R. and G.I. Coats. 1961. "The Determination of Airborne Radioactivity," Industrial
1915 Hygiene Journal, February, pp 64-69.
1916 Silva, RJ. and A.W. Yee. 1982. Geochemical Assessment of Nuclear Waste Isolation: Topical
1917 Report, Testing of Methods for the Separation of Soil and Aqueous Phases. Lawrence
1918 Berkeley Laboratory, Report LBL-14696, UC-70.
1919 Slobodine, MJ. and R.W. Granlund. 1974. Exuruded Expanded Polystyrene: A Smear Material
1920 for Use in Liquid Scintillation Counting. Health Physics 27:128-129.
1921 Stafford, R.B. 1973. Comparative Evaluation of Several Glass-Fiber Filter Media. Los Alamos
1922 Scientific Laboratory, LA-5297.
1923 Waite, D. A. and W.L. Nees. 1973. A Novel Particle Sizing Technique for Health Physics
1924 Application. Battelle, Pacific Northwest Laboratories, BNWL-SA-4658.
1925 Whittaker, E.L., J.D. Akridge, and J. Giovino. 1989. Two Test Procedures for Radon in Drinking
1926 Water: Interlaboratory Collaborative Study. EPA 600/2-87/082, Environmental Monitoring
1927 Systems Laboratory.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 10-59 DO NOT CITE OR QUOTE
-------
11 SAMPLE RECEIPT, INSPECTION, AND TRACKING
2 11.1 Introduction
3 This chapter provides guidance on laboratory sample receiving and screening, inspecting,
4 documenting custody, and assigning laboratory tracking numbers. These topics are presented in a
5 sequentially in this chapter, but they may be done in a different order. The chapter is directed
6 primarily at laboratory personnel (as are all of the Part II chapters), although the Project Manager
7 and field personnel need to be aware of the steps involved in sample receipt, inspection, and
8 tracking. For the purposes of MARLAP, the "sample receipt" process includes the screening of
9 the package and sample containers for radiological contamination. "Sample inspection" is used to
10 check the physical integrity of the package and samples, to confirm the identity of the sample, to
11 confirm field preservation (if necessary), and to record and communicate the presence of
12 hazardous materials. "Laboratory sample tracking" is a process starting with sample log-in and
13 assignment of a unique laboratory tracking number to be used to account for the sample through
14 analyses, storage, and shipment. Laboratory tracking continues the tracking that was initiated in
15 the field during sample collection.
16 Figure 11.1 presents an overview of the topics discussed in this chapter. Note that the flow
17 diagram in the field sample preparation chapter (Chapter 10, Field and Sampling Issues that
18 Affect Laboratory Measurements) leads into sample receipt. This chapter focuses on sample
19 receipt, inspection, and tracking of samples in the laboratory because these are the three modes of
20 initial control and accountability. Sample receipt and inspection activities need to be done in a
21 timely manner to allow the laboratory and field personnel to resolve any problems (e.g.,
22 insufficient material collected, lack of field preservation, etc.) with the samples received by the
23 laboratory as soon as is practical. An effective interface between field personnel and the
24 laboratory not only facilitates problem resolution but also prevents unnecessary delays in the
25 analytical process.
26 Other relevant issues, including the laboratory's license conditions and proper operating
27 procedures are also noted because these topics are linked to receipt, inspection, and tracking ities.
28 The end result of the sample receipt and inspection activities is to accept the samples as received
29 or to perform the necessary corrective action (which may include rejecting samples).
30 Health and safety information is not presented but can be found in NRC (1998a; 1998b).
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 11-1 DO NOT CITE OR QUOTE
-------
Sample Receipt, Inspection, and Tracking
Field Sample Shipment
(see Chapter 10)
Shipping Manifest
• Number and type of samples along
with field sample number
• Field processing and preservation
• Analysis reqested
Sample returned
or
disposed of
according to
appropriate regs
Short-term
sample storage
or sample
prep/analysis
laboratory
Sample received in designated area
• Authorized user notified for radiological
screening of package
• Check for evidence of breakage or
leakage of exterior of package then
sample containers — If yes,
radiological screening (swipe) and
decontaminate
• License requirements
• Corrective action if necessary
• COC procedures if required
In designated rad receiving/prep area:
• Check container labels against
shipping manifest
• Check sample content against
shipping manifest
• Check tamper seals
• Verify preservation against shipping
manifest
• Check preparation against shipping
manifest
• Corrective action if necessary
Any discrepancies in:
• Screening value
• Expected nuclide content
• Number and type of samples
will result in corrective action
FIGURE 11.1 — Overview of sample receipt, inspection, and tracking
MARLAP
DO NOT CITE OR QUOTE
11-2
JULY 2001
DRAFT FOR PUBLIC COMMENT
-------
Sample Receipt, Inspection, and Tracking
31 11.2 General Considerations
32 11.2.1 Communication Before Sample Receipt
33 Before the samples are received, the laboratory should know the relative numbers of samples that
34 will be received within a specific timeframe and the types of analyses that are expected for the
35 samples. Laboratory personnel should be provided with a contact in the field and with means of
36 contacting the person (telephone, FAX, e-mail). Communication between laboratory personnel
37 and project staff in the field allows the parties to coordinate activities, schedules, and sample
38 receipt. In particular, the Project Manager should provide to the laboratory special instructions
39 regarding the samples before shipment of samples. This information serves to notify the
40 laboratory of health and safety concerns and provides details that will affect analytical
41 procedures, sample disposition, etc. For example, without this communication, a laboratory
42 might receive a partial shipment and not realize that samples are missing. Furthermore, advance
43 communications allow laboratory staff to arrange for special handling or extra space for storage
44 should the need arise.
45 Planning for the samples to be received at the laboratory starts during the development of the
46 appropriate plan document and the statement of work (SOW) and continues through the
47 communication between the project staff in the field and the laboratory. For example, the
48 laboratory could pre-label and bar-code the appropriate containers to be used in the field. This
49 process would assist in assigning appropriate sample numbers for the laboratory tracking system,
50 which starts with sample receipt. The laboratory should instruct the field staff to place the
51 shipping manifest on the inside of the cooler lid for easy access and to include any other pertinent
52 information (field documentation, field screen information, etc.).
53 11.2.2 Standard Operating Procedures
54 A laboratory should have standard operating procedures (SOPs) for laboratory activities related
55 to sample receipt, inspection, and tracking. Some typical topics that might be addressed in
56 laboratory SOPs are presented in Table 11.1. For example, the laboratory should have an SOP
57 that describes what information should be included in the laboratory sample tracking system.
58 Laboratory SOPs should describe chain-of-custody procedures giving a comprehensive list of the
59 elements in the program such as signing the appropriate custody forms, storing samples in a
60 secure area, etc. (ASTM D4840; ASTM D5172; EPA, 1995).
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 11-3 DO NOT CITE OR QUOTE
-------
Sample Receipt, Inspection, and Tracking
61 TABLE 11.1 —Typical topics addressed in standard operating procedures related to sample receipt,
62 inspection, and tracking
63 Sample • Order and details for activities associated with receiving shipments of samples.
64 Receipt: • Screening methods.
65 Inspection: • pH measurement instructions.
• Confirm sample identification.
• Assign samples to laboratory information management system (LIMS).
• Check physical integrity.
• Identify/manage hazardous materials.
66 Tracking: • Ensure proper identification of samples throughout process.
• Procedures to quickly determine location and status of samples within laboratory.
• Maintain chain of custody and document sample handling during transfer from the field to
thg. !.?!?9^pry, then within the laboratory.
67 Custodian: • Execution^of responsibilities of the sample custodian.
68 Forms/Labels: • Examples of forms and labels used to maintain sample custody and document sample
handling in the lab.
69 The laboratory needs to establish corrective action guidelines (Section 11.3.3) as part of every
70 SOP for those instances when a nonconformance is noted. Early recognition of a nonconfor-
71 mance will allow the Project Manager and the laboratory more options for a quick resolution.
72 11.2.3 Laboratory License
73 Laboratory facilities with a few exceptions (e.g., certain DOE National Laboratories and DOD
74 laboratories) that handle radioactive materials are required to have a radioactive materials license
75 issued by the NRC or the Agreement State in which the laboratory operates. The radioactive
76 materials license lists the radionuclides that the laboratory can possess, handle, and store. In
77 addition, the license limits the total activity of specific radionuclides that can be in the possession
78 of the laboratory at a given time.
79 The laboratory needs to have specific information from the field staff to make sure they can
80 receive samples with the particular radionuclides expected to be present in the samples and that
81 the laboratories have the proper radioactive materials license. The information needed includes
82 the results of radiological field screening measurements. Both the laboratory and the Project
83 Manager need to be aware of the type of radionuclide(s) in the samples and the total number of
84 samples to be sent to the laboratory (this should be included in the appropriate plan document
85 and SOW prior to sampling).
MARLAP JULY 2001
DO NOT CITE OR QUOTE 11 -4 DRAFT FOR PUBLIC COMMENT
-------
Sample Receipt, Inspection, and Tracking
86 The laboratory is required by the license to maintain a current inventory of certain radioactive
87 materials present in the facility. The radioactive materials license also requires the laboratory to
88 develop and maintain a radiation protection plan (NRC, 1998b) that states how radioactive
89 samples will be received, stored, and disposed. The laboratory will designate an authorized user
90 (NRC, 1998b) to receive the samples. A Radiation Safety Officer (RSO) may be an authorized
91 user but not always. NRC (1998b) gives procedures for the receipt of radioactive samples during
92 working hours and non-working hours; part of these procedures are as follows:
93 During normal working hours, immediately upon receipt of any package of licensed material,
94 each package must be visually inspected for any signs of shipping damage such as crushed or
95 punctured containers or signs of dampness. Any obvious damage must be reported to the
96 RSO immediately. Do not touch any package suspected of leaking. Request the person
97 delivering the package to remain until monitored by the RSO.
98 Any packages containing radioactive material that arrive between (state times, e.g., 4:30 p.m.
99 and 7:00 a.m. or on Saturdays or Sundays) shall be signed for by the security guard (or other
100 designated trained individual) on duty and taken immediately to the designated receiving
101 area. Security personnel (or other designated trained individual) should unlock the door, place
102 the package in the designated secured storage area and re-lock the door.
103 Since certain packages of licensed material will have detectable external radiation, they
104 should be sent immediately to a designated storage area, where they will be checked for
105 contamination and external radiation level as soon as practical. They should not be allowed to
106 remain in the receiving area any longer than necessary, as they may be a source of exposure
107 for receiving personnel.
108 11.2.4 Sample Chain-of-Custody
109 "Sample chain-of-custody" (COC) is defined as a process whereby a sample is maintained under
110 physical possession or control during its entire life cycle, that is, from collection to disposal
ill (ASTM D4840—see Chapter 10). The purpose of COC is to ensure the security of the sample
112 throughout the process. COC procedures dictate the documentation needed to demonstrate that
113 COC is maintained. When a sample is accepted by the laboratory it is said to be in the physical
114 possession or control of the laboratory. ASTM D4840 says that a sample is under "custody" if it
115 is in possession or under control so as to prevent tampering or alteration of its characteristics.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 11-5 DO NOT CITE OR QUOTE
-------
Sample Receipt, Inspection, and Tracking
116 If the samples are transferred under COC the relinquisher and the receiver should sign the
117 appropriate parts of the COC form with the date and time of transfer. After receipt and inspection
118 the samples should be kept in a locked area or in an area with controlled access.
119 COC is not a requirement for all samples. COC is most often required when the sample data may
120 be used as legal evidence. The project plan should state whether COC will be required. The
121 paperwork received with the samples should also indicate whether COC has been maintained
122 from the time of collection and must be maintained in the laboratory. If the laboratory has been
123 informed that COC procedures should be followed, but it appears that appropriate COC
124 procedures have not been followed (before or after sample receipt at the laboratory) or there are
125 signs of possible sample tampering when the samples arrive, the Project Manager should be
126 contacted. The problem and resolution should be documented. Additional information on COC
127 can be found in EPA (1985).
128 11.3 Sample Receipt
129 Laboratory sample receipt occurs when a package containing samples is accepted, the package
130 and sample containers are screened for radiological contamination, and the physical integrity of
131 the package and samples is checked. Packages include the shipping parcel that holds the smaller
132 sample containers with the individual samples (see Section 11.3.2 on radiological screening).
133 Also note that topics and activities covered in Section 11.3 appear in a sequence but, in many
134 cases, these activities are performed simultaneously during initial receiving activities (i.e.,
135 package screening and observation of its physical integrity).
136 11.3.1 Package Receipt
137 Packages can be accepted only at a designated receiving area. Packages brought to any other
138 location by a carrier should be redirected to the appropriate receiving area. All packages labeled
139 RADIOACTIVE I, II, or in require immediate notification of the appropriate authorized user
140 (NRC, 1998b).
141 A sample packing slip or manifest is required and must be presented at the time of receipt, and
142 the approximate activity of the shipment should be compared to a list of acceptable quantities. If
143 known, the activity of each radionuclide contained in the shipment must be reviewed relative to
144 the total amount of that radionuclide currently on site to ensure that the additional activity will
145 not exceed that authorized by the NRC or Agreement State in the laboratory's license.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 11 -6 DRAFT FOR PUBLIC COMMENT
-------
Sample Receipt, Inspection, and Tracking
146 Screening measures described in Section 11.3.2 may indicate that the samples are more
147 radioactive than expected and that the radiation license limit may be exceeded. The laboratory
148 should take extra precautions with these samples, but the screening results should be verified.
149 The Federal, State, or local agency should be contacted immediately when verified license limits
150 are exceeded. The laboratory must respond quickly to stay in compliance with their license.
151 If the package is not accepted by the laboratory, the laboratory should follow corrective-action
152 procedures prescribed in the radiation materials license, the appropriate plan document (if this is
153 a reasonable possibility for the project), and the laboratory's SOPs.
154 11.3.2 Radiological Screening
155 In addition to ensuring compliance with the laboratory's license and verifying estimates of
156 radionuclide activity (Section 11.3.1), the radiological screening of packages during sample
157 receipt serves to identify and prevent the spread of external contamination. All packages
158 containing samples for analysis received by the laboratory should be screened for external
159 contamination and surface exposure rate. Exceptions may include known materials (types under
160 exclusion should be listed in the laboratory SOP) intended for analysis as: a) well-characterized
161 samples; b) bioassays; and c) radon and associated decay products in charcoal media. Screening
162 of packages and sample containers received in the laboratory should be conducted in accordance
163 with the laboratory's established, documented procedures and the laboratory radiation protection
164 and health and safety plan. The exterior of the package is screened first; if there is no evidence of
165 contamination or that the laboratory licence would be exceeded, the package is opened up and the
166 sample containers screened individually. These procedures should include the action level and
167 appropriate action as established by the facility. Personnel performing screening procedures
168 should be proficient in the use of portable radiation screening instruments and knowledgeable in
169 radiological contamination control procedures. Health and safety considerations are affected by
170 the suspected or known concentrations of radionuclides in a sample or the total activity of a
171 sample.
172 Radiation screening is normally conducted using Geiger-Mueller (GM) detectors, ionization
173 chambers, micro-R meters, or alpha scintillation probes, as appropriate. The laboratory should
174 refer to any information they obtained before receipt of samples or with the samples, especially
175 concerning the identity and concentration of radioactive and chemical constituents in the
176 samples. Radiological screening needs to be performed as soon as practical after receipt of the
177 package, but not later than three hours (10 CFR 20.1906) after the package is received at the
178 licensee's facility for packages received during normal working hours. For packages received
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 11-7 DO NOT CITE OR QUOTE
-------
Sample Receipt, Inspection, and Tracking
179 outside of normal working hours, the screening must be performed no later than three hours from
180 the beginning of the next workday.
181 Monitor the exterior of a labeled package for radioactive contamination (10 CFR 20.1906). If the
182 package is small (less than 100 cm2), the whole package should be wiped. Wipes are not always
183 used, but if there is reason to believe that something has leaked, then wipes should be used. An
184 external exposure rate determination of the package is also required within three hours after the
185 package is received (or three hours from beginning of the next business day for packages
186 received outside of normal working hours). This screening is performed to detect possible
187 violations of Department of Transportation (DOT) packaging and labeling regulations, as well as
188 to determine the possible presence of gamma- and some beta-emitting radionuclides that may
189 require special handling. Also, screening can help to avoid introducing a high-activity sample
190 into a low-activity area.
191 The Consolidated Guidance About Materials Licenses (NRC 1998b) gives the following sample
192 model for opening packages containing radioactive material:
193 • Wear gloves to prevent hand contamination.
194 • Visually inspect the package for any sign of damage (e.g. crushed, punctured). If damage is
195 noted, stop and notify the RSO.
196 • Check DOT White I, Yellow II, or Yellow HI label or packing slip for activity of contents, so
197 shipment does not exceed license possession limits.
198 • Monitor the external surfaces of a labeled package according to specifications in Table 8.4,
199 Section 13.14, Item 10.
200 • Open the outer package (following supplier's directions if provided) and remove packing
201 slip. Open inner package to verify contents (compare requisition, packing slip and label on
202 the bottle or other container). Check integrity of the final source container (e.g., inspecting
203 for breakage of seals or vials, loss of liquid, discoloration of packaging material, high count
204 rate on smear). Again check that the shipment does not exceed license possession limits. If
205 you find anything other than expected, stop and notify the RSO.
206 • Survey the packing material and packages for contamination before discarding. If
207 contamination is found, treat as radioactive waste. If no contamination is found, obliterate the
208 radiation labels prior to discarding in the regular trash.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 11-8 DRAFT FOR PUBLIC COMMENT
-------
Sample Receipt, Inspection, and Tracking
209 • Maintain records of receipt, package survey, and wipe test results.
210 • Notify the final carrier and by telephone, telegram, mailgram, or facsimile, the Administrator
211 of the appropriate NRC Regional Office listed in 10 CFR 20, Appendix D when removable
212 radioactive surface contamination exceeds the limits of 10 CFR 71.87(i); or external radiation
213 levels exceed the limits of 10 CFR 71.47.
214 11.3.3 Corrective Action
215 The laboratory's SOPs should specify corrective actions for routine and non-routine sample
216 problems, including deficiency in sample volume, leaking samples, and labeling errors. The
217 appropriate corrective action may require consulting the Project Manager and other laboratory
218 personnel. Timely response can allow for a broader range of options and minimize the impact of
219 the sample problem on the project. The laboratory should document the problem, the cause (if
220 known), the corrective action taken, and the resolution of each problem that requires corrective
221 action. The documentation should be included in the project files.
222 11.4 Sample Inspection
223 After sample receipt, the next steps are to confirm that the correct sample has been sent, to check
224 that the appropriate field preservation and processing have been performed, and to identify any
225 hazardous chemicals.
226 Documents accompanying the samples should be reviewed upon receipt of the samples at the
227 laboratory. If the proper paperwork is not present, the Project Manager should be notified. Data
228 recorded on the paperwork, such as collection dates, sample descriptions, requested analyses, and
229 field staff personnel, should be compared to data on the sample containers and other documen-
230 tation. Any deficiencies or discrepancies should be recorded by the laboratory and reported to the
231 Project Manager. The documents can provide data useful for health and safety screening,
232 tracking, and handling/processing of critical short-lived radionuclides.
233 11.4.1 Physical Integrity of Package and Sample Containers
234 This section discusses checking for leakage or breakage and tampering of packages and sample
235 containers. Sample containers should be thoroughly inspected for evidence of sample leakage.
236 Leakage can result from a loose lid, sample container puncture, or container breakage. Packages
237 suspected to contain leaking sample containers should be placed in plastic bags. The authorized
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 11-9 DO NOT CITE OR QUOTE
-------
Sample Receipt, Inspection, and Tracking
238 user or alternate authorized user must be notified immediately for assistance. If leakage has
239 occurred, appropriate radiological and chemical contamination controls should be implemented.
240 Sample materials that have leaked or spilled are normally not suitable for analysis and should be
241 properly disposed. In all cases, the laboratory's management and Project Manager should be
242 notified of leaks, breakage, spills, and the condition of sample materials that remain in the
243 original containers.
244 Containers that have leaked from a loose lid or puncture may still hold enough sample for the
245 requested analyses. The laboratory must first determine if there is sufficient sample and if this
246 material is representative of the original sample. An assessment should be made to determine the
247 quantity of sample that remains and if this material is likely to be contaminated. If the sample
248 was contaminated with the analyte of interest at the time when the container leaked, the sample is
249 normally not analyzed. Unless appropriate information is provided in the project plan or SOW,
250 the Project Manager should determine whether or not the sample materials can be used for
251 analysis or if new samples are required to replace those lost due to leakage or contamination.
252 Packages, cooler chests, or individual sample containers may arrive at the laboratory bearing
253 custody seals. These seals provide a means to detect unauthorized tampering. When packages or
254 samples arrive with custody seals, they should be closely inspected for evidence of tampering.
255 Custody seals are made from material that cannot be removed without tearing. If a custody seal is
256 torn or absent, sample tampering may have occurred. This evidence of possible tampering is
257 generally sufficient to preclude use of the sample for laboratory analyses. The Project Manager
258 should be notified of the condition of the custody seal to determine if new samples are needed.
259 Observations regarding the condition of the custody seals should be recorded according to the
260 laboratory's standard procedures.
261 11.4.2 Sample Identity Confirmation
262 Visual inspection is the means to confirm that the correct sample has been received. Verifying
263 the identity of a sample is a simple process where the appearance, sample container label, and
264 chain-of-custody record or shipping manifest are compared. If all three sources of information
265 identify the same sample, then the sample is ready for the next step. If the sample label indicates
266 the sample is a liquid and the container is full of soil, this discrepancy would indicate a
267 nonconformance. If the sample label states that there is 1,000 mL of liquid and there only appears
268 to be 200 mL in the container, there may be a nonconformance. Visual inspection can be used to:
269 • Verify identity of samples by matching container label IDs and sample manifest IDs;
MARLAP JULY 2001
DO NOT CITE OR QUOTE 11-10 DRAFT FOR PUBLIC COMMENT
-------
Sample Receipt, Inspection, and Tracking
270 • Verify that the samples are as described by matrix and quantity;
271 • Check the tamper seal (if used);
272 • Verify field preparation (for example, filtering, removing extraneous material ), if indicated;
273 and
274 • Note any changes to samples since shipping, such as a reaction with the preservative.
275 11.4.3 Confirmation of Field Preservation
276 For those liquid samples requiring acid preservation, pH measurements may be performed on all
277 or selected representative liquid samples to determine if acid has been added as a preservative.
278 The temperature of the sample may also be part of field preservation and the actual measured
279 temperature should be compared to the specified requirements in the documentation.
280 11.4.4 Presence of Hazardous Materials
281 The presence of hazardous materials in a sample typically creates the need for additional health
282 and safety precautions when handling, preparing, analyzing, and disposing samples. If there is
283 documentation on the presence of non-radiological hazardous constituents, the Project Manager
284 should notify the laboratory about the presence of these chemicals. These chemical contaminants
285 should be evaluated by the laboratory to determine the need for special precautions. The
286 laboratory can also perform preliminary sample screening for chemical contaminants using
287 screening devices such as a photoionization detector for volatile components. The presence of
288 suspected or known hazardous materials in a sample should be identified, if possible, during
289 project planning and documented in the plan document and SOW. Visual inspection can also be
290 used such as checking the color of the sample (i.e., a green-colored water sample may indicate
291 the presence of high chromium levels). The presence of suspected or known hazardous materials
292 determined in the field should be communicated to the laboratory prior to the arrival of samples
293 and noted on documentation accompanying the samples to the laboratory. If no documentation on
294 non-radiological hazardous constituents is available, the laboratory should review previous
295 experience concerning samples from the site to assess the likelihood of receiving samples with
296 chemical contaminants. The laboratory should notify the Health and Safety Officer and the
297 Project Manager about the presence of potentially hazardous chemical contaminants.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 11-11 DO NOT CITE OR QUOTE
-------
Sample Receipt, Inspection, and Tracking
298 11.4.5 Corrective Action
299 Visual inspection can also verify whether field sample preparation was performed as stated in
300 accompanying documentation. Samples that were not filtered in the field or that reacted with the
301 preservative to form a precipitate may represent a significant problem to the laboratory. If it
302 appears that the sample was filtered in the field (i.e., there is a corresponding filter sample for the
303 liquid sample), the liquid generally will be analyzed as originally specified. Laboratory personnel
304 should check the project plan or SOW to see if the filter and filtered materials require analyses
305 along with the filtered sample. If it appears that the sample was not filtered in the field (i.e., there
306 is no corresponding filter, there are obviously solid particles in a liquid sample), sample
307 documentation should be reviewed to determine if a deviation from the project plan was
308 documented for the sample. It may be appropriate to filter the sample in the laboratory. The
309 Project Manager should be notified immediately to discuss possible options such as filtering the
310 sample at the laboratory or collecting additional samples.
311 One example of a corrective action for inspection is, if the pH is out of conformance, it may be
312 possible to obtain a new sample. If it is not possible or practical to obtain a new sample, it may
313 be possible to acidify the sample in the laboratory.
314 Visual inspection can serve to check certain aspects of sample collection. For example, if the
315 SOP states that a soil sample is supposed to have twigs, grass, leaves, and stones larger than a
316 certain size removed during sample collection and some of this foreign material is still included
317 as part of the sample, this discrepancy results in a nonconformance.
318 11.5 Laboratory Sample Tracking
319 Sample tracking should be done to ensure that analytical results are reported for the "correct"
320 sample. A good sample tracking system helps to prevent sample mix-up. Sample tracking is a
321 process by which the location and status of a sample can be identified and documented. The
322 laboratory is responsible for sample tracking starting with receipt (at which time a unique
323 laboratory tracking number is assigned), during sample preparation, and after the performance of
324 analytical procedures until final sample disposition. The process of sample tracking begins the
325 moment a field worker assigns an identification number (based on the information provided in
326 the appropriate plan document) and documents how materials are collected. The way samples are
327 transported from the field to the laboratory should be documented. The sample receiving
328 procedures and documentation should be consistent when applicable with 10 CFR Part 20
329 Subpart J, and the client's requirements as stated in the appropriate plan document or statement
330 of work.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 11-12 DRAFT FOR PUBLIC COMMENT
-------
Sample Receipt, Inspection, and Tracking
331 11.5.1 Sample Log-In
332 Laboratory sample numbers should be assigned to each sample in accordance with the
333 laboratory's SOP on sample codes. Each sample should receive a unique tracking number by
334 which it can be logged into the laboratory tracking system, scheduled for analysis, tracked, and
335 disposed. Information to be recorded during sample log-in should include the field sample
336 identification number, laboratory sample tracking number, date and time samples were collected
337 and received, reference date for decay calculations, method of shipment, shipping numbers,
338 condition of samples, requested analyses, number and type of each sample, quality control
339 requirements, special instructions, and other information relevant to the analyzing and tracking of
340 samples at the laboratory. Laboratory sample tracking is a continuation of field sample tracking.
341 Documents generated for laboratory sample tracking must be sufficient to verify the sample
342 identity, that the sample may be reliably located, and that the right sample is analyzed for the
343 right analyte. The documentation should include sample log-in records, the analysis request form,
344 names of staff responsible for the work, when procedures are completed, and details concerning
345 sample disposal. The documentation must conform to the laboratory's SOPs.
346 During sample log-in, laboratory quality control (QC) samples may be scheduled for the analyses
347 requested. The type and frequency of QC samples should be provided by the plan document or
348 SOW and consistent with the laboratory's SOPs.
349 11.5.2 Sample Tracking During Analyses
350 At this point, samples are introduced into the laboratory's analytical processing system. The
351 information gathered during screening, along with the assigned tracking identification, passes to
352 the laboratory where specific preparation and analyses are performed. The sample may be further
353 sub-sampled. Each sub-sample, along with the original sample, requires tracking to account for
354 all materials handled and processed in the laboratory.
355 At the same time that samples are received at the laboratory, each set of samples should be
356 accompanied by documents listing requests for specific analyses. This documentation should be
357 compared to separate paperwork obtained before sample receipt. Laboratory management
358 personnel should be notified of any discrepancies. The requested analyses should be entered into
359 the laboratory's tracking system. Typically, only one sample container of sufficient volume or
360 quantity will be provided for a single or multiple set of different analyses. Each aliquant removed
361 from the original container may require tracking (and perhaps a different tracking number).
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 11-13 DO NOT CITE OR QUOTE
-------
Sample Receipt, Inspection, and Tracking
362 Aliquants used during the analytical process can be tracked using analysis laboratory notebooks,
363 forms, or bench sheets that record laboratory tracking numbers, analyte, reference date for decay
364 correction, aliquant size, and designated quality control samples. Bench sheets are loose-leaf or
365 bound pages used to record information during laboratory work. Bench sheets are used to assist
366 in sample tracking. Each sheet is helpful for identifying and processing samples in batches that
367 include designated quality control samples. The bench sheet, along with the laboratory log book,
368 can later be used to record analytical information for use during the data review process. Bench
369 sheets can also be used to indicate that sample aliquants were in the custody of authorized
370 personnel during the analytical process.
371 After receipt, verification of sample information and requested analyses, and assignment of
372 laboratory sample tracking numbers, the requested analyses can be scheduled for performance in
373 accordance with laboratory procedures. Using this system, the laboratory can formulate a work
374 schedule, and completion dates can be projected.
375 11.5.3 Storage of Samples
376 If samples are to be stored and analyzed at a later date, they must be placed in a secure area that
377 meets all custody requirements. Before storage, any special preservation requirements, such as
378 refrigeration or additives, should be determined.
379 The laboratory should keep records of the sample identities and the location of the sample
380 containers. Unused sample aliquants should be returned to the storage area for final disposition.
381 In addition, for some samples, depending on the level of radioactivity or hazardous constituents
382 present, the laboratory must record when the sample was disposed and the location of the
383 disposal facility. These records are necessary to ensure compliance with the laboratory's license
384 for radioactive materials and other environmental regulations.
385 Areas where samples are stored must be designated and posted as radioactive materials storage
386 areas. Depending on the activity level of the samples, storage areas may require special posting.
387 If additional storage space or shielding is needed, arrangements that are consistent with the
388 license must be made with the authorized user. See Chapter 20 on waste disposal for more
389 information.
390 11.6 References
391 American Society of Testing and Materials (ASTM) D4840. Standard Guide for Sampling
3 92 Chain-of-Custody Procedures.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 11-14 DRAFT FOR PUBLIC COMMENT
-------
Sample Receipt, Inspection, and Tracking
393 American Society of Testing and Materials (ASTM) D5172. Standard Guide for Documenting
3 94 the Standard Operating Procedures Used for the Analysis of Water.
395 U.S. Environmental Protection Agency (EPA). 1985. NEIC Policies and Procedures. EPA-
396 300/9-78DDI-R, June.
397 U.S. Environmental Protection Agency (EPA). 1995. QA/G-6, Guidance for the Preparation of
398 Standard Operating Procedures (SOPS) for Quality-Related Documents.
399 U.S. Nuclear Regulatory Commission. 1998b. Consolidated Guidance About Materials Licenses,
400 Volume 7. (NRC91). NUREG 1556.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 11-15 DO NOT CITE OR QUOTE
-------
12 LABORATORY SAMPLE PREPARATION
2 12.1 Introduction
3 On first impression, sample preparation may seem the most mundane aspect of an analytical
4 protocol. However, it is critical that the analyst realize and remember that a determination is only
5 as good as the sample preparation that has preceded it. If an aliquant taken for analysis does not
6 represent the original sample accurately, the results of this analysis are questionable. As a general
7 rule, the error in sampling and the sample preparation portion of an analytical procedure is
8 considerably higher than that in the methodology itself, as illustrated in Figure 12.1.
Sampling
Concentration,
Isolation, et
, \ /
1 \
\
SeparationX /
c. Steps \
. \
/
\ / Sample
/ Preparation
/
FIGURE 12.1—Degree of error in laboratory sample preparation (Scwedt, 1997)
9 One goal of laboratory sample preparation is to provide, without sample loss, representative
10 aliquants that are free of laboratory contamination that will be used in the next steps of the
11 protocol. Samples are prepared in accordance with applicable standard operating procedures
12 (SOPs) and laboratory SOPs using information provided by field sample preparation (Chapter 10,
13 Field and Sampling Issues that Affect Laboratory Measurements)., sample screening activities,
14 and objectives given in the appropriate planning documents. The laboratory sample preparation
15 techniques presented in this chapter include the physical manipulation of the sample (heating,
16 screening, grinding, mixing, etc.) up to the point of dissolution. Steps such as adding carriers
17 and tracers, followed by wet ashing or fusion, are discussed in Chapter 13 (Sample Dissolution)
18 and Chapter 14 (Separation Techniques).
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 12-1 DO NOT CITE OR QUOTE
-------
Laboratory Sample Preparation
19 This chapter presents some general guidance for sample preparation on the avoidance of sample
20 loss and of sample contamination. Owing to the physical nature of the media, sample preparation
21 for solids requires the most attention, and therefore is discussed at great length (Section 12.3).
22 General procedures for preparing solid samples (such as drying, obtaining a constant weight,
23 grinding, sieving, mixing, and subsampling) are discussed. Some sample preparation procedures
24 then are presented for typical types of solid samples (e.g., soil and sediment, biota, vegetation
25 including food, etc.). This chapter concludes with specific guidance for preparing samples of
26 filters (Section 12.4), wipes (Section 12.5), liquids (Section 12.6), gases (Section 12.7), and
27 bioassay (Section 12.8).
28 12.2 General Guidance for Sample Preparation
29 Some general considerations during sample preparation are to minimize sample losses and to
30 prevent contamination. Possible mechanisms for sample loss during preparation steps are
31 discussed in Section 12.2.1, and the contamination of samples from sources in the laboratory is
32 discussed in Section 12.2.2. Control of contamination through cleaning labware is important and
33 described in Section 12.2.3, and laboratory contamination control is discussed in Section 12.2.4.
34 12.2.1 Potential Sample Losses During Preparation
35 Materials may be lost from a sample during laboratory preparation. The following sections
36 discuss the potential types of losses and the methods used to control them. The addition of tracers
37 or carriers (Chapter 13) is encouraged at the earliest possible point and prior to any sample
38 preparation step where there might be a loss of analyte. Such preparation steps may include
39 homogenization or sample heating. The addition of tracers or carriers prior to these steps helps to
40 account for any analyte loss during sample preparation.
41 12.2.1.1 Losses as Dust or Particulates
42 When a sample is dry ashed, a fine residue (ash) is often formed. The small particles in the
43 residue are resuspended readily by any flow of air over the sample. Air flows are generated by
44 changes in temperature (e.g., opening the furnace while it is hot) or by passing a stream of gas
45 over the sample during heating to assist in combustion. These losses are minimized by ashing
46 samples at as low a temperature as possible, gradually increasing and decreasing the temperature
47 during the ashing process, using a slow gas-flow rate, and never opening the door of a hot
48 furnace (Section 12.3.1). If single samples are heated in a tube furnace with a flow of gas over
49 the sample, a plug of glass or quartz wool can be used to collect particulates or an absorption
MARLAP JULY 2001
DO NOT CITE OR QUOTE 12-2 DRAFT FOR PUBLIC COMMENT
-------
Laboratory Sample Preparation
50 vessel can be used to collect volatile materials. At a minimum, all ash or finely ground samples
51 should be covered before they are moved.
52 Solid samples are often ground to a fine particle size before they are fused or wet ashed (see
53 Chapters 13 and 14 on dissolution and separation) to increase the surface area and speed up the
54 reaction between the sample and the fluxing agent or acid. Since solid samples are frequently
55 heterogeneous, a source of error arises from the difference in hardness among the sample
56 components. The softer materials are converted to smaller particles more rapidly than the harder
57 ones, and therefore, any loss in the form of dust during the grinding process will alter the
58 composition of the sample. The finely ground particles are also susceptible to resuspension.
59 Samples may be moistened carefully with a small amount of water before adding other reagents.
60 Reagents should be added slowly to prevent losses as spray owing to reactions between the
61 sample and the reagents.
62 12.2.1.2 Losses Through Volatilization
63 Some radionuclides are volatile under specific conditions (e.g., heat, grinding, strong oxidizers),
64 and care should be taken to identify samples requiring analysis for these radionuclides. Special
65 preparation procedures should be used to prevent the volatilization of the radionuclide of interest.
66 The loss of volatile elements during heating is minimized by heating without exceeding the
67 boiling point of the volatile compound. Ashing aids can reduce losses by converting the sample
68 into less volatile compounds. These reduce losses but can contaminate samples. During the wet
69 ashing process, losses of volatile elements can be minimized by using a reflux condenser. If the
70 solution needs to be evaporated, the reflux solution can be collected separately. Volatilization
71 losses can be prevented when reactions are carried out in a sealed tube or a closed metal or
72 Teflon™ bomb. Table 12.1 lists some commonly analyzed radioisotopes, their volatile chemical
73 form, and the boiling point of that species at standard pressure (note that the boiling point may
74 vary based on solution, matrix, etc.).
75 Often the moisture content, and thus, the chemical composition of a solid is altered during
76 grinding and crushing (Dean, 1995). Decreases in water content are sometimes observed while
77 grinding solids containing essential water in the form of hydrates, likely as a result of localized
78 heating. (See Section 12.3.1.2 for a discussion of the types of moisture present in solid samples.)
79 Moisture loss is also observed when samples containing occluded water are ground and crushed.
80 The process ruptures some of the cavities, and exposes the water to evaporation. More
81 commonly, the grinding process results in an increase in moisture content owing to an increase in
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 12-3 DO NOT CITE OR QUOTE
-------
Laboratory Sample Preparation
82
83
84
85
86
87
surface area available for absorption of atmospheric water. Both of these conditions will affect
the analysis of 3H since 3H is normally present in environmental samples as 3HOH.
TABLE 12.1 — Examples of Volatile Radionuclides
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
Isotope
Chemical Form
Boiling Point (°C)
Tritium — 3H
Carbon — 14C
Iodine —131I,129I
Cesium — 134Cs, 135Cs,
137Cs
Ruthenium — 106Ru
Technetium — "Tc
H2O
CO2 (produced from CO3"2 or
oxidation of organic material)
Cs metal
Cs oxides
(nitrates decompose to oxides)
CsCl
RuO4
RuCl3-;sH2O
Tc207
TcCl4 _
100
-78.5
185.2 (sublimes readily)
678.4
-650
1290
40
decomposes above 500
310.6
Sublimes above 300
Source: Greenwood and Earnshaw (1984).
Additional elements that volatilize under specific conditions include arsenic, antimony, tin,
polonium, lead, selenium, mercury, germanium, and boron. Osmium is volatilized as the
tetroxide under oxidizing conditions similar to those for ruthenium. Carbon, phosphorus, and
silicon may be volatilized as hydrides, and chromium is volatilized under oxidizing conditions in
the presence of chloride.
12.2.1.3 Losses Owing to Reactions Between Sample and Container
Specific elements may be lost from sample materials owing to interaction with a container. Such
losses may be significant, especially for trace analyses used in radioanalytical work. Adsorption
reactions are discussed in Chapter 10 for glass and plastic containers. Losses owing to adsorption
may be minimized by using pretreated glassware with an established hydrated layer. Soaking new
glassware overnight in a dilute nitric or hydrochloric acid solution will provide an adequate
hydrated layer. Glassware that is used on a regular basis will already have established an
adequate hydrated layer. The use of strong acids to maintain a pH less than one also helps
minimize losses from adsorption.
MARLAP
DO NOT CITE OR QUOTE
12-4
JULY 2001
DRAFT FOR PUBLIC COMMENT
-------
Laboratory Sample Preparation
108 Reactions among analytes and other types of containers are described in Table 12.2. Leaving
109 platinum crucibles uncovered during dry ashing to heat samples will minimize reduction of
110 samples to base metals which form alloys with platinum. It is recommended that porcelain not be
ill used for analysis of lead, uranium, and thorium because the oxides of these elements react with
112 porcelain glazes. Increasing the amount of sample for dry ashing increases the amount of ash,
113 allowing trace elements to react with the ash instead of with the container.
114
115
116
117
118
119
120
TABLE 12.2 — Properties of Sample Container Materials
Material Recommended Use Properties
121
122
123
124
125
126
Borosilicate
Glass
Fused Quartz
Porcelain
Platinum
Nickel
Zirconium
Alumina
(A1203)
Polyethylene
Teflon™
General applications Transparent; good thermal properties; fragile; attacked by HF, H3PO4, and
alkaline solutions.
High temperature
applications
High temperature
applications
High temperature or
corrosive
applications
Molten alkali metal
hydroxide and na2o2
fusions
Peroxide fusions
Acids and alkali
melts at low
temperatures
Sample and reagent
storage
Corrosive
applications
Transparent; excellent thermal properties (up to 1100° C); fragile; more
expensive than glass; attacked by HF, H3PO4, and alkaline solutions.
Used at temperatures up to 1100° C; less expensive than quartz; difficult to
shape; attacked by HF, H3PO4, and alkaline solutions.
Virtually unaffected by acids, including HF; dissolves readily in mixtures of
HNO3 and HC1, C12 water or Br2 water; adequate resistance to H3PO4; very
expensive; forms alloys with Hg, Pb, Sn, Au, Cu, Si, Zn, Cd, As, Al, Bi, and
Fe, which may be formed under reducing conditions; permeable to H2 at red
heat, which serves as a reducing agent; may react with S, Se, Te, P, As, Sb,
B, and C to damage container; soft and easily deformed, often alloyed with
Ir, Au, or Rh for strength. Do not use with Na2CO3 for fusion.
Suitable for use with strongly alkaline solutions.
Less expensive alternative to platinum; extremely resistant to HC1; resistant
to HNO3; resistant to 50% H2SO4 and 60% H3PO4 up to 100° C; resistant to
molten NaOH; attacked by molten nitrate and bisulfate; usually available as
Zircaloy—98% Zr, 1.5% Sn, trace Fe, Cr, and Ni.
Resistant to acids and alkali melts; rapidly attacked by bisulfate melts;
brittle, requires thick walled containers.
Resistant to many acids; attacked by 16M HNO3 and glacial acetic acid;
begins to soften and lose shape at 60° C; appreciably porous to Br2, NH3,
H2S, H2O, and HNO3 (aqueous solutions can lose ~1% volume per year
when stored for extended periods of time).
Inert to almost all inorganic and organic compounds except F2; porosity to
gases is significantly less than that of polyethylene; safe to use at 250° C but
decomposes at 300° C; difficulty in shaping containers results in high cost;
low thermal conductivity (requires long periods of time to heat samples).
JULY 2001
DRAFT FOR PUBLIC COMMENT
12-5
MARLAP
DO NOT CITE OR QUOTE
-------
Laboratory Sample Preparation
127 12.2.2 Contamination from Sources in the Laboratory
128 Contamination leads to biased data that misrepresent the concentration or presence of
129 radionuclides in a specific sample. Therefore, laboratory personnel should take appropriate
130 measures to prevent the contamination of samples. Such precautions are most important when
131 multiple samples are processed together. Possible sources of contamination include:
132 • Airborne;
133 • Reagents (tracers are discussed in Chapter 13);
134 • Glassware/equipment; and
135 • Facilities.
136 The laboratory should use techniques that eliminate air particulates or the introduction of any
137 outside material (such as leaks from aerosols) into samples and that safeguard against using
138 contaminated glassware or laboratory equipment. Contamination of samples can be controlled by
139 adhering to established procedures for equipment preparation and decontamination before and
140 after each sample is prepared. Additionally, the results of blank samples (e.g., sand), which are
141 run as part of the internal quality assurance program, should be closely monitored, particularly
142 following the processing of samples with elevated activity.
143 "Cross-contamination" is the contamination of one sample by another sample that is being
144 processed concurrently or that was processed prior to the current sample leaving a residue on the
145 equipment being used. Simply keeping samples covered whenever practical is one technique to
146 minimize cross-contamination. Another technique is to order the processing of samples
147 beginning with the lowest contamination samples first. It is not always possible to know the
148 exact rank of samples, but historical or field screening data may be useful.
149 Laboratory personnel should be wary of using the same equipment (gloves, tweezers for filters,
150 contamination control mats, etc.) for multiple samples. Countertops and other preparation areas
151 should be routinely monitored for contamination.
152 12.2.2.1 Airborne Contamination
153 Airborne contamination is most likely to occur when grinding or pulverizing solid samples. Very
154 small particles (-10 microns) may be produced, suspended in air, and transported in the air
155 before settling onto a surface. Other sources of potential airborne contamination include samples
156 that already consist of very small particles, or radionuclides that decay through a gaseous
157 intermediate (i.e., 226Ra decays to 222Rn gas and eventually decays to 210Pb). Therefore, the
MARLAP JULY 2001
DO NOT CITE OR QUOTE 12-6 DRAFT FOR PUBLIC COMMENT
-------
Laboratory Sample Preparation
158 grinding or pulverizing of solid samples or the handling of samples that could produce airborne
159 contamination should be carried out under a laboratory hood to prevent dispersal or deposition in
160 the laboratory of contaminated air particulates. These particles easily can contaminate other
161 samples stored in the area. To prevent such cross-contamination, other samples should be
162 covered or removed from the area while potential sources of airborne contamination are being
163 processed.
164 If contamination from the ambient progeny of 222Rn is a concern, this can be avoided by
165 refraining from the use of suction filtration in chemical procedures, prefiltering of room air
166 (Lucas, 1967), and use of radon traps (Lucas, 1963; Sedlet, 1966). The laboratory may have
167 background levels of radon progeny from its construction materials.
168 12.2.2.2 Contamination of Reagents
169 Contamination from radiochemical impurities in reagents is especially troublesome in low-level
170 work (Wang et al., 1975). Care must be taken in obtaining reagents with the lowest contamina-
171 tion possible. Owing to the ubiquitous nature of uranium and thorium, they and their progeny are
172 frequently encountered in analytical reagents. For example, Yamamoto et al. (1989) found
173 significant 226Ra contamination in common barium and calcium reagents. Other problematic
174 reagents include the rare earths (especially cerium salts), cesium salts which may contain 40K or
175 87Rb, and potassium salts. Precipitating agents such as tetraphenyl borates and chloroplatinates
176 may also suffer from contamination problems. In certain chemical procedures, it is necessary to
177 replace inert carriers of the element of interest with non-isotopic carriers when it is difficult to
178 obtain the inert carrier in a contamination-free condition. Devoe (1961) has written an extensive
179 review article on the radiochemical contamination of analytical reagents.
180 12.2.2.3 Contamination of Glassware/Equipment
181 Other general considerations in sample preparation include the cleaning of glassware and
182 equipment (Section 12.2.3). Criteria established in the planning documents or laboratory SOPs
183 should give guidance on proper care of glassware and equipment (i.e., scratched glassware
184 increases the likelihood of sample contamination and losses owing to larger surface area).
185 Glassware should be routinely inspected for scratches, cracks, etc., and discarded if damaged.
186 Blanks and screening should be used to monitor for contamination of glassware.
187 Whenever possible, the use of new or disposable containers or labware is recommended. For
188 example, disposable weigh boats can be used to prevent contamination of a balance. Disposable
189 plastic centrifuge tubes are often less expensive to use than glass tubes that require cleaning after
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 12-7 DO NOT CITE OR QUOTE
-------
Laboratory Sample Preparation
190 every use. If non-disposable containers or labware are used, it may be necessary to use new
191 materials for each new project to reduce the potential for contamination. Blanks can be used to
192 detect cross-contamination. Periodic rinsing with a dilute solution of nitric acid can aid in
193 maintaining clean glassware. However, Bernabee et al. (1980) could not easily remove nuclides
194 sorbed onto the walls of plastic containers by washing with strong mineral acids. They report that
195 nuclides can be wiped from the walls, showing the importance of the physical action of a brush
196 to the cleaning process.
197 12.2.2.4 Contamination of Facilities
198 In order to avoid contamination of laboratory facilities and possible contamination of samples or
199 personnel, good laboratory practices must be constantly followed and the laboratory must be kept
200 in clean condition. The laboratory should establish and maintain a Laboratory Contamination
201 Control Program (Section 12.2.4) to avoid contamination of facilities and to deal with it
202 expeditiously if it occurs.
203 12.2.3 Cleaning of Labware, Glassware, and Equipment
204 12.2.3.1 Labware and Glassware
205 Some labware is too expensive to be used only once (e.g., crucibles, Teflon™ beakers, separately
206 funnels). Labware that will be used for more than one sample should be subjected to thorough
207 cleaning between uses. A typical cleaning protocol includes a detergent wash, an acid soak (HC1,
208 HNO3, or citric acid), and a rinse with deionized or distilled water. As noted in Chapter 10, the
209 use of a brush to physically scrub glassware aids in the removal of contaminates.
210 The Chemical Technician's Ready Reference Handbook (Shugar and Ballinger, 1996) offers
211 practical advice on washing and cleaning laboratory glassware:
212 • Always clean your apparatus immediately after use. It is much easier to clean the glassware
213 before the residues become dry and hard. If dirty glassware cannot be washed immediately, it
214 should be left in water to soak.
215 • Thoroughly rinse all soap or other cleaning agent residue after washing glassware to prevent
216 possible contamination. If the surface is clean, the water will wet the surface uniformly; if the
217 glassware is still soiled, the water will stand in droplets.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 12-8 DRAFT FOR PUBLIC COMMENT
-------
Laboratory Sample Preparation
218 • Use brushes carefully and be certain that the brush has no exposed sharp metal points that can
219 scratch the glass. Scratched glassware increases the likelihood of sample contamination and
220 losses owing to larger surface areas. Moreover, scratched glassware is more easily broken,
221 especially when heated.
222 Automatic laboratory dishwashers and ultrasound or ultrasonic cleaners are also used in many
223 radiochemical laboratories. It is important to note that cleaning labware in an automatic
224 laboratory dishwasher alone may not provide adequate decontamination. Contaminated glassware
225 may need to be soaked in acid or detergent to ensure complete decontamination. Ultrasonic
226 cleaning in an immersion tank is an exceptionally thorough process that rapidly and efficiently
227 cleans the external, as well as the internal, surfaces of glassware or equipment. Ultrasonic
228 cleaners generate high-frequency sound waves and work on the principle of "cavitation," the
229 formation and collapse of submicron bubbles. These bubbles form and collapse about 25,000
230 times each second with a violent microscopic intensity which produces a scrubbing action
231 (Shugar and Ballinger, 1996). This action effectively treats every surface of the labware because
232 it is immersed in the solution and the sound energy penetrates wherever the solution reaches.
233 The Manual for the Certification of Laboratories Analyzing Drinking Water (EPA, 1992)
234 contains a table of glassware cleaning and drying procedures for the various methods given in the
235 manual (including methods for the analysis of radionuclides in water). The suggested procedure
236 for cleaning glassware for metals analysis is to wash with detergent, rinse with tap water, soak
237 for 4 hours in 20 percent (v/v) HNO3 or dilute HNO3 (8 percent)/HCl (17 percent), rinse with
238 reagent water, then air dry. Shugar and Ballinger (1996) suggest treating acid-washed glassware
239 by soaking it in a solution containing 2 percent NaOH and 1 percent disodium ethylenediamine
240 tetraacetate for 2 hours, followed by a number of rinses with distilled water to remove metal
241 contaminants.
242 More specifically to radionuclides, in their paper discussing the simultaneous determination of
243 alpha-emitting nuclides in soil, Sill et al. (1974) examined the decontamination of certain
244 radionuclides from common labware and glassware:
245 By far the most serious source of contamination is the cell, electrode, and "O" ring used
246 in the electrodeposition step. Brief rinsing with a strong solution of hydrochloric acid
247 containing hydrofluoric acid and peroxide at room temperature was totally ineffective in
248 producing adequate decontamination. Boiling anode and cell with concentrated nitric acid
249 for 10 to 15 minutes removed virtually all of the activity resulting from the analysis of
250 samples containing less than 500 disintegrations per minute (dpm). When larger
251 quantities of activity such as the 2.5 x 104 counts per minute (cpm) used in the material
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 12-9 DO NOT CITE OR QUOTE
-------
Laboratory Sample Preparation
252 studies ... had been used, a second boiling with clean acid was generally required.
253 However, boiling nitric acid precipitates polonium and other procedures have to be used
254 in its presence. When such high levels of activity have been used, a blank should be run
255 to ensure that decontamination was adequate before the system is permitted to be used in
256 the analysis of subsequent low-level samples. Prudence suggests that a separate system
257 should be reserved for low-level samples and good management exercised over the level
258 of samples permitted in the low-level system to minimize the number of blanks and full-
259 length counting times required to determine adequate decontamination.
260 ...Beakers, flasks, and centrifuge tubes in which barium sulfate has been precipitated must
261 be cleaned by some agent known to dissolve barium sulfate, such as boiling perchloric or
262 sulfuric acids or boiling alkaline DTPA [diethylenetriaminepentacetate]. This is a
263 particularly important potential source of contamination, particularly if hot solutions
264 containing freshly-precipitated barium sulfate are allowed to cool without stirring. Some
265 barium sulfate post-precipitates after cooling and adheres to the walls so tenaciously that
266 chemical removal is required. Obviously, the barium sulfate will contain whichever
267 actinide is present, and will not dissolve even in solutions containing hydrofluoric acid.
268 Beakers or flasks in which radionuclides have been evaporated to dryness will invariably
269 contain residual activity which generally requires a pyrosulfate fusion to clean completely
270 and reliably. Separately funnels can generally be cleaned adequately by rinsing them with
271 ethanol and water to remove the organic solvent, and then with hydrochloric-hydrofluoric
272 acids and water to remove traces of hydrolyzed radionuclides...
273 However, one should note that current laboratory safety guidelines discourage the use of
274 perchloric acid (Schilt, 1979).
275 12.2.3.2 Equipment
276 In order to avoid cross-contamination, grinders, sieves, mixers and other equipment should be
277 cleaned before using them for a new sample. Cleaning equipment prior to use is only necessary if
278 the equipment has not been used for some time. The procedure can be as simple or as
279 complicated as the analytical objectives warrant as illustrated by Obenhauf et al. (website
280 reference) in the SPEX Certiprep Handbook of Sample Preparation and Handling. In some
281 applications, simply wiping down the equipment with ethanol may suffice. Another practical
282 approach is to brush out the container, and briefly process an expendable portion of the next
283 sample and discard it. For more thorough cleaning, one may process one or more batches of pure
284 quartz sand through the piece of solid processing equipment, and then wash it carefully. The
MARLAP JULY 2001
DO NOT CITE OR QUOTE 12-10 DRAFT FOR PUBLIC COMMENT
-------
Laboratory Sample Preparation
285 efficacy of the decontamination is determined by monitoring this sand for radionuclide
286 contamination.
287 An effective cleaning procedure for most grinding containers is to grind pure quartz sand
288 together with hot water and detergent, then to rinse and dry the container. This approach
289 incorporates a safety advantage in that it controls respirable airborne dusts. It is important to note
290 that grinding containers become more difficult to clean with age because of progressive pitting
291 and scratching of the grinding surface. Hardened steel containers can also rust, and therefore
292 should be dried thoroughly after cleaning and stored in a plastic bag containing a desiccating
293 agent. If rust does occur, the iron oxide coating can be removed by a warm dilute oxalic acid
294 solution or by abrasive cleaning.
295 12.2.4 Laboratory Contamination Control Program
296 The laboratory should establish a general program to prevent the contamination of samples.
297 Included in the program should be ways to detect contamination from any source during the
298 sample preparation steps if contamination of samples occurs. The laboratory contamination
299 control program should also provide the means to correct procedures to eliminate or reduce any
300 source of contamination. Some general aspects of a control program include:
301 • Appropriate engineering controls, such as ventilation, shielding, etc., should be in place.
302 • The laboratory should be kept clean and good laboratory practices should be followed.
303 Personnel should be well-trained in the safe handling of radioactive materials.
304 • Counter tops and equipment should be cleaned and decontaminated following spills of
305 liquids or dispersal of finely powdered solids. Plastic-backed absorbent benchtop coverings
306 or trays help to contain spills.
307 • There should be an active health physics program that includes frequent monitoring of
308 facilities and personnel.
309 • Wastes should be stored properly and not allowed to accumulate in the laboratory working
310 area. Satellite accumulation areas should be monitored.
311 • Personnel should be mindful of the use of proper personnel protection equipment and
312 practices (e.g., habitual use of lab coats, frequent glove changes, routine hand washes).
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 12-11 DO NOT CITE OR QUOTE
-------
Laboratory Sample Preparation
313 • Operations should be segregated according to activity level. Separate equipment and facilities
314 should be used for elevated and low-level samples whenever possible.
315 • SOPs describing decontamination and monitoring of labware, glassware, and equipment
316 should be available.
317 • Concentrated standard stock solutions should be kept isolated from the general laboratory
318 working areas.
319 As an example, Kralian et al. (1990) have published the guidelines for effective low-level
320 contamination control.
321 12.3 Solid Samples
322 This section discusses laboratory preparation procedures for solid samples as illustrated in
323 Figure 12.2. General procedures such as exclusion of unwanted material in the sample; drying,
324 charring, and ashing of samples; obtaining a constant weight (if required); and homogenization
325 are discussed first. Examples of preparative procedures for solid samples are then presented.
326
Solid samples may consist of a wide variety of materials, including:
327 • Soil and sediment;
328 • Biota (plants and animals); and
329 • Other materials (metal, concrete, asphalt, solid waste, etc.).
330 Before a solid sample is prepared, the specific procedures given in the planning documents
331 should be reviewed. This review should result in a decision that indicates whether materials other
332 than those in the intended matrix should be removed, discarded, or analyzed separately. Any
333 material removed from the sample should be identified, weighed, and documented.
334 To ensure that a representative aliquant of a sample is analyzed, the sample should first be dried
335 or ashed and then blended or ground thoroughly (Appendix F). Homogenization should result in
336 a uniform distribution of analytes and particles throughout the sample. The size of the particles
337 that make up the sample will have a bearing on the representativeness of each aliquant.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 12-12 DRAFT FOR PUBLIC COMMENT
-------
Laboratory Sample Preparation
Does the Sample
Require Heating?
No •*-
Yes-
Refer to Planning Documents to
Determine What Should be Removed,
Discarded, Analyzed, and/or Saved for
Future Analyses
Yes-
Options:
Heat to Remove Moisture
Heat to Char Organics
Dry Ash to Destroy Organics
Are Separate
Aliquots Needed
for Each
Analyte?
Prepare Separate Aliquots
for Each Analyte
FIGURE 12.2—Laboratory Sample Preparation Flowchart (for Solid Samples)
JULY 2001
DRAFT FOR PUBLIC COMMENT
12-13
MARLAP
DO NOT CITE OR QUOTE
-------
Laboratory Sample Preparation
338 12.3.1 General Procedures
339 The following sections discuss the general procedures for exclusion of material, heating solid
340 samples (drying, charring, and ashing), obtaining a constant weight, mechanical manipulation
341 grinding, sieving, and mixing), and subsampling. Not every step is done for all solid sample
342 categories (soil/sediment, biota, and other) but are presented here to illustrate the steps that could
343 be taken during preparation.
344 12.3.1.1 Exclusion of Material
345 Exclusion of Material by Size and Particles. During solid preparation, some particles may be
346 identified in the sample that are not a part of the matrix intended for analysis. Examples of such
347 particles are rocks and pebbles or fragments of glass and plastic. Depending on the specific
348 procedures given in the planning documents on the constitution of the sample taken, rocks and
349 pebbles can be removed and analyzed separately if desired. The sample should be weighed before
350 and after any material is removed. Other materials that are not a part of the required matrix can
351 also be removed and analyzed separately. If analysis of the material removed is necessary,
352 applicable SOPs should be used to prepare the material for analysis.
353 Exclusion of Organic Material. Leaves, twigs, and grass can easily be collected inadvertently
354 along with samples of soil or sediment. Because these are not usually intended for analysis, they
355 are often removed and stored for future analysis, if necessary. The material removed should be
356 identified, if possible, and weighed.
357 12.3.1.2 Principles of Drying Techniques
358 Applying elevated temperatures during sample preparation is a widely used technique for the
359 following reasons:
360 • To remove moisture or evaporate liquids (raise the temperatures to 60° to 110° C).
361 • To prepare organic material for subsequent wet ashing or fusion ("char" the material by
362 heating to medium temperature of 200° to 300° C).
363 • To prepare the sample for subsequent determination of nonvolatile constituents (dry ash at
364 high temperature of 450° to 750° C).
MARLAP JULY 2001
DO NOT CITE OR QUOTE 12-14 DRAFT FOR PUBLIC COMMENT
-------
Laboratory Sample Preparation
365 Once a decision is made to use elevated temperatures during sample preparation, several
366 questions should be considered:
367 • What material should be used for the sample container?
368 • What should serve as the heat source?
369 • How quickly should the temperature be raised? (Rate of stepwise temperature increase)
370 • What is the maximum temperature to which the sample should be exposed?
371 • How long should the sample be heated at the maximum temperature?
372 • How quickly should the sample be cooled afterward?
373 The following sections provide information related to these questions.
374 Note that there are times during sample preparation when samples should not be heated. For
375 example, samples to be prepared for 3H or 14C determination should not be heated. Since 3H is
376 normally present as tritiated water in environmental samples, heating will remove the 3H.
377 Similarly, 14C is usually present in environmental samples as carbonates or 14CO2 dissolved in
378 water, and heating will release 14C as a gas. Samples to be analyzed for iodine, mercury,
379 antimony, or other volatile elements should be heated only under conditions specified in the
380 planning documents. If both volatile and nonvolatile elements are determined from the same
381 sample, aliquants of the original sample should be removed for determination of the volatile
382 elements.
383 Ovens, furnaces, heat lamps, and hot plates are the traditional means to achieve elevated
384 temperatures in the laboratory. However, more recently, microwave ovens have added an
385 additional tool for elevating temperature during sample preparation. Walter et al. (1997) and
386 Kingston and lassie (1988) give an overview of the diverse field of microwave-assisted sample
387 preparation. A dynamic database of research articles related to this topic can be found at the
388 SamplePrep Web™ at http://www.sampleprep.duq.edu/index.html. As microwave sample
389 preparation has developed, numerous standard methods with microwave assistance have been
390 approved by the American Society for Testing and Materials (ASTM), Association of Official
391 Analytical Chemists (AOAC), and the U.S. Environmental Protection Agency (EPA). The
392 majority of the microwave-assisted methods are for acid-dissolution (Chapter 13), but several are
393 for drying samples.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 12-15 DO NOT CITE OR QUOTE
-------
Laboratory Sample Preparation
394 Alternatives to heating samples include drying them slowly in a vacuum desiccator, air-drying, or
395 freeze-drying. ASTM D3974 describes three methods of preparing soils, bottom sediments,
396 suspended sediments, and waterborne materials: (1) freeze-drying; (2) air-drying at room
397 temperature; and (3) accelerated air-drying.
398 Drying Samples
399 It must be determined at the start of an analytical procedure if the results are to be reported on an
400 as-received or dry-weightbasis. Most analytical results for solid samples should be reported on a
401 dry-weight basis, which denotes material dried at a specified temperature to a constant weight or
402 corrected through a "moisture" determination made on an aliquant of the sample taken at the
403 same time as the aliquant taken for sample analysis.
404 Typically, samples are dried at temperatures of 105° to 110° C. Sometimes it is difficult to
405 obtain constant weight at these temperatures, then higher temperatures must carefully be used.
406 Alternatively, for samples that are extremely heat sensitive and decompose readily, vacuum
407 desiccation or freeze-drying techniques are applicable.
408 The presence of water in a sample is a common problem frequently facing the analyst. Water
409 may be present as a contaminant (i.e., from the atmosphere or from the solution in which the
410 substance was formed) or be bonded as a chemical compound (i.e., a hydrate). Regardless of its
411 origin, water plays a role in the composition of the sample. Unfortunately, especially in the case
412 of solids, water content is variable and depends upon such things as humidity, temperature, and
413 the state of subdivision. Therefore, the make-up of a sample may change significantly with the
414 environment and the method of handling.
415 Traditionally, chemists distinguish several ways in which water is held by a solid (Dean, 1995).
416 • Essential water is an integral part of the molecular or crystal structure and is present in
417 stoichiometric quantities, for example, CaC2O4»2H2O.
418 • Water of constitution is not present as such in the solid, but is formed as a product when the
419 solid undergoes decomposition, usually as a result of heating. For example, Ca(OH)2 ^ CaO
420 + H2O.
421 • Nonessential water is retained by physical forces, is non-stoichiometric, and is not necessary
422 for the characterization of the chemical composition of the sample.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 12-16 DRAFT FOR PUBLIC COMMENT
-------
Laboratory Sample Preparation
423 • Absorbed water is retained on the surface of solids in contact with a moist environment, and
424 therefore, is dependent upon the humidity, temperature, and surface area of the solid.
425 • Sorbed water is encountered with many colloidal substances such as starch, charcoal, zeolite
426 minerals, and silica gel and may amount to as much as 20 percent or more of the solid.
427 Sorbed water is held as a condensed phase in the interstices or capillaries of the colloid and it
428 is greatly dependent upon temperature and humidity.
429 • Occluded water is entrapped in microscopic pockets spaced irregularly throughout solid
430 crystals. These cavities frequently occur naturally in minerals and rocks.
431 • Water also may be present as a solid solution in which the water molecules are distributed
432 homogeneously throughout the solid. For example, natural glasses may contain several
433 percent moisture in this form.
434 Heat Source. There are several choices when heating to dryness. The heat source is often
435 determined by the amount of time available for drying and the potential for the sample to spatter
436 or splash during drying. When time is not a primary concern and there is little or no chance of
437 sample cross-contamination, samples are heated uncovered in a drying oven at the minimum
438 temperature needed to remove moisture. If time is of concern, samples with high moisture
439 content can usually be dried or evaporated faster using a hot plate. Heating on a hot plate
440 significantly increases the chance of cross-contamination by spattering or splashing during
441 boiling. However, ribbed watch glasses, which cover the sample yet still allow for evaporation,
442 can be used to minimize cross contamination in this approach. Samples may also be placed under
443 a heat lamp. This method reduces the risk of cross-contamination by applying heat to the surface
444 where vaporization occurs, minimizing splashing during boiling. However, the elevated
445 temperature is difficult to measure or control, and spattering still may be a problem when the
446 sample reaches dryness.
447 Microwave systems may also be used to dry samples. ASTM El358 and ASTM D4643 use
448 microwave energy to dry either wood or soil to a constant weight. In a similar fashion, AOAC
449 Methods 985.14 and 985.26 use microwave energy to dry fat from meat or water from tomato
450 juice. Other examples include Beary (1988) who has compared microwave drying to
451 conventional techniques using NIST solid standards (coal, clays, limestone, sediment) and foods
452 and food materials (rice and wheat flour) standards and Koh (1980) who discusses microwave
453 drying of biological materials.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 12-17 DO NOT CITE OR QUOTE
-------
Laboratory Sample Preparation
454 Container Material. A sample container's material composition typically poses no problem.
455 Borosilicate glass is generally recommended because it is inexpensive, transparent, reusable, and
456 has good thermal properties. Platinum, Teflon™ (polytetrafluoroethylene—PTFE), porcelain, or
457 aluminum foil containers are acceptable and may be preferable in certain situations. Polyethylene
458 and other plastics of low melting point are only useful in hot water baths or ovens where the
459 temperature is closely monitored. If polyethylene is going to be used, be aware that it is affected
460 by heat applied directly to the container. The properties of several common materials used for
461 sample containers are presented in Table 12.2 (on page 12-5). Note that the sample containers
462 commonly received from the field will be those suitable for bulk samples rather than containers
463 used during sample preparation. The plan will identify the type of container material to be used
464 for field activities for samples to be shipped to the laboratory and the type of container material
465 to be used during the various steps of sample preparation.
466 Heating Rate. The heating rate is generally not considered when removing moisture, because the
467 maximum temperature typically is very low (60° to 110° C). Samples simply are placed inside
468 the preset oven. Hot plates may be preheated to the desired temperature before heating the
469 sample or turned on and gradually heated with the sample in place.
470 Maximum Temperature. The maximum temperature used for drying samples typically is just
471 above the boiling point of water—105 ° to 110° C. Higher temperatures will not dry the samples
472 significantly faster and may result in accidents or cross-contamination due to uneven heating.
473 Lower temperatures will not reduce the chance of cross-contamination, but will significantly
474 increase the drying time. One exception to this rule occurs when the physical form of the sample
475 needs to be preserved. Many minerals and chemicals have waters of hydration that affect the
476 structure and may also affect the chemical and physical properties. Samples heated at 60° C will
477 retain the waters of hydration in most chemicals and minerals and still provide dry samples in a
478 reasonable period of time (e.g., 12 to 15 hrs.).
479 Time. The duration a sample is heated to remove moisture depends on the size of the sample, the
480 amount of moisture in the sample, the air flow around the sample, and the temperature applied to
481 the sample. If heating the sample is to provide a constant dry weight, it is more difficult to
482 determine how long to heat the sample. One convenient approach, especially when working with
483 numerous samples, is to dry all materials overnight, or occasionally longer. This amount of
484 heating is usually more than sufficient for drying samples for radiochemical analysis. If time is a
485 critical factor or if a quantitative assessment of the uncertainty in the sample weight is required
486 by the planning documents, the sample can be subjected to repeated cycles of drying and
487 weighing until a series of weights meet the specified requirements (Section 12.3.1.3). For
488 example, one such requirement might be to obtain three consecutive weights with a standard
MARLAP JULY 2001
DO NOT CITE OR QUOTE 12-18 DRAFT FOR PUBLIC COMMENT
-------
Laboratory Sample Preparation
489 deviation less than 5 percent of the mean. While repeated cycles of drying and weighing can
490 provide a quantitative measure of the uncertainty in the sample weight over time, a single weight
491 after an overnight drying cycle typically provides a similar qualitative level of confidence with
492 significantly less working time. Another time-saving step is to use microwave techniques rather
493 than conventional heating sources during sample preparation (ANL/ACL, 1992; Walter et al.,
494 1997).
495 Alternatives to Heating. (1) Vacuum-desiccation. A desiccator is a glass or aluminum container
496 that is filled with a substance that absorbs water, a "desiccant." The desiccator provides a dry
497 atmosphere for objects and substances. Dried materials are stored in desiccators while cooling in
498 order to minimize the uptake of ambient moisture. The ground-glass or metal rim of the desicca-
499 tor should be greased lightly with petroleum jelly or silicone grease to improve performance.
500 Calcium sulfate, sodium hydroxide, potassium hydroxide, and silica gel are a few of the common
501 desiccants. The desiccant must be renewed frequently to keep it effective. Surface caking is a
502 signal to renew or replace the desiccant. Some desiccants contain a dye that changes color upon
503 exhaustion.
504 Vacuum desiccators are equipped with a side-arm so that they may be connected to a vacuum to
505 aid in drying. The contents of the sealed evacuated desiccator are maintained in a dry, reduced-
506 pressure atmosphere. Care must be exercised when applying a vacuum as a rapid pressure
507 reduction, for high water content samples can result in "boiling" with subsequent sample loss and
508 potential cross-contamination.
509 (2) Freeze-drying. Certain substances (i.e., biological materials, pharmaceuticals), which are
510 extremely heat sensitive and cannot be dried at atmospheric conditions, can be freeze-dried
511 (Cameron and Murgatroyd, 1996). Freeze-drying, also known as "lyophilization," is the process
512 by which substances are frozen, then subjected to high vacuum. Under these conditions ice
513 (water) sublimes and other volatile liquids are removed. The non-sublimable material is left
514 behind in a dry state.
515 To freeze-dry effectively, dilute solutions are used. In order to increase the surface area, the
516 material is spread out on the inner surface of the container as it is frozen. Once the solution or
517 substance to be dried is frozen solid, the primary drying stage begins in which a high vacuum is
518 applied, and the ice sublimes, desorbing the free ice and some of the bound moisture. During
519 secondary drying, a prolonged drying stage, the sorbed water which was bound strongly to the
520 solids, is converted to vapor. This can be a slow process because the remaining bound water has
521 a lower pressure than the free liquid at the same temperature, making it more difficult to remove.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 12-19 DO NOT CITE OR QUOTE
-------
Laboratory Sample Preparation
522 Secondary drying actually begins during the primary drying phase, but it must be extended after
523 the total removal of free ice to achieve low levels of residual moisture.
524 Commercial freeze-drying units are self contained. Simple units consist of a vacuum pump,
525 adequate vapor traps, and a receptacle for the material to be dried. More sophisticated models
526 include refrigeration units to chill the solutions, instrumentation to designate temperature and
527 pressure, heat and cold controls, as well as vacuum-release valves. The vacuum pump should be
528 protected from water with a dry-ice trap and from corrosive gases with chemical gas-washing
529 towers.
530 Charring of Samples to Partially Oxidize Organic Material
531 Heating samples at a moderate temperature (200° to 300° C) is sometimes used as a method of
532 preparing a sample for subsequent decomposition using wet ashing or fusion techniques. Large
533 amounts of organic material can react violently or even explosively during decomposition.
534 Heating the sample to partially oxidize—or "char"—the organic material may limit reactivity
535 during subsequent preparation.
536 Heat Source. Heat lamps, muffle furnaces, or hot plates may be used as a heat source for charring
537 samples. Heat lamps are often selected because they can also be used to dry the sample before
538 charring. Once dried, the sample can be moved closer to the lamp to raise the temperature and
539 char the sample (confirmed by visual inspection). Heat lamps also reduce the potential for cross-
540 contamination by minimizing spattering and splashing. Hot plates can be used similarly to heat
541 lamps. The sample is dried and the temperature is raised to char the sample; however, hot plates
542 increase the probability of spattering and splashing. Muffle furnaces can be used when the
543 charring is performed as part of dry ashing instead of part of the drying process. In this case, the
544 muffle furnace temperature is first raised slowly.
545 Sample Container. The choice of sample container depends primarily on the next step in the
546 sample preparation process. When dry ashing or fusing, the sample container will usually be a
547 platinum or porcelain crucible. Zirconium or nickel crucibles may also be used. If the sample will
548 be dissolved using wet ashing techniques, the container may be borosilicate glass or a platinum
549 crucible. Care should be taken to prevent ignition of samples in glass containers. Ignited samples
550 may burn at temperatures high enough to cause damage to the container and loss of sample.
551 Polyethylene and Teflon™ generally are not acceptable because of the increased temperature and
552 risk of melting the container.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 12-20 DRAFT FOR PUBLIC COMMENT
-------
Laboratory Sample Preparation
553 Heating Rate. Heating rate becomes a concern when charring samples because of the increased
554 temperatures. The general rule is to raise the temperature slowly to heat the sample evenly and
555 prevent large increases in temperature within the sample, which could lead to ignition. Typically,
556 a rate of 50° to 100° C per hour is considered appropriate. Samples containing large quantities of
557 organic material may require slower heating rates.
558 Maximum Temperature. One of the primary goals of charring a sample is to oxidize the materials
559 slowly and gently. Gentle oxidation is accomplished by slowly raising the temperature close to
560 the ignition point and letting the sample smolder. Many organic compounds ignite in the range of
561 200° to 300° C (e.g., paper burns at 230° C), so this is usually the range of temperatures where
562 charring takes place. Ignition results in rapid oxidation accompanied by large volumes of
563 released gases and potential sample loss. This reaction can raise the temperature of the sample to
564 several hundred degrees above the desired maximum and result in significant losses during off-
565 gassing. The progress of the reaction can be monitored visually by observing the volume of gas
566 or smoke released. Thin wisps of smoke are usually allowable; clouds of smoke and flames are
567 not. Visual inspection is easily accomplished when hot plates or heat lamps are used as heat
568 sources. Some muffle furnaces are fitted with viewing windows to allow visual inspection. Never
569 open a muffle furnace just to check on the progress of a reaction. This will cause a sudden
570 change in temperature, increase the oxygen level and possibly ignite the sample, and disrupt air
571 currents within the furnace to increase potential sample loss.
572 Time. The duration required to char a sample depends on the sample size, the amount of organic
573 material in the sample, the ignition point of the organic material, the temperature of the sample,
574 and the oxygen supply. Samples usually are heated until smoke begins to appear and allowed to
575 remain at that temperature until no more smoke is evident. This process is repeated until the
576 temperature is increased and no more smoke appears. Charring samples may require a significant
577 amount of time and effort to complete. The duration may be reduced by improving the flow of air
578 to the sample or mixing HNO3 or nitrate salts with the sample before drying. However, this
579 approach is recommended only for well-characterized samples, those previously evaluated for the
580 applicability of this technique, because nitrated organic compounds can oxidize in a violent or
581 explosive manner.
582 Dry Ashing Samples
583 The object of dry ashing is to combust all of the organic material and to prepare the sample for
584 subsequent treatment using wet ashing or fusion techniques. This procedure involves heating a
585 sample in an open dish or crucible in air, usually in a muffle furnace to control the temperature
586 and flow of air. Microwave techniques are also available for dry ashing samples.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 12-21 DO NOT CITE OR QUOTE
-------
Laboratory Sample Preparation
587 Dry ashing is used to determine ash weight as well as nonvolatile constituents. The associated
588 chemistry is very complex, with oxidizing and reducing conditions varying throughout the
589 sample and over time. During the combustion process, temperatures in the sample may reach
590 several hundred degrees above the desired temperature, particularly if there is good air flow at
591 the beginning of the ashing process (Bock, 1979). Covering samples during heating is not
592 recommended, especially when using platinum crucibles. The lack of air produces a reducing
593 atmosphere that results in reduction of metals that alloy with the crucible (Table 12.2 on page 12-
594 5). This reaction results in loss of sample and potential for contamination of subsequent samples
595 when using the same crucible.
596 Heat Source. The traditional heat sources for dry ashing are muffle furnaces or burner flames.
597 Electronic muffle furnaces are recommended for all heating of platinum crucibles because
598 burners produce significant levels of hydrogen gas during combustion, and platinum is permeable
599 to hydrogen gas at elevated temperatures. Hydrogen gas acts as a reducing agent that can result in
600 trace metals becoming alloyed to the platinum.
601 Microwave ovens have also proved to be quick and efficient when dry ashing plant tissue
602 samples, with results comparable to conventional resistance muffle furnaces (Zhang and Dotson,
603 1998). The microwave units are fitted with ashing blocks (a ceramic insert) which absorb
604 microwave energy and quickly heats to high temperatures. This, in combination with the
605 microwave energy absorbed directly by the sample, allows for rapid dry ashing of most materials.
606 The units are designed for increased air flow which further accelerates combustion of the
607 samples.
608 Sample Container. Platinum, zirconium, or porcelain are usually used to form crucibles for dry
609 ashing. Nickel may also be appropriate for some applications (Table 12.2). Platinum generally is
610 recommended when available and is essentially inert and virtually unaffected by most acids.
611 Zirconium and porcelain crucibles are resistant to most acids, are more resistant to HC1, and are
612 significantly less expensive than platinum. Glass and plastic containers should not be used for
613 dry ashing because the elevated temperatures exceed the melting point of these materials.
614 Crucibles fabricated from ceramic, graphite, and platinum can be used in microwave
615 applications. Quartz fiber crucibles can accelerate the ashing process since this material rapidly
616 cools and allows many sample types to be reweighed in 60 seconds or less after removal from the
617 microwave unit.
618 Heating Rate. Samples should be dried before dry ashing and placed in an unheated furnace;
619 then, the furnace temperature is gradually increased. The sample should be spread as thinly and
MARLAP JULY 2001
DO NOT CITE OR QUOTE 12-22 DRAFT FOR PUBLIC COMMENT
-------
Laboratory Sample Preparation
620 evenly as possible on the bottom of the container to allow for its equal heating. To ensure even
621 heating of the sample and to minimize the chance of ignition, the temperature of the furnace is
622 raised slowly. If the sample was previously charred, a rate of approximately 100° C per hour is
623 typical. This rate is slow enough that small amounts of organic material or water can be removed
624 from the sample without violent reactions. If the sample is not charred and contains a significant
625 amount of organic material, a slower rate may be necessary to control the oxidation of organic
626 material.
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
Maximum Temperature. The maximum temperature is determined by the sample matrix and the
volatility of the elements to be analyzed. Generally, the temperature should be as low as possible
to reduce the loss of volatile compounds, but high enough to ensure complete combustion of the
sample. A minimum temperature of 450° C is often used to ensure complete combustion (Bock,
1979). The upper limit for dry ashing is usually determined by the sample container and the
elements being analyzed and is generally considered to be 750° C, but sample-specific conditions
may use temperatures up to 1,100° C. However, in practice, some components which are
normally considered to be nonvolatile may be lost at temperatures above 650° C (Bock, 1979).
Ashing aids may be added to samples to accelerate oxidation, prevent volatilization of specific
elements, and prevent reaction between the sample and the container. Examples include adding
nitrate before drying to assist oxidation and loosen the ash during combustion, adding sulfate to
prevent volatilization of chlorides (e.g., PbCl2, CdCl2, NaCl) by converting them to the higher
boiling sulfates, and adding alkaline earth hydroxides or carbonates to prevent losses of anions
(e.g., Cl", As"3, P"3, B). Table 12.3 lists dry ashing procedures using a platinum container material
for several elements commonly determined by radiochemical techniques.
TABLE 12.3 — Examples of Dry-Ashing Temperatures (Platinum Container)
Element Temperature/Matrix
Cobalt 450° to 600° C for biological material; some losses reported owing to reactions with crucible;
increased volume of sample increases volume of ash and limits loss of sample.
Cesium 400° to 450° C for food and biological material; CsCl and CsNO3 volatilize at temperatures above
500° C.
Iodine 400° to 500° C with an alkaline ashing aid to prevent volatilization; losses reported as low as 450°
C; total volatilization >600° C.
Lead 450° to 500° C acceptable for most samples; bone or coal (lead phosphate) may be ashed as high as
900° C without significant losses; PbO2 reacts with silica in porcelain glaze at low temperatures;
PbCl2 is relatively volatile and nitrate or sulfate ashing aids have been used to good effect.
Plutonium 450° C with nitric acid ashing aid for biological material, 550° C for dust on air filters, 700° C for
soil; high temperature leads to adsorption onto carbon particles and incomplete dissolution of ash.
Strontium 450 ° to 550 ° C for plants, 600 ° C for meat, 700 ° C for milk and bone.
JULY 2001
DRAFT FOR PUBLIC COMMENT
12-23
MARLAP
DO NOT CITE OR QUOTE
-------
Laboratory Sample Preparation
650
651
652
Element Temperature/Matrix
Technetium 725 ° to 750 ° C for plants treated with ammonia.
Thorium 750° C.
Uranium 600 ° C for coal, 750 ° C for biological material; uranium reacts with porcelain glaze resulting in
sample losses.
653 Source: Bock (1979).
654 (Note that reducing conditions for platinum containers are given in Table 12.2)
655 Time. The duration required to completely combust a sample depends on the size of the sample,
656 the chemical and physical form of the sample before and after ashing, and the maximum
657 temperature required to ash the sample. In many cases, it is convenient to place the sample in an
658 unheated furnace and gradually raise the temperature during the day until the maximum
659 temperature is achieved. The furnace is then left at the maximum temperature overnight (12
660 hours). The furnace is allowed to cool during the next day, and samples are removed from a cold
661 oven. This procedure helps prevent sudden changes in temperature that could cause air currents
662 which may potentially disturb the ash. An alternative is to leave the sample at maximum
663 temperature for 24 hours and let the sample cool in the oven the second night to ensure complete
664 combustion of the sample.
665 The elapsed time for dry ashing samples can be significant (greater than 36 hours), but the actual
666 time required by laboratory personnel is minimal.
667 12.3.1.3 Obtaining a Constant Weight
668 If required, constant weight is obtained by subjecting a sample to repetitive cycles of drying and
669 weighing until a series of weights meets specified requirements. Project-specific planning
670 documents or laboratory SOPs should define the acceptance criteria. For example, in the methods
671 for Total Dissolved Solids and Total Suspended Solids, from the Standard Methods for the
672 Examination of Water and Wastewater (Greenberg et al., 1992), solids are repetitively heated for
673 an hour, then weighed until successive weighings agree within 4 percent of the mass or within
674 0.5 mg. In the ASTM guidelines for the preparation of biological samples (ASTM D4638), an
675 accurately weighed sample (1 to 2 g ± 0.1 mg, 5 to 10 g ± 1 mg, >10 g ± 10 mg) is heated for 2
676 hours, cooled in a desiccator, and weighed. Drying is repeated at hourly intervals to attain a
677 constant weight within the same accuracy.
678 Laboratory conditions, calibration of the balance, dynamic range of the balance, drying
679 techniques, drying vessel material, sample transfer techniques, weighing technique, taring and re-
680 taring the balance, recording techniques, data manipulation and modeling all impact the
MARLAP JULY 2001
DO NOT CITE OR QUOTE 12-24 DRAFT FOR PUBLIC COMMENT
-------
Laboratory Sample Preparation
681 uncertainty of determining the mass of the sample. At the National Institute of Science and
682 Technology (NIST), samples are dried to constant weight when taking a sample for analysis,
683 calibrating carriers, and recovering carriers after radiochemical separations. To minimize the
684 weighing uncertainty:
685 • Balances are calibrated annually with weights traceable to NIST.
686 • The balance is on a vibration resistant table in a low air turbulence area and, when possible,
687 the weighing room is on grade level.
688 • Clean plastic, glass, or metal weighing containers are used at temperatures well below their
689 softening point.
690 • Samples are ground to provide consistent particle size and surface area for reproducible
691 drying.
692 • Temperature or vacuum ramp and duration are designed to provide stoichiometric
693 consistency and cooled in desiccators.
694 • Once desiccators are opened, masses are corrected for absorption of moisture from the air.
695 • Weighings are conducted in a temperature and humidity controlled room.
696 • The calibrated 5 decimal place balance is double shielded from drafts with a lucite enclosure
697 box.
698 • The operator sits in front of the balance for about 10 minutes before weighing begins to warm
699 the balance with body heat.
700 • Temperature, pressure and relative humidity are recorded periodically to correct for air
701 buoyancy.
702 • Disposable gloves are used to prevent transfer of finger oils to the sample container.
703 • "Zero" or tares are periodically taken to monitor fluctuations and drift in the balance
704 operation.
705 • The operator tries to sight the readings off the scale reproducibly.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 12-25 DO NOT CITE OR QUOTE
-------
Laboratory Sample Preparation
706 • After a weight has been put on or taken off of the pan, the balance is allowed to settle for
707 about 1 minute before a reading is taken.
708 • When possible, an electronic balance is used for recording the masses through a cable linked
709 to a computer. In the absence of an electronic balance, two operators are used to read and
710 record the masses.
711 • Moisture absorption models are assessed for realistic extrapolation to the time when the
712 sample was taken from the desiccator, and uncertainties are assigned at t = 0 and for
713 differences among models.
714 • All decimal places are carried until the final value is reported.
715 • Replicate drying of several samples are used to validate the efficacy of the drying protocol
716 before routine use.
717 Among this list, major contributors to uncertainty are reaching stoichiometric consistency in the
718 dried sample, calibration of the balance, fluctuation and drift in the balance operation, and curve-
719 fitting moisture absorption corrections. These sources of measurement uncertainties should be
720 quantified by measurement, inference, or judgement, then combined in quadrature as the mass
721 uncertainty.
722 12.3.1.4 Sub sampling
723 Laboratories routinely receive larger samples than required for analysis. The challenge then
724 becomes to prepare a sample that is representative and large enough for analysis, but not so large
725 as to cause needless work in its final preparation. Generally, a raw sample first is crushed to a
726 reasonable particle size and a portion of the crushed material is taken for analysis. This step may
727 be repeated with intermittent sieving of the material until an appropriate sample size is obtained.
728 Then, this final portion is crushed to a size that minimizes sampling error and is fine enough for
729 the dissolution method (Dean 1995; Pitard, 1993).
730 French geologist Pierre Gy (1992) has developed a theory of particulate sampling, which is
731 applicable to subsampling in the laboratory. Appendix F summarizes important aspects of the
732 theory and includes applications to radiochemistry. Some of the important points to remember
733 include the following:
MARLAP JULY 2001
DO NOT CITE OR QUOTE 12-26 DRAFT FOR PUBLIC COMMENT
-------
Laboratory Sample Preparation
734 • For most practical purposes, a subsample is guaranteed to be unbiased only if every particle
735 in the sample has the same probability of being selected for the subsample.
736 • The weight of the subsample should be many times greater than the weight of the largest
737 particle in the sample.
738 • The variance associated with subsampling may be reduced either by increasing the size of the
739 subsample or by reducing the particle sizes before subsampling.
740 • Grouping and segregation of particles tends to increase the subsampling variance.
741 • Grouping and segregation can be reduced by increment sampling, splitting, or mixing.
742 Increment samplingis a technique in which the subsample is formed from a number of smaller
743 portions selected from the sample. A subsample formed from many small increments will
744 generally be more representative than a subsample formed from only one increment. The more
745 increments the better. An example of increment sampling is the one-dimensional "Japanese slab-
746 cake" method (Appendix F).
747 Splittingis a technique in which the sample is divided into a large number of equal-sized
748 portions and several portions are then recombined to form the subsample. Splitting may be
749 performed by a manual procedure, such as fractional shoveling (Appendix F), or by a mechanical
750 device, such as a riffle splitter. A riffle splitter consists of a series of chutes directed alternately to
751 opposite sides. The alternating chutes divide the sample into many portions, which are then
752 recombined into two. The riffle may be used repeatedly until the desired sample size is obtained.
753 Riffle splitters are normally used with free-flowing materials such as screened soils.
754 Another traditional method for splitting is coning and quartering (Appendix F). Gy (1992) and
755 Pitard (1993) do not recommend coning and quartering because with similar tools and effort, one
756 can do fractional shoveling, which is a more reliable method.
757 If proper techniques and tools are used and adequate care is taken, samples of the sizes typically
758 encountered in the laboratory can be mixed effectively. However, the effects of mixing tend to be
759 short-lived because of the constant influence of gravity. Heterogeneous material may begin to
760 segregate immediately after mixing.
761 The method and duration needed to mix a sample adequately depends on the volume and type of
762 material to be mixed. Small volumes can be mixed by shaking for a relatively short time. Large
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 12-27 DO NOT CITE OR QUOTE
-------
Laboratory Sample Preparation
763 volumes may require hours. Pitard (1993) describes dynamic and discontinuous processes for
764 mixing samples including:
765 • Mechanical mixing of test tube samples is useful for small sample size and can be performed
766 on many samples at once. Some examples are a pipette shaker with a motor-activated,
767 rocking controlled motion; a nutator mixer with the test tubes fixed to an oscillating plate;
768 and a tube rotator where tubes are attached to a rotating plate mounted at an angle.
769 • Mechanical mixing of closed containers by rotating about a tumbling axis. A turbula
770 mechanical mixer is an example.
771 • Magnetic stirrers are commonly used to homogenize the contents of an open beaker.
772 • V-blenders are used to homogenize samples from several hundred grams to kilogram size.
773 • Stirrers coupled with propellers or paddles are used to mix large volumes of slurries or pulp.
774 • Sheet mixing or rolling technique, in which the sample is placed on a sheet of paper, cloth, or
775 other material, and the opposite corners are held while rolling the sample (see ASTM C702
776 for aggregates).
777 • Ball and rod mills homogenize as well as grind the sample (see ASTM C999 for soils).
778 When dealing with solid samples, it is often necessary to grind the sample to reduce the particle
779 size in order to ensure homogeneity and to facilitate attack by reagents. The SPEX CertiPrep
780 Handbook of Sample Preparation and Handling (Obenauf et al., website reference) is an
781 excellent resource for information regarding grinding and blending.
782 For hand grinding, boron carbide mortars and pestles are recommended. For samples which can
783 be pulverized by impact at room temperature, a shatterbox, a mixer-mill, or a Wig-L-Bug™ is
784 appropriate, depending on the sample size. For brittle materials—such as wool, paper, dried
785 plants, wood, and soft rocks—which require shearing as well as impact, a hammer-cutter mill is
786 warranted. For flexible or heat-sensitive samples such as polymers, cereal grains, and biological
787 materials, cryogenic grinding is necessary. Methods are described below:
788 • A shatterbox spins the sample, a puck, and a ring inside a dish-shaped grinding container in a
789 tight, high-speed horizontal circle. Within two to five minutes, approximately 100 grams of
790 brittle material can be reduced to less than 200 mesh. Shatterboxes are typically used to grind
MARLAP JULY 2001
DO NOT CITE OR QUOTE 12-28 DRAFT FOR PUBLIC COMMENT
-------
Laboratory Sample Preparation
791 soils, cement mix, rocks, slags, ceramics, and ores. They have also been used for hundreds of
792 other materials including dried marsh-grass, pharmaceuticals, fertilizers, and pesticides.
793 When used in a cryogenic atmosphere, this approach can be used to grind rubber, polymers,
794 bone, hair, and tissue.
795 • A mixer-mill grinds samples by placing them in a container along with one or more grinding
796 elements and imparting motion to the container. The containers are usually cylindrical, and
797 the grinding elements are ordinarily balls, but may be rods, cylinders or other shapes. As the
798 container is rolled, swung, vibrated or shaken, the inertia of the grinding elements causes
799 them to move independently into each other and against the container wall, thus, grinding the
800 sample. Mixer-mills are available for a wide-range of sample sizes. The length of time
801 necessary to grind a sample depends on the hardness of the material and the fineness desired
802 in the final product.
803 • The Wig-L-Bug™ is an effective laboratory mill for pulverizing and blending very small
804 samples, typically in the range of 0.1 to 1 mL.
805 • A hammer-cutter mill utilizes high-speed revolving hammers and a serrated grinding
806 chamber lining to combine both shearing and impact. A slide at the bottom of the hopper
807 feeds small portions of the sample (up to 100 mL) into the grinding chamber. After the
808 sample is adequately pulverized, it passes through a perforated-steel screen at the bottom of
809 the grinding chamber and is then collected. With this approach, dried plants and roots, soils,
810 coal and peat, chemicals, and soft rocks all grind quickly with little sample loss.
811 • Many analytical samples—such as polymers, rubber, and tissues that are too flexible or
812 susceptible to degradation to be impact-ground at room temperature—can be embrittled by
813 chilling, and then pulverized. Samples can be frozen and placed in a traditional grinder, or
814 alternatively, a freezer mill can be utilized. In a freezer mill, the grinding vial is immersed in
815 liquid nitrogen and an alternating magnetic field shuttles a steel impactor against the ends of
816 the vial to pulverize the brittle material. Researchers at Los Alamos National Laboratory
817 (LANL) developed a method of cryogenic grinding of samples to homogenize them and
818 allow the acquisition of a representative aliquant of the materials (LANL, 1996).
819 When samples agglomerate or "cake" during grinding, further particle size reduction is
820 suppressed. Caking can be caused from moisture, heat, static charge accumulation, the fusing of
821 particles under pressure, etc. When it occurs, caking is a serious challenge. There are two main
822 approaches to this problem, slurry grinding and dry grinding.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 12-29 DO NOT CITE OR QUOTE
-------
Laboratory Sample Preparation
823 • In slurry grinding, particles are suspended in solution during grinding. Water, alcohol, or
824 other liquids are added to the sample before grinding, and have to be removed afterwards.
825 Slurry grinding is a fairly reliable way of grinding a sample to micron-sized particles, but it is
826 sloppy and time-consuming.
827 • Dry grinding is often simpler and quicker, but requires careful matching of the technique to
828 the sample. If caking is due to moisture, as in many soils or cements, the sample should be
829 dried before grinding. Grinding aids such as lubricants, antistatic agents, abrasives, and
830 binding agents can also be used. Examples of grinding aids include dry soap or detergent (a
831 lubricant), graphite (an antistatic agent as well as a lubricant), polyvinyl alcohol, phenyl
832 acetate, propylene glycol, and aspirin. For example, propylene glycol (one drop for up to ten
833 grams of sample) is used for laboratory fine grinding of Portland cement and many minerals.
834 Grinding efficiency can be improved through intermittent screening of the material. The ground
835 sample is placed upon a wire or cloth sieve that passes particles of the desired size. The residual
836 particles are reground and this process is repeated until the entire sample passes through the
837 screen. Sieves with large openings can be used in the initial stages of sample preparation to
838 remove unwanted large rocks, sticks, etc.
839 12.3.2 Soil/Sediment Samples
840 For many studies, the majority of the solid samples will be soil/sediment samples or samples that
841 contain some soil. The definition of soil is given in Chapter 10 {Field and Sampling Issues that
842 Affect Laboratory Measurements). Size is used to distinguish between soils (consisting of sands,
843 silts, and clays) and gravels.
844 The procedures to be followed to process a raw soil sample to obtain a representative subsample
845 for analysis depend, to some extent, upon the size of the sample, the amount of processing
846 already undertaken in the field, and more importantly, the radionuclide of interest and the nature
847 of the contamination. Global fallout is relatively homogeneous in particle size and distribution in
848 the sample, and therefore, standard preparation procedures should be adequate for this
849 application. However, when sampling accidental or operational releases, the standard procedures
850 may be inadequate. Transuranic elements, especially plutonium, are notorious for being present
851 as "hot-spots" ions (Eberhardt and Gilbert, 1980; Sill, 1975) and great care must be employed so
852 that the subsample taken for analysis accurately represents the total sample. This will depend on
853 the size and the degree of homogeneity. Multiple subsampling, larger aliquants, and multiple
854 analysis may be the only techniques available to adequately define the content of radionuclides in
MARLAP JULY 2001
DO NOT CITE OR QUOTE 12-30 DRAFT FOR PUBLIC COMMENT
-------
855
856
Laboratory Sample Preparation
heterogeneous samples. Therefore, it is imperative that the analyst choose a preparation approach
appropriate to the nature of the sample.
857 12.3.2.1 Soils
858 ASTM has developed a Standard Practice for the preparation of soil samples (ASTM C999).
859 Guidance is given in this ASTM method for the preparation of a homogenous soil sample from
860 composited core samples. The soil samples are dried at 110° C until at constant weight, ground
861 and mixed in a ball mill, and processed through a U.S. Series No. 35 (500-jim or 32-mesh) sieve.
862 This method is intended to produce a homogeneous sample from which a relatively small
863 aliquant (10 g) may be drawn for radiochemical analyses.
864 A similar procedure for homogenizing soil samples is given in HASL-300 (DOE, 1997).
865 Unwanted material (e.g, vegetation, large rocks) is removed as warranted, and the sample is
866 dried. If the sample contains small rocks or pebbles, the entire soil sample is crushed to 6.35 mm,
867 or the entire sample is sieved through a 12.7-mm screen. The sample is blended, then reduced in
868 size by quartering. This subsample of soil is processed through a grinder, ball mill, sieve, or
869 pulverizer until the soil is reduced to < 1.3 mm (15 mesh equivalent).
870 Sill et al. (1974) described a procedure where they dried raw soil samples for two to three hours
871 at 120° C and then ground the cooled sample lightly in a mortar and pestle. All rocks larger than
872 1A inch were removed. The sample was charred at 400° C for two to three hours, cooled and
873 passed though a No. 35 U.S. standard sieve, and then blended prior to aliquanting (10.0 g are
874 taken for the analysis).
875 12.3.2.2 Sediments
876 ASTM D3976 is a standard practice for the preparation of sediment samples for chemical
877 analysis. This ASTM practice describes the preparation of test samples collected from streams,
878 rivers, ponds, lakes, and oceans. The procedures are applicable to the determination of volatile,
879 semivolatile, and nonvolatile constituents of sediments. Samples are first screened to remove
880 foreign objects and then mixed by stirring. The solids are allowed to settle and the supernatant
881 liquid is decanted. To minimize stratification effects due to differential rates of settling, the
882 sample is mixed again before aliquanting for drying and analysis.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 12-31 DO NOT CITE OR QUOTE
-------
Laboratory Sample Preparation
883 12.3.3 Biota Samples
884 12.3.3.1 Biological Samples
885 ASTM D4638 is a standard guide for the preparation of biological samples for inorganic
886 chemical analysis. This ASTM guide gives procedures for the preparation of test samples of
887 plankton, mollusks, fish, and plants. The preparation techniques are applicable for the
888 determination of volatile, semivolatile, and nonvolatile inorganic compounds in biological
889 materials. However, different preparation steps are involved for the three classes of inorganic
890 compounds. In the case of nonvolatile compounds, the first step is to remove foreign objects and
891 most of the occluded water. For large samples such as fish, samples are homogenized using a
892 tissue disrupter, blender, or equivalent, and a moisture determination is performed on a one to
893 two gram aliquant. The samples then are dried by heating in an oven, by dessication, by air
894 drying, by freeze drying, or by low-temperature drying using an infrared lamp, hot plate, or a low
895 setting on a muffle furnace. Finally, the samples are dry ashed.
896 12.3.3.2 Food
897 The International Atomic Energy Agency has provided a guidebook for the measurement of
898 radionuclides in food and the environment (IAEA, 1989). Sample preparations for milk and other
899 foods such as meat, fish, fruit, vegetables, and grains are given in this guidebook. Additionally,
900 methods are presented in HASL-300 for the preparation of milk, vegetables, composite diets, etc.
901 These methods generally involve dry ashing. The samples first are dried thoroughly at 125° C.
902 Then, the temperature is raised at intervals over an 8-hour period through the critical range where
903 ignition occurs, and finally to 500° C for 16 hours. If only a portion of ash is to be used for
904 analysis, it is ground and sieved prior to aliquanting.
905 12.3.3.3 Vegetation
906 There are several DOE site references that contain examples of sample preparation for
907 vegetation. Los Alamos National Laboratory (LANL, 1997) recently grew pinto beans, sweet
908 corn, and zucchini squash in a field experiment at a site that contained observable levels of
909 surface gross gamma radioactivity within Los Alamos Canyon. Washed edible and nonedible
910 crop tissues (as well as the soil) were prepared for analysis for various radionuclides.
911 Brookhaven National Laboratory has also evaluated the effect of its operation on the local
912 environment. Their site environmental report (DOE, 1995) gives sample preparation steps for
913 radionuclide analysis of vegetation and fauna (along with ambient air, soil, sewage effluent,
MARLAP JULY 2001
DO NOT CITE OR QUOTE 12-32 DRAFT FOR PUBLIC COMMENT
-------
Laboratory Sample Preparation
TABLE 12.4—Preliminary Ashing Temperature for
Food Samples
(Method Sr-02-RC, HASL-300 [DOE, 1997])
Material Temperature (° C)
Eggs 150-250
Meat Burning
Fish Burning
175-325
175-325
914 surface water, and groundwater). HASL-300 also
915 describes sample preparation techniques for
916 vegetation samples for a variety of radionuclides.
917 12.3.3.4 Bone and Tissue
918 Bone and tissue samples can be dry ashed in a
919 muffle furnace (HASL-300, Fisenne, 1994;
920 Fisenne et al.,1980), wet ashed with nitric acid
921 and peroxide (Fisenne and Perry, 1978) or
922 alternately dry ashed and wet ashed with nitric
923 acid until all visible signs of carbonaceous
924 material has disappeared (Mclnroy et al., 1985).
925 12.3.4 Other Samples
926 The category "other" includes such matrices as
927 concrete, asphalt, coal, plastic, etc. The sample
928 preparation procedures applied to soils are
929 generally applicable for the "other" category,
930 except for more aggressive grinding and blending
931 in the initial step. For example, items such as
932 plastic or rubber which are too flexible to be
933 impact-ground at room temperature must be
934 ground cryogenically. They are embrittled by
935 chilling and then pulverized. ASTM Cl 14 describes the sample preparation steps for the
936 chemical analysis of hydraulic cement, whereas ASTM C702 describes the sample preparation of
937 aggregate samples, and is also applicable to lime and limestone products as noted in ASTM C50.
938 Additionally, ASTM D2013 describes the preparation of coal samples for analysis.
939 12.4 Filters
Fruit (fresh)
Fruit (canned)
Milk (dry)
Milk (wet)
Buttermilk (dry)
Vegetables (fresh) . . .
Vegetables (canned) . .
Root vegetables
Grass
Flour
Dry beans 175-250
Fruit juices 175-225
Grains 225-325
Macaroni 225-325
Bread 225-325
175-325
175-225
175-250
200-325
225-250
Burning
940 Filters are used to collect analytes of interest from large volumes of liquids or gases. The exact
941 form of the filter depends on the media (e.g., air, aqueous liquid, nonaqueous liquid), the analyte
942 matrix (e.g., sediment, suspended particulates, radon gas), and the objectives of the project (e.g.,
943 volume of sample passing through the filter, flow rate through the filter, detection limits, etc. (see
944 the section on filtration in Chapter 10, Field and Sampling Issues that Affect Laboratory
945 Measurements).
JULY 2001
DRAFT FOR PUBLIC COMMENT
12-33
MARLAP
DO NOT CITE OR QUOTE
-------
Laboratory Sample Preparation
946 Filter samples from liquids usually consist of the filter with the associated solid material. For
947 samples with a large amount of sediment, the solid material may be removed from the filter and
948 analyzed as a solid. When there is a relatively small amount of solid material, the filter may be
949 considered as part of the sample for analytical purposes. When large volumes of liquid are
950 processed at high flow rates, filter cartridges often are used. Typically, the cartridge case is not
951 considered part of the sample, and laboratory sample preparation includes removing the filter
952 material and sample from the cartridge case. Any special handling instructions should be
953 included as SOPs in the planning documents.
954 Air filters may be particulate filters, which are prepared in the same manner as liquid filters, or
955 they may be cartridges of absorbent material. Filters that absorb materials are typically designed
956 for a specific analysis. For example, activated charcoal cartridges are often used to collect
957 samples of iodine or radon. Silver zeolite cartridges may be used for noble gases such as argon,
958 krypton, or xenon. These cartridges are often designed to be analyzed intact, so no special sample
959 preparation is needed. If the cartridges need to be disassembled for analysis, a special SOP for
960 preparing these samples is usually required.
961 Homogenization is rarely an issue when preparing filter samples. Typically, the entire filter is
962 digested and analyzed. However, obtaining a representative sample of a filter does become an
963 issue when the entire filter is not analyzed. The planning document should give the details of
964 sample preparation for portions of a filter (e.g., sample size reduction through quartering). Steps
965 such as using tweezers for holding filters and using individual sample bags should be taken to
966 prevent the loss of material collected on the filter during handling and processing.
967 12.5 Wipe Samples
968 Wipe samples (also referred to as "swipes" or "smears") are collected to indicate the presence of
969 removable surface contamination. The loose contamination is transferred from the surface to a
970 sample of wipe material. The wipe material can be virtually anything, but common materials
971 include Whatman filter paper and nylon membrane. The greatest challenge in preparing wipe
972 samples is homogenizing the sample to obtain a representative portion for analysis, although
973 usually the entire wipe is analyzed. Wipe samples are commonly digested prior to analysis, but
974 they can be analyzed directly through appropriate counting techniques (McFarland, 1998a,
975 1998b).
976 Many wipe samples are collected using filter paper or disc smears. In many cases, the
977 contamination on these samples is simply assumed to be fairly evenly distributed, and the wipe
978 samples are prepared like filter samples. Sometimes, a specific analytical procedure is anticipated
MARLAP JULY 2001
DO NOT CITE OR QUOTE 12-34 DRAFT FOR PUBLIC COMMENT
-------
Laboratory Sample Preparation
979 and a special wipe material is used. For example, Styrofoam generates static electricity and can
980 attract dust particles from a relatively clean surface. Styrofoam is dissolved easily in most (if not
981 all) commercially available liquid scintillation cocktails. These wipe samples can be very easily
982 collected, stored, and transported in liquid scintillation vials. Have the cocktail added by the
983 laboratory and be counted for gross alpha and gross beta activity using a liquid scintillation
984 counter.
985 12.6 Liquid Samples
986 Liquid samples are commonly classified as aqueous, nonaqueous, and mixtures. Aqueous liquids
987 are most often surface water, groundwater, drinking water, precipitation, effluent, or runoff.
988 Nonaqueous liquids may include solvents, oils, or other organic liquids. Mixtures may be
989 combinations of aqueous and nonaqueous liquids, but may include solid material mixed with
990 aqueous or nonaqueous liquids or both.
991 Preliminary sample measurements (e.g., conductivity, turbidity) may be performed to provide
992 information about the sample and to confirm field processing (see measurement of pH to confirm
993 field preservation in Chapter 11). These measurements are especially useful when there is no
994 prior historical information available from the sample collection site. In addition, this
995 information can also be helpful in the performance of certain radiochemical analyses. These
996 preliminary measurements typically require little or no sample preparation. In many cases, the
997 results of preliminary measurements can be used to determine the quantity of sample to be used
998 for a specific analysis.
999 12.6.1 Conductivity
1000 In radiochemistry, conductivity measurements typically are used as a surrogate to estimate
1001 dissolved solids content for gross-alpha and gross-beta measurements. Because the preservation
1002 of samples with acid prevents the measurement of conductivity, the recommendation is to
1003 perform the QC checks for conductivity in the field when the original measurements are
1004 performed. If the sample is not preserved in the field, the measurement can be done in the
1005 laboratory.
1006 ASTM Dl 125 is the standard test method for determining the electrical conductivity of water.
1007 The method is used for the measurement of ionic constituents, including dissolved electrolytes in
1008 natural and treated water.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 12-35 DO NOT CITE OR QUOTE
-------
Laboratory Sample Preparation
1009 12.6.2 Turbidity
1010 The presence of dissolved or suspended solids, liquids, or gases causes turbidity in water.
1011 Measurement of turbidity provides a means to determine if removal of suspended matter is
1012 necessary in order to meet the specifications for liquid samples as given in the plan document.
1013 ASTM D1889 is the standard test method for the determination of turbidity of water and
1014 wastewater in the range from 0.05 to 40 nephelometric turbidity units (NTU). In the ASTM
1015 method, a photoelectric nephelometer is used to measure the amount of light that a sample
1016 scatters when the light is transmitted through the sample.
1017 12.6.3 Filtration
1018 The filtration of samples is based on the appropriate plan document which should also give the
1019 selection of the filter material to be used. If samples have not been filtered in the field, the
1020 laboratory can perform the filtration. Guidance on filtration of liquid samples is provided in
1021 Section 10.3.3. Filtering is normally done in the field so that preservatives can be added without
1022 promoting the dissolution of undissolved solids in the sample at the time of collection.
1023 12.6.4 Aqueous Liquids
1024 Aqueous liquids are a common matrix analyzed by laboratories, and are often referred to as water
1025 samples. Examples of possible aqueous liquids requiring radionuclide analysis include the
1026 following:
1027 • Drinking water;
1028 • Surface water;
1029 • Ground water;
1030 • Soil pore water;
1031 • Storage tank water;
1032 • Oil production water or brine;
1033 • Trench or landfill leachate; and
1034 • Water from vegetation.
1035 For certain samples that are not filtered, inversion is a form of homogenization. Typically, the
1036 sample is homogenized by inverting the container several times to mix the sample thoroughly. If
1037 there is some air in the container, the passage of air bubbles through the sample will create
1038 sufficient turbulence to mix the sample thoroughly with three or four inversions of the sample
1039 container. If the sample contains zero headspace (so there is no air in the sample container), the
MARLAP JULY 2001
DO NOT CITE OR QUOTE 12-36 DRAFT FOR PUBLIC COMMENT
-------
Laboratory Sample Preparation
1040 sample should be inverted and allowed to stay inverted for several seconds before the next
1041 inversion. Ten to twenty inversions of the sample container may be required to ensure that the
1042 sample is mixed thoroughly under zero headspace conditions. Simply shaking the container will
1043 not mix the contents as thoroughly as inverting the sample container. Mechanical shakers,
1044 mixers, or rotators may be used to homogenize aqueous samples thoroughly.
1045 Filtration and acidification performed in the field is typically the only preparation required for
1046 aqueous liquids (Chapter 10). A general discussion concerning preparation of water samples for
1047 the measurement of radioactivity is presented in NCRP (1976). Analytical Chemistry Laboratory
1048 Sample Preparation Methods (ACL, 1992) gives a number of sample preparation methods for
1049 various materials, including water samples.
1050 ASTM gives standard test methods for the preparation of water samples for the determination of
1051 alpha and beta radioactivity (ASTM D1943 and D1890, respectively). After collecting the water
1052 sample in accordance with ASTM D3370, the sample is made radioactively homogeneous by
1053 addition of a reagent in which the radionuclides present in the sample are soluble in large
1054 concentrations. Acids, complexing agents, or chemically similar stable carriers may be used to
1055 obtain homogeneity. The chemical nature of the radionuclides and compounds present and the
1056 subsequent steps in the method will indicate the action to be taken. Different preparation
1057 techniques for freshwater and seawater samples are illustrated in Radiochemical Analytical
1058 Procedures for Analysis of Environmental Samples (EPA, 1979) and for drinking water in EPA
1059 (1980).
1060 12.6.5 Nonaqueous Liquids
1061 Nonaqueous liquids can be substances other than water such as organic solvents, oil, or grease.
1062 Many organic solvents are widely used to clean oil, grease, and residual material from electrical
1063 and mechanical equipment. The resulting waste liquid may contain a significant amount of solid
1064 material. It may be necessary to filter such liquids to determine (1) if the analyte is contained in
1065 the filtrate and is soluble, or (2) if the analyte is contained in the solids and therefore is insoluble.
1066 The appropriate plan document should be reviewed to determine if filtration is necessary. ASTM
1067 C1234, Standard Test Method for Preparation of Oils and Oily Waste Samples byHigh-
1068 Pressure, High-Temperature Digestion for Trace Element Determinations., describes the
1069 preparation of homogeneous samples from nuclear processing facilities.
1070 Homogenization of nonaqueous samples is accomplished in a manner similar to that for aqueous
1071 samples. Visual inspection is typically used as a qualitative measure of homogeneity in
1072 nonaqueous samples. If a quantitative measure of mixing is desired, turbidity measurements can
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 12-37 DO NOT CITE OR QUOTE
-------
Laboratory Sample Preparation
1073 be performed after a predetermined amount of mixing (e.g., every 10 inversions, every 2 minutes,
1074 etc.) until a steady level of turbidity is achieved (e.g., 1 to 10 percent variance, depending on the
1075 project objectives—see ASTM D1889, Standard Test Method for Turbidity of Water).
1076 DOE (ANL/ACL, 1995) evaluated sample preparation techniques used for the analysis of oils. In
1077 evaluating the performance of a sample preparation technique, DOE considered the following
1078 qualities to be important:
1079 • Thorough sample decomposition;
1080 • Retention of volatile analytes;
1081 • Acceptable analyte recovery;
1082 • Minimal contamination from the environment or the digestion vessel;
1083 • Low reagent blanks; and
1084 • Speed.
1085 One of the preparation methods involved combustion of oil under oxygen at 25 atm pressure
1086 (ASTM E926) and another used nitric acid decomposition of the oil in a sealed vessel heated
1087 with a microwave (EPA, 1990).
1088 Many nonaqueous liquids present a health hazard (e.g., carcinogenicity) or require special safety
1089 considerations (e.g., flammability). Any special handling requirements based on health and safety
1090 considerations should be documented in the planning documents.
1091 12.6.6 Mixtures
1092 Some common examples of mixtures that may be encountered by the laboratory are water with
1093 lots of total dissolved solids and undissolved solids or water and oil in separate layers. The
1094 following sections discuss preparation procedures for these types of mixtures.
1095 12.6.6.1 Liquid-Liquid Mixtures
1096 When aqueous and nonaqueous liquids are combined, they usually form an immiscible mixture,
1097 such as oil and water.1 In most cases, a separately funnel helps in separating the liquids into two
1 It is often necessary to determine which liquid is aqueous and which liquid is nonaqueous. Never assume that the
top layer is always nonaqueous, or the bottom layer is always aqueous. The density of the bottom layer is always
greater than the density of the top layer. Halogenated solvents (e.g., carbon tetrachloride, CC14) tend to have
(continued...)
MARLAP JULY 2001
DO NOT CITE OR QUOTE 12-38 DRAFT FOR PUBLIC COMMENT
-------
Laboratory Sample Preparation
1098 samples. Each sample then is analyzed separately. If, in the rare case, both liquids must be
1099 processed together, there is greater difficulty in preparing the combined liquids for analysis.
1100 Obtaining a homogenous aliquant is a key consideration in this case. Often times, the entire
1101 sample should be analyzed. This approach avoids processing problems and yields the desired
1102 result.
1103 12.6.6.2 Liquid-Solid Mixtures
1104 Mixtures of liquids and solids are usually separated by filtering, centrifuging, or decanting, and
1105 the two phases are analyzed separately. If the mixture is an aqueous liquid and a solid, and will
1106 be analyzed as a single sample, the sample is often treated as a solid. Completely drying the
1107 sample followed by dry ashing before any attempt at wet ashing is recommended to reduce the
1108 chance of organic solids reacting with strong oxidizing acids (e.g., H2SO4, HNO3, etc.). If the
1109 mixture includes a nonaqueous liquid and a solid, it is suggested that the phases be separated by
1110 filtration and the solid rinsed thoroughly with a volatile solvent such as ethanol or methanol
1111 before continuing with the sample preparation process.
1112 In rare cases where a sample contains a mixture of aqueous liquid, nonaqueous liquid, and solid
1113 material, the sample can be separated into three different phases before analysis. The sample
1114 should be allowed to settle overnight and the liquids decanted. The liquids can then be separated
1115 in a separatory funnel without the solid material clogging the funnel. Each liquid should be
1116 filtered to remove any remaining solid material. The solid should be filtered to remove any
1117 remaining liquid and rinsed with a volatile solvent. This rinse removes any traces of organic
1118 liquids to reduce problems during subsequent dissolution activities. The three phases are then
1119 analyzed separately. If necessary, the results can be added together to obtain a single result for the
1120 mixture after the separate analyses are completed.
1121 12.7 Gases
1122 Sample preparation steps are usually not required for gas samples. Lodge (1988) gives general
1123 techniques, including any necessary sample preparation, for the sampling and storage of gases
l(.. .continued)
densities greater than about 1 g/mL, so they typically represent the bottom layer. Other organic liquids (e.g., diethyl
ether, oil, etc.) tend to have densities less than 1 g/mL, so they typically represent the top layer. Mixtures of organic
liquids may have almost any density. To test the liquids, add a drop of water to the top layer. If the drop dissolves in
the top layer, the top layer is aqueous. If the drop settles through the top layer and dissolves in the bottom layer, the
bottom layer is aqueous.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 12-39 DO NOT CITE OR QUOTE
-------
Laboratory Sample Preparation
1124 and vapors. The determination of the tritium content of water vapor in the atmosphere is one of
1125 the example procedures. ASTM D3442 is a standard test method for the measurement of total
1126 tritium activity in the atmosphere. Sample preparation is covered in this test method.
1127 EPA has prepared "Background Information Document: Procedures Approved for Demonstrating
1128 Compliance with 40 CFR Part 61" (EPA, 1989) for use in demonstrating compliance with the
1129 radionuclide National Emission Standards for Hazardous Air Pollutants (NESHAP). This
1130 document includes references to air sampling and sample preparation. Table 3-1 of EPA (1989)
1131 lists numerous references to radionuclide air sampling and preparation; examples include:
1132 • A Study of Airborne Radioactive Effluents from the Pharmaceutical Industry (Cehn, 1979).
1133 • "The Fraction of Material Released as Airborne Activity During Typical Radioiodinations,"
1134 (Eichling, 1983).
1135 • "Application for Renewal of Source Material License: SUB-526, Docket 40-3392," (Allied
1136 Chemical, 1982).
1137 • "Airborne Concentrations of 1-131 in a Nuclear Medicine Laboratory" (Browning et al.,
1138 1978).
1139 12.8 Bioassay
1140 Analyses of bioassay samples are necessary to monitor the health of employees involved in
1141 radiological assessment work. Normally these types of samples include urine and fecal
1142 specimens.
1143 Urine samples are typically wet ashed with nitric acid (DOE, 1997; HASL-300) or with nitric
1144 acid and peroxide (RESL, 1982). Alternatively, there are procedures which co-precipitate the
1145 target analytes in urine by phosphate precipitation (Horwitz et al., 1990; Stradling and
1146 Popplewell, 1974; Elias, 1997). Fecal samples are normally dry ashed in a muffle furnace
1147 (HASL-300), or prepared by lyophilization, "freeze drying" (Dugan and McKibbin, 1993).
1148 It is important to note that although ANSI 13.30 indicates that aliquanting a homogeneous
1149 sample to determine the activity present in the total sample is acceptable, this standard dictates
1150 that the entire sample should be prepared for analysis and the aliquant taken after the sample
1151 preparation has been completed.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 12-40 DRAFT FOR PUBLIC COMMENT
-------
Laboratory Sample Preparation
1152 12.9 References
1153 12.9.1 Cited Sources
1154 Allied Chemical UF6 Conversion Plant. 1982. "Application for Renewal of Source Material
1155 License: SUB-526, Docket 40-3392," Metropolis, Illinois.
1156 American National Standards Institute (ANSI). 1996. ANSI 13.30 Performance Criteria for
1157 Radiobioassay. Health Physics Society.
1158 Analytical Chemistry Laboratory (ACL) Procedure Compendium. 1992. Volume 2: Sample
1159 Preparation Methods. Battelle Pacific Northwest Laboratories, PNL-MA-559.
1160 ANL/ACL. 1992. Innovative Methods for Inorganic Sample Preparation. Argonne National
1161 Laboratory, April 1992.
1162 ANL/ACL. 1995. Preparation of Waste Oil for Analysis to Determine Hazardous Metals.
1163 Argonne National Laboratory, July.
1164 Association of Official Analytical Chemists International (AOAC). 1995. Official Method
1165 985.14. "Moisture in Meat and Poultry Products," in Official Methods of Analysis of AOAC
1166 International, Cuniff, P., Ed., Arlington, VA.
1167 Association of Official Analytical Chemists International (AOAC). 1995. Official Method
1168 985.26. "Solids (Total) in Processed Tomato Products", In Official Methods of Analysis of
1169 AOAC International, Cuniff, P., Ed.; Association of Official Analytical Chemists
1170 International: Arlington, VA.
1171 American Society for Testing and Materials (ASTM) C50. Standard Practice for Sampling,
1172 Inspection, Packing, and Marking of Lime and Limestone Products.
1173 American Society for Testing and Materials (ASTM) Cl 14. Standard Test Method for Chemical
1174 Analysis of Hydraulic Cement.
1175 American Society for Testing and Materials (ASTM) C702. Standard Practice for Reducing
1176 Samples of Aggregate to Testing Size, Vol 04.02.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 12-41 DO NOT CITE OR QUOTE
-------
Laboratory Sample Preparation
llll American Society for Testing and Materials (ASTM) C999. Standard Practice for Soil Sample
1178 Preparation for the Determination of Radionuclides.
1179 American Society for Testing and Materials (ASTM) C1234. Standard Test Method for
1180 Preparation of Oils and Oily Waste Samples by High-Pressure, High-Temperature Digestion
1181 for Trace Element Determinations.
1182 American Society for Testing and Materials (ASTM) Dl 125. Standard Test Method for
1183 Determining the Electrical Conductivity of Water.
1184 American Society for Testing and Materials (ASTM) D1889. Standard Test Method for Turbidity
1185 of Water.
1186 American Society for Testing and Materials (ASTM) D1890. Standard Test Method for Beta
1187 Particle Radioactivity of Water.
1188 American Society for Testing and Materials (ASTM) D1943. Standard Test Method for Alpha
1189 Particle Radioactivity of Water.
1190 American Society for Testing and Materials (ASTM) D2013. Standard Method of Preparing
1191 Coal Samples for Analysis.
1192 American Society for Testing and Materials (ASTM) D3370. Standard Practices for Sampling
1193 Water from Closed Conduits.
1194 American Society for Testing and Materials (ASTM) D3442. Standard Test Method for Gaseous
1195 Tritium Content of the Atmosphere.
1196 American Society for Testing and Materials (ASTM) D3974. Standard Practice for Extraction of
1197 Trace Elements from Sediments.
1198 American Society for Testing and Materials (ASTM) D3975, Standard Practice for Development
1199 and Use (Preparation) of Samples for Collaborative Testing of Methods for Analysis of
1200 Sediments.
1201 American Society for Testing and Materials (ASTM) D3976. Standard Practice for Preparation
1202 of Sediment Samples for Chemical Analysis.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 12-42 DRAFT FOR PUBLIC COMMENT
-------
Laboratory Sample Preparation
1203 American Society for Testing and Materials (ASTM) D4638. Standard Guide for Preparation of
1204 Biological Samples for Inorganic Chemical Analysis.
1205 American Society for Testing and Materials (ASTM) D4643. Standard Test Method for
1206 Determination of Water (Moisture) Content in Soil by the Microwave Oven Method.
1207 American Society for Testing and Materials (ASTM) E926. Standard Practices for Preparing
1208 Refuse-Derived Fuel (RDF) Samples for Analyses of Metals.
1209 American Society for Testing and Materials (ASTM) E1358. Standard Test Method for
1210 Determination of Moisture Content of Particulate Wood Fuels Using a Microwave Oven.
1211 Beary, E.S. 1988. "Comparison of Microwave Drying And Conventional Drying Techniques For
1212 Reference Materials." Anal. ChemVol. 60, pp. 742-746.
1213 Bernabee, R. P., D. R. Percival, and D. B. Martin. 1980. "Fractionation of Radionuclides in
1214 Liquid Samples from Nuclear Power Facilities." Health Physics Vol. 39, pp. 57-67.
1215 Bock, R. 1979. A Handbook of Decomposition Methods in Analytical Chemistry. International
1216 Textbook Company, Limited. T. & A. Constable Ltd., Great Britain.
1217 Browning, E.J., K. Banerjee, and W.E. Reisinger. 1978. "Airborne Concentrations of 1-131 in a
1218 Nuclear Medicine Laboratory," Journal of Nuclear Medicine, Vol. 19, pp. 1078-1081.
1219 Cameron, P., and Murgatroyd, K. 1996. Good Pharmaceutical Freeze-Drying Practice.
1220 Interpharm Press.
1221 Cehn, J.I. 1979. A Study of Airborne Radioactive Effluents from the Pharmaceutical Industry,
1222 Fianl Report, Prepared by Teknekron, Inc., for the U.S. EPA Eastern Environmental Research
1223 Facility, Montgomery, AL.
1224 Dean, J.A. 1995. Analytical Chemistry Handbook, McGraw-Hill, Inc., New York.
1225 DeVoe, J.R. 1961. Radioactive Contamination of Materials Used in Scientific Research.
1226 Publication 895, NAS-NRC.
1227 Dugan, J.P. and T.T. McKibbin. 1993. "Preparation of Fecal Samples for Radiobioassay by
1228 Lyophilization," Radioactivity & Radiochemistry4:3, pp. 12-15.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 12-43 DO NOT CITE OR QUOTE
-------
Laboratory Sample Preparation
1229 Eberhardt, L.L., and R.O. Gilbert. 1980. "Statistics and Sampling in Transuranic Studies," in
1230 Transuranic Elements in the Environment, edited by W.C. Hanson, U.S. Department of
1231 Energy. DOE/TIC-22800.
1232 Eichling, J. 1983. "The Fraction of Material Released as Airborne Activity During Typical
1233 Radioiodinations," Proceedings of the 9th Biennial Conference of Campus Radiation Safety
1234 Officers, University of Missouri-Columbia, June 6-8, 1983.
1235 Elias, G. 1997. "A Rapid Method for the Analysis of Plutonium and Uranium in Urine Samples,"
1236 Radioactivity & RadiochemistryK:!, pp. 20-24.
1237 Fisenne, I.M. and P. Perry 1978. "The Determination of Plutonium in Tissue by Aliquat-336
1238 Extraction," Radiochem Radioanal. Letters Vol. 33, pp. 259-264.
1239 Fisenne, I.M., P. Perry and G.A. Welford. 1980. "Determination of Uranium Isotopes in Human
1240 Bone Ash," Anal. Chem. Vol. 52, pp. 777-779.
1241 Fisenne, I.M. 1994. "Lead-210 in Animal and Human Bone: A New Analytical Method," Env.
1242 Int. Vol. 20, pp. 627-632.
1243 Greenberg, A.E., L.S. Clesceri, and A.D. Eaton (Eds). 1992. Standard Methods for the
1244 Examination of Water and Wastewater. American Public Health Association.
1245 Greenwood, N. N. and A. Earnshaw. 1984. Chemistry of the Elements. Pergamon Press, Inc.
1246 Elmsford, New York.
1247 Gy, Pierre M. 1992. Sampling of Heterogeneous and Dynamic Material Systems: Theories of
1248 Heterogeneity, Sampling, and Homogenizing. Elsevier, Amsterdam, The Netherlands.
1249 Horwitz, E.P., M.L. Dietz, D.M. Nelson, JJ. LaRosa, and W.D. Fairman 1990. "Concentration
1250 and Separation of Actinides from Urine using a Supported Bifunctional Organophosphorous
1251 Extractant," Analytica Chimica Acta. Vol. 238, pp. 263-271.
1252 IAEA. 1989. Measurement of Radionuclides in Food and the Environment—A Guidebook.
1253 Technical Reports Series No. 295, International Atomic Energy Agency, Vienna.
1254 Kingston, H.M., and lassie, L.B. 1988. Introduction to Microwave Sample Preparation: Theory
1255 and Practice, American Chemical Society, Washington, DC.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 12-44 DRAFT FOR PUBLIC COMMENT
-------
Laboratory Sample Preparation
1256 Koh, T.S. 1980. AnalChewVol. 52, pp. 1978-1979.
1257 Kralian, M.A., MJ. Atkins, and S.A. Farber. 1990. "Guidelines for Effective Low-Level
1258 Contamination Control in a Combination Environmental/Radioactive Waste Analysis
1259 Facility," Radioactivity & Radiochemistry. 1:3, pp. 8-18.
1260 Lodge, J. 1988. Methods of Air Sampling and Analysis. Third Edition, CRC Press, Florida.
1261 LANL. 1996. Application of Cryogenic Grinding to Achieve Homogenization of Transuranic
1262 Waste, Atkins, W. H., LANL-13175.
1263 Los Alamos National Laboratory (LANL). 1997. Radionuclide Concentration in pinto beans,
1264 sweet corn, and zucchini squash grown in Los Alamos Canyon at Los Alamos National
1265 Laboratory, Fresquez, P. R., M. A. Mullen, L. Naranjo, and D. R. Armstrong, May 1997.
1266 Lucas, H.F., Jr. 1963. "A Fast and Accurate Survey Technique for Both Radon-222 and Radium-
1267 226," The Natural Radiation Environment, Proceedings of the International Symposium,
1268 William Rice University, Houston, TX, 315-319.
1269 Lucas, H.F. 1967. "A Radon Removal System for the NASA Lunar Sample Laboratory: Design
1270 and Discussion," Argonne National Laboratory Radiological Physics Division Annual
1271 Report, ANL-7360.
1272 McFarland, R.C. 1998a. "Determination of Alpha-Particle Counting Efficiency for Wipe-Test
1273 Samples," Radioactivity & Radiochemistry. 9:1, pp. 4-8.
1274 McFarland, R.C. 1998b. "Determination of Counting Efficiency for Wipe-Test Samples
1275 Containing Radionuclides that Emit High-Energy Beta Particles," Radioactivity &
1276 Radiochemistry'9:1, pp. 4-9.
1277 Mclnroy, J.F., H.A. Boyd, B.C. Eutsler, and D. Romero. 1985. "Part IV: Preparation and
1278 Analysis of the Tissue and Bones," Health Physics, 49:4, pp. 585-621.
1279 NCRP Report No. 50. 1976. Environmental Radiation Measurements.
1280 Obenhauf, R.H., R. Bostwick, W. Fithian, M. McCann, J.D. McCormack, and D. Selem SPEX
1281 CertiPrep Handbook of Sample Preparation and Handling, SPEX CertiPrep, Inc., Metuchen,
1282 NJ. Also, at http://www.spexcsp.com/spmain.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 12-45 DO NOT CITE OR QUOTE
-------
Laboratory Sample Preparation
1283 Pitard, F. F. 1993. Pierre Gy's Sampling Theory and Practice. CRC Press, Inc., Boca Raton, FL.
1284 Second Edition.
1285 RESL Analytical Chemistry Branch Procedures Manual, DOE-ID, collected and published in
1286 1982 and containing procedures beginning in 1969.
1287 Schilt, A. 1979. Perchloric Acid and Perchlorates. The G. Frederick Smith Chemical Company,
1288 Columbus, Ohio.
1289 Scwedt, G. 1997. The Essential Guide to Analytical Chemistry (Translation of the revised and
1290 updated German Second Edition. Translated by Brooks Haderlie), John Wiley & Sons,
1291 England.
1292 Sedlet, J. 1966. "Radon and Radium," in Treatise on Analytical Chemistry, Part II, Vol. IV,
1293 p219-366, edited by I.M. Kolthoff and PJ. Elving, John Wiley & Sons, Inc, New York.
1294 Shugar, G.J. and J.T. Ballinger. 1996. Chemical Technicians' Ready Reference Handbook.
1295 McGraw-Hill, New York.
1296 Sill, C.W., K.W. Puphal, and F.D. Hindman. 1974. "Simultaneous Determination of Alpha-
1297 Emitting Nuclides of Radium through Californium in Soil," Anal. Chem 46:12, pp. 1725-
1298 1737.
1299 Sill, C.W. 1975. "Some Problems in Measuring Plutonium in the Environment," Health Physics.
1300 Vol. 29, pp. 619-626.
1301 Stradling, G.N. and D.S. Popplewell. 1974. "Rapid Determination of Plutonium in Urine by
1302 Ultrafiltration," Int. J. Appl. Radiation Isotopes. Vol. 25, p 217.
1303 U.S. Department of Energy (DOE). 1990. EML Procedures Afam/a/(HASL-300-Ed.27), G. de
1304 Planque, Environmental Measurements Laboratory.
1305 U.S. Department of Energy (DOE). 1995. Brookhaven National Laboratory Site Environmental
1306 Report for Calendar Year 1995, Naidu, J. R., D. E. Paquette, G. L. Schroeder, BNL
1307 December, 1996.
1308 U.S. Department of Energy. 1997 (DOE).EML Procedures Manual (HASL-300-Ed.28), Edited
1309 by N. A. Chieco, Environmental Measurements Laboratory.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 12-46 DRAFT FOR PUBLIC COMMENT
-------
Laboratory Sample Preparation
1310 U.S. Environmental Protection Agency (EPA). 1979. Radiochemical Analytical Procedures for
1311 Analysis of Environmental Samples; F. B. Johns, P. B. Hahn, D. J. Thome, and E. W.
1312 Bretthauer, EMSL, March 1979.
1313 U.S. Environmental Protection Agency (EPA). 1980. Prescribed Procedures for Measurement of
1314 Radioactivity in Drinking Water. H. L. Krieger and E. L. Whittaker, EPA 600-4-80-032,
1315 August 1980.
1316 U.S. Environmental Protection Agency (EPA). 1989. Background Information Document:
1317 Procedures Approved for Demonstrating Compliance with 40 CFR Part 61, Subpart I. EPA
1318 520-1-89-001, Office of Radiation Programs, October, 1989.
1319 U.S. Environmental Protection Agency (EPA). 1990. Test Methods for Evaluating Solid Waste—
1320 Physical/Chemical Methods. SW-846, Third Edition, Method 3051.
1321 U.S. Environmental Protection Agency (EPA). 1992. Manual for the Certification of
1322 Laboratories Analyzing Drinking Water: Criteria and Procedures. Fourth Edition, EPA 814-
1323 B-92-002, Office of Ground Water and Drinking Water, Cincinnati, Ohio.
1324 Walter, P., S. Chalk, and H. Kingston. 1997. "Overview of Microwave-Assisted Sample
1325 Preparation." Chapter 2, Microwave-Enhanced Chemistry., H. Kingston and S. Haswell,
1326 editors, American Chemical Society, Washington, DC.
1327 Wang, C.H., Willis, D.L., and Loveland W.D. 1975. Radiotracer Methodology in the Biological,
1328 Environmental, and Physical Sciences. Prentice-Hall, Inc., New Jersey.
1329 Yamamoto, M., Komura, K., and Ueno, K. 1989. "Determination of Low-Level 226Ra in
1330 Environmental Water Samples by Alpha-Ray Spectrometry," Radiochimica Acta Vol. 46, pp.
1331 137-142.
1332 Zhang, H. and P. Dotson. 1998. The Use of Microwave Muffle Furnace for Dry Ashing Plant
1333 Tissue Samples. CEM Corporation. Also, Commun. Soil Sci. Plant Anal. 25:9&10, pp. 1321-
1334 1327 (1994).
1335 12.9.2 Other Sources
1336 American Society for Testing and Materials (ASTM) D5245. Standard Practice for Cleaning
1337 Laboratory Glassware, Plasticware, and Equipment Used in Microbiological Analyses.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 12-47 DO NOT CITE OR QUOTE
-------
Laboratory Sample Preparation
1338 American Society for Testing and Materials (ASTM) El 157. Standard Specification for
1339 Sampling and Testing of Reusable Laboratory Glassware.
1340 Delherz, R. 1957. Chemist-Analyst. Vol. 46, p. 11.
1341 Kahn, B., Shleien, B., and Weaver, C. 1972. "Environmental Experience with Radioactive
1342 Effluents From Operating Nuclear Power Plants," page 559 in Peaceful Uses of Atomic
1343 Energy, Vol. 11 (United Nations, New York). Also, Kahn, B. 1973. "Determination of
1344 Radioactive Nuclides in Water," page 1357 in Water and Water Pollution Handbook, Vol. 4
1345 Ciaccio, L. L., Ed. (M. Decker, New York).
1346 Krieger, H.L. and E.L. Whittaker. 1980. "Prescribed Procedures for Measurement of
1347 Radioactivity in Drinking Water," Environmental Monitoring and Support Laboratory,
1348 Cincinnati, OH, EPA-600/4-80-032.
1349 Laug, E.P. 1934. Ind. Eng. Chem., Anal Ed. Vol. 13, pp. 419.
1350 Shugar and Dean. 1990. The Chemist's Ready Reference Handbook, McGraw-Hill.
1351 U.S. Environmental Protection Agency. 1987. Eastern Environmental Radiation Facility
1352 Radiochemistry Procedures Manual. Compiled and edited by R. Lieberman, EPA 520-5-84-
1353 006, Office of Radiation Programs, August, 1987.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 12-48 DRAFT FOR PUBLIC COMMENT
-------
13 SAMPLE DISSOLUTION
2 13.1 Introduction
3 The overall success of any analytical procedure depends upon many factors, including proper
4 sample preparation, appropriate sample dissolution, and adequate separation and isolation of the
5 target analytes. This chapter describes sample dissolution techniques and strategies. Some of the
6 principles of dissolution are common to those of radiochemical separation that are described in
7 the next chapter, but their importance to dissolution is reviewed in this chapter.
8 Sample dissolution can be one of the biggest challenges facing the analytical chemist, because
9 most samples consist mainly of unknown compounds with unknown chemistries. There are many
10 factors for the analyst to consider: What are the data quality objective requirements for bias and
11 precision to meet the data quality objectives of the program? What is the nature of the sample; is
12 it refractory or is there only surface contamination? How effective is the dissolution technique?
13 Will any analyte be lost? Will the vessel be attacked? Will any of the reagents interfere in the
14 subsequent analysis or can any excess reagent be removed? What are the safety issues involved?
15 What are the labor and material costs? How much and what type of wastes are generated? The
16 challenge for the analyst is to balance these factors and to choose the method that is most
17 applicable to the material to be analyzed.
18 The objective of sample dissolution is to mix a solid or nonaqueous liquid sample quantitatively
19 with water to produce an aqueous solution (homogeneous mixture),so that subsequent separation
20 and analyses may be performed. Because very few natural or organic materials are water-soluble,
21 these materials routinely require the use of acids or fusion salts to bring them into solution. These
22 reagents typically achieve dissolution through an oxidation-reduction process that leaves the
23 constituent elements in a more soluble form. Moreover, because radiochemists routinely add
24 carriers or use the technique of isotope dilution to determine certain radioisotopes, dissolution
25 helps to ensure exchange between the carrier or isotopic tracer and the element or radioisotope to
26 be determined, although additional chemical treatment might be required to ensure exchange.
27 There are three main techniques for sample decomposition discussed in this chapter:
28 • Fusion;
29 • Wet ashing, acid leaching, or acid dissolution; and
30 • Microwave digestion.
31 Fusion and wet ashing techniques are used singly or in combination to decompose most samples
32 analyzed in radioanalytical laboratories. Generally, fusion techniques are used when a total
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 13-1 DO NOT CITE OR QUOTE
-------
Sample Dissolution
33 dissolution of a difficult sample matrix is required. Leaching techniques are used to determine
34 the soluble fraction of the radionuclide of interest under specific conditions. Because recent
35 advances in microwave vessel design have allowed for the use of larger samples, microwave
36 dissolution is becoming an important tool in the radiochemistry laboratory.
37 Because of the potential for injury and explosions, it is essential that proper laboratory safety
38 procedures be in place, the appropriate safety equipment be available, a safe work space be
39 provided, and that the laboratory personnel undergo the necessary training to ensure a safe
40 working environment before any of these methods are used.
41 Aspects of proper sample preparation, such as moisture removal, oxidation of organic matter, and
42 homogenization, were discussed in Chapter 12, Laboratory Sample Preparation. Fundamental
43 separation principles and techniques, such as complexation, solvent extraction, ion exchange, and
44 co-precipitation, are reviewed in Chapter 14, Separation Techniques.
45 There are many excellent references on the topic of sample dissolution, including A Handbook of
46 Decomposition Methods in Analytical Chemistry (Bock, 1979), Analytical Chemistry Handbook
47 (Dean, 1995), Methods for Decomposition in Inorganic Analysis (Sulcek and Povondra, 1989),
48 and "A Decomposition and Dissolution of Samples: Inorganic" (Bogen, 1978).
49 13.2 The Chemistry of Dissolution
50 In order to dissolve a sample completely, each insoluble component must be converted into a
51 soluble form. Several basic chemical methods are employed to accomplish complete dissolution
52 of the sample, but usually the tracer is added to the sample. An outline of the principles of these
53 chemical methods is provided in this section, but a complete description is available in Chapter
54 14 (Chemical Separations), where the principles are applied to a broader range of topics.
55 13.2.1 Solubility and the Solubility Product Constant, Ksp
56 The solubility data of many compounds, minerals, ores, and elements are available in reference
57 manuals. Solubilities typically are expressed in grams of substance per 100 mL of solvent,
58 although other units are used sometimes. The information is more complete for some substances
59 than others, and for many substances solubility is expressed only in general terms, such as
60 "soluble," "slightly soluble," or "insoluble." Many environmental samples consist of complex
61 mixtures of elements, compounds, minerals, or ores, most of which are insoluble and must be
62 treated chemically to dissolve completely. In some cases, the sample constituents are known to
63 the analyst, but often they are not. Solubility data might not be available even for known
MARLAP JULY 2001
DO NOT CITE OR QUOTE 13 -2 DRAFT FOR PUBLIC COMMENT
-------
Sample Dissolution
64 constituents, or the available data might be inadequate. Under these circumstances, sample
65 dissolution is not a simple case of following the solubilities of known substances. For known
66 constituents with solubility data, the solubilities indicate those that must be treated to complete
67 dissolution. This, in turn, provides a guide to the method of treatment of the sample. Given the
68 potential complexity of environmental samples, it is difficult to describe conditions for
69 dissolving all samples. Sometimes one method is used to dissolve one part of the sample while
70 another is used to dissolve the residue.
71 The solubility of many compounds in water is very low, on the order of small fractions of a
72 grams per 100 mL. Instead, the solubility is often expressed by a solubility product constant
73 (Ksp), an equilibrium constant for dissolution of the compound in water (see Section 14.8.3.1,
74 "Solubility and Solubility Product Constant"). The solubility product constant for strontium
75 carbonate, a highly insoluble salt (0.0006 g/100 mL), is the equilibrium constant for the process:
76 SrCO3(s) - Sr+2(aq) + CCV2(aq)
77 and is represented by:
78 Ksp = [Sr+2][CO3-2] = 1.6 x ID'9
79 The brackets indicate the molar concentration (moles/liter) of the respective ions dissolved in
80 water. The very small value of the constant results from the low concentration of dissolved ions,
81 and the compound is referred to as "insoluble." Chemical treatment is necessary sometimes to
82 dissolve the components of a compound in water. In this example, strontium carbonate requires
83 the addition of an acid to solubilize Sr+2. The next section describes chemical treatment to
84 dissolve compounds.
85 13.2.2 Chemical Exchange, Decomposition, and Simple Rearrangement Reactions
86 Chemical exchange, decomposition, and simple rearrangement reactions refer to one method for
87 solubilizing components of a sample. In this chemical process, the sample is treated to convert
88 insoluble components to a soluble chemical species using chemical exchange (double displace-
89 ment), decomposition, or simple rearrangement reactions rather than oxidation-reduction
90 processes or complex formations. Some fluxes solubilize sample components using chemical
91 exchange. Radium or strontium cations in radium or strontium carbonate (RaCO3 or SrCOj)
92 exchange the carbonate anion for the chloride ion on acid treatment with HC1 to produce the
93 soluble chlorides; the carbonic acid product decomposes to carbon dioxide and water:
94 RaCO3 + 2 HC1 - RaCl2 + H2CO3
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 13-3 DO NOT CITE OR QUOTE
-------
Sample Dissolution
95 H2CO3 - CO2 + H2O
96 and the net reaction is as follows:
97 RaCO3 + 2 HC1 - RaCl2 + CO2 + H2O
98 Sodium pyrosulfate fusion, for example, converts zirconia (ZrO2) into zirconium sulfate
99 [Zr(SO4)2], which is soluble in acid solution by a simple (nonoxidative) rearrangement of oxygen
100 atoms (Hahn, 1961, p. 81; Steinberg, 1960, p. 4):
101 ZrO2 + 2 Na2S2O7 - 2 Na2SO4 + Zr(SO4)2
102 Many environmental samples contain insoluble silicates, such as aluminum silicate [Al2(SiO3)3 or
103 A12O3 • 3SiO2], which can be converted into soluble silicates by fusion with sodium carbonate:
104 Al2(SiO3)3 + 4 Na2CO3 - 3 Na2SiO3 + 2 NaAlO2 + 4 CO2
105 Dissolution of radium from some ores depends on the exchange of anions associated with the
106 radium cation (sulfate for example) to generate a soluble compound. Extraction with nitric acid is
107 partly based on this process, generating soluble radium nitrate.
108 13.2.3 Oxidation-Reduction Processes
109 Oxidation-reduction (redox) processes play an important role in sample dissolution because
110 solubility is highly dependent not only on the chemical form of the element, but also on oxidation
ill state. Moreover, many radiochemical procedures require the addition of a carrier and isotope
112 tracer, and to achieve quantitative yields, there must be complete equilibration (isotopic
113 exchange) between the added isotopes and all chemical species present. Dissolution of the
114 sample in the presence of the appropriate carrier and/or tracer is one way to promote
115 equilibration by exposing all components of the analytical mixture to the same redox conditions.
116 An oxidation-reduction reaction is a reaction that redistributes electrons among the atoms,
117 molecules, or ions in the reaction. In some redox reactions, electrons actually are transferred from
118 one reacting species to another. In other redox reactions, electrons are not transferred completely
119 from one reacting species to another; the electron density about one atom decreases, while it
120 increases about another atom. A complete discussion of oxidation and reduction is found in
121 Section 14.2, "Oxidation/Reduction Processes."
122 Many oxidizing agents used in sample dissolution convert metals to a stable oxidation state
123 displacing hydrogen from hydrochloric, nitric, sulfuric, and perchloric acids. (This redox process
MARLAP JULY 2001
DO NOT CITE OR QUOTE 13 -4 DRAFT FOR PUBLIC COMMENT
-------
Sample Dissolution
124 often is referred to in the literature as nonoxidative hydrogen replacement by an active metal, but
125 it is a redox process where the metal is oxidized to a cation, usually in its highest oxidation state,
126 and the hydrogen ion is reduced to it elemental form.) Dissolution of uranium for analysis is an
127 example of hydrogen-ion displacement to produce a soluble substance (Grindler, 1962, p. 252):
128 U + 8 HNO3 - UO2(NO3)2 + 6 NO2 + 4 H2O
129 Prediction of the reactivity of a metal with acids is dependent on its position in the electromotive
130 force series (activity series). A discussion of the series appears in Section 13.4.1, "Acids and
131 Oxidants." In general, metals below hydrogen in the reduction series will displace hydrogen from
132 acid solution and be dissolved, while acids above the series will not. Perchloric acid offers a
133 particular advantage because very soluble perchlorate salts are formed.
134 Other important oxidizing processes depend on either oxidizing a lower, less soluble oxidation
135 state of a metal to a higher, more soluble state or oxidizing the counter anion to generate a more
136 soluble compound. Oxidation to a higher oxidation state is common when dissolving uranium
137 samples in acids or during treatment with fluxes. The uranyl ion (UO2+2) forms soluble salts—
138 such as chloride, nitrate, and perchlorate—with anions of the common acids (Grindler, 1962, p.
139 255 and pp. 9-14). (Complex-ion formation also plays a role in these dissolutions; see the next
140 section). Dissolution of oxides, sulfides, or halides of technetium by alkaline hydrogen peroxide
141 converts all oxidation states to the soluble technate salts (Cobble, 1964,p. 418):
142 2 TcO2 + 2 NaOH + 3 H2O2 - 2 NaTcO4 + 4 H2O
143 13.2.4 Complexation
144 The formation of complex ions (see also Section 14.3, "Complexation") is important in some
145 dissolution processes and usually occurs in conjunction with treatment by an acid, but can also
146 occur during fusion. Complexation increases solubility in the dissolution mixture and helps to
147 minimize hydrolysis of the cation. The solubility of radium sulfate in concentrated sulfuric acid
148 is the result of forming a complex-ion, Ra(SO4)2"2. The ability of both hydrochloric and
149 hydrofluoric acids to act as a solubilizing agent is dependent on their abilities to form stable
150 complex ions with cations. Refractory plutonium samples are solubilized in a nitric acid-
151 hydrofluoric acid solution forming cationic fluorocomplexes such as PuF+3 (Booman and Rein,
152 1962, p. 244). Numerous stable complexes of anions from solubilizing acids (HC1, HF, HNO3,
153 H2SO4, HC1O4) contribute to the dissolution of other radionuclides, such as americium, cobalt,
154 technetium, thorium, uranium, and zirconium (see Section 14.10, "Radiochemical Equilibrium").
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 13-5 DO NOT CITE OR QUOTE
-------
Sample Dissolution
155 The process effusion with sodium carbonate to solubilize uranium samples is also based on the
156 formation of UO2(CO3)2"4 after the metal is oxidized to U+6 (Grindler, 1962, p. 256).
157 13.2.5 Equilibrium: Carriers and Tracers
158 Carriers and tracers that are required for radiochemical separation and detection procedures
159 usually are added to samples before dissolution in order to subject them to the same chemical
160 treatment as the analyte. Addition as soon as practical promotes equilibration with the analyte.
161 The dissolution process tends to bring the carrier and tracers to the same oxidation state as the
162 analyte and ensures intimate mixing of all the components in solution. Acid mixtures also create
163 a large hydrogen-ion concentration that minimizes the tendency of cations to hydrolyze and
164 subsequently form insoluble complexes. Detailed discussions of carriers and tracers as well as
165 radiochemical equilibration are found in Section 14.9, "Carriers and Tracers" and Section 14.10,
166 "Radiochemical Equilibration." Knowledge of the behavior of carriers and tracers and of the
167 principles behind radiochemical equilibrium is very important, because the final form of the
168 analyte in solution is crucial to understanding their behavior, not only during solubilization of the
169 sample but also in the separation and detection steps of the analysis. During each of the steps in
170 the method, the analyst should be aware of the expected oxidation states of the analyte and its
171 tendency to hydrolyze, polymerize, and form complexes and radiocolloids, and other issues
172 during each step of the procedure. Knowledge of these processes will ensure that the analyst will
173 be able to recognize and address problems if they arise.
174 13.3 Fusion Techniques
175 Sample decomposition through fusion is most employed often for samples that are difficult to
176 dissolve in acids such as soils, sludges, silicates, and some mineral oxides. Fusion is accomp-
177 lished by heating a salt (the flux) mixed with a small amount of sample. The mixture is heated to
178 a temperature above the melting point of the salt, and the sample is allowed to react in the molten
179 mixture. When the reaction is completed, the mixture is allowed to cool to room temperature.
180 The fused sample is then dissolved, and the analysis is continued. Any residue remaining may be
181 treated by repeating the fusion with the same salt, performing a fusion with a different salt, wet
182 ashing, or any combination of the three.
183 Decomposition of the sample matrix depends on the high temperatures required to melt a flux
184 salt and the ratio of the flux salt to the sample. For a fusion to be successful, the sample must
185 contain chemically bound oxygen as in oxides, carbonates, and silicates. Samples that contain no
186 chemically bound oxygen, such as sulfides, metals, and organics, must be oxidized before the
187 fusion process.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 13-6 DRAFT FOR PUBLIC COMMENT
-------
Sample Dissolution
188 Samples to be fused should be oven-dried to remove moisture. Charring to remove organic
189 material is not usually necessary because samples with significant amounts of organic material
190 are typically dry ashed or wet ashed before fusion. Solid samples are ground mesh size to
191 increase the surface area, allowing the fusion process to proceed more readily. The sample must
192 be thoroughly mixed with the flux in an appropriate ratio. Generally, the crucible should never be
193 more than half-filled at the outset of the fusion process. Fusions may be performed using sand or
194 oil baths on a hot plate, in a muffle furnace, or over a burner. Crucibles are made of platinum,
195 zirconium, nickel, or porcelain (Table 13.1). The choice of heat source and crucible material
196 generally depends on the salt used for the fusion.
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
TABLE 13.1 — Common fusion fluxes
Flux
(mp, °C)
Na2S2O7 (403) or
K2S207(419)
NaOH (321)
or
KOH (404)
Na2CO3 (853) or
K2CO3 (903)
Na2O2
H3BO3 (169)
Na2B4O7 (878)
Li2B4O7 (920)
or
LiBO2 (845)
NH4HF2 (125) NaF
(992)
KF (857)
or
KHF, (239)
Fusion
Temperature, °C
Up to red heat
450-600
900-1,000
600
1,000-1,200
1,000-1,100
900
Type of
Crucible
Pt, quartz,
porcelain
Ni, Ag, glassy
carbon
Ni
Pt for short
periods (use lid)
Ni; Ag, Au, Zr;
Pt (<500 °C)
Pt
Pt
Pt, graphite
Pt
Types of Sample Decomposed
For insoluble oxides and oxide-containing samples,
particularly those of Al, Be, Ta, Ti, Zr, Pu, and the
rare earths.
For silicates, oxides, phosphates, and fluorides.
For silicates and silica-containing samples (clays,
minerals, rocks, glasses), refractory oxides, quartz,
and insoluble phosphates and sulfates.
For sulfides; acid-insoluble alloys of Fe, Ni, Cr, Mo,
W, and Li; Pt alloys; Cr, Sn, and Zn minerals.
For analysis of sand, aluminum silicates, titanite,
natural aluminum oxide (corundum), and enamels.
For A12O3; ZrO2 and zirconium ores, minerals of the
rare earths, Ti, Nb, and Ta, aluminum-containing
materials; iron ores and slags.
For almost anything except metals and sulfides. The
tetraborate salt is especially good for basic oxides and
some resistant silicates. The metaborate is better
suited for dissolving acidic oxides such as silica and
TiO2 and nearly all minerals.
For the removal of silicon, the destruction of silicates
and rare earth minerals, and the analysis of oxides of
Nb, Ta, Ti, and Zr.
Source: Dean (1995) and Bock (1979).
Fusions are heated slowly and evenly to prevent ignition of the sample before the reaction with
the molten salt can begin. It is especially important to raise the temperature slowly when using a
JULY 2001
DRAFT FOR PUBLIC COMMENT
13-7
MARLAP
DO NOT CITE OR QUOTE
-------
Sample Dissolution
221 gas flame because the evolution of water and gases is a common occurrence at the beginning of
222 the fusion, and hence a source of spattering. The crucible can be covered with a lid as an added
223 precaution. Sand and oil baths provide the most even source of heat, but they are difficult to
224 maintain at very high temperatures. Muffle furnaces provide an even source of heat, but when
225 using them it is difficult to monitor the progress of the reaction and impossible to work with the
226 sample during the fusion. Burners are used often as a convenient heat source although they make
227 it difficult to heat the sample evenly.
228 The maximum temperature employed varies considerably and depends on the sample and the
229 flux. In order to minimize attack of the crucible and decomposition of the flux, excessive
230 temperatures should be avoided. Once the salt has melted, the melt is swirled gently to monitor
231 the reaction. The fusion continues until visible signs of reaction are completed (e.g., formation of
232 gases, foaming, fumes). It is frequently difficult to decide when heating should be discontinued.
233 In ideal cases, a clear melt serves to indicate the completeness of sample decomposition. In other
234 cases, it is not as obvious, and the analyst must base the heating time on past experience with the
235 sample type.
236 The melt is swirled during cooling to spread it over the inside of the crucible. Thin layers of salt
237 on the sides of the crucible often will crack and flake into small pieces during cooling. These
238 small fragments are easier to dissolve.
239 After the sample has returned to room temperature, the fused material is dissolved. The solvent is
240 usually warm water or a dilute acid solution, depending on the salt. For example, dilute acid
241 typically would not be used to dissolve a carbonate fusion because of losses to spray caused by
242 release of CO2. The aqueous solution from the dissolution of the fusion melt should be examined
243 carefully for particles of undissolved sample. If undissolved particles are present, they should be
244 separated from solution by centrifugation or filtration, and a second fusion should be performed.
245 Several types of materials are used for crucibles, but platinum, other metals (Ni, Zr, Ag), and
246 graphite are most common. Graphite crucibles are a cost-effective alternative to metal crucibles;
247 they are disposable, which eliminates the need for cleaning and the possibility of cross-sample
248 contamination. Graphite crucibles are chemically inert and heat-resistant, although they do
249 oxidize slowly at temperatures above 430 °C. Graphite is not recommended for extremely
250 lengthy fusions or for reactions where the sample may be reduced. Platinum is probably the most
251 commonly used crucible material. It is virtually unaffected by any of the usual acids, including
252 hydrofluoric, and it is attacked only by concentrated phosphoric acid at very high temperatures,
253 and by sodium carbonate. However, it dissolves readily in mixtures of hydrochloric and nitric
254 acids (aqua regia), nitric acid containing added chlorides, or chlorine water or bromine water.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 13-8 DRAFT FOR PUBLIC COMMENT
-------
Sample Dissolution
255 Platinum offers adequate resistance toward molten alkali metal, borates, fluorides, nitrates, and
256 bisulfates. When using a platinum crucible, one should avoid using aqua regia, sodium peroxide,
257 free elements (C, P, S, Ag, Bi, Cu, Pb, Zn, Se, and Te), ammonium, chlorine and volatile
258 chlorides, sulfur dioxide, and gases with carbon content. Platinum crucibles can be cleaned in
259 boiling HC1, by hand cleaning with sea sand, or by performing a blank fusion with sodium
260 hydrogen sulfate.
261 Many kinds of salts are used for fusions. The lowest melting flux capable of reacting completely
262 with the sample is usually the optimum choice. Basic fluxes, such as the carbonates, the
263 hydroxides, and the borates, are used to attack acidic materials. Sodium or potassium nitrate may
264 be added to furnish an oxidizing agent when one is needed, as with the sulfides, certain oxides,
265 ferroalloys, and some silicate materials. The most effective alkaline oxidizing flux is sodium
266 peroxide; it is both a strong base and a powerful oxidizing agent. Because it is such a strong
267 alkali, sodium peroxide is often used even when no oxidant is required. Alternatively, acid fluxes
268 are the pyrosulfates, the acid fluorides, and boric acids. Table 13.1 lists several types effusions,
269 examples of salts used for each type of fusion, and the melting points of the salts.
270 SULFATE FUSION is useful for the conversion of ignited oxides to sulfates, but is generally an
271 ineffective approach for silicates. Sulfate fusion is particularly useful for BeO, Fe2O3, Cr2O3,
272 MoO3, TeO2, TiO2, ZrO2, Nb2O5, Ta2O5, PuO2, and rare earth oxides (Bock, 1979, pp. 77-82).
273 Pyrosulfate fusions are prepared routinely in the laboratory by heating a mixture of sodium or
274 potassium sulfate with a stoichiometric excess of sulfuric acid:
275 Na2SO4 + H2SO4 - [2NaHSO4] - Na2S2O7 + H2O
276 Na2S2O7 - Na2SO4 + SO31
277 Na2SO4 etc.
278 The rate of heating is increased with time until the sulfuric acid has volatilized and a clear
279 pyrosulfate fusion is obtained. It is important to note that pyrosulfate fusions are reversible and,
280 if needed, the fusion can be cooled, additional sulfuric acid added, and the fusion repeated as
281 many times as needed to dissolve the sample. The analyst must distinguish between insoluble
282 material that has not yet or will not dissolve, and material that has precipitated during the final
283 stages of a prolonged pyrosulfate fusion. In the latter situation the fusion must be cooled,
284 additional sulfuric acid added, and the sample refused until the precipitated material redissolves
285 and a clear melt is obtained. Otherwise, the precipitated material will be extremely difficult, if
286 not impossible, to dissolve in subsequent steps. Platinum or quartz crucibles are recommended
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 13-9 DO NOT CITE OR QUOTE
-------
Sample Dissolution
287 for this type effusion, with quartz being preferred for analysis of the platinum group metals.
288 After the melt is cooled and solidified, it should be dissolved in dilute sulfuric or hydrochloric
289 acid rather than in water to avoid hydrolysis and precipitation of Ti, Zr, etc. Niobium and
290 tantalum may precipitate even in the presence of more concentrated acid. In order to avoid
291 precipitation of Nb or Ta, concentrated sulfuric acid, tartaric acid, ammonium oxalate, hydrogen
292 peroxide, or hydrofluoric acid must be used. Mercury and the anions of volatile acids are largely
293 volatilized during these fusion procedures.
294 13.3.1 Alkali-Metal Hydroxide Fusions
295 Alkali metal hydroxide fusions are used for silicate analysis of ash and slag; for decomposition of
296 oxides, phosphates, and fluorides (Bock, 1979, pp. 102-108); and for dissolution of soils for
297 actinide analyses (Smith et al., 1995). Sodium hydroxide (NaOH) generally is used because of its
298 lower melting point, but potassium hydroxide (KOH) is just as effective. These fusions generally
299 are rapid, the melts are easy to dissolve in water, and the losses because of volatility are reduced
300 because of the low temperature of the melt. Nickel, silver, or glassy carbon crucibles are
301 recommended for this type of fusion. The maximum suggested temperature for nickel crucibles is
302 600 °C, but silver crucibles can be used up to 700 °C. Generally, crucibles made of platinum,
303 palladium, and their alloys should not be used with hydroxide fusions because the crucibles are
304 easily attacked in the presence of atmospheric oxygen. The weight ratio of fusion salt to sample
305 is normally 5-10:1. Typically, these fusions are carried out below red heat at 450 to 500 °C for
306 15 to 20 minutes, or sometimes at higher temperatures between 600 to 700 °C for 5 to 10
307 minutes. The solidified melt dissolves readily in water; and therefore, this step may be carried out
308 directly in the crucible, or alternatively in a nickel dish. Under no circumstances should the
309 dissolution be carried out in a glass vessel because the resulting concentrated hydroxide solution
310 attacks glass quite readily.
311 FUSION WITH SODIUM CARBONATE (Na2CO3) is a common procedure for decomposing silicates
312 (clays, rocks, mineral, slags, glasses, etc.), refractory oxides (magnesia, alumina, beryllia,
313 zirconia, quartz, etc.), and insoluble phosphates and sulfates (Bogen, 1978). The fusion may
314 result in the formation of a specific compound such as sodium aluminate, or it may simply
315 convert a refractory oxide into a condition where it is soluble in hydrochloric acid—this is the
316 method of choice when silica in a silicate is to be determined, because the fusion converts an
317 insoluble silicate into a mixture that is easily decomposed by hydrochloric acid ("M" represents a
318 metal in the equations below):
319 MSiO3 + Na2CO3 - Na2SiO3 + MCO3 (or MO + CO2),
MARLAP JULY 2001
DO NOT CITE OR QUOTE 13-10 DRAFT FOR PUBLIC COMMENT
-------
Sample Dissolution
320 followed by acidification to form a more soluble chloride salt,
321 Na2SiO3 + MCO3 + 4 HC1 + *H2O - H2SiO3 • *H2O + MC12 + CO2 + H2O + NaCl.
322 Carbonate fusions provide an oxidizing melt for the analysis of chromium, manganese, sulfur,
323 boron, and the platinum group metals. Organic material is destroyed, sometimes violently.
324 Na2CO3 generally is used because of its lower melting point. However, despite its higher melting
325 point and hygroscopic nature, K2CO3 is preferred for niobium and tantalum analyses because the
326 resulting potassium salts are soluble, whereas the analogous sodium salts are insoluble.
327 The required temperature and duration of the fusion depend on the nature of the sample as well
328 as particle size. In the typical carbonate fusion, 1 g of the powdered sample is mixed with 4 to 6 g
329 of sodium carbonate and heated at 900 to 1,000 °C for 10 to 30 minutes. Very refractory
330 materials may require heating at 1,200 °C for as long as 1 to 2 hours. Silica will begin to react at
331 500 °C, while barium sulfate and alumina react at temperatures above 700 °C. Notably, volatility
332 is a problem at these temperatures. Mercury and thallium are lost completely, while selenium,
333 arsenic, and iodine suffer considerable losses. Non-silicate samples should be dissolved in water,
334 while silicate samples should be treated with acid (Bock, 1979, p. 1 1 1).
335 Platinum crucibles are recommended, even though there is a 1 to 2 mg loss of platinum per
336 fusion. Attack on the crucible can be reduced significantly by covering the melt with a lid during
337 the fusion process, or virtually eliminated by working in an inert atmosphere. Moreover, nitrate is
338 often added to prevent the reduction of metals and the subsequent alloying with the platinum
339 crucibles. The platinum crucibles may be seriously attacked by samples containing high
340 concentrations of Fe2+, Fe3+, Sn4+, Pb2+, and compounds of Sb and As, because these ions are
341 reduced easily to the metallic state and then form intermetallic alloys with platinum that are not
342 easily dissolved in mineral acids. This problem is especially prevalent when fusion is carried out
343 in a gas flame. Porcelain crucibles are corroded rapidly and should be discarded after a single
344 use.
345 13.3.2 Boron Fusions
346 Fusions with boron compounds are recommended for analysis of sand, slag, aluminum silicates,
347 alumina (A12O3), iron and rare earth ores, zirconium dioxide, titanium, niobium, and tantalum.
348 Relatively large amounts of flux are required for these types effusions. The melts are quite
349 viscous and require swirling or stirring, so they should not be performed in a furnace. Platinum
350 crucibles should be used for these fusions because other materials are rapidly attacked by the
351 melt, even though some platinum is lost in each fusion.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 13-11 DO NOT CITE OR QUOTE
-------
Sample Dissolution
352 BORIC ACID (H3BO3) can be used to fuse a number of otherwise rather inert substances such as
353 sand, aluminum silicates, titanite, natural aluminum oxide (corundum), and enamels. Boric acid
354 fusions generally require 4 to 8 times as much reagent as sample. Initially, the mixture should be
355 heated cautiously while water is being driven off, then more strongly until gas evolution is
356 completed, and then more vigorously if the sample has yet to be fully decomposed. Normally, the
357 procedure is complete within 20 to 30 minutes. The cooled and solidified melt usually is
358 dissolved in dilute acid. Additionally, boric acid has one great advantage over all other fluxes in
359 that it can be completely removed by addition of methanol and subsequent volatilization of the
360 methyl ester.
361 Because MOLTEN SODIUM TETRABORATE (Na2B4O7) dissolves so many inorganic compounds, it is
362 an important analytical tool for dissolving very resistant substances. Fusions with sodium
363 tetraborate alone are useful for A12O3, ZrO2 and zirconium ores, minerals of the rare earths,
364 titanium, niobium, and tantalum, aluminum-containing materials, and iron ores and slags (Bock,
365 1979). Relatively large amounts of borax are mixed with the sample, and the fusion is carried out
366 at a relatively high temperature (1,000 to 1,200 °C) until the melt becomes clear. Thallium,
367 mercury, selenium, arsenic, and the halogens are volatilized under these conditions. Boric acid
368 can be removed from the melt as previously described. By dissolving the melt in dilute
369 hydrofluoric acid, calcium, thorium, and the rare earths can be separated from titanium, niobium,
370 and tantalum as insoluble fluorides.
371 Fluxes of LITHIUM TETRABORATE (Li2B4O7) are well suited for dissolving basic oxides such as
372 alumina (SiO2) and some resistant silicates. However, lithium metaborate, LiBO2, (or a mixture
373 of meta- and tetraborate) is more basic and better suited for dissolving acidic oxides such as
374 silica or titanium dioxide, although it is capable of dissolving nearly all minerals (Dean, 1995).
375 Platinum dishes normally are used for this type of fusion, but occasionally graphite crucibles are
376 advantageous because they can be heated rapidly by induction heating and because they are not
377 wetted by Li2B4O7 melts. The fusion melt typically is dissolved in dilute acid, usually nitric but
378 sometimes sulfuric. When easily hydrolyzed metal ions are present, it is recommended that
379 dissolution be carried out in the presence of EDTA or its sodium salt in 0.01 M HC1 (Bock, 1979,
380 p. 92). Moreover, when titanium is present, hydrogen peroxide can be used to help maintain the
381 titanium in solution.
382 13.3.3 Fluoride Fusions
383 Fluoride fusions are used for the removal of silicon, the destruction of silicates and rare earth
384 minerals, and the analysis of oxides of niobium, tantalum, titanium, and zirconium. Sill et al.
385 (1974) and Sill and Sill (1995) has described a method using potassium fluoride/potassium
MARLAP JULY 2001
DO NOT CITE OR QUOTE 13-12 DRAFT FOR PUBLIC COMMENT
-------
Sample Dissolution
386 pyrosulfate fusion for determining alpha-emitting nuclides in soil (see Sect. 13.8). Sulcek and
387 Povondra (1989) describe the isolation of the rare earth elements (REE) and thorium from
388 silicate materials and their minerals, especially monazite, through potassium hydrofluoride
389 fusion. The silicate matrix is first degraded by evaporation with HF, then the residue is fused
390 with tenfold excess flux, and finally the melt is digested with dilute acid. The resulting fluorides
391 (REE + Th + Ca + U) are filtered off, dissolved, and further separated by chromatography.
392 Platinum crucibles are recommended for fluoride fusions. Silicon and boron are volatilized
393 during these fusion procedures, and if the temperature is high enough, some molybdenum,
394 tantalum, and niobium also are lost. Residual fluoride can be a problem for subsequent analysis
395 of many elements such as aluminum, tin, beryllium, and zirconium. This excess fluoride usually
396 is removed by evaporation with sulfuric acid.
397 13.4 Wet Ashing and Acid Dissolution Techniques
398 "Wet ashing" and "acid dissolution" are terms used to describe sample decomposition using hot,
399 concentrated acid solutions. Because many inorganic matrices such as oxides, silicates, nitrides,
400 carbides, and borides can be difficult to dissolve completely, geological or ceramic samples can
401 be particularly challenging. Therefore, different acids are used alone or in combination to
402 decompose specific compounds that may be present in the sample. Few techniques will
403 completely decompose all types of samples. Many decomposition procedures use wet ashing to
404 dissolve the major portion of the sample but leave a minor fraction as residue. Whether or not
405 this residue requires additional treatment (by wet ashing or fusion) depends on the amount of
406 residue and whether it is expected to contain the radionuclides of interest. The residue should not
407 be discarded until all of the results have been reviewed and determined to be acceptable.
408 13.4.1 Acids and Oxidants
409 Numerous acids are commonly used in wet ashing procedures. Table 13.2 lists several acids and
410 the types of compounds they generally react with during acid dissolution. The electromotive
411 force series (Table 13.3) is a summary of oxidation-reduction half-reactions arranged in
412 decreasing oxidation strength and is also useful in selecting reagent systems (Dean, 1995). The
413 table allows one to predict which metals will dissolve in nonoxidizing acids, such as
414 hydrochloric, hydrobromic, hydrofluoric, phosphoric, dilute sulfuric, and dilute perchloric acid
415 The dissolution process is simply a replacement of hydrogen by the metal (Dean, 1995). In
416 practice, however, what actually occurs is influenced by a number of factors, and the behavior of
417 the metals cannot be predicted from the potentials alone. Generally, metals below hydrogen in
418 Table 13.3 displace hydrogen and dissolve in nonoxidizing acids with the evolution of hydrogen.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 13-13 DO NOT CITE OR QUOTE
-------
Sample Dissolution
419 Notable exceptions include the very slow dissolution by hydrochloric acid of lead, cobalt, nickel,
420 cadmium, and chromium. Also, lead is insoluble in sulfuric acid because of the formation of a
421 surface film of insoluble lead sulfate.
422 TABLE 13.2 — Examples of acids used for wet ashing
423 Acid
Typical Uses
TT , _ . . . , TT_ Removal of silicon and destruction of silicates; dissolves oxides of Nb, Ta, Ti, and Zr,
424 Hydrofluoric Acid, HF ,,_,,_
and Nb, and Ta ores.
... TT , .... . , TT^, Dissolves many carbonates, oxides, hydroxides, phosphates, borates, and sulfides;
425 Hydrochloric Acid, HC1 ,-,
dissolves cement.
426 Hydrobromic Acid, HBr Distillation of bromides (e.g., As, Sb, Sn, Se).
427 Hydroiodic Acid, HI Effective reducing agent; dissolves Sn (IV) oxide and Hg (II) sulfide.
.„„ „ ,,, . . ., TT „„ Dissolves oxides, hydroxides, carbonates, and various sulfide ores; hot concentrated
428 Sulfunc Acid, H2SO4 ., .„ ... • ,
acid will oxidize most organic compounds.
429 Phosphoric Acid, H3PO4 Dissolves A12O3, chrome ores, iron oxide ores, and slag.
430 Nitric Acid, HNO3 Oxidizes many metals and alloys to soluble nitrates; organic material oxidized slowly.
_ ,, . . . , TT™~ Extremely strong oxidizer; reacts violently or explosively to oxidize organic
431 Perchloric Acid, HC1O4 . , , ,, i
compounds; attacks nearly all metals.
432 TABLE 13.3 — Standard reduction potentials of
433 selected half- reactions at 25 °C
434 Half-Reaction
435 Ag^+ + e" = Ag+
436 S2O82" + 2e" = 2SO42"
437 HN3 + 3H+ + 2e" = NH4+ + N2
438 Ce4+ + e = Ce3+
439 MnO4" + 4H+ + 3e" = MnO2 (c) + 2H2O
440 2HC1O + 2H+ + 2e" = C12 + 2H2O
44 1 2HBrO + 2H+ + 2e" = Br2 + 2H2O
442 NiO2+ 4H+ + 2e = Ni2+ + 2H2O
443 Bi2O4 (bismuthate) + 4H+ + 2e = 2BiO+ + 2H
444 MnO4" + 8H+ + 5e" = Mn2+ + 4H2O
445 2BrO3' + 12H+ + lOe" = Br2 + 6H2O
446 PbO2 + 4H+ + 2e" = Pb2+ + 2H2O
447 Cr2O72' + 14H+ + 6e" = 2Cr3+ + 7H2O
448 C12 + 2e" = 2C1"
449 2HNO2 + 4H+ + 4e = N2O + 3H2O
E° (volts)
1.980
1 96
1 96
1.72
1.70
1630
1604
1.593
2O 1.59
1 51
1478
1.468
1.36
1 3583
1.297
MARLAP
DO NOT CITE OR QUOTE
13-14
JULY 2001
DRAFT FOR PUBLIC COMMENT
-------
Sample Dissolution
Half-Reaction
E° (volts)
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
MnO2 + 4H++ 2e = Mn2+ + 2H2O .......................... 1.23
O2 + 4H+ + 4e = 2H2O ................................... 1.229
C1O4- + 2H+ + 2e = ClO-3 + H2O ........................... 1.201
2IO3" + 12H+ + 10e" = I2 + 3H2O ........................... 1.195
N2O4 + 2H+ + 2e = 2HNO3 ............................... 1.07
2ICl-2 + 2e- = 4C1' + 12 ................................... 1.07
Br2 (Iq) + 2e" = 2Bf ..................................... 1.065
N2O4 + 4H+ + 4e = 2ND + 2H2O ........................... 1.039
HNO2 + H++ e = NO + H2O .............................. 0.996
NO3- + 4H+ + 3e = NO + 2H2O ............................ 0.957
NO3- + 3H+ + 2e = HNO2 + H2O ........................... 0.94
2Hg2+ + 2e- = Hg22+ ...................................... 0.911
Cu2+ + T + e" = Cul ...................................... 0.861
OsO4 (c) + 8H+ + 8e = Os + 4H2O ......................... 0.84
Ag+ + e=Ag .......................................... 0.7991
Hg22+ + 2e = 2Hg ....................................... 0.7960
Fe3+ + e" = Fe2+ ......................................... 0.771
H2SeO3 + 4H+ + 4e = Se + 3H2O ........................... 0.739
HN3+llH+ + 8e- = 2NH4+ ................................ 0.695
O2 + 2H+ + 2e= H2O2 ................................... 0.695
Ag2SO4 + 2e = 2Ag + SO42' ............................... 0.654
Cu2+ + Br + e- = CuBr (c) ................................ 0.654
2HgCl2 + 2e = Hg2Cl2 (c) + 2C1' ........................... 0.63
Sb2O5 + 6H+ + 4e = 2SbO+ + 3H2O ......................... 0.605
H3AsO4 + 2H+ + 2e = HAsO2 + 2 H2O ...................... 0.560
TeOOH+ + 3H+ + 4e" = Te + 2H2O ......................... 0.559
Cu2+ + Cr + e = CuCl (c) ................................. 0.559
r2 + 2e = 3r ........................................... 0.536
I2 + 2e" = 21 ........................................... 0.536
Cu+ + e = Cu .......................................... 0.53
4H2SO3 + 4H+ + 6e' = S4O62- + 6H2O ........................ 0.507
Ag2CrO4 + 2e- = 2Ag + CrO42 ............................. 0.449
2H2SO3 + 2H+ + 4e" = S2O32' + 3H2O ........................ 0.400
UO2+ + 4H+ + e" = U4+ + 2H2O ............................. 0.38
Cu2+ + 2e- = Cu ......................................... 0.340
VO2+ + 2H+ + e' = V3+ + H2O .............................. 0.337
BiO+ + 2H+ + 3e = Bi + H2O .............................. 0.32
UO22+ + 4H+ + 2e = U4+ + 2H2O ........................... 0.27
Hg7Cl7 (c) + 2e" = 2Hg + 2C1" ............................. 0.2676
JULY 2001
DRAFT FOR PUBLIC COMMENT
13-15
MARLAP
DO NOT CITE OR QUOTE
-------
Sample Dissolution
Half-Reaction E° (volts)
489 AgCl + e" = Ag + Cr 0.2223
490 SbO+ + 2H+ + 3e = Sb + H2O 0.212
491 CuCl32' + e = Cu + 3C1' 0.178
492 SO/' + 4H+ + 2e- = H2SO3 + H2O 0.158
493 Sn4+ + 2e = Sn2+ 0.15
494 CuCl + e-= Cu + Cr 0.121
495 TiO2++ 2H++ e-= Ti3++ H2O 0.100
496 S4O62- + 2e = 2S2O32- 0.08
497 2H+ + 2e" = H2 0.0000
498 Hg2I2 + 2e = 2Hg + 21- -0.0405
499 Pb2+ + 2e = Pb -0.125
500 Sn2+ + 2e = Sn -0.136
501 Agl + e = Ag + T -0.1522
502 V3+ + e = V2+ -0.255
503 Ni2+ + 2e- = Ni -0.257
504 Co2+ + 2e- = Co -0.277
505 PbSO4 + 2e = Pb + SO42' -0.3505
506 Cd2+ + 2e- = Cd -0.4025
507 Cr* + e' = Cr* -0.424
508 Fe2+ + 2e" = Fe -0.44
509 H3PO3 + 2H+ + 2e" = HPH2O2 + H2O -0.499
510 U4+ + e=U3+ -0.52
511 Zn2+ + 2e = Zn -0.7626
512 Mn2++ 2e = Mn -1.18
513 A13+ + 3e" = Al -1.67
514 Mg2+ + 2e" = Mg -2.356
515 Na+ + e=Na -2.714
516 K+ + e=K -2.925
517 Li+ + e" = Li -3.045
518 Source: Dean, 1995.
519 Oxidizing acids, such as nitric acid, hot concentrated sulfuric acid, or hot concentrated perchloric
520 acid, are used to dissolve metals above hydrogen. For nitric acid, the potential of the nitrate ion-
521 nitric oxide couple can be employed as a rough estimate of the solvent power. For aqua regia, the
522 presence of free chlorine ions allows one to make predictions based upon the potential of the
523 chlorine-chloride couple, although NOC1 also plays a significant role. Some oxidizing acids
524 exhibit a passivating effect with transition elements such as chromium and pure tungsten,
525 resulting in a very slow attack because of the formation of an insoluble surface film of the oxide
MARLAP JULY 2001
DO NOT CITE OR QUOTE 13-16 DRAFT FOR PUBLIC COMMENT
-------
Sample Dissolution
526 in the acid (Bogen, 1978). Moreover, oxides are often resistant to dissolution in oxidizing acids
527 and, in fact, dissolve much more readily in nonoxidizing acids. A common example is ferric
528 oxide, which is readily soluble in hydrochloric acid but is relatively inert in nitric acid.
529 However, insoluble oxides of the lower oxidation states of an element sometime dissolve in
530 oxidizing acids with concurrent oxidation of the element. For example, UO2 and U3O8 dissolve
531 readily in nitric acid to produce a solution of uranyl ion (UO2+2).
532 HYDROFLUORIC ACID. The most important property of HF is its ability to dissolve silica and
533 other silicates. For example:
534 SiO2 + 6HF - H2SiF6 + 2H2O
535 whereby the fluorosilicic acid formed dissociates into gaseous silicon tetrafluoride and hydrogen
536 fluoride upon heating:
537 H2SiF6-SiF4t+2HF
538 HF also exhibits pronounced complexing properties that are widely used in analytical chemistry.
539 Hydrofluoric acid prevents the formation of sparingly soluble hydrolytic products in solution,
540 especially of compounds of elements from the IVth to VIth groups of the periodic table (Sulcek
541 and Povondra, 1989). In the presence of fluoride, soluble hydrolytic products that are often
542 polymeric depolymerize to form reactive monomeric species suitable for further analytical
543 operations. Formation of colloidal solutions is avoided and the stability of solutions is increased
544 even with compounds of elements that are hydrolyzed easily in aqueous solution (e.g., Si, Sn, Ti,
545 Zr, Hf, Mb, Ta, and Pa).
546 HF also exhibits pronounced complexing properties that are widely used in analytical chemistry.
547 Hydrofluoric acid prevents the formation of sparingly soluble hydrolytic products in solution,
548 especially of compounds of elements from the IVth to VIth groups of the periodic table (Sulcek
549 and Povondra, 1989). In the presence of fluoride, soluble hydrolytic products that are often
550 polymeric depolymerize to form reactive monomeric species suitable for further analytical
551 operations. Formation of colloidal solutions is avoided and the stability of solutions is increased
552 even with compounds of elements that are easily hydrolyzed in aqueous solution (e.g., Si, Sn, Ti,
553 Zr, Hf, Nb, Ta, and Pa).
554 HF should never be used or stored in glass containers. Platinum containers are preferred, and
555 Teflon is acceptable as long as the temperature does not exceed 250 °C; the constant boiling
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 13-17 DO NOT CITE OR QUOTE
-------
Sample Dissolution
556 azeotrope boils at 112 °C. HF works most effectively when used alone. Samples should be
557 ground to a fine powder to increase the surface area and moistened with water to prevent losses
558 as dust and spray when the acid is added to the sample. After the addition of HF, the sample is
559 allowed to stand overnight to dissolve the silicates. However, the reaction can be sped up by
560 heating the solution. Because it is such a strong complexing agent, excess fluoride ion can cause
561 problems with many chemical reactions. Residual fluoride is usually removed by evaporation to
562 fumes in a low-volatility acid (e.g., H2SO4, HNO3, HC1O4) or, in extreme cases, excess fluoride
563 ion can be removed by fusing the residue with K2S2O7 or by the addition of quartz (SiO2).
564 HYDROCHLORIC ACID (HC1) is one of the most widely used acids for wet ashing samples because
565 of the wide range of compounds it reacts with and the low boiling point of the azeotrope (110
566 °C); after a period of heating in an open container, a constant boiling 6M solution remains. HC1
567 forms strong complexes with gold (HI), titanium (HI), and mercury (II). The concentrated acid
568 will also complex iron (in), gallium (HI), indium (HI), and tin (IV). Most chloride compounds are
569 readily soluble in water except for silver chloride, mercury chloride, titanium chloride, and lead
570 chloride. HC1 can be oxidized to form chlorine gas by manganese dioxide, permanganate, and
571 persulfate. While HC1 dissolves many carbonates, oxides, hydroxides, phosphates, borates,
572 sulfides, and cement, it does not dissolve the following:
573 • Most silicates or ignited oxides of Al, Be, Cr, Fe, Ti, Zr, or Th;
574 • Oxides of Sn, Sb, Nb, or Ta;
575 • Zr phosphate;
576 • Sulfates of Sr, Ba, Ra, or Pb;
577 • Alkaline earth fluorides;
578 • Sulfides of Hg; or
579 • Ores of Nb, Ta, U, or Th.
580 The dissolution behavior of specific actinides by hydrochloric acid is discussed by Sulcek and
581 Povondra(1989):
582 "The rate of decomposition of oxidic uranium ores depends on the U(VI)/U(IV) ratio. The so-
583 called uranium blacks with minimal contents of U(IV) are even dissolved in dilute
584 hydrochloric acid. Uraninite (UO2) requires an oxidizing mixture of hydrochloric acid with
585 hydrogen peroxide, chlorate, or nitric acid for dissolution. Uranium and thorium compounds
586 cannot be completely leached from granites by hydrochloric acid. Natural and synthetic
587 thorium dioxides are highly resistant toward hydrochloric acid and must be decomposed in a
588 pressure vessel. Binary phosphates of uranyl and divalent cations, e.g., autunite and tobernite,
589 are dissolved without difficulties. On the other hand, phosphates of thorium, tetravalent
MARLAP JULY 2001
DO NOT CITE OR QUOTE 13-18 DRAFT FOR PUBLIC COMMENT
-------
Sample Dissolution
590 uranium, and the rare earths (monazite and xenotime) are only negligibly attacked, even with
591 the concentrated acid."
592 Arsenic (in), antimony (HI), germanium (HI), and selenium (IV) are easily volatilized in HC1
593 solutions, while mercury (II), tin (IV), and rhenium (VII) are volatilized in the latter stages of
594 evaporation. Glass is the preferred container for HC1 solutions.
595 HYDROBROMIC ACID (HBr) has no important advantages over HC1 for wet ashing samples. HBr
596 forms an azeotrope with water containing 47.6 percent w/w of HBr, boiling at 124.3 °C. HBr is
597 used to distill off volatile bromides of arsenic, antimony, tin, and selenium. HBr can also be used
598 as a complexing agent for liquid-liquid extractions of gold, titanium, and indium.
599 HYDROIODIC ACID (HI) is readily oxidized and often appears as a yellowish-brown liquid
600 because of free iodine. HI is most often used as a reducing agent during dissolutions. HI also
601 dissolves tin (IV) oxide, and complexes and dissolves mercury (II) sulfide. HI forms an azeotrope
602 with water containing 56.9 percent w/w of HI, boiling at 127 °C.
603 SULFURIC ACID (H2SO4) is another widely used acid for sample decomposition. Part of its
604 effectiveness is due to its high boiling point (about 340 °C). Oxides, hydroxides, carbonates, and
605 sulfide ores can be dissolved in H2SO4. The boiling point can be raised by the addition of sodium
606 or potassium sulfate to improve the attack on ignited oxides, although silicates will still not
607 dissolve. H2SO4 is not appropriate when calcium is a major constituent because of the low
608 solubility of CaSO4. Other inorganic sulfates are typically soluble in water, with the notable
609 exceptions of strontium, barium, radium, and lead.
610 Dilute H2SO4 does not exhibit oxidizing properties, but the concentrated acid will oxidize many
611 elements and almost all organic compounds. Oxidation of organic compounds in H2SO4 is a slow
612 reaction with a tendency to form indestructible charred residues. Moreover, because of the high
613 boiling point of H2SO4, there is an increased risk of losses because of volatilization. Iodine can
614 be distilled quantitatively, and boron, mercury, selenium, osmium, ruthenium, and rhenium may
615 be lost to some extent. The method of choice is to oxidize the organic substances with HNO3,
616 volatilize the nitric acid, add H2SO4 until charred, followed by HNO3 again, repeating the process
617 until the sample will not char with either HNO3 or H2SO4. Dissolution is then continued with
618 HC1O4.
619 Glass, quartz, platinum, and porcelain are resistant to H2SO4 up to the boiling point. Teflon
620 decomposes at 300 °C, below the boiling point, and, therefore, is not recommended for
621 applications involving H2SO4 that require elevated temperature.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 13-19 DO NOT CITE OR QUOTE
-------
Sample Dissolution
622 PHOSPHORIC ACID (H3PO4) seldom is used for wet ashing because the residual phosphates
623 interfere with many procedures. H3PO4 attacks glass, although glass containers are usually
624 acceptable at temperatures below 300 °C. Alumina, chromium ores, iron oxide ores, and slags
625 can be dissolved in H3PO4. The acid also has been used to dissolve silicates selectively without
626 attacking quartz.
627 NITRIC ACID (HNO3) is one of the most widely used oxidizing acids for sample decomposition.
628 Most metals and alloys are oxidized to nitrates, which are usually very soluble in water, although
629 many metals exhibit a pronounced tendency to hydrolyze in nitric acid solution. Nitric acid does
630 not attack gold, hafnium, tantalum, zirconium, and the metals of the platinum group (except
631 palladium). Aluminum, boron, chromium, gallium, indium, niobium, thorium, titanium, calcium,
632 magnesium, and iron form a layer of insoluble oxide when treated with HNO3 and are thereby
633 pacified and do not dissolve in the concentrated acid. However, calcium, magnesium, and iron
634 will dissolve in more dilute acid.
635 Complexing agents (e.g., Cl", F", citrate, tartrate) can assist HNO3 in dissolving most metals. For
636 example, Sulcek and Povondra (1989) describe the decomposition of thorium and uranium
637 dioxides in nitric acid, which is catalytically accelerated by the addition of 0.05 to 0.1 M FTP.
638 They report that a solid solution of the mixed oxides (Pu, U)O2 or PuO2 ignited at temperatures
639 below 800 °C behaves analogously.
640 Although nitric acid is a good oxidizing agent, it usually boils away before sample oxidation is
641 complete. Oxidation of organic materials proceeds slowly and is usually accomplished by
642 repeatedly heating the solution to HNO3 fumes. Refluxing in the concentrated acid can help
643 facilitate the treatment, but HNO3 is seldom used alone to decompose organic materials.
644 PERCHLORIC ACID (HC1O4). Hot concentrated solutions of HC1O4 act as a powerful oxidizer, but
645 dilute aqueous solutions are not oxidizing. Hot concentrated HC1O4 will attack nearly all metals
646 (except gold and platinum group metals) and oxidize them to the highest oxidation state, except
647 for lead and manganese, which are oxidized only to the +2 oxidation state. Perchloric acid is an
648 excellent solvent for stainless steel, oxidizing the chromium and vanadium to the hexavalent and
649 pentavalent acids, respectively. Many nonmetals also will react with HC1O4. Because of the
650 violence of the oxidation reactions, HC1O4 is rarely used alone for the destruction of organic
651 materials. H2SO4 or HNO3 are used to dilute the solution and break down easily oxidized material
652 before HC1O4 becomes an oxidizer above 160 °C.
653 The concentrated acid is a dangerous oxidant that can explode violently. The following are
654 examples of some reactions with HC1O4 that should never be attempted:
MARLAP JULY 2001
DO NOT CITE OR QUOTE 13-20 DRAFT FOR PUBLIC COMMENT
-------
Sample Dissolution
655 • Heating Bi metal and alloys with concentrated acid.
656 • Dissolving metals (e.g., steel) in concentrated acid when gas-phase hydrogen becomes
657 heated.
658 • Heating uranium turnings or powder in concentrated acid.
659 • Heating finely divided aluminum and silicon in concentrated acid.
660 • Heating antimony or antimony (HI) compounds in HC1O4
661 • Mixing HC1O4 with hydrazine or hydroxylamine.
662 • Mixing HC1O4 with hypophosphates.
663 • Mixing HC1O4 with fats, oils, greases, or waxes.
664 • Evaporating solutions of metal salts to dryness in HC1O4.
665 • Evaporating alcoholic filtrates after collection of KC1O4 precipitates.
666 • Heating HC1O4 with cellulose, sugar, and polyhydroxy alcohols.
667 • Heating HC1O4 with N-heterocyclic compounds.
668 • Mixing HC1O4 with any dehydrating agent.
669 Perchloric acid vapor should never be allowed to come in contact with organic materials such as
670 rubber stoppers. The acid should be stored only in glass bottles. Splashed or spilled acid should
671 be diluted with water immediately and mopped up with a woolen cloth, never cotton. HC1O4
672 should only be used only in specially designed fume hoods incorporating a washdown system.
673 Acid dissolutions involving HC1O4 should only be performed by analysts experienced in working
674 with this acid. When any procedure is designed, the experimental details should be recorded
675 exactly. These records are used to develop a detailed SOP that must be followed exactly to
676 ensure the safety of the analyst (Schilt, 1979).
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 13-21 DO NOT CITE OR QUOTE
-------
Sample Dissolution
677 AQUA REGIA. One part concentrated HNO3 and 3 parts concentrated HC1 (volume/volume) are
678 combined to form aqua regia:
679 3HC1 + HNO3 - NOC1 + C12 + 2H2O
680 However, the interaction of these two acids is much more complex than indicated by this simple
681 equation. Both the elemental chlorine and the trivalent nitrogen of the nitrosyl chloride exhibit
682 oxidizing effects, as do other unstable products formed during the reaction of these two acids.
683 Coupled with the catalytic effect of C12 and NOC1, this mixture combines the acidity and
684 complexing power of the chloride ions. The solution is more effective if allowed to stand for 10
685 to 20 minutes after it is prepared.
686 Aqua regia dissolves sulfides, phosphates, and many metals and alloys including gold, platinum,
687 and palladium. Ammonium salts are decomposed in this acid mixture. Aqua regia volatilizes
688 osmium as the tetroxide; has little effect on rhodium, iridium, and ruthenium; and has no effect
689 on titanium. Oxidic uranium ores with uraninite and synthetic mixed oxides (U3O8) are dissolved
690 in aqua regia, with oxidation of the uranium (VI) to UO22+ ions (Sulcek and Povondra, 1989).
691 However, this dissolution procedure is insufficient for poor ores; the resistant, insoluble fraction
692 must be further attacked (e.g., by sodium peroxide or borate fusion) or by mixed-acid digestion
693 with HF, HNO3, and HC1O4.
694 Oxysalts, such as KMnO4 (potassium permanganate) and K2Cr2O7 (potassium dichromate), are
695 commonly not used to solubilize or wet ash environmental samples for radiochemical analysis
696 because of their limited ability to oxidize metals and the residue that they leave in the sample
697 mixture. These oxysalts are more commonly used to oxidize organic compounds.
698 POTASSIUM PERMANGANATE (KMnO4) is a strong oxidizer whose use is limited primarily to the
699 decomposition of organic substances and mixtures, although it oxidizes metals such as mercury
700 to the ionic form. Oxidation can be performed in an acid, neutral, or basic medium; near-neutral
701 or basic solutions produce an insoluble residue of manganese dioxide (MnO2) that can be
702 removed by filtration. Oxidation in acid media leaves the manganese (II) ion in solution, which
703 might interfere with additional chemical procedures or analyses. Extreme caution must be taken
704 when using this reagent because KMnO4 reacts violently with some organic substances such as
705 acetic acid and glycerol, with some metals such as antimony and arsenic, and with common
706 laboratory reagents such as hydrochloric acid and hydrogen peroxide.
707 POTASSIUM BICHROMATE (K2Cr2O7) is a strong oxidizing agent for organic compounds but is not
708 as strong as KMnO4. K2Cr2O7 has been used to determine carbon and halogen in organic
MARLAP JULY 2001
DO NOT CITE OR QUOTE 13 -22 DRAFT FOR PUBLIC COMMENT
-------
Sample Dissolution
709 materials, but the procedure is not used extensively. K2Cr2O7 is commonly mixed with sulfuric
710 acid and heated. The chromium (HI) ion remains after sample oxidation and this might interfere
711 with other chemical procedures or analyses. K2Cr2O7 can react violently with certain organic
712 substances such as ethanol and might ignite in the presence of boron. Caution also must be
713 observed in handling this oxidizing agent because of human safety concerns, particularly with the
714 hexavalent form of chromium.
715 SODIUM BROMATE (NaBrO3) is an oxidizing agent for organic compounds but is not used for
716 metals. Unlike KMnO4 and K2Cr2O7, the bromate ion can be removed from solution after sample
717 oxidation by boiling with excess HC1 to produce water and Br2. Caution must be observed when
718 using this oxidizing agent because it can react violently with some organic and inorganic
719 substances.
720 13.4.2 Acid Digestion Bombs
721 Some materials that would not be totally dissolved by acid digestion in an open vessel on a
722 hotplate, can be completely dissolved in an acid digestion bomb. These pressure vessels hold
723 strong mineral acids or alkalies at temperatures well above normal boiling points, thereby
724 allowing one to obtain complete digestion or dissolution of samples that would react slowly or
725 incompletely at atmospheric pressure. Sample dissolution is obtained without losing volatile
726 elements and without adding contaminants from the digestion vessel. Ores, rock samples, glass
727 and other inorganic samples can be dissolved quickly using strong mineral acids such as HF,
728 HC1, H2SO4, HNO3, or aqua regia.
729 These sealed pressure vessels are lined with Teflon, which offers resistance to cross-
730 contamination between samples and to attack by HF. In all reactions, the bomb must never be
731 completely filled; there must be adequate vapor space above the contents. When working with
732 inorganic materials, the total volume of sample plus reagents must never exceed two-thirds of the
733 capacity of the bomb. Moreover, many organic materials can be treated satisfactorily in these
734 bombs, but critical attention must be given to the nature of the sample as well to possible
735 explosive reactions with the digestion media.
736 13.4.3 Is it Dissolved?
737 Following aggressive acid digestion and even fusion, the analyst often must determine if the
738 sample has indeed been dissolved. This determination is first made through visual inspection for
739 particulate matter in the acid leachate or dissolved fusion melt. If a residue is observed, this
740 residue can be physically separated and subsequently fused or treated in an acid digestion bomb
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 13-23 DO NOT CITE OR QUOTE
-------
Sample Dissolution
741 to determine if any analyte was left behind. Sometimes these residues are inconsequential and
742 contain no analyte of interest. In other cases, residues may consist of materials such as zircons or
743 other minerals that can contain trapped uranium, thorium, etc. Even if no particles are readily
744 observed, small undissolved particles that are invisible to the naked eye may be present.
745 Therefore, the analyst may choose to filter the sample through a 0.22 to 0.45 jim filter, and then
746 count the filter for gross a, p, and y activity to determine if any activity has been left behind in
747 the residue. However, this approach is applicable only for samples that contain elevated levels of
748 radioactivity. Finally, for those cases where the laboratory has decided to perform an acid
749 digestion rather than a total dissolution fusion, it is advisable to perform a total dissolution on a
750 subset of the samples and compare the results to those obtained from the acid digestion. This
751 check will help to substantiate that the acid digestion approach is adequate for the particular
752 sample matrix.
753 13.5 Microwave Digestion
754 Microwave energy as a heat source for sample digestion was first described more than 20 years
755 ago (Abu-Samra et al., 1975). Its popularity is derived from the fact that it is faster, cleaner, more
756 reproducible, and more accurate than traditional hot-plate digestion. However, until recently, this
757 technology has had limited application in the radiochemical laboratory because of constraints on
758 sample size resulting from vessel pressure limitations. Because of this drawback, microwave
759 dissolution was not practical for many radiochemical procedures where larger sample sizes are
760 dictated to achieve required detection limits. However, recent advances in vessel design and
761 improved detection methods, such as ICP-MS (inductively coupled plasma-mass spectrometry)
762 and ion chromatography have eliminated this disadvantage, and microwave dissolution is
763 becoming an important tool for today's radiochemists (Smith and Yaeger, 1996; Alvarado et al.,
764 1996). A series of articles in the journal Spectroscopy describes recent advances in microwave
765 dissolution technology (Kammin and Brandt, 1989; Grillo, 1989 and 1990; Oilman and
766 Engelhardt, 1989; Lautenschlager, 1989; Noltner et al., 1990), and Dean (1995) presents a
767 synopsis of current microwave theory and technology in the Analytical Chemistry Handbook.
768 Moreover, Introduction to Microwave Sample Preparation: Theory and Practice by Kingston
769 and lassie (1988) and Microwave-Enhanced Chemistry—Fundamentals, Sample Preparation,
770 and Applications by Kingston and Haswell (1997), are excellent resources for this topic.
771 Some example protocols for various media are given in ASTM standards: "Standard Practice for
772 Acid-Extraction from Sediments Using Closed Vessel Microwave Heating" (ASTM D5258)
773 describes the decomposition of soil and sediment samples for subsequent analyte extraction;
774 "Standard Practice for Sample Digestion Using Closed Vessel Microwave Heating Technique for
775 the Determination of Total Metals in Water" (ASTM D4309) addresses the decomposition of
MARLAP JULY 2001
DO NOT CITE OR QUOTE 13 -24 DRAFT FOR PUBLIC COMMENT
-------
Sample Dissolution
776 surface, saline, domestic, and industrial waste water samples; and "Standard Practice for
777 Microwave Digestion of Industrial Furnace Feedstreams for Trace Element Analysis" (ASTM
778 D5513) covers the multistage decomposition of samples of cement raw feed materials, waste-
779 derived fuels, and other industrial feedstreams for subsequent trace metal analysis. A method for
780 acid digestion of siliceous and organically based matrices is given in EPA (1996).
781 There are various brands and models of microwave instruments that may be satisfactory
782 depending on sample preparation considerations. The three main approaches to microwave
783 dissolution are: focused open-vessel, low-pressure closed-vessel, and high-pressure closed-
784 vessel. Each has certain advantages and disadvantages and the choice of system depends upon the
785 application.
786 13.5.1 Focused Open-Vessel Systems
787 A focused open-vessel system has no oven but consists of a magnetron to generate microwaves, a
788 waveguide to direct and focus the microwaves and a cavity to contain the sample (Grillo, 1989).
789 Because of the open-vessel design, there is no pressure buildup during processing, and reagents
790 may be added during the digestion program. These systems are quite universal in that any reagent
791 and any type of vessel (glass, Perfluoroalcoholoxil™ [PFA], or quartz) can be used.
792 The waveguide ensures that energy is directed only at the portion of the vessel in the path of the
793 focused microwaves thereby allowing the neck of the vessel and refluxer to remain cool and
794 ensuring refluxing action. Because of this refluxing action, the system maintains all elements,
795 even selenium and mercury. The focused microwaves cause solutions to reach higher
796 temperatures faster than with conventional hotplates or block-type digesters and do so with
797 superior reproducibility. An aspirator removes excess acid vapors and decomposition gases.
798 Depending on the system, up to 20 g of solids or 50 to 100 mL of liquids can be digested within
799 10 to 30 minutes on average.
800 13.5.2 Low-Pressure, Closed-Vessel Systems
801 These systems consist of a microwave oven equipped with a turntable, a rotor to hold the sample
802 vessels, and a pressure-control module (Grillo, 1990). The PFA vessels used with these systems
803 are limited to approximately 225 °C, and, therefore, low-boiling reagents or mixtures of reagents
804 should be used. However, waste is minimized in these systems because smaller quantities of acid
805 are required. Moreover, because little or no acid is lost during the digestion, additional portions
806 of acid may not be required and blank values are minimized. Additionally, these sealed vessels
807 are limited to 100 to 300 psi, depending on the model thereby limiting the size of organic
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 13-25 DO NOT CITE OR QUOTE
-------
Sample Dissolution
808 samples utilized. However, inorganic materials such as metals, water and waste waters, minerals,
809 and most soils and sediments are easily digested without generating large amounts of gaseous by-
810 products. Typical sample sizes are on the order of 0.5 g for solids and 45 mL for waters.
811 The pressure control module regulates the digestion cycle by monitoring, controlling, and
812 dwelling at several preferred pressure levels for specified time periods in order to obtain
813 complete dissolution and precise recoveries in the minimum amount of time. As the samples are
814 irradiated, temperatures in the vessels rise thereby increasing the pressure. The pressure
815 transducer will cycle the magnetron to maintain sufficient heat to hold the samples at the
816 programmed pressure level for a preset dwell time. The vessels are designed to vent safely in
817 case of excessive internal pressure.
818 13.5.3 High-Pressure, Closed-Vessel Systems
819 Recent advances in vessel design have produced microwave vessels capable of withstanding
820 pressures on the order of 1,500 psi (Lautenschlager, 1989), allowing for larger sample sizes on
821 the order of 1 to 2 g for soil (Smith and Yaeger, 1996) or 0.5 to 3 g for vegetation (Alvarado et
822 al., 1996) and, consequently, better detection limits. These high-pressure vessels are used to
823 digest organic and inorganic substances, such as coals, heavy oils, refractories, and ceramic
824 oxides, which cannot easily be digested with other techniques. Additionally, vessel composition
825 continues to improve. Noltner et al. (1990) have demonstrated that Tetrafluorometoxil™ (TFM)
826 vessels exhibit significantly lower blank background values from residual contamination and
827 reuse than vessels produced with the more traditional PFA. This lower "memory" results in lower
828 detection limits, a clear advantage for environmental laboratories.
829 13.6 Special Matrix Considerations
830 13.6.1 Liquid Samples
831 13.6.1.1 Aqueous Samples
832 Aqueous samples are usually considered to be in solution. This may not always be true, and,
833 based on the objectives of the project, additional decomposition of aqueous samples may be
834 requested.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 13-26 DRAFT FOR PUBLIC COMMENT
-------
Sample Dissolution
835 13.6.1.2 Nonaqueous Samples
836 Most radiochemical analyses are performed in aqueous solutions. Because nonaqueous liquids
837 are incompatible with this requirement, these samples must be converted into an aqueous form.
838 In most cases, the nonaqueous liquid is simply a solvent that does not contain the radionuclide of
839 interest, and the nonaqueous solvent simply can be removed and the residue dissolved as
840 described in Sections 13.3 and 13.4.
841 Occasionally, the nonaqueous phase must be analyzed. A procedure for the decomposition of
842 petroleum products is described by Coomber (1975). There are restrictions on how many
843 nonaqueous liquids can be disposed of, even as laboratory samples. Evaporation of volatile
844 solvents may initially be an attractive alternative, but the legal restrictions on evaporating
845 solvents into the air should be investigated before this method is implemented. Burning flam-
846 mable liquids such as oil may also initially appear attractive, but legal restrictions on incineration
847 of organic liquids may need to be considered. A liquid-liquid extraction or separation using ion
848 exchange resin may be the only alternative for transferring the radionuclide of interest into an
849 aqueous solution. Unfortunately, these methods require extensive knowledge of the sample
850 matrix and chemical form of the contaminant, which is seldom available. Often, gross
851 radioactivity measurements using liquid scintillation counting techniques or broad spectrum
852 direct measurements such as gamma spectroscopy are the only measurements that can be
853 practically performed on nonaqueous liquids.
854 13.6.2 Solid Samples
855 Decomposition of solid samples is accomplished by applying fusion, wet ashing, leaching, or
856 combustion techniques singly or in some combination. A discussion of each of these techniques
857 is included in this chapter.
858 13.6.3 Filters
859 Air filter samples generally have a small amount of fine paniculate material on a relatively small
860 amount of filter media. In many cases, filters of liquid samples also have limited amounts of
861 sample associated with the filter material. This situation may initially appear to make the sample
862 decomposition process much easier, the small amount of sample appears to dissolve readily in a
863 simple acid dissolution. The ease with which many filters dissolve in concentrated acid does not
864 always mean that the sample has dissolved, and the fine particles are often impossible to see in
865 an acid solution. If the radionuclides of concern are known to be in the oxide form, or if the
866 chemical form of the contaminants is unknown, a simple acid dissolution will not completely
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 13-27 DO NOT CITE OR QUOTE
-------
Sample Dissolution
867 dissolve the sample. In these cases, the sample may be dry ashed to destroy the filter and the
868 residue subjected to fusion or other decomposition of oxides in the sample.
869 13.6.4 Wipe Samples
870 If oxides and silicates are not present in wipe samples, acid dissolutions are generally acceptable
871 for sample decomposition. In many cases, it is not the sample but the material from which the
872 wipe is constructed that causes problems with acid dissolution. Paper wipes are decomposed
873 easily in sulfuric-nitric solutions or in perchloric nitric solutions or by combustion, and it may be
874 necessary to dry ash the sample before dissolution. If volatile isotopes are expected, precautions
875 must be taken to prevent loss when heating (see Section 14.5, "Volatilization and Distillation).
876 "Sticky" smears can be more difficult to dissolve—the glue can be especially troublesome and
877 should be watched closely if perchloric acid is used. Other materials used for wipe samples
878 should be evaluated on an individual basis to determine the best method for sample
879 decomposition. In some cases, the sample will be a problem to decompose as well. Oil and
880 grease are often collected on wipe samples from machinery, and these samples are usually dry
881 ashed before acid dissolution to remove the organic material. If large amounts of solid material
882 (i.e., soil, dust, etc.) are collected with the wipe, it is recommended that the sample be treated as
883 a solid (the analytical protocol specification or the project manager should be consulted before
884 removing the wipe and simply analyzing the solid sample).
885 13.6.5 Liquid Scintillation Samples
886 Sample oxidation is used in association with liquid scintillation counting to enhance the
887 solubility of samples, decolorize samples to limit quenching, separate radionuclides, concentrate
888 the analyte from bulk material, or for a combination of these reasons.
889 13.6.5.1 Wet Oxidation
890 Wet oxidation reagents are used to liberate 14CO2,3H2O, and 35SO3 from samples containing 14C,
891 3H, and 35S, respectively, with limited success (Gibbs et al., 1978; Peng, 1977). Nitric acid, nitric
892 acid with perchloric acid, fuming sulfuric acid with periodate and chromic acid, and perchloric
893 acid with hydrogen peroxide are employed. However, a consequence of using these strong
894 reagents is the production of chemiluminescence. Moreover, these reagents also suppress
895 counting efficiency because they are strong quenching agents.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 13-28 DRAFT FOR PUBLIC COMMENT
-------
Sample Dissolution
896 13.6.5.2 Dry Oxidation
897 Dry oxidation refers to combustion of the sample in an oxygen atmosphere to yield the highest
898 oxides, e.g., H2O, CO2, SO3. Sample oxidizers are currently available for liquid scintillation
899 based upon this approach. The current system uses a continuous flow of oxygen to ensure
900 complete oxidation of the sample and to force the gaseous products through the H2O and CO2
901 collection regions and any untrapped gases to vented waste. The sample is loaded into a
902 platinum-rhodium wire basket and then is sealed into the combustion flask. Oxygen begins to
903 flow as an electric current passes through the wire basket to ignite the sample. The continuous
904 flow of O2 sweeps the gaseous combustion products into the air-cooled condenser. The collection
905 of the combustion products consists of two consecutive stages. First, the water produced in the
906 combustion process is condensed at 2 °C and collected. Second, the CO2 produced in the
907 combustion is isolated by a CO2 absorber. Each fraction is then mixed with liquid scintillation
908 cocktail and counted. This instrument is designed to give highly reproducible recoveries of 3H
909 and 14C while eliminating chemiluminescence and various quenching problems.
910 13.7 Total Dissolution and Leaching
911 Sample dissolution can be one of the biggest challenges facing the analyst because the adequacy
912 of the dissolution has direct and profound effects on the resultant data. The analyst must balance
913 numerous factors such as the nature of the sample and the analyte (e.g., is it refractory or
914 volatile?), the effects of excess reagents during subsequent analyses, the accuracy and precision
915 requirements for the data, and the costs associated with effort, materials, and waste generation.
916 Consequently, the question of total dissolution through fusion or digestion, or through acid
917 leaching, is under constant debate, and it is important for the analyst to be aware of the
918 limitations of both methods.
919 The MARLAP process enables one to make a decision concerning the dissolution required
920 through its process of establishing data quality objectives, analytical protocol specification, and
921 measurement quality objectives. During this process, all pertinent information is available to the
922 radioanalytical specialist who then evaluates the alternatives and assists with the decision. The
923 following discussion on acid leaching focuses on its use for the complete dissolution of the
924 analyte of interest and not for such procedures as the Environmental Protection Agency's
925 "Toxicity Characteristic Leaching Procedure" (TCLP; 40 CFR 261), which are intended to
926 determine the teachability of a chemical.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 13-29 DO NOT CITE OR QUOTE
-------
Sample Dissolution
927 13.7.1 Acid Leaching
928 "Acid leaching" has no accepted definition, but will be defined here as the use of nitric or
929 hydrochloric acid to put the radionuclide into solution. The acid concentration may vary up to
930 and include concentrated acid. Normally, the use of hydrofluoric acid and aqua regia are not
931 included in this definition. Sample size is usually relatively much larger than that used for fusion.
932 Although mineral acids might not totally break down all matrices, they have been shown to be
933 effective leaching solvents for metals, oxides, and salts in some samples. In some cases, leaching
934 requires fewer chemicals and less time to accomplish than complete sample dissolution. For
935 matrices amenable to leaching, multiple samples are easily processed simultaneously using a
936 hotplate or microwave system, and excess reagents can be removed through evaporation.
937 Complete dissolution of a sample is not necessary if it can be demonstrated confidently that the
938 radionuclide of interest is completely leached from the sample medium. However, if complete
939 dissolution of the analyte cannot be so demonstrated, then it may be necessary to compare
940 leaching data with data from totally dissolved samples in order for the analyst to determine the
941 appropriate method for total analyte content of a specific set of samples. When leaching is a
942 viable option for analyte removal, as an alternative to complete dissolution, the samples can be
943 treated with strong acids to leach all or a large fraction of the radionuclides of interest from solid
944 media. It may be possible to complete the dissolution of leach residue with hot aqua regia and
945 then followed by hot hydrofluoric acid. The use of these acids is usually used on relatively small
946 sample residues and may also be used on small samples.
947 Sill and Sill (1995) point out that:
948 "In many cases, the mono-, di-, and small tervalent elements can be leached fairly
949 completely from simple solids by boiling with concentrated hydrochloric or nitric acids.
950 However, even these elements cannot necessarily be guaranteed to be dissolved
951 completely by selective leaching. If they are included in a refractory matrix, they will not
952 be removed completely without dissolution of the matrix. If the samples have been
953 exposed to water over long periods of time, such as with sediments in a radioactive waste
954 pond, small ions such as divalent cobalt will have diffused deeply into the rock lattice
955 from which they cannot be removed without complete dissolution of the host matrix. In
956 contrast, because of its large size, ionic cesium has a marked tendency to undergo
957 isomorphous replacement in the lattice of complex silicates from which it too cannot be
958 removed completely. In some unpublished work by the present authors, 15% of the 137Cs
959 and 5% of the 60Co in some pond sediments remained in the residue after extensive
960 leaching, and could not be removed by further boiling for two hours with either
MARLAP JULY 2001
DO NOT CITE OR QUOTE 13-30 DRAFT FOR PUBLIC COMMENT
-------
Sample Dissolution
961 concentrated nitric or hydrochloric acids. The fraction remaining in the residue was
962 obviously much greater with shorter, more reasonable leaching times."
963 13.7.2 Total Dissolution through Fusion
964 There are those within the radiochemistry community who maintain that leaching techniques are
965 always inadequate. Sill and Sill (1995), longtime proponents of total dissolution, state, "Any
966 procedure that fails to obtain complete sample dissolution for whatever reasons of economy,
967 speed, sample load, or other expediency is untrustworthy at best, and will inevitably give low and
968 erratic results." They go on to support their argument:
969 "The large ter-, quadri-, and pentavalent elements are extremely hydrolytic and form
970 hydroxides, phosphates, silicates, carbides, etc., that are very insoluble and difficult to
971 dissolve in common acids, particularly if they have been heated strongly and converted to
972 refractory forms. For example, eight samples of soil taken in the vicinity of a plutonium-
973 handling facility were analyzed in the facility's own laboratory for 239Pu by their routine
974 procedure involving leaching with nitric acid in the presence of 236Pu tracer. The insoluble
975 residues were then analyzed for the same radionuclide by one of the present authors using
976 a procedure involving complete dissolution in a potassium fluoride fusion in the presence
977 of 236Pu tracer. Four of the residues contained more 239Pu than the corresponding
978 leachates, three residues contained about half as much as the leachates, and only one
979 contained as little as 22%, largely because that sample contained relatively high activity
980 of the radionuclide (Sill, 1981). None of the water-soluble 236Pu tracer used in the original
981 leach determination was present in any of the residues, showing that heterogeneous
982 exchange did not occur (Sill, 1975). The original results from leaching were, therefore,
983 grossly inaccurate."
984 However, there are also disadvantages and challenges associated with the fusion approach.
985 Fusions are frequently more labor intensive than the leaching approach. More often than not, it is
986 one sample at a time using a burner. Large quantities of the flux are generally required to
987 decompose most substances, often 5 tolO times the sample weight. Therefore, contamination of
988 the sample by impurities in the reagent is quite possible. Furthermore, the aqueous solutions
989 resulting from the fusions will have a very high salt content, which may lead to difficulties in
990 subsequent steps of the analysis, i.e., difficulties of entrainment, partial replacements, etc. The
991 high temperatures associated with these fusions increase the danger of loss of certain analytes by
992 volatilization. Finally, the crucible itself is often attacked by the flux, once again leading to
993 possible contamination of the sample. The typical sample size for fusions ranges from typically
994 one to ten grams. The analyst must consider whether a this sample is representative.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 13-31 DO NOT CITE OR QUOTE
-------
Sample Dissolution
995 13.7.3 Acid Digestion — Fusion Combined Approach
996 Clearly, the sample history, as well as the analytical protocol specifications of a study, should
997 play a significant role in the choice of analytical method. The analyst must be certain that the
998 chosen dissolution technique will provide adequate data for the problem at hand, whether it be
999 through acid leaching or total dissolution. However, as a compromise, it is common practice to
1000 employ a combination of the two approaches when the majority of the material to be analyzed is
1001 acid-soluble. First, an acid leach is applied to the bulk of the sample. Then any undecomposed
1002 residue is isolated by filtration and fused with a relatively small quantity of suitable flux. Finally,
1003 the melt is dissolved and combined with the rest of the sample.
1004 Through this approach, the total matrix is decomposed, but the problems, such as reagent
1005 quantity, and sample and fusion vessel size (commonly associated with fusions), are limited. The
1006 quantities of added salt are less; therefore, the sources of contamination or of subsequent
1007 chemical interferences are reduced. Moreover, losses because of volatility tend to be less because
1008 only a small fraction of the sample is exposed to the high temperatures associated with the fusion
1009 process.
1010 13.8 Examples of Decomposition Procedures
1011 DECOMPOSITION OF ORGANIC MATERIAL WITH SULFURIC AND NITRIC ACIDS. Add H2SO4 to the
1012 sample and heat to fumes in a Kjeldahl flask. Add concentrated HNO3 by drops to the flask,
1013 allowing the reaction to subside after each addition. Periodically heat to fuming to remove water
1014 and to keep the temperature high. When the solution is clear and colorless, the reaction is
1015 complete. Very reactive material can be left overnight in a 1:1 solution of the acids. Red or white
1016 fuming nitric acid can be used to speed up the reaction, if necessary.
1017 DECOMPOSITION OF ORGANIC MATERIAL WITH PERCHLORIC AND NITRIC ACIDS. The acids can be
1018 added to the sample as a mixture or the sample can be treated with concentrated FDSTO3 first to
1019 destroy any highly reactive material. The solution is heated to drive off the HNO3 and to raise the
1020 temperature to 160 °C, where the HC1O4 begins to oxidize the organic material. The reaction is
1021 generally accompanied by foaming, and HNO3 is used to cool the solution and to control the
1022 formation of foam. The solution should be cooled immediately if any layer of material begins to
1023 separate and turn brown. HNO3 is added to the sample before it is returned to the hot plate. The
1024 transition into HC1O4 continues until the foaming is completed and dense white fumes are
1025 evolved, indicating that HC1O4 is being evaporated. The volume is reduced and the solution
1026 converted to FDSTO3 by repeated addition of FDSTO3 and evaporation to near dryness.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 13-32 DRAFT FOR PUBLIC COMMENT
-------
Sample Dissolution
1027 DECOMPOSITION OF A SAMPLE OF UNKNOWN COMPOSITION (Noyes and Bray, 1927/1943; Bock,
1028 1979, Appendix 1). First, destroy the organic material with perchloric and nitric acids and then
1029 perform an oxidizing dissolution with HBr and Br2. Separate the residue and oxidize it with nitric
1030 acid. Subsequently, heat to fumes with perchloric acid and HF to destroy any silicates present.
1031 Combine with the Fffir solution and distill off the bromides of arsenic, germanium, and selenium.
1032 Oxidize the residue with nitric acid, add sodium peroxide, and distill off osmium as the tetroxide.
1033 Add perchloric acid and distill off ruthenium as the tetroxide. Reduce the contents of the flask
1034 with formic acid. Separate the residue, and leach with FTP to dissolve niobium, tantalum, and
1035 tungsten. Separate the residue, and fuse with sodium carbonate to convert fluorides to carbonates;
1036 then dissolve the melt in dilute perchloric acid. Separate the residue, and treat with aqua regia to
1037 dissolve the gold group metals. Separate the residue, and treat with ammonia to dissolve silver.
1038 Separate the residue, and fuse with K2S2O7; then dissolve the melt in water. Separate the residue,
1039 and fuse with sodium peroxide.
1040 DECOMPOSITION OF SOIL FOR ACTINIDE ANALYSIS (Sill et al., 1974; Sill and Sill, 1995). Sill has
1041 described a potassium fluoride-potassium pyrosulfate fusion technique that can be used before
1042 elemental separation for the alpha-emitting nuclides of radium through californium. The organic
1043 matter of the soil is initially destroyed by heating the sample with nitric acid in a platinum
1044 crucible. To a 1 g sample, potassium fluoride is added and mixed well. The potassium fluoride
1045 fusion is carried out using a blast burner at approximately 900 °C. After the melt is cooled,
1046 concentrated sulfuric acid is added, and the mixture is heated to decompose the potassium
1047 fluoride cake, with the simultaneous volatilization of hydrogen fluoride and silicon tetrafluoride.
1048 After the cake is completely transformed, anhydrous sodium sulfate is added and the pyrosulfate
1049 fusion is performed. The resultant cake is then dissolved in dilute HC1 before subsequent
1050 elemental separation.
1051 13.9 References
1052 13.9.1 Cited References
1053 Abu-Samra, A., Morris, J.S., and Koirtyohann, S.R. 1975. "Wet Ashing of Some Biological
1054 Samples in a Microwave Oven," Analytical Chemistry, 47:8, pp 1475-1477.
1055 Alvarado, J.S., Neal, T.J., Smith, L.L., and Erickson, M.D. 1996. "Microwave Dissolution of
1056 Plant Tissue and the Subsequent Determination of Trace Lanthanide and Actinide Elements
1057 by Inductively Coupled Plasma-Mass Spectrometry," Analytica Chimica Acta, Vol. 322, pp.
1058 11-20.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 13-33 DO NOT CITE OR QUOTE
-------
Sample Dissolution
1059 American Society of Testing Materials (ASTM) D4309. "Standard Practice for Sample Digestion
1060 Using Closed Vessel Microwave Heating Technique for the Determination of Total Metals in
1061 Water," in 1994 Annual Book of ASTM Standards, Vol. 11.01, 1996.
1062 American Society of Testing Materials (ASTM) D5258. "Standard Practice for Acid-Extraction
1063 from Sediments Using Closed Vessel Microwave Heating," in 1992 Annual Book of ASTM
1064 Standards, Vol. 11.02, 1992.
1065 American Society of Testing Materials (ASTM) D5513. "Standard Practice for Microwave
1066 Digestion of Industrial Furnace Feedstreams for Trace Element Analysis," in 1994 Annual
1067 Book of ASTM Standards, Vol. 11.04, 1994.
1068 Bock, R. 1979. A Handbook of Decomposition Methods in Analytical Chemistry., Halsted Press,
1069 John Wiley and Sons, New York.
1070 Bogen, DC. 1978. "Decomposition and Dissolution of Samples: Inorganic," in Kolthoff, I.M. and
1071 Elving, P.J., Eds., Treatise on Analytical Chemistry, Part I, Vol. 5, Wiley-Interscience, New
1072 York, pp. 1-22.
1073 Booman, G.L. and Rein, I.E. 1962. "Uranium," in Treatise on Analytical Chemistry, Kolthoff,
1074 I.M. and Elving, P.J., Eds., Part II, Volume 9, John Wiley and Sons, New York, pp. 1-188.
1075 40 CFR 261, Appendix II, Method 1311, "Toxicity Characteristic Leaching Procedure (TCLP)."
1076 Cobble, J.W. 1964. "Technetium," in Treatise on Analytical Chemistry, Kolthoff, I.M. and
1077 Elving, P.J., Eds., Part II, Volume 6, John Wiley and Sons, New York, pp. 404-434.
1078 Coomber, D.I. 1975. "Separation Methods for Inorganic Species," in RadiochemicalMethods in
1079 Analysis, Coomber, D.I., Ed., Plenum Press, New York, pp. 175-218.
1080 Dean, J. 1995. Analytical Chemistry Handbook, McGraw-Hill, New York.
1081 Environmental Protection Agency (EPA). December, 1996. "Microwave Assisted Digestion of
1082 Siliceous and Organically Based Materials," in Test Methods for Evaluating Solid Waste,
1083 Physical/Chemical Methods, SW-846, Method 3052.
1084 Gibbs, J., Everett, L., and Moore, D. 1978. Sample Preparation for Liquid Scintillation
1085 Counting, Packard Instrument Co., Downers Grove, IL., pp 65-78.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 13-34 DRAFT FOR PUBLIC COMMENT
-------
Sample Dissolution
1086 Oilman, L.B., and Engelhardt, W.G. 1989. "Recent Advances in Microwave Sample
1087 Preparation," Spectroscopy, 4:8, pp. 4-21.
1088 Grille, A.C. 1989. "Microwave Digestion by Means of a Focused Open-Vessel System,"
1089 Spectroscopy, Vol. 4, No. 7, pp. 16-21.
1090 Grille, A.C. 1990. "Microwave Digestion Using a Closed Vessel System," Spectroscopy, Vol. 5,
1091 No. 1, pp. 14, 16, 55.
1092 Grindler, I.E. 1962. The Radiochemistry of Uranium, National Academy of Sciences-National
1093 Research Council (NAS-NS), NAS-NS 3050, Washington, DC.
1094 Hahn, R.B. 1961. "Zirconium and Hafnium," in Treatise on Analytical Chemistry, Kolthoff, I.M.
1095 and Elving, P.J., Eds., Part n, Volume 5, John Wiley and Sons, New York, pp. 61-138.
1096 Kammin, W.R., and Brandt, M.J. 1989. "The Simulation of EPA Method 3050 Using a High-
1097 Temperature and High-Pressure Microwave Bomb," Spectroscopy, 4:6, pp. 22, 24.
1098 Kingston, H.M., and lassie, L.B. 1988. Introduction to Microwave Sample Preparation: Theory
1099 and Practice, American Chemical Society, Washington, DC.
1100 Kingston, H.M., and SJ. Haswell. 1997. Microwave-Enhanced Chemistry: Fundamentals,
1101 Sample Preparation, and Applications, American Chemical Society, Washington, DC.
1102 Lautenschlager, W. 1989. "Microwave Digestion in a Closed-Vessel, High-Pressure System,"
1103 Spectroscopy, 4:9, pp. 16-21.
1104 Noltner, T., Maisenbacher, P., and Puchelt, H. 1990. "Microwave Acid Digestion of Geological
1105 and Biological Standard Reference Materials for Trace Element Analysis by ICP-MS,"
1106 Spectroscopy, 5:4, pp. 49-53.
1107 Noyes, A.A. and Bray, W.C. 1927, reprint 1943. A System of Qualitative Analysis for the Rarer
1108 Elements, MacMillan, New York.
1109 Peng, T. 1977. Sample Preparation in Liquid Scintillation Counting, Amersham Corporation,
1110 Arlington Heights, IL., pp. 48-54.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 13-35 DO NOT CITE OR QUOTE
-------
Sample Dissolution
1111 Schilt, A. 1979. Perchloric Acids and Perchlorates, The G. Frederick Smith Company,
1112 Columbus, Ohio.
1113 Sill, C.W., Puphal, K.W., and Hindman, F.D. 1974. "Simultaneous Determination of Alpha-
1114 Emitting Nuclides or Radium through Californium in Soil," Analytical Chemistry, 46:12, pp.
1115 1725-1737.
1116 Sill, C.W. 1975. "Some Problems in Measuring Plutonium in the Environment," Health Physics,
1117 Vol. 29, pp. 619-626.
1118 Sill, C.W. 1981. "A Critique of Current Practices in the Determination of Actinides," in
1119 Actinides in Man and Animal, Wren, M.E., Ed., RD Press, Salt Lake City, Utah, pp. 1-28.
1120 Sill, C.W. and Sill, D.S. 1995. "Sample Dissolution," Radioactivity and Radiochemistry, 6:2, pp.
1121 8-14.
1122 Smith, LL., Crain, J.S., Yaeger, J.S., Horwitz, E.P., Diamond, H., and Chiarizia, R. 1995.
1123 "Improved Separation Method for Determining Actinides in Soil Samples," Journal of
1124 RadioanayticalNuclear Chemistry, Articles, 194:1, pp. 151-156.
1125 Smith, L.L. and Yaeger, J.S. 1996. "High-Pressure Microwave Digestion: A Waste-Minimization
1126 Tool for the Radiochemistry Laboratory," Radioactivity and Radiochemistry, 7:2, pp. 35-38.
1127 Steinberg, E.O. 1960. The Radiochemistry of Zirconium and Hafnium, National Academy of
1128 Sciences-National Research Council (NAS-NRC), NAS-NRC 3011, Washington, DC.
1129 Sulcek, Z., and Povondra, P. 1989. Methods for Decomposition in Inorganic Analysis, CRC
1130 Press, Inc., Boca Raton, Florida.
1131 13.9.2 Other Sources
1132 Bishop, C.T., Sheehan, W.E., Gillette, R.K., and Robinson, B. 1971. "Comparison of a Leaching
1133 Method and a Fusion Method for the Determination of Plutonium-23 8 in Soil," Proceedings
1134 of Environmental Symposium, Los Alamos Scientific Laboratory, Los Alamos, NM, U.S.
1135 Atomic Energy Commission, Document LA-4756, December, pp. 63-71.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 13-36 DRAFT FOR PUBLIC COMMENT
-------
Sample Dissolution
1136 Department of Energy (DOE). 1990. EML Procedures Manual, Chieco, N.A., Bogen, DC., and
1137 Knutson, E.O., Eds., HASL-300, 27th Edition, DOE Environmental Measurements
1138 Laboratory, New York.
1139 Environmental Protection Agency (EPA). 1992. Guidance for Preforming Site Inspections Under
1140 CERCLA, EPA/540-R-92-021, Office of Solid Waste and Emergency Response, Washington,
1141 DC.
1142 Environmental Protection Agency (EPA). 1997. Multi-Agency Radiation Survey and Site
1143 Investigation Manual (MARSSIM), Petullo, C.F., Chair, NUREG-1575, EPA 402-R-97-016,
1144 Washington, DC.
1145 Grimaldi, F.S. 1961. "Thorium," in Treatise on Analytical Chemistry, Kolthoff, I.M. and Elving,
1146 P.J., Eds., Part II, Volume 5, John Wiley and Sons, New York, pp. 142-216.
1147 Krey, P.W. and Bogen, DC. 1987. "Determination of Acid Leachable and Total Plutonium in
1148 Large Soil Samples," Journal of Radioanalytical and Nuclear Chemistry, 115:2 pp. 335-355.
1149 Smith, L.L., Markun, F., and TenKate, T. 1992. "Comparison of Acid Leachate and Fusion
1150 Methods to Determine Plutonium and Americium in Environmental Samples," Argonne
1151 National Laboratory, ANL/ACL-92/2.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 13-37 DO NOT CITE OR QUOTE
-------
14 SEPARATION TECHNIQUES
2 14.1 Introduction
3 The methods for separation, collection, and detection of radionuclides are similar to ordinary
4 analytical procedures and employ many of the chemical and physical principles that apply to their
5 nonradioactive isotopes. However, some important aspects of the behavior of radionuclides are
6 significantly different, resulting in challenges to the radiochemist to find a means for isolation of
7 a pure sample for analysis (Friedlander et al., 1981, pp. 292-293).
8 The contents of Chapter 14 provide in one reference document: (1) a review of the important
9 chemical principles that constitute the foundation of radiochemical separations, (2) a survey of
10 the important separation methods used in radiochemistry with a discussion of the advantages and
11 disadvantages of each method, and (3) an examination of the particular features of radioanalytical
12 chemistry that differentiate it from ordinary analytical chemistry. Extensive examples have been
13 employed throughout the chapter to illustrate various principles, practices, and procedures in
14 radiochemistry. Many were purposely selected from agency procedural manuals to provide
15 illustrations from familiar and available documents. Others were taken from the classical and
16 recent radiochemical literature to afford a broad, general overview of the subject.
17 The material in this chapter is presented in three topic areas. It begins with a review of oxidation-
18 reduction processes and complex-ion formation, two subjects that constitute the principal
19 foundation of radiochemistry procedures and provide background for the topics to follow. The
20 chapter continues with a description of separation techniques commonly found in radiochemical
21 procedures: solvent extraction, volatilization and distillation, electrodeposition, chromatography,
22 and precipitation and coprecipitation. It concludes with two subjects unique to radioanalytical
23 chemistry: carriers and tracers, and radiochemical equilibrium. This organization is designed to
24 provide a developmental approach to the description of each topic area. Explanation of the
25 separation techniques, for example, is dependent on basic chemical principles generally known to
26 the reader, as well as the specific principles developed in the preceding sections. Descriptions of
27 carriers and tracers, and radiochemical equilibrium are contingent on an adequate knowledge of
28 preceding topics, and their explanation makes extensive use of the principles developed in these
29 sections. In all sections of Chapter 14, specific radionuclide examples are used to illustrate the
30 principles and practices involved. Practical guidance is also provided for the practicing
31 radiochemist.
32 Because the radiochemist detects atoms by their radiation, the success or failure of a radio-
33 chemical procedure often depends on the ability to separate extremely small quantities of
34 radionuclides (e.g., 10"6 to 10"12 g) that might interfere with detection of the analyte. For example,
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 14-1 DO NOT CITE OR QUOTE
-------
Separation Techniques
35 isolation of trace quantities of a radionuclide that will not precipitate on their own with a counter-
36 ion requires judicious selection of a carrier and careful technique to produce a coprecipitate
37 containing the pure radionuclide, free of interfering ions. In detection procedures, the differences
38 in the behavior of radionuclides provide unique opportunities not available in the traditional
39 analytical chemistry of nonradioactive elements. Radionuclides can often be detected by their
40 unique radiation regardless of the chemical form of the element. There is also a time factor
41 involved, because of the short half-lives of some radionuclides. Traditional procedures involving
42 long digestions or slow filtrations cannot be used for short-lived radionuclides, thereby requiring
43 that rapid separations be developed. Another distinction is the hazards associated with
44 radioactive materials. At very high activity levels (radiolysis), chemical effects of the radiation,
45 such as decomposition of solvents and heat effects, can affect the procedures. Equally important,
46 even at lower activity levels, is the radiation dose (especially with gamma-emitters) that the
47 radiochemist can receive unless protected by shielding or distance. Even at levels where the
48 health concerns are minimal, special care needs to be taken to guard against laboratory and
49 equipment contamination. Moreover, the modern radiochemist should be concerned about the
50 type and quantity of the waste generated by the chemical procedures employed, because the costs
51 and difficulties associated with the disposal of low-level and mixed radioactive waste continue to
52 rise. A review of the basic chemical principles that apply to the analysis of radionuclides is
53 presented in this chapter with an emphasis on the unique behavior of radionuclides.
54 14.2 Oxidation/Reduction Processes
55 14.2.1 Introduction
56 Oxidation and reduction (redox) processes play an important role in radioanalytical chemistry,
57 particularly from the standpoint of the dissolution, separation, and detection of analytes, tracers,
58 and carriers. Ion exchange, solvent extraction, and solid-phase extraction separation techniques,
59 for example, are highly dependent upon the oxidation state of the analytes. Moreover, most
60 radiochemical procedures involve the addition of a carrier or isotope tracer, and to achieve
61 quantitative yields, there should be complete equilibration (isotopic exchange) between the added
62 isotope(s) and all the analyte species present. The oxidation number of a radionuclide can affect
63 its (1) chemical stability in the presence of water, oxygen, and other natural substances in
64 solution; (2) reactivity with reagents used in the radioanalytical procedure; (3) solubility in the
65 presence of other ions and molecules; and (4) behavior in the presence of carriers and tracers.
66 The oxidation numbers of radionuclides in solution and their susceptibility to change, because of
67 natural or induced redox processes, are critical, therefore, to the physical and chemical behavior
68 of radionuclides during these analytical procedures. The differences in mass number of all
MARLAP JULY 2001
DO NOT CITE OR QUOTE 14-2 DRAFT FOR PUBLIC COMMENT
-------
Separation Techniques
69 radionuclides of an element are so small that elements with the same oxidation number will
70 exhibit the same chemical behavior during radiochemical analysis.
71 14.2.2 Oxidation-Reduction Reactions
72 An oxidation-reduction reaction (redox reaction) is a reaction in which electrons are
73 redistributed among the atoms, molecules, or ions in the reaction. In some redox reactions,
74 electrons are actually transferred from one reacting species to another. Oxidation under these
75 conditions is defined as the loss of electron(s) by an atom or other chemical species, whereas
76 reduction is the gain of electron(s). Two examples will illustrate this type of redox reaction:
77 U + 3 F2 - U+6 + 6 F1
78 Pu+4 + Fe+2 - Pu+3 + Fe+3
79 In the first reaction, uranium (U) loses electrons, becoming a cation, and fluorine (F) gains an
80 electron, becoming an anion. In the second reaction, the reactants are already ions, but the
81 plutonium cation (Pu+4) gains electrons, becoming Pu+3, and the ferrous ion (Fe+2) loses electrons,
82 becoming Fe+3.
83 In other redox reactions, electrons are not completely transferred from one reacting species to
84 another: the electron density about an atom decreases, while it increases about another atom. The
85 change in electron density occurs as covalent bonds, in which electrons are shared between two
86 atoms, are broken, and/or are made during a chemical reaction. In covalent bonds between two
87 atoms of different elements, one atom is more electronegative than the other atom. Electronega-
88 tivity is the ability of an atom to attract electrons in a covalent bond. One atom, therefore, attracts
89 the shared pair of electrons more effectively, causing a difference in electron density about the
90 atoms in the bond. An atom that ends up bonded to a more electronegative atom at the end of a
91 chemical reaction loses net electron density. Conversely, an atom that ends up bonded to a less
92 electronegative atom gains net electron density. Electrons are not transferred completely to other
93 atoms, and ions are not formed because the electrons are still shared between the atoms in the
94 covalent bond. Oxidation, in this case, is defined as the loss of electron density, and reduction is
95 defined as the gain of electron density. When carbon (C) is oxidized to carbon dioxide (CO2) by
96 oxygen (O2):
97 C + O2 - CO2
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 14-3 DO NOT CITE OR QUOTE
-------
Separation Techniques
98 the electron density associated with the carbon atom decreases, and that of the oxygen atoms
99 increases, because the electronegativity of oxygen is greater than the electronegativity of carbon.
100 In this example, carbon is oxidized and oxygen is reduced. Another example from the chemistry
101 of the preparation of gaseous uranium hexafluoride (UF6) illustrates this type of redox reaction:
102 3UF4 + 2C1F3 - 3UF6 + Cl
2
103 Because the order of electronegativity of the atoms increases in the order U
-------
Separation Techniques
132 specific exception would be nitrogen in the cation and anion in ammonium nitrate, NH4NO3.) It
133 is important to remember that in many cases, oxidation numbers are not actual electrical charges,
134 but only a helpful bookkeeping method for following redox reactions or examining various
135 oxidation states. As we will see below, the oxidation number of atoms in isolated elements and
136 monatomic ions are actually the charge on the chemical species. The priority rules are:
137 1. The sum of oxidation numbers of all atoms in a chemical species adds up to equal the
138 charge on the species. This is zero for elements and compounds because they are
139 electrically neutral species and are the total charge for a monatomic or polyatomic ion.
140 2. The alkali metals (the IA elements, Li, Na, K, Rb, Cs, and Fr) have an oxidation number
141 of+1; the alkaline earth metals (the HA elements, Be, Mg, Ca, Sr, Ba, and Ra) have an
142 oxidation number of +2.
143 3. Fluorine (F) has an oxidation number of -1; hydrogen (H) has an oxidation number of +1.
144 4. Oxygen has an oxidation number of -2.
145 5. The halogens (the VIIA elements, F, Cl, Br, I, and At) have an oxidation number of -1.
146 6. In binary compounds (compounds containing elements), the oxidation number of the
147 oxygen family of elements (the VIA elements, O, S, Se, Te, and Po) is -2; for the nitrogen
148 family of elements (the VA elements except Bi, N, P, As, and Sb), it is -3.
149 Applying these rules illustrates their use:
150 1. The oxidation number of metallic uranium and molecular oxygen is 0. Applying rule one,
151 the charge on elements is 0.
152 2. The oxidation number of Pu+4 is +4. Applying rule one again, the charge is +4.
153 3. The oxidation numbers of carbon and oxygen in CO2 are +4 and -2, respectively.
154 Applying rule one, the oxidation numbers of each atom must add up to the charge of 0
155 because carbon dioxide is a molecule. The next rule that applies is rule four. Therefore,
156 the oxidation number of each oxygen atom is -2. The oxidation number of carbon is
157 determined by C + 2(-2) = 0, or +4. Notice that there is no charge on carbon and oxygen
158 in carbon dioxide because the compound is molecular and does not consist of ions.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 14-5 DO NOT CITE OR QUOTE
-------
Separation Techniques
159 4. The oxidation numbers of calcium and hydrogen in calcium hydride (CaH2) are +2 and -1,
160 respectively. The compound is neutral, and the application of rule one requires that the
161 oxidation numbers of all atoms add up to 0. By rule two, the oxidation number of calcium
162 is +2. Applying rule one, the oxidation number of hydrogen is: 2H + 2=0, or -1. Notice
163 that in this example, the oxidation number as predicted by the rules does not agree with
164 rule three, but the number is determined by rules one and two, which take precedence
165 over rule three.
166 5. The oxidation numbers of uranium and oxygen in the uranyl ion, UO2+2, are +6 and -2,
167 respectively. Applying rule one, the oxidation numbers of each atom must add up to the
168 charge of+2. Rule four indicates that the oxygen atoms are -2 each. Applying rule one,
169 the oxidation number of uranium is U + 2(-2) = +2, and uranium is +6. In this example,
170 the charges on uranium and oxygen are not actually +6 and -2, respectively, because the
171 polyatomic ion is held together through covalent bonds. The charge on the ion is the
172 result of a deficiency of two electrons.
173 Oxidation numbers (states) are commonly represented by zero and positive and negative
174 numbers, such as +4, -2, etc. They are sometimes represented by Roman numerals for metals,
175 especially the oxidation numbers of atoms participating in covalent bonds of a molecule or those
176 of polyatomic ions. Uranium in UO2+2 can be represented as U(VI) instead of U+6, or chromium
177 in Cr
-------
Separation Techniques
192 Each of the transition metals has at least two stable oxidation states, except for Sc, Y, and La
193 (Group TUB), which exhibit only the +3 oxidation state. Generally, negative oxidation states are
194 not observed for these metallic elements. The large number of oxidation states exhibited by the
195 transition elements leads to an extensive, often complicated, oxidation-reduction chemistry. For
196 example, oxidation states from -1 through +7 have been observed for technetium, although the
197 +7 and +4 are most common (Anders, 1960, p. 4). In an oxidizing environment, Tc exists
198 predominantly in the heptavalent state as the pertechnetate ion, TcO^1, which is water soluble,
199 but which can yield insoluble salts with large cations. Technetium forms volatile heptoxides and
200 acid-insoluble heptasulfides. Subsequently, pertechnetate is easily lost upon evaporation of acid
201 solutions unless a reducing agent is present or the evaporation is conducted at low temperatures.
202 Technetium(VII) can be reduced to lower oxidation states by reducing agents such as bisulfite
203 (HSO34). This process proceeds through several intermediate steps, some of which are slow;
204 therefore, unless precautions are taken to maintain technetium in the appropriate oxidation state,
205 erratic results can be obtained. The +7 and +4 ions behave very differently in solution. For
206 instance, pertechnetate does not coprecipitate with ferric hydroxide, while Tc(IV) does.
207 The oxidation states of the actinide elements have been comprehensively discussed by Ahrland
208 (1986, pp. 1480-1481) and Cotton and Wilkinson (1988, pp. 985-987 and pp. 1000-1014). The
209 actinides exhibit an unusually broad range of oxidation states, of from +2 to +7 in solution.
210 Similar to the lanthanides, the most common oxidation state is +3 for actinium (Ac), americium
211 (Am), and curium (Cm). The +4 state is common for thorium and plutonium, whereas +5 is most
212 common for protactinium (Pa) and neptunium (Np). The most stable state for uranium is the +6
213 oxidation state.
214 In compounds of the +3 and +4 oxidation states, the elements are present as simple M+3 or M+4
215 cations; but for higher oxidation states, the most common forms in compounds and in solution
216 are the oxygenated actinyl ions, MO2+1 and MO2+2:
217 • M+3. The +3 oxidation state is the most stable condition for actinium, americium, and curium.
218 It is easy to produce Pu+3. This stability is of critical importance to the radiochemistry of
219 plutonium. Many separation schemes take advantage of the fact that Pu can be selectively
220 maintained in either the +3 or +4 oxidation state. Unlike Pu and Np, U+3 is such a strong
221 reducing agent that it is difficult to keep in solution.
222 • M+4. The only oxidation state of thorium that is experienced in radiochemical separations is
223 +4. Pa+4, U+4, and Np+4 are stable, but they are easily oxidized by O2 In acid solutions with
224 low plutonium concentrations, Pu+4 is stable. Americium and curium can be oxidized to the
225 +4 state with strong oxidizing agents such as persulfate.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 14-7 DO NOT CITE OR QUOTE
-------
Separation Techniques
226 • M+5. The actinides from protactinium through americium form MO2+1 ions in solution. PuO2+1
227 can be the dominant species in solution at low concentration in natural waters that are
228 relatively free of organic material.
229 • M+6. M+6 is the most stable oxidation state of uranium, which exist as the UO2+2 species.
230 Neptunium, plutonium, and americium also form MO2+2 ions in solution. The bond strength,
231 as well as the chemical stability toward reduction for these MO2+2 ions, decrease in the order
232 U > Np > Pu > Am.
233 Reactions that do not involve making or breaking bonds, M+3 ^ M+4 or MO2+1 ^ MO2+2, are fast
234 and reversible, while reactions that involve chemical bond formation, M+3 -> MO2+1 or
235 M+4 ^ MO2+2, are slow and irreversible.
236 Plutonium exhibits redox behavior unmatched in the periodic table. It is possible to prepare
237 solutions of plutonium ions with appreciable concentrations of four oxidation states, +3, +4, +5,
238 and +6, as Pu+3, Pu+4, PuO2+1, and PuO2+2, respectively. [Detailed discussions can be found in
239 Cleveland (1970), Seaborg and Loveland (1990), and in the Coleman (1965) monograph.]
240 According to Cleveland (1970), this polyvalent behavior occurs because of the tendency of Pu+4
241 and Pu+5 to disproportionate:
242 3Pu+4 + 2H2O - 2Pu+3 + PuO2+2 + 4H+1
243 3PuO2+1 + 4H+1 - Pu+3 + 2PuO2+2 + 2H2O
244 and because of the slow rates of reaction involving formation or rupture of Pu-O bonds (such as
245 PuO2+ andPuO22+) compared to the much faster reactions involving only electron transfer. The
246 distribution depends on the type and concentration of acid used for dissolution, the method of
247 solution preparation, and the initial concentration of the different oxidation states. In HC1, HNO3,
248 and HC1O4, appreciable concentrations of all four states exist in equilibrium. Seaborg and
249 Loveland (1990, p. 88) report that in 0.5 M HC1 at 25 °C, the equilibrium percentages of
250 plutonium in the various oxidation states are found to be as follows:
251
252
253
254
MARLAP JULY 2001
DO NOT CITE OR QUOTE 14-8 DRAFT FOR PUBLIC COMMENT
Pu+3
Pu+4
Pu+5
Pu+6
27.2%
58.4%
-0.7%
13.6%
-------
Separation Techniques
255
256
257
258
259
260
261
262
263
264
265
266
267
Apart from the disproportionation reactions, the oxidation state of plutonium ions in solution is
affected by its own decay radiation or external gamma and X-rays. Radiolysis products of the
solution can oxidize or reduce the plutonium, depending on the nature of the solution and the
oxidation state of plutonium. Therefore, the stated oxidation states of old plutonium solutions,
particularly old HC1O4 and H2SO4 solutions, should be viewed with suspicion. Plutonium also
tends to hydrolyze and polymerize in solution, further complicating the situation (see Section
14.10, Radiochemical Equilibrium).
Tables 14.1 and 14.2 summarize the common oxidation number(s) of some important elements
encountered in the radioanalytical chemistry of environmental samples and the common
chemical form of the oxidation state.
TABLE 14.1 — Oxidation states of elements(1)
Element
Am
Cs
Co
Fe
3H
I
Ni
Nb
Po
Oxidation
State(2)
+3
+4
+5
+6
+1
+2
+3
+2
+3
+1
-1
-1/3
+5
+7
+2
+3
+5
+4
Chemical
Form
Am+3
Am+4
AmO2+1
Am02+2
Cs(H20)x+1
Co(H20)6+2
Co(H20)6+3
Fe(H20)6+2
Fe(H20)6+3
3HOH and
3HOH2O+1
r1
V1
io3-'
KV1
Ni(H20)6+2
Unknown
HNb6019-7
Notes
Pink; stable; difficult to oxidize
Pink-red; unstable in acid
Pink-yellow; disproportionates in strong acid; reduced by products
of its own radiation
Rum color; stable
Colorless; x probably is 6
Pink to red; oxidation is very unfavorable in solution
Rapidly reduced to +2 by water unless acidic
Green
Pale violet; hydrolyses in solution to form yellow or brown
complexes
Exchange of tritium is extremely rapid in samples that have water
introduced.
Colorless
Brown; commonly in solutions of I"1 exposed to air
Colorless; formed in vigorously oxidized solutions
Colorless
Green
In sulfuric acid solutions of Nb2O5
268
269
270
271
272
273
274
275
JULY 2001
DRAFT FOR PUBLIC COMMENT
14-9
MARLAP
DO NOT CITE OR QUOTE
-------
Separation Techniques
Element
Pu
Ra
Sr
Tc
Th
U
Zr
Oxidation
State*2'
+3
+4
+5
+6
+7
+2
+2
+4
+5
+7
+4
+3
+4
+5
+6
+4
Chemical
Form
Pu(H20)x+3
Pu(H20)x+4
Pu(H20)x+5
or
PuO2+5
Pu02+5
Pu05-3
or
PuO4(OH)2-3
Ra(H2O)x+2
Sr(H20)x+2
Tc03-2
TcO3-'
TcO4-'
Th(H20)8+4
U(H20)X+3
U(H20)8.9+4
U02+1
U02(l^20)5+
Zr(H20)6+4
Zr4(OH)8(H2
O^,/2
Notes
Violet; stable to air and water; easily oxidized to +4
Tan; first state formed in freshly prepared solutions; stable in 6 M
acid; disproportionates in low acidity to +3 and +6
Never observed alone; always disproportionates; most stable in low
acidity
Purple
Yellow-pink; stable but fairly easy to reduce
Green
PuO4(OH)2-3 more likely form
Colorless; behaves chemically like Sr and Ba
Colorless
Colorless; at pH>3 forms complex hydrolysis products
Red-brown; slowly oxidized by water and rapidly by air to +4
Green; stable but slowly oxidized by air to +6
Unstable but more stable at ph 2-4; disproportionates to +4 and +6
Yellow; only form stable in solution containing air; difficult to
reduce
Only at very low ion concentrations and high acidity
At typical concentrations in absence of complexing agents
276
277
278
279
280
281
282
283
284
285
(1) Compiled from: Booman and Rein, 1962;
Earnshaw, 1984; Grinder, 1962; Hampel,
(2) Most common form is in bold.
Cotton and Wilkinson, 1988; Emsley, 1989; Greenwood and
1968; Katzin, 1986; Latimer, 1952; and Pauling, 1970.
286
287
288
289
290
291
292
293
294
295
296
297
TABLE 14.2 — Stable oxidation states of selected elements
(1,2)
Element + 1
Titanium
Vanadium
Chromium
Manganese
Iron
Cobalt
Nickel
Strontium
Yttrium
Molybdenum
+2 +3 +4 +5 +6 +7 +8
0 0 •
o o • •
• o o •
0 • 0 0 •
• 0 0
•
o o
•
o o • • •
MARLAP
DO NOT CITE OR QUOTE
14-10
JULY 2001
DRAFT FOR PUBLIC COMMENT
-------
Separation Techniques
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
Element
Technetium
Silver
Cesium
Barium
Lanthanides
Lead
Polonium
Radium
Actinium
Thorium
Protactinium
Uranium
Neptunium
Plutonium
Americium
Curium
+1 +2 +3 +4 +5 +6 +7 +8
o o • o o •
• o o
•
•
•
• o
0 • 0
•
•
•
0 •
0 0 0 •
o o • o o
o • o o
• 0 0 0
• 0
(1) The stable nonzero oxidation states are indicated. The more common oxidation
states are indicated by solid black circles.
(2) Data compiled from Seaborg and Loveland (1990) and the NAS-NRC
monographs listed in the references.
14.2.4 Oxidation State in Solution
319 For the short-lived isotopes that decay by alpha emission or spontaneous fission, high levels of
320 radioactivity cause heating and chemical effects that can alter the nature and behavior of the ions
321 in solution and produce chemical reactions not observed with longer-lived isotopes. Decompo-
322 sition of water by radiation (radiolysis) leads to H and OH free radicals and formation of H2 and
323 H2O2, among other reactive species, and higher oxidation states of plutonium and americium are
324 produced.
325 The solutions of some ions are also complicated by disproportionation, the autooxidation-
326 reduction of a chemical species in a single oxidation state to higher and lower oxidation states.
327 The processes are particularly dependent on the pH of the solution. Oxidation of iodine, uranium,
328 americium, and plutonium are all susceptible to this change in solution. The disproportionation
329 of UO2+1, for example, is represented by the chemical equation:
330 2 UO2+1 + 4 H+1 - U+4 + UO2+2 + 2 H2O (K = 1.7 x 106)
JULY 2001
DRAFT FOR PUBLIC COMMENT
14-11
MARLAP
DO NOT CITE OR QUOTE
-------
Separation Techniques
331 The magnitude of the equilibrium constant reflects the instability of the +5 oxidation state of
332 uranium in UO2+1 described in Table 14.1, and the presence of hydrogen ions reveals the
333 influence of acidity on the redox process. An increase in acidity promotes the reaction.
334 14.2.5 Common Oxidizing and Reducing Agents
335 HYDROGEN PEROXIDE. Hydrogen peroxide (H2O2) has many practical applications in the
336 laboratory. It is a very strong oxidizing agent that will spontaneously oxidize many organic
337 substances, and water samples are frequently boiled with peroxide to destroy organic compounds
338 before separation procedures. When hydrogen peroxide serves as an oxidizing reagent, each
339 oxygen atom changes its oxidation state from -1 to -2. For example, the reaction for the oxidation
340 of ferrous ion is as follows:
341 H2O2 + 2H+1 + 2Fe+2 - 2H2O + 2Fe+3
342 Hydrogen peroxide is frequently employed to oxidize Tc+4 to the pertechnetate:
343 4H2O2 + Tc+4 - Tc
-------
Separation Techniques
357 on the experimental conditions. For example, on reduction in acidic solutions, the permanganate
358 ion accepts five electrons, forming the manganous ion Mn+2:
359 MnCV1 + 5e4 + 8H+1 - Mn+2 + 4H2O
360 In neutral or basic solution, permanganate accepts 3 electrons, and forms manganese dioxide
361 (MnO2), which precipitates:
362 MnCV1 + 3e4 + 4H+1 - MnO2 I + 2H2O
363 These oxidizing agents are discussed further in Section 12.4, "Wet Ashing and Acid Dissolution
364 Techniques."
365 NITRITE. Nitrite ion (NO^1), plays an important role in the manipulation of Pu oxidation states in
366 solution. It is capable of oxidizing Pu+3 to Pu+4 and of reducing Pu+6 to Pu+4. Because most
367 aqueous processes center around Pu+4, sodium nitrite (NaNO2) is frequently used as a valence
368 adjuster to convert all Pu to the +4 state. And because the Pu+6^ Pu+4 reaction by nitrite is slow,
369 another reducing agent, such as the ferrous ion, often is added to increase the rate of reaction.
370 PERCHLORIC ACID. The use of perchloric acid (HC1O4) as an oxidizing agent is covered in depth
371 in Section 12.4, "Wet Ashing and Acid Dissolution Techniques."
372 METALS IONS. Generally, metals ions (Ti+3, Cr+2, Fe+2, etc.) are strong reducing agents. For
373 example, both Ti+3 and Cr+2 have been shown to reduce Pu+4 to Pu+3 rapidly in acidic media.
374 Fe+2 rapidly reduces Np+5 to Np+4 in H2SO4.
375 Ti+3 is used extensively as a reducing agent in both inorganic and organic analyses. Ti+3 is
376 obtained by reducing Ti+4, either electrolytically or with zinc. Ti+4 is the most stable and common
377 oxidation state of titanium. Compounds in the lower oxidation states (-1, 0, +2, and +3) are quite
378 readily oxidized to Ti+4 by air, water, or other reagents.
379 ASCORBIC ACID. Commonly known as vitamin C, ascorbic acid is an important reducing agent
380 for the radiochemist. Because the ferric ion interferes with the uptake of Am+3 in several popular
381 extraction schemes, ascorbic acid is frequently used to reduce Fe+3 to Fe+2 to remove this
382 interference. Ascorbic acid is also used to reduce Pu+4 to Pu+3.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 14-13 DO NOT CITE OR QUOTE
-------
Separation Techniques
383 14.2.6 Oxidation State and Radiochemical Analysis
384 Most radiochemical analyses require the radionuclide be in aqueous solution. Thus, except for
385 water samples, the first step of an analysis is the complete dissolution of the sample so that all
386 components remaining at the end of the process are in a true solution. Dissolution of many
387 samples requires vigorous conditions to release the radionuclides from its natural matrix. Strong
388 mineral acids or strong bases, which also serve as powerful oxidizing agents, are used in boiling
389 mixtures or under fusion conditions to decompose the matrix—evaporating portions of the acid
390 or base from the mixture and oxidizing the radionuclide to a common oxidation state. The final
391 state depends, generally, on the radionuclide, oxidizers used, and pH of the solution (see notes in
392 Table 14.1, Section 14.2.3, "Common Oxidation States"). Even water samples might contain
393 radionuclides at various states of oxidation because of their exposure to a variety of natural
394 oxidizing conditions in the environment and the pH of the sample.
395 Once the analyte is in solution, the radioelement and the tracers and carriers used in the
396 procedure must be in the same oxidation state to ensure the same chemical behavior (Section
397 14.10.2, "Oxidation State"). For radionuclides that can exist in multiple oxidation states, one
398 state must be achieved; for those such as plutonium, which disproportionates, a reproducible
399 equilibrium mixture of all oxidation states can be established. Oxidizing or reducing agents are
400 added to the reaction mixture to establish the required conditions. Table 14.3 contains a summary
401 of several chemical methods for the oxidation and reduction of select radionuclides.
402
403
404
405
406
407
408
409
410
411
412
TABLE 14.3 — Redox reagents for radionuclides(
Redox Reaction
Am+3 - AmO2+2
Am+4 - AmO2+2
AmO2+1 - AmO2+2
AmO2+2 - AmO2+1
AmO2+2 - Am+3
Am+4 - Am+3
Co+2 - Co+3
Co+3 - Co+2
Fe+2 - Fe+3
Reagent
Ag+2, Ag+/S208
03
Ce+4
03
Br ', Cl'1
Na2CO3
T1, H2O2, NO,'1, SO2
alpha radiation effects
03
02, H202
H20
02
Ce+4, MnO4-'f NCV1, NO2
Conditions
13MNH4F
HC104
Heated HNO3 or HC1O4
Heat to precipitate NaAmO2CO3; dissolve in H+1
Spontaneous
Cold HC1O4
Complexed cobalt
Rapid with evolution of H2
Faster in base; slower in neutral and acid solution; decreases
withH+1
MARLAP
DO NOT CITE OR QUOTE
14-14
JULY 2001
DRAFT FOR PUBLIC COMMENT
-------
Separation Techniques
Redox Reaction
Fe+3 - Fe+2
r1 - 12
r1 - ro3-'
r1 - KV
KV' - 12
KV - r1
i2 - r1
Pu+3 - Pu+4
Pu+4 - Pu02+2
Pu02+1 - Pu02+2
PuO,+2 - PuO,+1
Reagent
Cr2072
H2S, H2SO3
Zn, Cd, Al, Ag amalgams
Sn+2, 1'1, Cu+1, Ti+3
NH2OH
HNO2 (NaNO2 in acid)
MnO2 in acid
6M HNO3
NaHSO3orNaHSO3in
Na2SO3; Na2S2O3
KMnO4
50% CrO3 in 9M H2SO4
NaClOinbase
NH2OH-HC1
H2C2O4
NaHSO3 in acid
SO2; NaHSO3
Br03-'
Ce+4
Cr2O7 2, KV, MnO4-'
NO,'1
N03-'
HNO2
NaBiO3
Br03-'
Ce+4
HOC1 (KC1O)
MnO4-'
03
Ag(H)
Cr2072
C12
NO3"'
Ag20
io3-
HNO3
NH2OH-HC1
r1
SO2
V+3orTi+3
r1
Conditions
HClorH2SO4
Excess removed by boiling
Boiling solution
Does not affect other halides
Well suited for lab work
(9 M H2SO4
Dilute H+1
HC1 of H2SO4 solution
Dilute H+1
HNO3
HNO3 or dilute HC1 (100°C)
HNO3
Dilute HNO3 at 85 °C
Dilute HNO3 or HCLO4
pH 4.5 at 80°C or 45% K2CO3 at 40°C
Dilute HNO3
Ce+4 or Ag+1 catalyst or dil. H2SO4/60°C
Ag+1/S2O8"' in dil. HNO3
Dilute H2SO4
Dilute H2SO4 at 80°C or dil.HClO./Cr1
Dilute HNO3 at 95 °C
43%K2CO3at75°C
Dilute; slow
Slow
pH 2; slow
Dilute H+1; slow
HC1O4; slow
pH2
JULY 2001
DRAFT FOR PUBLIC COMMENT
14-15
MARLAP
DO NOT CITE OR QUOTE
-------
Separation Techniques
Redox Reaction
Pu02+2 - Pu+4
PuO2+1 - Pu+4
Pu+4 - Pu+3
Tc+4 - Tc(V
TcO2(hydrated) -
Tc04-'
TcCl6-2 - TcO;1
TcO4-' - Tc+4 or
TcO2(hyd)
Reagent
S02
Fe+2
V+3orU+4
HNO2
Ag
C2042
r1
Fe+2
Sn+2
H2O2
Ti+3
Cu2O
HNO2
Zn
HNO2
NH2OH-HC1
hydroquinone
H2/Pt
r1
HS03-'
NH2OH-HC1
Zn
SO2
Ti+3
ascorbic acid
u+4
H2S
HNO3
H202
O2 (air)
Ce+4
H2O2
HNO3
H202
C12
Ce+4
MnO4J
N2H4
NH2OH
ascorbic acid
Conditions
H+1
HClO4orHCl
HC1O4
dil. HNO3NaNO3
dil. HC1
75°C;RTwithdil.HCl
HNO3
HC1,HNO3, orH2SO4
HC1/HC1O4
HNO3; continues to Pu+3 in absence of Fe+3
HC1O4
45%K2CO375°C
HNO3/75°C
dil. HC1
slow
dil. HC1, slow
dil. HNO3
HC1
dil. HC1
dil. HNO3
dil. HC1
dil. HNO3
HC1, dil. H2SO4, or dil. HNO3/H2SO4
HNO3
dil. HC1O4
dil. acid
dil. H2SO4
dil. H2SO4
dil. H,S(X
MARLAP
DO NOT CITE OR QUOTE
14-16
JULY 2001
DRAFT FOR PUBLIC COMMENT
-------
Separation Techniques
Redox Reaction
U+3 _ U+4
U+4 - U02+2
U02+1 - U02+2
UO2+2 - U+4
UO2+2 - U+3
U02+1 - U+4
Reagent
Sn+2
Zn
Cone. HC1
CIO;1
Co+3 complexes
Cr+3 and Cr+3 complexes
H2O
U02+1
U02+2
O2 (air)
Br2
BrCV1
Ce+4
cicv1
Fe+3
HC1O2
HCrCV
HNO2
HNO3
H202
02
Pu+4
PuO2+2
MnO2
Fe+3
Cr+2
Eu+2
Np+3
Ti+3
V+2 and V+3
Zn(Hg)
Cr+2
H2
Zn(Hg)
Conditions
dil. H2SO4
dil. HC1
to TcCV2
dil. HC1O4
dil. HClO4orLiClO4
dil. HClO4orLiClO4
dil. or cone. HClorH2SO4
dil. HC1O4
dil. HC1O4
catalyzed by Fe+3 or Mn+2
HC1O4
dil. HC1O4
catalyzed by Fe+2 or V+5
phenol
catalyzed by Fe+2
434
436
437
438
439
440
441
442
443
(1) Compiled from: Anders, 1960; Bailar et al., 1984; Bate and Leddicotte, 1961; Cobble, 1964; Coleman, 1965;
Cotton and Wilkinson, 1988; Greenwood and Earnshaw, 1984; Hassinsky and Adloff, 1965; Kleinberg and
Cowan, 1960; Kolthoff et al., 1969; Latimer, 1952; Metz and Waterbury, 1962; Schulz and Penneman, 1986;
Weigel, 1986; and Weigel et al., 1986.
JULY 2001
DRAFT FOR PUBLIC COMMENT
14-17
MARLAP
DO NOT CITE OR QUOTE
-------
Separation Techniques
444 In some radioanalytical procedures, establishing different states at different steps in the procedure
445 is necessary to ensure the requisite chemical behavior of the analyte.
446 One method for the analysis of 129I in aqueous solutions illustrates the use of oxidation and
447 reduction chemistry to bring the radionuclide to a specific oxidation state so that it can be
448 isolated from other radionuclides and other elements (DOE, 1995, Method RP230). Iodine
449 species in the water sample are first oxidized to iodate (KV1) by sodium hypochlorite (NaCIO),
450 and then reduced to iodide (I"1) by sodium bisulfite. The iodine is finally oxidized to molecular
451 iodine (I2) and extracted from most other radionuclides and elements in solution by a nonpolar
452 organic solvent such as carbon tetrachloride (CC14) or chloroform (CHC13) (see Section 14.4,
453 "Solvent Extraction").
454 Plutonium and its tracers can be equilibrated in a reproducible mixture of oxidation states by the
455 rapid reduction of all forms of the ion to the +3 state, momentarily, with iodide ion (I"1) in acid
456 solution. Disproportionation begins immediately, but all radionuclide forms of the analyte and
457 tracer begin at the same time from the same oxidation state, and a true equilibrium mixture of the
458 radionuclide and its tracer is achieved. All plutonium radionuclides in the same oxidation state
459 can be expected to behave the same chemically in subsequent separation and detection
460 procedures.
461 In addition to dissolution and separation strategies, oxidation-reduction processes are used in
462 several quantitation steps of radiochemical analyses. These processes include titration of the
463 analyte and electrochemical deposition on a target for counting.
464 The classical titrimetric method is not commonly employed in the quantitation of environmental
465 level samples because the concentrations of radionuclides in these samples are typically too low
466 for detection of the endpoint of the titration, even by electrometric or spectroscopic means.
467 However, the method is used for the determination of radionuclides in other samples containing
468 larger quantities of long-lived radionuclides. Millimole quantities of uranium and plutonium in
469 nuclear fuels have been determined by titration using methods of endpoint detection as well as
470 chemical indicators (IAEA, 1972). In one method, uranium in the +6 oxidation state is reduced to
471 +3 and +4 with Ti+3, and that in the +3 state is oxidized to +4 with air bubbles (Baetsel and
472 Demildt, 1972). The solution is then treated with a slight excess of Ce+4 solution of known
473 concentration, which oxidizes U+4 to U+6 (as UO2+2) while being reduced, as follows:
474 U+4 + 2 Ce+4 - U+6 + 2 Ce+3
475 (U+4 + 2 Ce+4 +2 H2O - UO2+2 + 2 Ce+3 + 4 H+1)
MARLAP JULY 2001
DO NOT CITE OR QUOTE 14-18 DRAFT FOR PUBLIC COMMENT
-------
Separation Techniques
476 The excess Ce+4 is back-titrated with Fe+2 solution, using ferrion as indicator for the endpoint of
477 the titration:
478 Fe+2 + Ce+4 - Fe+3 + Ce+3
479 Electrochemical methods are typically used in radiochemistry to reduce ions in solution, plating
480 them onto a target metal for counting. Americium ions (Am+3) from soil samples are ultimately
481 reduced from solution onto a platinum (Pt) electrode by application of an electrical current in an
482 electrolytic cell (DOE, 1990 and 1997, Method Am-01). The amount of americium on the
483 electrode is determined by alpha spectrometry.
484 In some cases, the deposition process occurs spontaneously without the necessity of an applied
485 current. Polonium (Po) and lead (Pb) spontaneously deposit from a solution of hydrochloric acid
486 (HC1) onto a nickel (Ni) disk at 85 °C (Blanchard, 1966). Alpha and beta counting are used to
487 determine 210Po and 210Pb. Wahl and Bonner (1951, pp. 460-465) contains a table of
488 electrochemical methods used for the oxidation and reduction of carrier-free tracers.
489 14.3 Complexation
490 14.3.1 Introduction
491 A complex ion is formed when a metal atom or ion bonds with one or more molecules or anions
492 through an atom capable of donating one or more electron pairs. A ligandis any molecule or ion
493 that has at least one electron pair that can be donated to the metal. The bond is called a
494 coordination bond, and a compound containing a complex ion is a coordination compound. The
495 following are several examples of the formation of complex ions:
496 Th+4 + 2 NO34 - Th(NO3)2+2
497 Ra+2 + EDTA'4 * - Ra(EDTA)'2
498 U+4 + 5 CO,'2 - U(CO3)5-6
499 * EDTA-4 = Ethylenediaminetetraacetate, C1OOC)2-NH-CH2-CH2-NH-(COO-1)2
500 In a fundamental sense, every ion in solution can be considered complexed; there are no free or
501 "naked" ions. Dissolved ions are surrounded by solvent molecules. In aqueous solutions, the
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 14-19 DO NOT CITE OR QUOTE
-------
Separation Techniques
502
503
504
505
506
507
508
509
510
complexed water molecules, referred to as the inner hydration sphere, form aquo ions that can be
either weakly or strongly bound:
Fe+2 + 6 H2O - Fe(H2O)6+2
From an elementary standpoint, the process of complexation is simply the dynamic process of
replacing one set of ligands, the solvent molecules, with another. The complexation of a metal
ion in aqueous solution with a ligand, L, can be expressed as:
M(H2O)n+x
M(H2O)n.1Lx-y + H2O
Successive aquo groups can be replaced by other ligand groups until the complex MLnx"ny is
formed as follows:
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
^L + L - M(H2O)n.2L2 + H2O, etc.
In the absence of other complexing agents, in dilute aqueous solution solvated metal ions are
simply written as M+n for simplicity.
Ligands are classified by the number of electrons they donate to the metal to form coordination
bonds to the metal. If only one atom in the ligand is bonded to the metal, it is called a unidentate
ligand (dentate is from the Latin word for teeth. It is a categorization of ligands that describe the
number of atoms with electron pairs a ligand has available for donation in complex-ion
formation; if two atoms, bidentate, and so on for tridentate, tetradentate, pentadentate, and
hexadentate.) The term coordination number is also used to indicate the number of atoms
donating electrons to the metal atom. The coordination number is five in U(CO3)5"6, as illustrated
above. EDTA, also illustrated above, is a hexadentate ligand, because it bonds to the metal
through the four oxygen atoms and two nitrogen atoms.
Table 14.4 lists some common ligands arranged by type.
_ TABLE 14.4 — Common ligands _
Ligand Type'
Examples
Unidentate
Bidentate
Tridentate
Water (H2O), halides (X'1), hydroxide (OH'1), ammonia (NH3)
cyanide (CN1), nitrite (NO/1), thiocyanate (SCN'1), carbon
monoxide (CO)
Oxalate, ethylenediamine, citrate
Diethylenetriamine, 1,3,5 triaminocyclohexane
MARLAP
DO NOT CITE OR QUOTE
14-20
JULY 2001
DRAFT FOR PUBLIC COMMENT
-------
Separation Techniques
529
530
531
532
533
534
535
536
537
538
Ligand Type'
Examples
Polydentate
8-hydroxyquinoline, p-diketones (acetylacetone-2-
thenoyltrifluoroacetone [TTA]), ethylenediaminetetraacetic acid
(EDTA), diethylenetriaminepentaacetic acid (DTPA),
organophosphates: (octyl(phenyl)-A^A^-diiso-butylcarbamoyl-
methylphosphine oxide [CMPO]); tributyl phosphate (TBP),
trioctylphosphinic oxide (TOPO), quaternary amines (tricaprylyl-
methylammonium chloride [Aliquat-336]), triiso-octylamine
(TIOA), tri-n-octylamine (TnOA), macrocyclic polyethers (crown
ethers such as [181-crown-6X cryptates
(1) Ligands are categorized by the number of electron pairs available for donation. Unidentate ligands donate
one pair of electrons; bidentate donate two pairs, etc.
A ligand can be characterized by the nature and basicity of its ligand atom. Oxygen donors and
the fluoride ion are general complexing agents; they combine with any metal ion (cation) with a
charge of more than one. Acetates, citrates, tartrate, and p-diketones generally complex all
metals. Conversely, cyanide (CN"1), the heavy halides, sulfur donors, and—to a smaller
extent—nitrogen donors, are more selective complexing agents than the oxygen donors. These
ligands do not complex the A-metals of the periodic table; only the cations of the B-metals and
the transition metals coordinate to carbon, sulfur, nitrogen, chlorine, bromine, and iodine.
14.3.2 Chelates
540 When a multidentate ligand is bound to the metal atom or ion by two or more electron pairs,
541 forming a ring structure, it is referred to as a chelate and the multidentate ligand is called a
542 chelating agent or reagent. Chelates are organic compounds containing two, four, or six
543 carboxylic acid (RCOOH) or amine (RNH2) functional groups. A chelate is effective at a pH
544 where the acid groups are in the anionic form as carboxylates, RCOO"1, but the nitrogen is not
545 protonated so that its lone pair of electrons is free for bonding. The chelate bonds to the metal
546 through the lone pair of electrons of these groups as bi, tetra, or hexadentate ligands, forming a
547 coordination complex with the metal. Binding through multiple sites wraps up the metal in a
548 claw-like fashion, thus the name chelate, which means claw. Practically all chelates form five- or
549 six-membered rings on coordinating with the metal. Chelates are much more stable than complex
550 compounds formed by unidentate reagents. Moreover, if multiple ring systems are formed with a
551 single metal atom or ion, stability improves. For example, ethylenediaminetetraacetic acid
552 (EDTA), a hexadentate ligand, forms especially stable complexes with most metals. As
553 illustrated in Figure 14.1, EDTA has two donor pairs from the nitrogen atoms, and four donor
554 pairs from the oxygen atoms.
JULY 2001
DRAFT FOR PUBLIC COMMENT
14-21
MARLAP
DO NOT CITE OR QUOTE
-------
Separation Techniques
H
H
H U u u
rlo rlo rl
II L L
. c-c c- cv
-O7 X /
:: N .
/
.. ^r - r
\-s \-s
H
1 H2
:O:
:O :
o Ho
L L II
C - C \ ••
\ / O-H
. N
\ 6-H
c - c ^ .-
H2 1
:O:
FIGURE 14.1 — Ethylenediaminetetraacetic Acid(1) (EDTA)
555
556
557
558
559
560
561
(1) EDTA forms very stable complexes with most metal atoms because it has two pairs of electrons available from
the nitrogen atoms, and four pairs of electrons from the oxygen atoms. It is often used as a complexing agent in
a basic solution. Under these conditions, the four carboxylic-acid groups ionize with the loss of a hydrogen ion
(H+1), forming ethylenediaminetetraacetate (EDTA"4), a stronger complexing agent. EDTA is often used as a
food additive to increase shelf life, because it combines with transition metal ions that catalyze the decompo-
sition of food. It is also used as a water softener to remove calcium (Ca+2) and magnesium (Mg+2) ions from
hard water.
562 Various chelating agents bind more readily to certain cations, providing the specificity for
563 separating ions by selective bonding. Usually, the complex is insoluble under the solvent
564 conditions used, allowing the collection of the complex by precipitation. Selectivity of a chelate
565 can be partially controlled by adjusting the pH of the medium to vary the net charge on its
566 functional groups. Different chelates provide specificity through the number of functional groups
567 available for bonding and the size of claw formed by the molecular structure, providing a select
568 fit for the diameter of a specific cation. The electron-donating atoms of the chelate form a ring
569 system with the metal atom when they participate in the coordination bond. In most cases,
570 chelates form much more stable complexes than unidentate ligands. For example, the complex
571 ion formed between Ni+2 and the bidentate ligand ethylenediamine (H2N-CH2-CH2-NH2, or en),
572 Ni(en)3+2, is almost 108 times more stable than the complex ion formed between the metal ion
573 and ammonia, Ni(NH3)+2.
MARLAP
DO NOT CITE OR QUOTE
14-22
JULY 2001
DRAFT FOR PUBLIC COMMENT
-------
Separation Techniques
574
575
576
577
Another class of ligands that is becoming increasingly important to the radiochemist doing
laboratory analyses is the macrocyclic polyethers, commonly called crown ethers (Horwitz et al.,
1991 and 1992; Smith et al., 1996 and 1997). These compounds are cyclic ethers containing a
number of regularly spaced oxygen atoms. Some examples are given in Figure 14.2.
O
0>
[18]-crown-6
[12]-crown-4
dibenzo[ 18]-crown-6
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
FIGURE 14.2 — Crown ethers
First identified in 1967, crown ethers have been shown to form particularly stable coordination
complexes. The term, "crown ether," was suggested by the three-dimensional shape of the
molecule. In the common names of the crown ethers, the ring size is given in brackets, and the
number of oxygen atoms follows the word "crown."
Crown ethers have been shown to react rapidly and with high selectivity (Gokel, 1991; Hiraoka,
1992). This property is particularly significant when a separation requires high selectivity and
efficiency in removing low-level species from complex and concentrated matrices, a situation
frequently encountered in environmental or mixed-waste analyses. Because crown ethers are
multidentate chelating ligands, they have very high formation constants. Moreover, because the
metal ion must fit within the cavity, crown ethers demonstrate some selectivity for metal ions
according to their size. Crown ethers can be designed to be very selective by changing the ring
size, the ring substituents, the ring number, the donor atom type, etc. For example, dibenzo-18-
crown-6 forms a strong complex with potassium; weaker complexes with sodium, cesium, and
rubidium; and no complex with lithium or ammonium, while 12-crown-4, with its smaller cavity,
specifically complexes with lithium.
Other crown ethers are selective for radionuclide ions such as radium and UO2+2. Addition of 18-
crown-6 to solutions containing NpO2+2 causes the reduction of neptunium to Np(V) as NpO2+1,
which is encircled by the ether ligand (Clark et al., 1998).
JULY 2001
DRAFT FOR PUBLIC COMMENT
14-23
MARLAP
DO NOT CITE OR QUOTE
-------
Separation Techniques
596 14.3.3 The Formation (Stability) Constant
597 The stability of the complex is represented by the magnitude of an equilibrium constant
598 representing its formation. The complex ion, [Th(NO3)2+2], forms in two equilibrium steps:
599 Th+4 + NO,4 - Th(NO3)+3
600 Th(NO3)+3 + NO,'1 - Th(NO3)2+2
601 The stepwise formation (stability) constants are:
[Th(NQ3)+3]
602 K' = [Th"][N03-']
603 and
v [Th(N03)2+2]
604
[Th(N03)+3][NO,
605 The overall formation (stability) constantly.
[Th(N03)2+2]
606 K =
[Th+4][N03~1]2
607 which can be calculated from Kx and K2:
608 K = Kj x K2.
609 In the Ni+2 examples cited in the preceding section, the relative stabilities of the complex ions are
610 represented by the values of K; for Ni(en)3+2 it is 1018'28, and for Ni(NH3)+2 it is 108'61 (Cotton and
611 Wilkinson, 1988, p. 45).
612 Many radionuclides form stable complex ions and coordination compounds that are important to
613 the separation and determination steps in radioanalytical chemistry. Formation of a complex
614 changes the properties of the ion in several ways. For example:
615 • Complexation of UO2+2 with carbonate to form UO2(CO3)'4 increases the solubility of the
616 uranium species in groundwater (Lindsay, 1988, p. 9.2-19).
MARLAP JULY 2001
DO NOT CITE OR QUOTE 14-24 DRAFT FOR PUBLIC COMMENT
-------
Separation Techniques
617 • Th+2 forms Th(NO3)6"2 in nitric acid solution (optimally at 7 M) that is the basis for separation
618 of thorium from other actinides and thorium progeny, because they do not form anionic
619 complexes under these conditions (Hyde, 1960, p. 25).
620 • Ra+2 form a very insoluble compound with sulfate (RaSO4) but is soluble in hot concentrated
621 sulfuric acid because of the formation of Ra(SO4)2"2 (Kirby and Salutsky, 1964, p. 9).
622 In addition, the complex ion in solution is in equilibrium with the free (hydrated) ion, and the
623 equilibrium mixture might, therefore, contain sufficient concentration of the free ion for it to be
624 available for other reactions, depending on the stability of the complex ion:
625 Cu+2 + 4 NH3 - Cu(NH3)4+2
626 14.3.4 Complexation and Radiochemical Analysis
627 Property changes also accompany the formation of complex ions and coordination compounds
628 from simple radionuclide ions. These changes provide a valuable approach in radiochemistry for
629 isolating, separating, and measuring radionuclide concentrations, and are important in several
630 areas of radiochemistry.
631 14.3.4.1 Extraction of Laboratory Samples and Ores
632 Uranium ores are leached with alkaline carbonates to dissolve uranium as the UO2(CO3)2"4
633 complex ion after oxygen is used to convert U+4 to U+6 (Grindler, 1962, p. 256). Samples
634 containing refractory plutonium oxides are dissolved with the aid of a nitric acid-hydrofluoric
635 acid solution to produce the complex cation PuF+3 and similar cationic fluorocomplexes
636 (Booman and Rein, 1962, p. 244). Refractory silicates containing niobium (Nb) also yield to
637 fluoride treatment. Potassium bifluoride (KF2 J) is used as a low-temperature flux to produce a
638 fluoride complex NbFg"1 (Willard and Rulfs, 1961, p. 1046; Greenwood and Earnshaw, 1984,
639 p. 1158).
640 14.3.4.2 Separation by Solvent Extraction and Ion-Exchange Chromatography
641 Many ion-exchange separations of radionuclides are based on the formation of complex ions
642 from the metal ions in solution or the displacement of ions bound to an exchanger by complex
643 formation. Uranium in urine samples, for example, is partly purified by forming a chlorocomplex
644 of U+4 and UO2+2 ions, UC16"2 and UC^CV1, that bind preferentially to the anion-exchange ligands
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 14-25 DO NOT CITE OR QUOTE
-------
Separation Techniques
645 in 7 M HC1. Other cations pass through the column under these conditions. Uranium is
646 subsequently eluted with 1 M HC1 (DOE, 1990 and 1997, Method U-01).
647 For separation on a larger scale—such as in an industrial setting—chelates are often used in a
648 column chromatography or filtration unit. They are immobilized by bonding to an inert matrix,
649 such as polystyrene or an alumina/silica material. A solution containing the ions to be separated
650 is passed continuously through the column or over the filter, where the select cations are bonded
651 to the chelate as the other ions pass through. Washing the column or filter with a solution at
652 alternate pH or ionic strength will permit the elution of the bound cation.
653 Tetrapositive thorium (Th) is adsorbed more strongly by cation exchangers than most other
654 cations (Hyde, 1960, pp. 21-23). The adsorbed thorium is separated from most other ions by
655 washing the column with mineral acids or other eluting agents. Even the tetrapositive plutonium
656 ion, Pu+4, and the uranyl ion, UO2+2, are washed off with high concentrations of HC1 because they
657 form chlorocomplexes, PuCl6"2 and UC^CV1, respectively. Thorium is then removed by eluting
658 with a suitable complexing agent such as oxalate, which reduces the effective concentration of
659 Th+4, reversing the adsorption process. Using oxalate, Th(C2O4)4"4 forms and the anion is not
660 attracted to the cation exchanger.
661 14.3.4.3 Formation and Dissolution of Precipitates
662 A classical procedure for the separation and determination of nickel (Ni) is the precipitation of
663 Ni+2 with dimethylglyoxime, a bidentate ligand that forms a highly selective, stable chelate
664 complex with the ion, Ni(C4H7N2O2'1)2 (DOE, 1995, Method RP300). Uranium in the +4
665 oxidation state can also be precipitated from acidic solutions with a chelating agent, cupferron
666 (ammonium nitrosophenylhydroxylamine, C8H5(NO)O"1NH4+1) (Grindler, 1962, p. 256). In
667 another procedure, Co+2 can be selectively precipitated from solution as K3Co(NO2)6. In this
668 procedure, cobalt, which forms the largest number of complexes of all the metals, forms a
669 complex anion with six nitrite ligands, Co(NO2)6"3 (EPA, 1973, pp. 53-58).
670 In radiochemical separations and purification procedures, precipitates of radionuclides are
671 commonly redissolved to release the metal ion for further purification or determination. In the
672 determination of 90Sr, strontium (Sr+2) is separated from the bulk of the solution by direct
673 precipitation of the sulfate, SrSO4. The precipitate is redissolved by forming a complex ion with
674 EDTA, Sr(EDTA)'2, to separate it from lanthanides and actinides (DOE, 1995, Method RP520).
675 Radium also forms a very stable complex with EDTA. Solubilization of radium, Ra+2,
676 coprecipitated with barium sulfate (BaSO4) is used in the 228Ra determination of drinking water
677 by using EDTA (EPA, 1980, pp. 49-57).
MARLAP JULY 2001
DO NOT CITE OR QUOTE 14-26 DRAFT FOR PUBLIC COMMENT
-------
Separation Techniques
678 14.3.4.4 Stabilization of Ions in Solution
679 In some radiochemical procedures, select radionuclides are separated from other elements and
680 other radionuclides by stabilizing the ions as complex ions, while the other substances are
681 precipitated from solution. In a procedure extensively used at Oak Ridge National Laboratory
682 (ORNL), 95Nb is determined in solutions by taking advantage of complex-ion formation to
683 stabilize the ion (Nb+5) in solution during several steps of the procedure (Kallmann, 1964,
684 pp. 343-344). The niobium sample and carrier are complexed with oxalic acid in acidic solution
685 to prevent precipitation of the carrier and to promote interchange between the carrier and 95Nb.
686 Niobium is precipitated as the pentoxide after warming the solution to destroy the oxalate ion,
687 separating it from the bulk of other ions in solution. Niobium is also separated specifically from
688 zirconium by dissolving the zirconium oxide in hydrofluoric acid.
689 14.3.4.5 Detection and Determination
690 Compleximetric titration of metal ions with EDTA using colorimetric indicators to detect the
691 endpoint can be used for determination procedures. Uranium does not form a selective complex
692 with EDTA, but this chelate has been used to titrate pure uranium solutions (Grindler, 1962,
693 p. 94). The soluble EDTA complex of thorium is the basis of a titrimetric determination of small
694 amounts of thorium (Hyde, 1960, p. 9).
695 Spectrometric determinations are also based on the formation of complex ions. Microgram
696 quantities of uranium are determined by the absorbance at 415 nm (a colorimetric determination)
697 of the uranyl chelate complex with dibenzoylmethane, C6H5-CO-CH2-CO-C6H5 (Grindler, 1962,
698 pp. 271-276).
699 14.4 Solvent Extraction
700 14.4.1 Extraction Principles
701 Since the early days of the Manhattan Project, when scientists extracted uranyl nitrate into diethyl
702 ether to purify the uranium used in the first reactors, solvent extraction has been an important
703 separation technique for radiochemists. Solvent extraction, or liquid-liquid extraction, is a
704 technique used both in the laboratory and on the industrial scale. However, current laboratory
705 trends are away from this technique, mainly because of the costs of materials and because it is
706 becoming more difficult and costly to dispose of the mixed waste generated from the large
707 volumes of solvents required. The technique also tends to be labor intensive because of the need
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 14-27 DO NOT CITE OR QUOTE
-------
Separation Techniques
708 for multiple extractions using separately funnels. Nonetheless, solvent extraction remains a
709 powerful separation technique worthy of consideration.
710 Solvent extraction refers to the process of selectively removing a solute from a liquid mixture
711 with a solvent. As a separation technique, it is a partitioning process based on the unequal
712 distribution of the solute (A) between two immiscible solvents, usually water (aq) and an organic
713 liquid (org):
714 Aaq * Aorg
715 The solute can be in a solid or liquid form. The extracting solvent can be water, a water-miscible
716 solvent, or a water-immiscible solvent; but it must be insoluble in the solvent of the liquid
717 mixture. Solutes exhibit different solubilities in various solvents. Therefore, the choice of
718 extracting solvent will depend upon the properties of solute, the liquid mixture, as well as other
719 requirements of the experimental procedure. The solvents in many applications are water and a
720 nonpolar organic liquid, such as hexane or diethyl ether, but other solvent pairs are commonly
721 used. In general terms, the solute to be removed along with impurities or interfering analytes to
722 be separated are already dissolved in one of the solvents (water, for example). In this example, a
723 nonpolar organic solvent is added and the two are thoroughly mixed, usually by shaking in a
724 separately funnel. Shaking produces a fine dispersion of each solvent in the other that will
725 separate into two distinct layers after standing for several minutes. The more dense solvent will
726 form as the bottom layer. Separation is achieved because the solute and accompanying impurities
727 or analytes have different solubilities in the two solvents. The solute, for example, might
728 preferentially remain in the aqueous phase, while the impurities or analyte selectively dissolve in
729 the organic phase. The impurities and analyte are extracted from the aqueous layer into the
730 organic layer. Alternatively, the solute might be more soluble in the organic solvent and will be
731 extracted from the aqueous layer into the organic layer, leaving the impurities behind in the
732 aqueous layer.
733 14.4.2 Distribution Coefficient
734 The different solubilities of a solute in the solvent pairs of an extraction system are described by
735 the distribution or partition coefficient, Kd. The coefficient is an equilibrium constant that
736 represents the solubility of the solute in one solvent relative to its solubility in another solvent.
737 Once equilibrium is established, the concentration of solute in one phase has a direct relationship
738 to the solute concentration in the other phase. This is expressed mathematically by:
739 Kd=[Aorg]/[Aaq]
MARLAP JULY 2001
DO NOT CITE OR QUOTE 14-28 DRAFT FOR PUBLIC COMMENT
-------
Separation Techniques
740 where [Aorg] and [Aaq] are the concentration of the solute in the organic and aqueous phase
741 respectively, and Kd is a constant. The concentrations are typically expressed in units of moles/kg
742 (molality) or g/g; therefore, the constant is unitless. These solubilities usually represent saturated
743 concentrations for the solute in each solvent. Because the solubilities vary with temperature, the
744 coefficient is temperature-dependent, but not by a constant factor. Wahl and Bonner (195 1, pp.
745 434-439) contains a table of solvent extraction systems for carrier-free tracers containing
746 laboratory conditions and distribution coefficients.
747 A distribution coefficient of 90 for a solute in a hexane/water system, for example, means that
748 the solute is 90 times more soluble at saturation conditions in hexane than in water, but note that
749 some of the water still contains a small amount of the solute. Solvent extraction selectively
750 dissolves the solute in one solvent, but it does not remove the solute completely from the other
751 solvent. A larger coefficient would indicate that, after extraction, more solute would be
752 distributed in hexane relative to water, but a small quantity would still be in the water. Solvent
753 extraction procedures often use repeated extractions to extract a solute quantitatively from a
754 liquid mixture.
755 The expression of the distribution law is only a very useful approximation; it is not thermo-
756 dynamically rigorous, nor does it account for situations in which the solute is involved in a
757 chemical reaction, such as dissociation or association, in either phase. Consider, for example,
758 dimerization in the organic phase:
759 2Aorg*(A)2j0rg
760 where the distribution ratio, D, is an alternate form of the distribution coefficient expressed by:
761 D — (|_AorgJmonomer + LAorgJdimer)/|_AaqJ
762 or
763 D = ([Aorg]+2[(A)2i0rg])/[Aaq]
764 Because the concentration of the monomer that represents the dimeric form of the solute is twice
765 that of the concentration of the dimer:
766 [Aorg]dimer = 2[(A)2j0rg]
767 Substitution of ^produces:
768 D
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 14-29 DO NOT CITE OR QUOTE
-------
Separation Techniques
769 where K2 is the dimerization constant, K2 = [(A)2 org]/[Aorg]2. Because dimerization decreases the
770 concentration of the monomer, the species that takes part directly in the phase partition, the
771 overall distribution increases.
772 14.4.3 Extraction Technique
773 There is extensive literature on the topic of extraction technique, but only a few sources are listed
774 here. The theory of solvent extraction is covered thoroughly in Irving and Williams (1961), Lo et
775 al. (1983), and Dean (1995). Moreover, the journal Solvent Extraction and Ion Exchange is an
776 excellent source for current advances in this field. A practical discussion on the basics of solvent
777 extraction is found in Korkisch (1969). The discussion applies to a metallic element in solution
778 as a cation extracted by a nonpolar solvent:
779 "In solvent extraction, the element which is to be separated, contained in an aqueous solution,
780 is converted to a compound which is soluble in an organic solvent. The organic solvent must
781 be virtually immiscible with water. By shaking the aqueous solution with the organic solvent
782 (extractant) in a separating funnel, the element is extracted into the organic phase. After
783 allowing the aqueous and organic phases to separate in the funnel, the organic extract is
784 removed from contact with the aqueous layer. This single-stage batch extraction method is
785 employed when Kd is relatively large and for a simple separation it is essential that the
786 distribution coefficients of the metal ions to be separated be sufficiently different. As in the
787 case of ion exchange, the effectiveness of separation is usually expressed by means of the
788 separation factor which is given by the ratio of the distribution coefficients of two different
789 elements which were determined under identical experimental conditions. This ratio
790 determines the separability of two elements by liquid-liquid extraction. Separations can only
791 be achieved if this ratio shows a value which is different from unity and they are clean and
792 can be quickly and easily achieved where one of the distribution coefficients is relatively
793 large and the other very small (high separation factor).
794 "In those extractions where the separation factor approaches unity, it is necessary to employ
795 continuous extraction or fractionation methods. With the latter techniques distribution,
796 transfer and recombination of various fractions are performed a sufficient number of times to
797 achieve separation. In continuous extraction use is made of a continuous flow of immiscible
798 solvent through the solution or a continuous counter-current flow of both phases. In
799 continuous extraction the spent solvent is stripped and recycled by distillation, or fresh
800 solvent is added continuously from a reservoir. Continuous counter-current extraction
801 involves a process where the two liquid phases are caused to flow counter to each other.
802 Large-scale separations are usually performed using this technique.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 14-30 DRAFT FOR PUBLIC COMMENT
-------
Separation Techniques
803 "When employing liquid-liquid extraction techniques, one of the most important
804 considerations is the selection of a suitable organic solvent. Apart from the fact already
805 mentioned that it must be virtually immiscible with water, the solubility of the extracted
806 compound in the solvent must be high if a good separation is to be obtained. Furthermore, it
807 has to be selective, i.e., has to show the ability to extract one component of a solution in
808 preference to another. Although the selectivity of a solvent for a given component can be
809 determined from phase diagrams, it is a little-used procedure in analytical chemistry. The
810 principal difficulty is simply that too few phase diagrams exist in the literature. The result is
811 that the choice of an extractant is based on either experience or semi-empirical
812 considerations. As a rule, however, polar solvents are used for the extraction of polar
813 substances from nonpolar media, and vice versa. Certainly the interactions of solute and
814 solvent will have an effect on the selectivity of the solvent. If the solute is readily solvated by
815 a given solvent, then it will be soluble in that solvent. Hydrogen bond formation between
816 solute and solvent influences solubility and selectivity.
817 "Almost as important as the selectivity of the extractant is the recovery of the solute from the
818 organic extract. Recovery can be achieved by distillation or evaporation of the solvent,
819 provided that the solute is nonvolatile and thermally stable. This technique is, however, less
820 frequently used than the principle of back extraction (stripping) which involves the treatment
821 of the organic extract with an aqueous solution containing a reagent which causes the
822 extracted solute to pass quantitatively into the aqueous layer...
823 "In solvent extraction the specific gravity of the extractant in relation to the aqueous phase is
824 important. The greater the difference in the solvent densities, the faster will be the rate at
825 which the immiscible layers separate. Emulsions are more easily produced when the densities
826 of the two solvents are similar. Sometimes troublesome emulsions can be broken by
827 introducing a strong electrolyte into the system or by the addition of small quantities of an
828 aliphatic alcohol" (Korkisch, 1969, pp. 20-22).
829 Korkisch continues:
830 "Liquid-liquid extraction can be applied to the analysis of inorganic materials in two different
831 ways.
832 (a) Where the element or elements to be determined are extracted into the organic phase.
833 (b) Where the interfering elements are removed by extraction, leaving the element or
834 elements to be determined in the aqueous phase.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 14-31 DO NOT CITE OR QUOTE
-------
Separation Techniques
835 "Solvent extraction separations are mainly dependent for their successful operation upon the
836 distribution ratio of the species between the organic and aqueous phase and the pH and salt
837 concentration of the aqueous phase. Much of the selectivity which is achieved in liquid-liquid
838 extraction is dependent upon adequate control of the pH of the solution. The addition of
839 masking agents such as EDTA and cyanide can greatly improve selectivity, but they too are
840 dependent upon the pH of the solution to exert their full effect. In many cases complete
841 extractions and separations are obtained only in the presence of salting-out agent. An
842 example is the extraction of uranyl nitrate. In the presence of additional nitrate, the increase
843 in the concentration of the nitrate ion in the aqueous solution shifts the equilibrium between
844 the uranyl ion and the nitrate complexes toward the formation of the latter, and this facilitates
845 a more complete extraction of the uranium into the organic solvent. At the same time, the
846 salting-out agent has another, more general, effect: as its affinity for water is large, it
847 becomes hydrated by the water molecules so that the substance to be extracted is really
848 dissolved in a smaller amount of water, and this is the same as if the concentration in the
849 solution were increased. As a result, the distribution coefficient between the aqueous and the
850 organic phases is increased. As a rule the salting-out agent also lowers the solubility of the
851 extractant in the aqueous phase, and this is often important in separations by extraction. The
852 efficiency of the salting-out action depends upon the nature and the concentration of the
853 salting-out agent. For the same molar concentration of the salting-out agent its action
854 increases with an increase in the charge and decrease in the radius of its cation" (Korkisch,
855 1969, pp. 23-24).
856 A hydrated metal ion will always prefer the aqueous phase to the organic phase because of
857 hydrogen bonding and dipole interaction in the aqueous phase. Therefore, to get the metal ion to
858 extract, some or all of the inner hydration sphere must be removed. The resulting complex must
859 be neutrally charged and organophilic. Removal of the hydration sphere is accomplished by
860 coordination with an anion to form a neutral complex. Neutral complexes will generally be more
861 soluble in an organic phase. Larger complexing anions favor the solubility in the organic phase.
862 Extracting agents are thus divided into three classes: polydentate organic anions, neutral organic
863 molecules, and large organic cations. Many of the multidentate ligands discussed previously are
864 used in solvent extraction systems.
865 The radioanalytical procedure for uranium (U) and thorium (Th) employs solvent extraction to
866 separate the analytes before alpha counting (EPA, 1984, pp. U/Th-01-1-14). An aqueous solution
867 of the two is extracted with a 10 percent solution of triisooctylamine (TIOA) in para-xylene to
868 remove uranium, leaving thorium in the water (Grinder, 1962, pp. 175-180). Each solution is
869 further processed to recover the respective radionuclides for separate counting.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 14-32 DRAFT FOR PUBLIC COMMENT
-------
Separation Techniques
870 14.4.4 Solvent Extraction and Radiochemical Analysis
871 In many purification procedures, separated solutions are used directly in further isolation steps. If
872 necessary, the substances can be collected by distillation or evaporation of the respective
873 solvents. In the uranium/thorium procedure described above, the aqueous layer containing
874 thorium is evaporated, and the thorium is redissolved in an alternate solution before it is purified
875 further. In other cases, the solution is extracted again to take up the solute in another solvent
876 before the next step in the procedure. Uranium in TIOA//>-xylene, for example, is extracted back
877 into a nitric acid solution for additional purification (EPA, 1984, pp. U/Th-01-l-U/Th-01-14).
878 In some solvent-extraction procedures, more than one extraction step is required for the
879 quantitative removal of a solute from its original solvent. The solute is more soluble in one
880 component of the solvent pair, but not completely insoluble in the other component, so
881 successive extractions of the aqueous solution of the solute by the organic solvent will remove
882 more and more of the solute from the water until virtually none remains in the aqueous layer.
883 Extraction of uranium with TIOA//>-xylene, for example, requires two extractions before
884 quantitative removal is achieved (EPA, 1984, pp. U/Th-01-l-U/Th-01-14). The organic layers
885 containing the uranium are then combined into one solution for additional processing.
886 Solvent extraction is greatly influenced by the chemical form (ionic or molecular) of the solute to
887 be extracted, because different forms of the solute can have different solubilities in the solvents.
888 In the uranium/thorium procedure described above, uranium is extracted from water by
889 TIOA/hydrochloric acid, but it is stripped from the amine solution when extracted with nitric
890 acid. Simply changing the anion of uranium and TIOA from chloride to nitrate significantly alters
891 the complex stability of uranium and TIOA.
892 Organic amines are sometimes converted to their cationic forms, which are much more soluble in
893 water and much less soluble in organic solvents. The amine is converted to the corresponding
894 ammonium salt by an acid, such as hydrochloric acid:
895 RNH2 + HC1 - RMV'Cr1
896 Correspondingly, carboxylic acids are converted to their carboxylates that are more soluble in
897 water and less soluble in organic solvents. They are produced by treating the carboxylic acid with
898 a base, such as sodium hydroxide:
899 RCOOH + NaOH - RCOO'W1 + H9O
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 14-33 DO NOT CITE OR QUOTE
-------
Separation Techniques
900 Multidentate organic anions that form chelates are important extracting agents. These reagents,
901 such as the p-diketonates and thenoyltrifluoroacetone (TTA) (Ahrland, 1986, pp. 1518-1521), are
902 commonly used for extracting the actinide elements. When the aqueous solution and organic
903 phase come into contact with one another, the chelating agent dissolves in the aqueous phase,
904 ionizes, and complexes the metal ion; the resulting metal chelate subsequently dissolves in the
905 organic phase.
906 A number of organophosphorus compounds are also efficient extractants because they and their
907 complexes are readily soluble in organic solvents. The actinide MO2+2 and actinide +4 ions are
908 very effectively extracted by reagents such as monobasic diethylhexylphosphoric acid (HDEHP)
909 and dibutylphosphoric acid (HDBP) (Cadieux and Reboul, 1996).
910 Among the neutral compounds, alcohols, ethers, and ketones have been commonly employed as
911 extractants. Methyl isobutyl ketone was used in one of the early large-scale processes (the Redox
912 process) to recover uranium and plutonium from irradiated fuel (Choppin et al., 1995, p. 607).
913 However, the most widely used neutral extractants are the organophosphorus compounds such as
914 TBP (tributyl phosphate). The actinide elements thorium, uranium, neptunium, and plutonium
915 easily form complexes with TBP (Choppin et al., 1995, p. 607). Salting-out agents such as HNO3
916 and A1(NO3)3 are commonly employed to increase extraction in these systems. This chemistry is
917 the basis of the Purex process used to reprocess spent nuclear fuel (Choppin et al., 1995, pp. 608-
918 610).
919 An important addition to the Purex process is the solvent extraction procedure known as TRUEX
920 (7/ans Uranium .Extraction). This process uses the bifunctional extractant CMPO
921 ([octyl(phenyl)]-N,N-diisobutylcarbonylmethylphosphine oxide) to remove transuranium
922 elements from the waste solutions generated in the Purex process. This type of compound
923 extracts actinides at high acidities, and can be stripped at low acidity or with complexing agents.
924 Many of the recent laboratory procedures for biological waste and environmental samples are
925 based upon this approach (see Section 14.4.5.1, "Extraction Chromatography Columns").
926 The amines, especially the tertiary and quaternary amines, are strong cationic extractants. These
927 strong bases form complexes with actinide metal cations. The extraction efficiency improves
928 when the alkyl groups have long carbon chains, such as in trioctylamine (TnOA) or
929 triisooctylamine (TIOA). The pertechnetate ion (TcO44) is also extracted by these cationic
930 extractants (Chen, 1990).
931 Table 14.5 lists common solvent extraction procedures for some radionuclides of interest and
932 includes the examples described above.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 14-34 DRAFT FOR PUBLIC COMMENT
-------
Separation Techniques
933
934
935
936
TABLE 14.5 — Radioanalytical methods employing solvent extraction
(i)
Analyte
89/90Cj.
"Tc04-
210pb
Radium through
Californium
Actinides
Thorium
Uranium
Extraction Conditions (Reference)
From soils and sediments with dicyclohexano-18-crown-6 in trichloromethane with back
extraction with EDTA (Pimpl, 1995)
From dilute H2SO4 solutions into a 5% TnOA in xylene mixture and back extracted with NaOH
(Golchert and Sedlet, 1969; Chen, 1990); from dilute H2SO4, HNO3, and HC1 solutions into a
5% TnOA in xylene (Dale et al., 1996); from HNO3 into 30% TnOA in xylene and back
extracted with NaOH (Hirano, 1989); from dilute H2SO4 solutions into TBP (Holm et al., 1984;
Garcia-Leon, 1990); the tetraphenyl arsonium complex of Tc into chloroform (Martin and
Hylko, 1987); from K2CO3 with MEK (Paducah R-46); from alkaline nuclear-waste media with
crown ethers (Bonnesen et al., 1995)
As lead bromide from bone, food, urine, feces, blood, air, and water with Aliquat-336 (DOE,
1990 and 1997, Method Pb-01; Morse and Welford, 1971)
From soil following KF-pyrosulfate fusion and concentration by barium sulfate precipitation
with Aliquat-336 in xylene (Sill et al., 1974)
From water following concentration by ferric hydroxide precipitation and group separation by
bismuth phosphate precipitation, uranium extracted by TOPO, plutonium and neptunium
extracted by TIOA from strong HC1, and thorium separated from americium and curium by
extraction with TOPO (EPA, 1980, Method 907.0)
And other metals from TOPO (NAS-NS 3 102) and from high-molecular weight amines such as
TIOA(NAS-NS3101).
Uranium and plutonium from HC1 with TIOA (Moore, 1958)
From nitric acid wastes using the TRUEX process with CMPO (Horwitz et al., 1985 and 1987)
With various extractive scintillators followed by PERALS® spectrometry (McDowell 1986 and
1992); with HDEHP after extraction chromatography followed by PERALS® spectrometry
(Cadieux and Reboul, 1996)
From aqueous samples after ion exchange with TTA, TIOA, or Aliquat-336 (DOE, 1995,
Method RP570)
From waters with ethyl acetate and magnesium nitrate as salting-out agent (EPA, 1980, Method
908. 1); with URAEX™ followed by PERALS® spectrometry (Leyba et al., 1995)
From soil, vegetation, fecal ash, and bone ash with Alamine-336 (DOE, 1990 and 1997,
Methods Se-Ol.U-03)
937
938
939
940
941
942
943
944
945
(1) This list is representative of the methods found in the literature. It is not an exhaustive compilation, nor does it
imply preference over methods not listed.
14.4.5 Solid-Phase Extraction
946 A technique closely related to solvent extraction is solid-phase extraction (SPE). SPE is a
947 solvent-extraction system in which one of the liquid phases is made stationary by adsorption onto
948 a solid support, usually silica, and the other liquid phase is mobile. Small columns or membranes
JULY 2001
DRAFT FOR PUBLIC COMMENT
14-35
MARLAP
DO NOT CITE OR QUOTE
-------
Separation Techniques
949 are used in the SPE approach. Many of the same extracting agents used in solvent extraction can
950 be used in these systems. SPE is becoming widely accepted as an excellent substitute for liquid-
951 liquid extraction because it is generally faster, more efficient, and generates less waste.
952 14.4.5.1 Extraction Chromatography Columns
953 Over the past decade, extraction chromatography methods have gained wide acceptance in the
954 radiochemistry community as new extraction chromatographic resins have become commercially
955 available, such as Sr, TRU, and TEVA resins (Eichrom Industries, Inc., Darien, IL) (Dietz and
956 Horwitz, 1993; Horwitz et al., 1991, 1992, and 1993). These resins are composed of extractant
957 materials, such as CMPO and 4,4'(5')-bis(t-butylcyclohexano)-18-crown-6, absorbed onto an
958 inert polymeric support matrix. They are most frequently used in a column, rather than a batch
959 mode.
960
961
962
963
964
965
966
967
968
969
970
971
Another example of the advances in the area are use of fibrous discs impregnated with high
molecular weight chelates, selective for certain nuclides such as Cs, Sr, and Tc (Empore Discs,
3M Company, and the TEVA Disc, Eichrom Industries, Inc.). Many of the traditional methods
based upon repetitive precipitations, or solvent extraction in separately funnels, have been
replaced by this strategy. This approach allows for the specificity of liquid-liquid extraction with
the convenience of column chromatography. Numerous papers detailing the determination of
radionuclides by this technique have been published recently, and examples are cited in Table
14.6.
TABLE 14.6 — Radioanalytical methods employing extraction chromatography
(i)
Analyte
Ni-59/63
Sr-89/90
Sr-90
Y-90
Tc-99
Ligand
dimethylgloxime
4,4'(5')-bis(t-butyl-cyclohexano)-18-
crown-6 in n-octanol
octyl(phenyl)-7V,yV-diisobutyl-
carbamoylmethylphosphine oxide
[CMPO] in tributyl phosphate
4,4'(5')-bis(t-butyl-cyclohexano)-18-
crown-6 in n-octanol
Aliquat-336N
Method Citations
Aqueous samples (DOE, 1997)
Biological, Environmental, and Nuclear Waste (Horwitz
et al., 1991 and 1992); Water (ASTM, D5811-95; DOE,
1995, Method RP500); Urine (Dietz and Horwitz, 1992;
Alvarez and Navarro, 1996); Milk (Jeter and Grob,
1994); Geological Materials (Pin and Bassin, 1992)
Brines (Bunzl et al., 1996)
Medical applications (Dietz and Horwitz, 1992)
Low-level radioactive waste (Banavali, 1995); Water
(Sullivan et al., 1993; DOE, 1993, Method RP550)
972
973
974
MARLAP
DO NOT CITE OR QUOTE
14-36
JULY 2001
DRAFT FOR PUBLIC COMMENT
-------
Separation Techniques
Analyte
Pb-210
Ra-228
Rare earths
Actinides
Ligand
4,4'(5')-bis(t-butyl-cyclohexano)-18-
crown-6 in isodecanol
octyl(phenyl)-7V,yV-diisobutyl-
carbamoylmethylphosphine oxide
[CMPO] in tributyl phosphate or
diethylhexyl-phosphoric acid [HDEHP]
impregnated in Amberlite XAD-7
diamyl,amylphosphonate
octyl(phenyl)-7V,yV-diisobutyl-
carbamoylmethylphosphine oxide [C-
MPO] in tributyl phosphate and dieth
ylhexyl-phosphoric acid [HDEHP]
impregnated in Amberlite XAD-7
octyl(phenyl)-7V,yV-diisobutyl-
carbamoylmethylphosphine oxide
[CMPO] in tributyl phosphate and
4,4'(5')-bis(t-butyl-cyclohexano)-18-
crown-6 in n-octanol
octyl(phenyl)-7V,yV-diisobutyl-
carbamoylmethylphosphine oxide
[CMPO] in tributyl phosphate
diamyl,amylphosphonate
tri-n-octylphosphine oxide [TOPO] and
di(2-ethylhexyl)phosphoric acid
[HDEHP]
Method Citations
Water (DOE, 1995, Method RP280); Geological
materials (Horwitz et al., 1994; Woittiez and Kroon,
1995); complex metal ores (Gale, 1996)
Natural waters (Burnett et al., 1995); Volcanic rocks
(Chabaux, 1994)
Actinide-containing matrices (Carney, 1995)
Sequential separation of light rare earths, U, and Th in
geological materials (Pin et al., 1996)
Concomitant separation of Sr, Sm, and Nd in silicate
samples (Pin etal., 1994)
Air filters (Berne, 1995); Waters (Berne, 1995); Group-
screening (DOE, 1997, Method RP725); Urine (Horwitz
et al., 1990; Nguyen et al., 1996); Acidic media
(Horwitz, 1993; DOE, 1997); Soil and sludge (Smith et
al., 1995; Kaye et al., 1995); Environmental (Bunzl and
Kracke, 1994)
Acidic media (Horwitz et al., 1992)
Environmental and industrial samples (Testa et al.,
1995)
975
976
977
978
979
980
981
(1) This list is representative of the methods found in the literature. It is not complete, nor does it imply preference
over methods not listed.
14.4.5.2 Extraction Membranes
982 SPE membranes have also become a popular approach to sample preparation for organic
983 compounds in aqueous samples over the past decade. As of 1995, 22 methods employing SPE
984 disks have been accepted by the U.S. Environmental Protection Agency. More recently, disks
985 have been developed for specific radionuclides, such as technetium, strontium, and radium
986 (DOE, 1990 and 1997; Orlandini, 1998; Smith et al., 1996 and 1997).
JULY 2001
DRAFT FOR PUBLIC COMMENT
14-37
MARLAP
DO NOT CITE OR QUOTE
-------
Separation Techniques
987 These SPE membranes significantly reduce extraction time and reagent use. Samples typically
988 are processed through the membranes at flow rates of at least 50 milliliters per minute; a one liter
989 sample can be processed in as little as 20 minutes. Moreover, these selective-membranes often
990 can be counted directly, thereby condensing sample preparation and counting source preparation
991 into a single step. Many of the hazardous reagents associated with more traditional methods are
992 eliminated in this approach, and these membrane-based extractions use up to 90 percent less
993 solvent than liquid-liquid extractions. The sorbent particles embedded in the membrane are
994 extremely small and evenly distributed, thereby eliminating the problem of channeling that is
995 associated with columns.
996 14.4.6 Advantages and Disadvantages of Solvent Extraction
997 14.4.6.1 Advantages
998 • Lends itself to rapid and very selective separations that are usually highly efficient.
999 • Partition coefficients are often approximately independent of concentration down to tracer
1000 levels and, therefore, can be applied to a wide range of concentrations.
1001 • Can usually be followed by back-extraction into aqueous solvents or, in some cases, the
1002 solution can be used directly in subsequent procedures.
1003 • Wide scope of applications—the composition of the organic phase and the nature of
1004 complexing or binding agents can be varied so that the number of practical combinations is
1005 virtually unlimited.
1006 • Can be performed with simple equipment, but can also be automated.
1007 • Column extraction is fast, very selective, generates a low volume of waste, can often be
1008 applied to samples from very acidic media, requires relatively inexpensive materials, and can
1009 often be correlated with liquid/liquid extraction.
1010 14.4.6.2 Disadvantages
1011 • Cumbersome for a large number of samples or for large samples.
1012 • Often requires toxic and/or flammable solvents.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 14-38 DRAFT FOR PUBLIC COMMENT
-------
Separation Techniques
1013 • Can be time consuming, especially if attainment of equilibrium is slow.
1014 • Can require costly amounts of organic solvents and generate large volumes of organic waste.
1015 • Can be affected by small impurities in the solvent(s).
1016 • Multiple extractions might be required, thereby increasing time, consumption of materials,
1017 and generati on of waste.
1018 • Formation of emulsions can interfere.
1019 • Counter-current process can be complicated and can require complicated equipment.
1020 • Alteration of chemical form can change, going from one phase to the other, thereby altering
1021 the distribution coefficient and effectiveness of the extraction.
1022 • Tracer-levels of analytes can form radiocolloids that cannot be extracted, dissociate into less
1023 soluble forms, or adsorb on the container surface or onto impurities in the system.
1024 • Extraction columns cannot be reused.
1025 14.5 Volatilization and Distillation
1026 14.5.1 Introduction
1027 Differences in vapor pressures of elements or their compounds can be exploited for the
1028 separation of radionuclides. Friedlander et al. (1981, p. 300), describes the process:
1029 "The most straightforward application is the removal of radioactive rare gases from aqueous
1030 solutions or melts by sweeping an inert gas or helium. The volatility of... compounds ... can
1031 be used to effect separations ...by distillation ... Distillation and volatilization methods often
1032 give clean separations, provided that proper precautions are taken to avoid contamination of
1033 the distillate by spray or mechanical entrapment. Most volatilization methods can be done
1034 without specific carriers, but some nonisotopic carrier gas might be required. Precautions are
1035 sometimes necessary to avoid loss of volatile radioactive substances during the dissolving of
1036 irradiated targets or during irradiation itself."
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 14-39 DO NOT CITE OR QUOTE
-------
Separation Techniques
1037 Similar precautions are also advisable during the solubilization of samples containing volatile
1038 elements or compounds (Chapter 13, Sample Dissolution).
1039 14.5.2 Volatilization Principles
1040 Volatilization particularly provides a rapid and often selective method of separation for a wide
1041 range of elements (McMillan, 1975, p. 306). A list of the elements that can be separated by
1042 volatilization and their chemical form(s) upon separation are given in Table 14.7 (McMillan,
1043 1975, p. 307).
1044 McMillan continues (1975, p. 306):
1045 "While many of the volatile species are commonly encountered and a large proportion can be
1046 produced from aqueous solutions, a significant number are rarely met. The volatilization of
1047 highly reactive materials and those with high boiling points are only used in special
1048 circumstances, e.g., for very rapid separations. ... Many other volatile compounds have been
1049 used to separate the elements, including sulphides, carbonyls, stable organic complexes
1050 and fluorinated p-diketones for the lanthanides."
1051 "Separation, ... , is achieved by differentiation during the volatilization process, fractionation
1052 by transfer, and selective collection. Gaseous evolution can be controlled by making use of
1053 differences in vapor pressure with temperature, adjustment of the oxidation state of the
1054 element in solution or by alteration of the matrix, in order to change the chemical
1055 combination of the element. Once gaseous, additional separation is possible and physical
1056 processes can be adopted such as gas chromatography, zone refining, fractional distillation,
1057 electrostatic precipitation, filtration of condensed phases and low temperature trapping.
1058 Chemical methods used are mainly based on the selective trapping of interfering substances
1059 by solid or liquid reagents. The methods of preferential collection of the species sought are
1060 similar to those used in the transfer stage."Both solid and liquid samples can be used in
1061 volatilization separations (Krivan, 1986, p. 377):
1062 "With solid samples, there are several types of separation methods. The most important of
1063 them are ones in which (1) the gas forms a volatile compound with only the trace elements
1064 and not the matrix, (2) the gas forms a volatile compound with the matrix but not the trace
1065 elements, and (3) volatile compounds are formed with both the matrix and the trace elements.
1066 Different gases have been used in separation by volatilization, including inert gases N2, He,
1067 and Ar and the reactive gases H2O, O2, H2, ... F2, and HF. The apparatus usually consists of
MARLAP JULY 2001
DO NOT CITE OR QUOTE 14-40 DRAFT FOR PUBLIC COMMENT
-------
Separation Techniques
TABLE 14.7 — Elements separable by volatilization as certain sp
ffi «
*Zt cd
pj
0 1
*1
ol
+
< «
0 I
.1
J
353
^ T3
m
t
£•8
fil
T3
H T3
0
CO
C3
O
^•8
M|
o "S
H ^
CO X>
S ""O
CO X>
M cd
S 03
bO
•^C cd
S
PH ca
(§"8
S-S
1.
g.
N T3
>
CO T3
5 -a
3-s
o-g
'« •§
£
H 03
£•8
•^C cs
fi
^ T3
0^
2 "8
^
HH .
HH .
^
C3
*o
OS
^
^
H
W
0
tn
>^
e
o
w
s
CO
PH
§
£
*
-------
Separation Techniques
1068 three parts: gas regulation and purification, oven with temperature programming and control,
1069 and condensation or adsorption with temperature regulation."
1070 "The radiotracer technique provides the best way to determine the recoveries of trace
1071 elements in the volatilization process and to optimize the separation with respect to the
1072 pertinent experimental parameters."
1073 14.5.3 Distillation Principles
1074 Distillation is the separation of a volatile component(s) of a mixture by vaporization at the
1075 boiling point of the mixture and subsequent condensation of the vapor. The vapor produced on
1076 boiling the mixture is richer in the more volatile component—the component with the higher
1077 vapor pressure (partial pressure) and correspondingly lower boiling point. The process of
1078 distillation, therefore, essentially takes advantage of the differences in the boiling points of the
1079 constituents to separate a mixture into its components. It is a useful separation tool if the analyte
1080 is volatile or can be transformed into a volatile compound. Most inorganic applications of
1081 distillation involve batch distillation, whereas most organic applications require some type of
1082 fractional distillation. In a simple batch distillation, the sample solution containing a single
1083 volatile component or components with widely separated boiling points is placed in a distillation
1084 flask, boiling is initiated, and the vapors are then continuously removed, condensed, and
1085 collected. Mixtures containing multiple volatile components require fractional distillation., which
1086 employs repeated vaporization-condensation cycles for separation, and is commonly performed
1087 in a fractionation column for that purpose. The column allows the cycles to occur in one
1088 operation, and the separated component is collected after the last condensation.
1089 Distillation has been widely used for separating organic mixtures but this approach has less
1090 applicability in inorganic analysis (Korkisch, 1969, p. 25). Korkisch states: "Nevertheless, some
1091 of the elements of interest to radiochemists can be very effectively separated by distillation as
1092 their volatile chlorides, bromides, and oxides .... these elements are germanium (Ge), selenium
1093 (Se), technetium (Tc), rhenium (Re), ruthenium (Ru), and osmium (Os) (Korkisch, 1969, p. 25;
1094 also see DOE, 1995 Method RP530). Two common analytes determined through distillation,
1095 tritium and 226Ra, by radon emanation are discussed below.
1096 Specific distillation principles are commonly found in chemistry reference and textbooks. For a
1097 theoretical discussion of distillation see Peters (1974) and Perry and Weisberger (1965, pp. 1-
1098 229). Distillation procedures are discussed for many inorganic applications in Dean (1995) and
1099 for less common radioanalytes in the NAS-NS 3108 Monograph, Application of Distillation
MARLAP JULY 2001
DO NOT CITE OR QUOTE 14-42 DRAFT FOR PUBLIC COMMENT
-------
Separation Techniques
1100 Techniques to Radiochemical Separation (DeVoe, 1962), and in NAS-NS 3104 Monograph,
1101 Rapid Radiochemical Separations (Kuska and Meinke, 1961).
1102 14.5.4 Separations in Radiochemical Analysis
1103 The best known use of distillation in radiochemical analysis is in the determination of tritium
1104 (EPA, 1984, pp. H-01-1-8; DOE, 1995, pp. RP580). Water is the carrier as simple distillation is
1105 used to separate tritium from water or soil samples. For determination of tritium, the aqueous
1106 sample is treated with a small amount of sodium hydroxide (NaOH) and potassium permanganate
1107 (KMnO4), and it is then distilled. The early distillate is discarded, and a portion of the distillate is
1108 collected for tritium determination by liquid scintillation counting. The alkaline treatment
1109 prevents other radionuclides, such as radioiodine or radiocarbon, from distilling over with the
1110 tritium (3H), and the permanganate (MnO44) treatment destroys trace organic material in the
1111 sample that could cause quenching during the counting procedure.
1112 Larger samples are distilled using a round-bottom flask, while a MICRO DIST® tube can be
1113 utilized for smaller samples (DOE, 1995, Method RP580). The distillate can be added directly to
1114 a liquid scintillation cocktail (EPA, 1980, Method 906.0), or further enriched by acid electrolysis
1115 (DOE, 1990 and 1997, Method 3H-01) or alkaline electrolysis (DOE, 1990 and 1997, Method 3H-
1116 02).
1117 Iodine (I2) is separated from aqueous samples by distillation from acidic solutions into alkaline
1118 solutions (EPA, 1973, pp. 73-76). Iodide (I"1) is added as carrier; but nitric acid (HNO3) as part of
1119 the acid solution, oxidizes the anion to molecular iodine as the mixture is heated for distillation.
1120 One determination of 79Se employs an optional purification step, distillation of the metal as
1121 selenous acid, H2SeO3 (DOE, 1995, Method RP530). The solution is maintained with excess
1122 bromine (Br2) and hydrobromic acid (HBr) to hold the selenium in the oxyacid form during the
1123 distillation. Technetium can be separated from other elements, or can be separated from
1124 ruthenium, osmium, or rhenium by distillation of their oxides (Friedlander et al., 1981, p 300).
1125 Metals are sometimes distilled in their elemental form—polonium in bismuth or lead (McMillan,
1126 1975, p. 308).
1127 226Ra in solution can be determined by de-emanating its gaseous progeny 222Rn into an ionization
1128 chamber or scintillation cell. Generally, the procedure initially involves the concentration of
1129 radium by coprecipitation with barium sulfate (BaSO4). The barium sulfate is then dissolved in
1130 an EDTA solution, transferred to a sealed bubbler, and stored to allow for the ingrowth of 222Rn.
1131 Following sufficient in-growth, the 222Rn is de-emanated by purging the solution with an inert
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 14-43 DO NOT CITE OR QUOTE
-------
Separation Techniques
1132 gas, such as helium (He) or argon (Ar), and is transferred via a drying tube to a scintillation cell
1133 or ionization chamber. After the short-lived 222Rn progeny have reached secular equilibrium with
1134 the 222Rn (approximately four hours), the sample is counted to determine alpha activity (EPA,
1135 1980, Method 903.1; DOE, 1990 and 1997, Methods Ra-01 through Ra-07; Sedlet, 1966; Lucas,
1136 1990).
1137 When processing samples containing radon, care should be taken to guard against the inadvertent
1138 loss of the gas or contamination of the distillation apparatus. Radon can be adsorbed on, or
1139 permeate through, materials used in its handling. Diffusion through rubber and plastic tubing or
1140 through polyethylene bottles has been observed. Since radon is soluble in many organic
1141 compounds, impurities, including greases used in ground-glass connections, can increase
1142 adsorption.
1143 14.5.5 Advantages and Disadvantages of Volatilization
1144 14.5.5.1 Advantages
1145 • Can be very selective, producing clean separations.
1146 • Very rapid, especially with high-vacuum equipment.
1147 • Can be performed from solid or liquid samples.
1148 • Most can be performed without a specific carrier gas.
1149 14.5.5.2 Disadvantages
1150 • Relatively few volatile elements or inorganic compounds are available.
1151 • Atmosphere can alter the nature of a volatile form of the tracer or surface material.
1152 • Effects of experimental parameters (carrier gas, gas flow, temperature, time, and recovery)
1153 are highly variable.
1154 • Precautions are sometimes necessary to avoid loss of volatile radionuclide substances during
1155 subsequent procedures.
1156 • Some systems require high-temperature, complex equipment.
1157 • Contamination of distillate by carrier, spray, or mechanical entrapment is a potential problem.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 14-44 DRAFT FOR PUBLIC COMMENT
-------
Separation Techniques
1158 14.6 Electrodeposition
1159 14.6.1 Electrodeposition Principles
1160 Radionuclides in solution as ions can be deposited (plated) by electrochemical reactions (redox
1161 reactions) onto an electrode, either by a spontaneous process (produced by a favorable electrode
1162 potential existing between the ion and electrode) or by a nonspontaneous process (requiring the
1163 application of an external voltage (potential) (Section 14.2, "Oxidation and Reduction
1164 Processes").
1165 Spontaneous electrochemical processes are described by the Nernst equation, which relates the
1166 electrode potential of the reaction to the activity of substances participating in a reaction:
1167 E=E° - RT/nF ln(ap/ar)
1168 where E is the electrochemical potential, E° is the standard potential for the process, R is the
1169 ideal gas constant, T is the absolute temperature, n is the number of electrons exchanged in the
1170 redox reaction, F is Faraday's constant, and 3p and ar are the activities of the products of the
1171 reaction and the reactants, respectively. The activity^ of ions in solution is a measure of their
1172 molar concentration (c in moles/L) under ideal conditions of infinite dilution. Expressing the
1173 activities in terms of the product of molar concentrations and activity coefficients, y (a measure
1174 of the extent the ion deviates from ideal behavior in solution; thus a=y • c, where y < 1), the
1175 Nernst equation becomes:
1176 E=E° - RT/nF ln(ypcp/yrcr)
1177 For dilute solutions of electrolytes (< 10"2 molar), the activity coefficient is approximately one
1178 (y~ 1; it approaches one as the solution becomes more dilute, becoming one under ideal
1179 conditions). Then, the Nernst equation is expressed in terms of the concentrations of ions in
1180 solution, the typical form in which the equation is found in most chemistry textbooks (see also
1181 Section 14.8.3.1, "Solubility and Solubility Product Constant, Ksp," for an application of activity
1182 to the solubility product constant):
1183 E=E° - RT/nF ln(cp/cr)
1184 At concentrations less than 10"6M, electrodeposition may show considerable deviations from
1185 behavior of macroamounts of elements whose behavior partly depends on the nature and
1186 previous treatment of the electrode (Adolff and Guillaumont, 1993, p. 275). Inconsistent
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 14-45 DO NOT CITE OR QUOTE
-------
Separation Techniques
1187 behavior is the result of heterogeneity of the surface metal, a very important consideration when
1188 electrodepositing radionuclides at very low concentrations. The spontaneity predicted by the
1189 Nernst equation for macroconcentrations of ions in solution at controlled potential is not always
1190 observed for microconcentrations (Choppin et al., 1995, p. 246). The activity of radionuclide ions
1191 is usually unknown at low concentrations even if the concentration is known, because the activity
1192 coefficient (y) is dependent on the behavior of the mixed electrolytic system. In addition, the
1193 concentration might not be accurately known because ions might adsorb on various surfaces,
1194 form complexes with impurities, or precipitate on the electrode, for example. (See ection
1195 14.9.3.7, "Oxidation and Reduction of Tracers," for another application of the Nernst equation.)
1196 Separation is limited partly because electrodeposition from very dilute solutions is slow, but it is
1197 also limited because it rarely leads to complete separation of one element from many others
1198 (Coomber, 1975, p. 313). Overall, the behavior of an element during an electrochemical process
1199 is determined by its electrochemical potential, which depends on the nature of the ion; its
1200 chemical form, its concentration, the general composition of the electrolyte, the current density,
1201 material and design of the electrode, and construction features of the electrochemical cell
1202 (Zolotov, 1990, pp. 94-95).
1203 Often trace elements are deposited on a solid cathode, but large separation factors between
1204 micro- and macro-components are required. This condition is met when electrochemically active
1205 metals are the main components or when the analyzed matrix does not contain macro-
1206 components that will separate on the cathode (Zolotov, 1990, p. 95). Deposition of heavy metals
1207 and actinides can be more difficult to control, for example, because of the decomposition of
1208 water and reactions of cations and anions at electrodes (Adolff and Guillaumont, 1993, p. 158).
1209 In some cases, deposition of matrix components can be avoided by selection of a suitable
1210 medium and composition of the electrolyte. Overall, the effectiveness of electrodeposition of
1211 trace components depends on the electrode potential, electrode material and its working surface
1212 area, duration of electrolysis, properties of the electrolyte (composition and viscosity),
1213 temperature, and mixing rate (Zolotov, 1990, pp. 95-96). Even so, published data are empirical
1214 for the most part, and conditions for qualitative reproducible separation are determined for each
1215 case. It is difficult, therefore, to make general recommendations for selecting concentration
1216 conditions. It is advisable to estimate and account for possible effects of different electrolysis
1217 factors when developing separation or concentration methodologies (Zolotov, 1990, p. 98).
1218 14.6.2 Separation of Radionuclides
1219 Although electrodeposition is not frequently used as a radiochemical separation technique,
1220 several radionuclides [including iron (Fe) (Hahn, 1945), cadmium (Cd) (Wright, 1947), and
1221 technetium (Tc) (Flagg, 1945)] have been isolated by electrodeposition on a metal electrode.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 14-46 DRAFT FOR PUBLIC COMMENT
-------
Separation Techniques
1222 Electrodeposition is, however, the standard separation technique for polonium (Po), copper (Cu),
1223 and platinum (Pt). Polonium is isolated through deposition on nickel from a strong hydrochloric
1224 acid (HC1) medium (DOE, 1990 and 1997, Method Po-01). This separation is very specific, and,
1225 therefore, can be accomplished in the presence of many other radionuclides. Electrodeposition at
1226 a mercury cathode has also been used to separate technetium from fission products and for group
1227 separation of fission products (Coomber, 1975, p. 198). Numerous metals have been deposited
1228 on thin metal films by electrolysis with a magnesium (Mg) cathode. According to Coomber,
1229 "Electrodeposition of metals can be sensitive to the presence of other substances" (Coomber,
1230 1975, p. 198). Deposition of polonium on silver (Ag) is inhibited by iron unless a reducing agent
1231 is present; and the presence of fluoride (F"1), trace amounts of rare earths, can inhibit the
1232 deposition of americium (Am). "In many cases the uncertainties of yield can be corrected by the
1233 use of another radioisotope as an internal standard" (Coomber, 1975, p. 198).
1234 14.6.3 Preparation of Counting Sources
1235 Electrodeposition is primarily used to prepare counting sources by depositing materials uniformly
1236 in an extremely thin layer. Because of potential self-absorption effects, this approach is ideal for
1237 the preparation of alpha sources. Numerous methods have been published for the electro-
1238 deposition of the heavy metals, e.g., the Mitchell method from hydrochloric acid (Mitchell,
1239 1960), the Talvitie method from dilute ammonium sulfate [(NH4)2SO4] (Talvitie, 1972), and the
1240 Kressin method from sodium sulfate-sodium bisulfate media (Kressin, 1977).
1241 Sill and Williams (1981) and Hindman (1983, 1986) contend that coprecipitation is the preferred
1242 method for preparation of sources for alpha spectrometry and that the it should be assessed when
1243 electrodeposition is being considered. Also see Section 16.7.2, "Coprecipitation," in this manual.
1244 14.6.4 Advantages and Disadvantages of Electrodeposition
1245 14.6.4.1 Advantages
1246 • Highly selective in some cases.
1247 • Deposits material in an extremely thin uniform layer resulting in excellent spectral resolution.
1248 • One of the common methods for preparing actinides for alpha spectrometry.
1249 14.6.4.2 Disadvantages
1250 • Not applicable to many radionuclides.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 14-47 DO NOT CITE OR QUOTE
-------
Separation Techniques
1251 • Sensitive to the presence of other substances.
1252 • For tracer-level quantities, the process is relatively slow, it seldom leads to complete
1253 separation of one element from many others, and there is usually no direct comparison of
1254 concentration in solution to deposited activity.
1255 • No further separations can be performed (see Section 16.7.2, "Coprecipitation," for methods
1256 using NdF3.).
1257 14.7 Chromatography
1258 14.7.1 Chromatographic Principles
1259 Chromatography is a separation technique that is based on the unequal distribution (partition) of
1260 substances between two immiscible phases, one moving past the other. A mixture of the
1261 substances (the analytical mixture) in the mobile phase passes over the immobile phase. Either
1262 phase can be a solid, liquid, or gas, but the alternate phase cannot be in the same physical state.
1263 The two most common phase pairs are liquid/solid and gas/liquid. Separation occurs as the
1264 components in the mixture partition between the two phases because, in a properly designed
1265 chromatographic system, the phases are chosen so that the distribution of the components
1266 between the phases is not equal.
1267 With the broad range of choices of phase materials, the number of techniques employed to
1268 establish differential distributions of components between the phases, and the various practical
1269 laboratory methods used to cause the mobile phases to pass over the immobile phases, there are
1270 many chromatographic techniques available in separation chemistry. The names of the
1271 chromatographic techniques themselves partially identify the methods or principles employed
1272 and suggest the variety of applications available using this approach to separation. They include
1273 paper chromatography, ion-exchange chromatography, adsorption chromatography, gas
1274 chromatography, high-pressure liquid chromatography, and affinity chromatography. Each aspect
1275 of chromatography used in separation chemistry will be described below, including the phases
1276 commonly employed, the principles used to establish differential distributions, and the laboratory
1277 techniques employed to run a chromatographic separation.
1278 The most common phase pairs used in chromatography are a mobile liquid phase in contact with
1279 a solid phase. The liquid phase can be a pure liquid, such as water or an organic solvent, or it can
1280 be a solution, such as methyl alcohol, sodium chloride in water, or hexane in toluene. The solid
1281 phase can be a continuous material such as paper, or a fine-grained solid such as silica, powdered
MARLAP JULY 2001
DO NOT CITE OR QUOTE 14-48 DRAFT FOR PUBLIC COMMENT
-------
Separation Techniques
1282 charcoal, or alumina. The fine-grained solid can also be applied to a supporting material, such as
1283 paper, plastic, or glass, to form a coat of continuous material. Alternatively, gas/liquid phase
1284 systems can consist of an inert gas, such as nitrogen (N2) or helium (He), in conjunction with a
1285 high-boiling point liquid polymer coated on the surface of a fine-grained inert material, such as
1286 firebrick. This system is called gas-liquid phase chromatography (GLPC), or simply gas
1287 chromatography(GC). In each system, both phases play a role in the separation by offering a
1288 physical or chemical characteristic that will result in differential distribution of the components
1289 of the analytical mixture being separated. Liquid/liquid phase systems are similar to gas/liquid
1290 phase systems in that one of the liquid phases is bound to an inert surface and remains stationary.
1291 These systems are often referred to as liquid partition chromatography or liquid-phase
1292 chromatography (LPC), because they are essentially liquid-liquid extraction systems with one
1293 mobile and one immobile phase (Section 14.4, "Solvent Extraction").
1294 Differential distributions are established between the separating phases by the combination of
1295 physical and chemical properties of the two phases in combination with those of the components
1296 of the analytical mixture. The properties that are most commonly exploited by separation
1297 chromatography are solubility, adsorption, ionic interactions, complementary interactions, and
1298 selective inclusion. One or more of these properties is acting to cause the separation to occur.
1299 14.7.2 Gas-Liquid and Liquid-Liquid Phase Chromatography
1300 In gas-liquid phase chromatography, the components of the analytical mixture are first converted
1301 to a vapor themselves and added to the flowing gas phase. They are then partitioned between the
1302 carrier gas and liquid phases primarily by solubility differences of the components in the liquid
1303 phase. As the gas/vapor mixture travels over the liquid phase, the more soluble components of
1304 the mixture spend more time in the liquid. They travel more slowly through the chromatography
1305 system and are separated from the less soluble, and therefore faster moving, components.
1306 Liquid/liquid phase chromatography provides separation based on the same principle of
1307 solubility in the two liquid phases, but the separation is performed at ambient temperatures with
1308 the components of the analytical mixture initially dissolved in the mobile phase. Partitioning
1309 occurs between the two phases as the mobile phase passes over the stationary liquid phase.
1310 Gas chromatography has been used to concentrate tritium, and to separate krypton and xenon
1311 fission products and fission-produced halogens (Coomber, 1975, p. 189). A large number of
1312 volatile metal compounds could be separated by gas chromatography, but few have been
1313 prepared. Lanthanides and trivalent actinides have been separated on glass capillary columns
1314 using volatile double halides formed with aluminum chloride (Coomber, 1975, p. 189).
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 14-49 DO NOT CITE OR QUOTE
-------
Separation Techniques
1315 14.7.3 Adsorption Chromatography
1316 Adsorption chromatography partitions components of a mixture by means of their different
1317 adsorption characteristics onto the surface of a solid phase and their different solubilities in a
1318 liquid phase. Adsorption phenomena are primarily based on intermolecular interactions between
1319 the chemical components on the surface of the solid and the individual components of the
1320 mixture. They include Van der Waals forces, dipole-dipole interactions, and hydrogen bonds.
1321 Silica is a useful adsorption medium because of the ability of its silyl OH groups to hydrogen
1322 bond or form dipole-dipole interactions with molecules in the mixture. These forces compete
1323 with similar intermolecular interactions—between the liquid phase and the components of the
1324 mixture—to produce the differential distribution of the components. This process causes
1325 separation to occur as the liquid phase passes over the solid phase.
1326 Many separations have been performed via paper and thin-layer chromatography. Modified and
1327 treated papers have been used to separate the various valence states of technetium (Coomber,
1328 1975, p. 189).
1329 14.7.4 Ion-Exchange Chromatography
1330 14.7.4.1 Principles of Ion Exchange
1331 Since the discovery by Adams and Holmes (1935) that synthetic resins can have ion-exchanging
1332 properties, ion exchange has become one of the most popular, predominant, and useful tech-
1333 niques for radiochemical separations, both with and without carriers. There are many excellent
1334 references available in the literature, e.g., Dean (1995), Dorfner (1972), Korkisch (1989), Rieman
1335 and Walton (1970), and NAS monographs (listed in References, under the author's name). The
1336 j ournal, Ion Exchange and Solvent Extraction, reports recent advances in this field of separation.
1337 Ion-exchange methods are based on the reversible exchange of metal ions between a liquid
1338 phase, typically water, and a solid ionic phase of opposite charge, the resin. The resin competes
1339 with the ion-solvent interactions in the liquid phase, primarily ion-dipole interactions and
1340 hydrogen bonding, to produce the selective partition of ions, causing separation. The solid phase
1341 consists of an insoluble, but permeable, inert polymeric matrixthat contains fixed charged groups
1342 (exchange sites) associated with mobile counter-ions of opposite charge. It is these counter-ions
1343 that are exchanged for other ions in the liquid phase. Resins are either naturally occurring sub-
1344 stances, such as zeolites (inorganic silicate polymers) or synthetic polymers. The synthetic resins
1345 are organic polymers with groups containing the exchange sites. The exchange sites are acid or
1346 base groups (amines, phenols, and carboxylic or sulfonic acids) used over a specific pH range
MARLAP JULY 2001
DO NOT CITE OR QUOTE 14-50 DRAFT FOR PUBLIC COMMENT
-------
Separation Techniques
1347 where they are in their ionic form. Typical exchange groups for cations (K+1, Ca+2, and UO2+2) are
1348 the sulfonate anion, RSO34, or the carboxylate anion, RCOO"1. The quaternary-amine cation,
1349 RNH3+1, or its derivative, is a common exchange group for anions (Cl"1, OH"1, and UO2(SO4)3"4).
1350 In a practical description of ion-exchange equilibria, the weight distribution coefficient, Kd, and
1351 the separation factor, a, are significant. The weight distribution coefficient is defined as:
1352 Kd = ((Vg^J / (C2 /mLsolution)
1353 where Cx is the weight of metal ion adsorbed on 1 g of the dry resin, and C2 is the weight of
1354 metal that remains in 1 mL of solution after equilibrium has been reached. The separation factor
1355 refers to the ratio of the distribution coefficients for two ions that were determined under
1356 identical experimental conditions:
1357 Separation factor (a) = Kd a / Kd b
1358 where a and b refer to a pair of ions. This ratio determines the separability of the two ions;
1359 separation will only be achieved if a * 1. The more that a deviates from unity, the easier it will
1360 be to obtain separation.
1361 An example of the separation process is the cation-exchange resin. It is usually prepared for
1362 separation procedures as a hydrogen salt of the exchange group. Separation occurs when an
1363 aqueous solution of another alkali-metal ion (i.e., Li+1, K+1, Rb+1, or Cs+1) comes in contact with
1364 the resin. Different ions bond selectively to the exchange group, depending on the separation
1365 conditions, displacing the counter-ion that is present in the prepared resin as follows:
1366 ResinS03-1 H+1 + Cs+1 - ResinS03-1Cs+1 + H+1
1367 Diffusion is an important process during ion exchange; the solute ions must penetrate the pores
1368 of the spherical resin beads to exchange with the existing ions. Equilibrium is established
1369 between each ion in the analyte solution and the exchange site on the resin. The ion least tightly
1370 bonded to the exchange site and most solvated in solution spends more time in solution. Selec-
1371 tive bonding is a factor of the size and charge of the ion, the nature of the exchange group, and
1372 the pH and ionic strength of the media. The order of strength of bonding at low acid concentra-
1373 tions in this example is H+1 or Li+1 < Na+1 < K+1 < Rb+1 < Cs+1 (Showsmith, 1984). Under the
1374 appropriate conditions, for example, Cs+1 will bond exclusively, or Cs+1 and Rb+1 will bond,
1375 leaving the remaining cations in solution. The process can be operated as a batch operation or via
1376 continuous-flow with the resin in an ion-exchange column. In either case, actual separation is
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 14-51 DO NOT CITE OR QUOTE
-------
Separation Techniques
1377 achieved as the equilibrated solution elutes from the resin, leaving select ions bonded to the resin
1378 and others in solution. The ion that spends more time in solution elutes first. The ability to "hold"
1379 ionic material is the resin capacity, measured in units of mg or meq per gram of resin. Eventually,
1380 most of the exchange groups are occupied by select ions. The resin is essentially saturated, and
1381 additional cations cannot bond. In a continuous-flow process, breakthrough will then occur. At
1382 this time, added quantities of select cations (Cs+1 or Cs+1 and Rb+1 in this example) will pass
1383 through the ion-exchange column and appear in the output solution (eluate). No further separa-
1384 tion can occur after breakthrough, and the bonded ions must be remove to prepare the column for
1385 additional separation. The number of bed volumes of incoming solution (eluanf) that passes
1386 through a column resin before breakthrough occurs provides one relative measure of the treat-
1387 ment capacity of the resin under the conditions of column use. The bonded cations are displaced
1388 by adjusting the pH of the medium to change the net charge on the exchange groups. This change
1389 alters the ability of the exchange groups to attract ions, thereby replacing the bonded cations with
1390 cations that bond more strongly. More commonly, the resin is treated with a more concentrated
1391 solution of the counter-ion—H+1 in this example. Excess H+1 favors the equilibrium that produces
1392 the initial counter-ion form of the exchange group. This process that returns the column to its
1393 original form is referred to as "regeneration."
1394 Overall, selectivity of the exchange resin determines the efficiency of adsorption of the analyte
1395 from solution, the ease with which the ions can be subsequently removed from the resin, and the
1396 degree to which two different ions of like charge can be separated from each other. The
1397 equilibrium distribution of ions between the resin and solution depends on many factors, of
1398 which the most important are the nature of the exchanging ions, the resin, and the solution:
1399 • In dilute solutions, the stationary phase will show preference for ions of higher charge.
1400 • The selectivity of ion exchangers for ions increases with the increase of atomic number
1401 within the same periodic group, i.e., Li+ < Na+ < K+ < Rb+ < Cs+.
1402 • The higher the polarizability and the lower the degree of solvation (favored by low charge
1403 and large size), the more strongly an ion will be adsorbed.
1404 • Resins containing weakly acidic and weakly basic groups are highly selective towards H+ and
1405 OH" ions. Ion-exchange resins that contain groups capable of complex formation with
1406 particular ions will be more selective towards those ions.
1407 • As cross-linking is increased (see discussion of resins below), resins become more selective
1408 in their behavior towards ions of different sizes.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 14-52 DRAFT FOR PUBLIC COMMENT
-------
Separation Techniques
1409 • No variation in the eluent concentration will improve the separation for ions of the same
1410 charge; however, for ions of different net charges, the separation does depend on the eluent
1411 concentration.
1412 14.7.4.2 Resins
1413 The most popular ion-exchange resins are polystyrenes cross-linked through divinylbenzene
1414 (DVB). The percentage of DVB present during polymerization controls the extent of cross-
1415 linking. Manufacturers indicate the degree of cross-linking by a number following an X, which
1416 indicates the percentage of DVB used. For instance, AG 1-X8 and AG 1-X2 are 8 percent and 2
1417 percent cross-linked resins, respectively. As this percentage is increased, the ionic groups effec-
1418 lively come into closer proximity, resulting in increased selectivity. However, increases in cross-
1419 linking decrease the diffusion rate in the resin particle. Because diffusion is the rate-controlling
1420 step in column operations, intermediate cross-linking in the range of 4 to 8 percent is commonly
1421 used.
1422 Particle diameters of 0.04-0.3 mm (50-400 mesh) are commonly used, but larger particles give
1423 higher flow rates. Difficult separations can require 200-400 mesh resins. Decreasing the particle
1424 size reduces the time required for attaining equilibrium; but at the same time, it decreases flow
1425 rate. When extremely small particle sizes are used, pressure must be applied to the system to
1426 obtain acceptable flow rates (see discussion of high pressure liquid chromatography in Section
1427 14.7.7, "Chromatographic Methods").
1428 Ion-exchange resins are used in batch operations, or more commonly, in column processes in the
1429 laboratory. Columns can be made in any size desired. The diameter of the column depends on the
1430 amount of material to be processed, and the length of the column depends primarily on the
1431 difficulty of separations to be accomplished. Generally, the ratio of column height to diameter
1432 should be 8:1. Higher ratios lead to reduced flow rate; lower ratios might not provide effective
1433 separations.
1434 Some other factors should be considered when using ion-exchange resins:
1435 • Resins should not be allowed to dry out, especially during analysis. Rehydration of dried
1436 resins will result in cracking; these resins should not be used.
1437 • Non-ionic and weakly ionic solutes may be absorbed (not exchanged) by the resin. These
1438 materials, if present during analysis, can alter the exchange characteristics of the resin for
1439 certain ions.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 14-53 DO NOT CITE OR QUOTE
-------
Separation Techniques
1440
1441
1442
1443
1444
1445
1446
1447
1448
1449
1450
1451
1452
1453
1454
1455
1456
1457
1458
1459
1460
1461
1462
1463
1464
1465
1466
1467
1468
1469
1470
• Particulate matter present in the analyte solution may be filtered by the resin. This material
will have several undesired effects, such as decreased flow rate, reduced capacity, and
ineffective separation.
• Organic solvents suspended in the analyte solution from previous separation steps can be
adsorbed by the resin creating separation problems.
Ion exchangers are classified as cationic or anionic (cation exchangers or anion exchangers,
respectively), according to their affinity for negative or positive counter-ions. They are further
subdivided into strongly or weakly ionized groups. Most cation exchangers (such as Dowex-50
and Amberlite IR-100) contain free sulfonic acid groups, whereas typical anion exchangers (such
as AG 1 and Dowex-1) have quaternary amine groups with replaceable hydroxyl ions (see Table
14.8).
TABLE 14.8 — Typical functional groups
of ion-exchange resins
Cation Exchangers
-SO3H
-COOH
-OH
-SH
Anion Exchangers
-NH2
-NHR
-NR2
-NIC
R=alkyl group
The sulfonate resins are known as strong acid cation (SAC) resins because the anion is derived
from a strong sulfonic acid (RSO3H). Likewise, the carboxylate resins are known as weak acid
cation (WAC) resins because the anion is derived from a weak carboxylic acid (RCOOH). R in
the formulas represents the inert matrix. The quaternary-amine cation (RNH3+1) or its derivatives,
represents the common exchange group for anions.
Several examples from the literature illustrate the use of ion-exchange chromatography for the
separation of radionuclides. Radium is separated from other alkaline-earth cations (Be+2, Mg+2,
Ca+2, Sr+2, and Ba+2) in hydrochloric solutions on sulfonated polystyrene resins (Kirby and
Salutsky, 1964, pp. 26-27), or converted to an anionic complex with citrate or EDTA and
separated on a quaternary ammonium polystyrene resin (Sedlet, 1966, p. 302).
Anion-exchange resins separate anions by an analogous process beginning with a prepared resin,
usually in the chloride form (RNH3+1C1"1), and adding a solution of ions. Anion-exchange
MARLAP
DO NOT CITE OR QUOTE
14-54
JULY 2001
DRAFT FOR PUBLIC COMMENT
-------
Separation Techniques
1471
1472
1473
1474
1475
1476
1477
1478
1479
1480
1481
1482
1483
1484
1485
1486
1487
1488
1489
1490
1491
1492
1493
1494
1495
1496
1497
1498
1499
1500
1501
1502
chromatography is used in one step of a procedure to isolate thorium for radioanalysis by alpha
counting (EPA, 1984, pp. U/Th-01-1-14). Thorium cations (Th+4) form anionic nitrate complexes
that bind to an anion-exchange resin containing the quaternary complex, R-CH2-N(CH3)3+1. Most
metal ion impurities do not form the complex and, as cations, they do not bind to the exchanger,
but remain with the liquid phase. Once the impurities are removed, thorium itself is separated
from the resin by treatment with hydrochloric acid (HC1) that destroys the nitrate complex,
leaving thorium in its +4 state, which will not bind to the anionic exchanger.
A selection of commercially available resins commonly employed in the radiochemistry
laboratory is given in Table 14.9.
The behavior of the elements on ani on- and cation-exchange resins is effectively summarized for
several resins in Paris and Buchanan (Paris and Buchanan, 1964), Kraus and Nelson (Kraus and
Nelson, 1956), and Nelson et al. (1964). The behavior in concentrated hydrochloric acid is
illustrated for cations on cation-exchange resins in Figure 14.3 (Dorfner, 1972, p. 208) and for
cations on anion-exchange resins in Figure 14.4 (Dorfner, 1972, p. 210).
TABLE 14.9 — Common ion-exchange resins
(i)
Resin type &
nominal %
cross-link
Minimum wet
capacity
meq» ml/1
Density
(nominal)
g» ml/1
Description
Anion-exchange resins — gel type — strongly basic — quaternary ammonium functionality
Dowex, AG or
Eichrom
1-X4
Dowex, AG or
Eichrom
1-X8
1.0
1.2
0.70
0.75
Strongly basic anion exchanger with S-DVB matrix for
separation of organic acids, nucleotides, and other anions.
Molecular weight exclusion < 1400.
Strongly basic anion exchanger with S-DVB matrix for
separation of inorganic and organic anions with molecular
weight exclusion < 1000. 100-200 mesh is standard for
analytical separations.
Anion-exchange resins — gel type — intermediate basicity
Bio-Rex 5
1.1
0.70
Intermediate basic anion exchanger with primary tertiary amines
on an polyalkylene-amine matrix for separation of organic
acids.
Anion-exchange resins — gel type — weakly basic — polyamine functionality
Dowex or AG
4-X4
Amberlite
IRA-68
0.8
1.6
0.7
1.06
Weakly basic anion exchanger with tertiary amines on an
acrylic matrix. Suitable for use with high molecular weight
organic compounds.
Acrylic -DVB with unusually high capacity for large organic
molecules.
JULY 2001
DRAFT FOR PUBLIC COMMENT
14-55
MARLAP
DO NOT CITE OR QUOTE
-------
Separation Techniques
Resin type &
nominal %
cross-link
Minimum wet
capacity
meq» ml/1
Density
(nominal)
g* ml/1
Description
Cation-exchange resins - gel type - strongly acidic - sulfonic acid functionality
Dowex, AG or
Eichrom
SOW- X4
Dowex, AG or
Eichrom
SOW- X8
Amberlite
IR-120
1.1
1.7
1.9
0.80
0.80
1.26
Strongly acidic cation exchanger with S-DVB matrix for
separation of amino acids, nucleosides and cations. Molecular
weight exclusion is < 1400.
Strongly acidic cation exchanger with S-DVB matrix for
separation of amino acids, metal cations, and cations. Molecular
weight exclusion is < 1000. 100-200 mesh is standard for
analytical applications.
8% styrene-DVB type; high physical stability.
Selective ion-exchange resins
Duolite GT-73
Amberlite
IRA-743A
Amberlite
IRC-718
Chelex® 100
Eichrom
Diphonix®
1.3
0.6
1.0
0.4
1.30
1.05
1.14
0.65
Removal of Ag, Cd, Cu, Hg, and Pb.
Boron-specific.
Removal of transition metals.
Weakly acidic chelating resin with S-DVB matrix for heavy
metal concentration.
Chelating ion-exchange resin containing geminally substituted
diphosphonic groups chemically bonded to a styrenic -based
polymer matrix. Extraordinarily strong affinity for actinides in
the tetra- and hexavalent oxidation states from highly acidic
media.
Anion exchanger — macroreticular type — strongly basic — quaternary ammonium functionality
AGMP-1
1.0
0.70
Strongly basic macroporous anion exchanger with S-DVB
matrix for separation of some enzymes, and anions of
radionuclides.
Cation-exchange resin — macroreticular type — sulfonic acid functionality
AG MP-50
1.5
0.80
Strongly acidic macroporous cation exchanger with S-DVB
matrix for separation of cations of radionuclides and other
applications.
Microcrystalline exchanger
AMP-1
4.0
Microcrystalline ammonium molybophosphate with cation
exchange capacity of 1.2 meq/g. Selectively adsorbs larger
alkali-metal ions from smaller alkali-metal ions, particularly
cesium.
1527 (1) Dowex is the trade name for Dow resins; AG and Bio-Rex are the trade names for Bio-Rad Laboratories resins;
1528 Amberlite is the trade name of Rohm & Haas resins. MP is the acronym for macroporous resin; S-DVB is the
1529 acronym for styrene-divinylbenzene.
MARLAP
DO NOT CITE OR QUOTE
14-56
JULY 2001
DRAFT FOR PUBLIC COMMENT
-------
Separation Techniques
7
-, £Q —
**S
tel
5P?
-M
c Z
J1J
0)1 6o1
. __, i.
ml 1
n
13:
fel
fel
*~>- v
ca <\j
z
M If._
fel
s:\
fel
•*r
: jfej
IM
-bA
fe)
^
-E
-=?^
^K
^
FIGURE 14.3 — The behavior of elements in concentrated
hydrochloric acid on cation-exchange resins
JULY 2001
DRAFT FOR PUBLIC COMMENT
14-57
MARLAP
DO NOT CITE OR QUOTE
-------
Separation Techniques
ads. ~ no exchange 0.1 < JW HCI
acfc. = some exchonge m 72 M HCI
, ads. — stronger exchange Cy »
fei
ax
M;
~fei
L
^i
B)
7
-e
N ?
§'
-e
Figure 14.4 —The behavior of elements in concentrated
Hydrochloric acid on anion-exchange resins
MARLAP
DO NOT CITE OR QUOTE
14-58
JULY 2001
DRAFT FOR PUBLIC COMMENT
-------
Separation Techniques
1530 14.7.5 Affinity Chromatography
1531 Several newer types of chromatography are based on highly selective and specific attractive
1532 forces that exist between groups chemically bound to an inert solid matrix (ligands) and molecu-
1533 lar or ionic components of the analytical mixture. Affinity chromatography is an example of this
1534 separation technique, which is used in biochemistry to isolate antigenic materials, such as
1535 proteins. The proteins are attracted to their specific antibody that is bonded to a solid matrix.
1536 These attractive forces are often called complementary interactions because they are based on a
1537 lock-and-key type of fit between the two constituents. The interaction is complementary because
1538 the two components match (fit) each other in size and electrical nature.
1539 Crown ethers bonded to solid matrices serve as ligands in a chromatographic separation of
1540 radium ions from aqueous solutions containing other cations (see Section 14.4.5.1, "Extraction
1541 Chromatography Columns"). Even other alkaline-earth cations with the same +2 charge, such as
1542 strontium (Sr+2) and barium (Ba+2), offer little interference with radium binding because the
1543 cyclic nature of the crown ether creates a ring structure with a cavity that complements the radius
1544 of the radium ion in solution. In addition, the oxygen atoms of the cyclic ether are inside the ring,
1545 allowing these electron-dense atoms to form effective ion-dipole interactions through water
1546 molecules with the radium cation. Radionuclides analyzed by this method include 89Sr/90Sr, "Tc,
1547 90Y, and 210Pb.
1548 14.7.6 Gel-Filtration Chromatography
1549 Another physical property that is used to separate molecules by a chromatographic procedure is
1550 the effective size (molecular weight) of the molecule. High molecular-weight ions can also be
1551 separated by this procedure. The method is known by several names, including gel-filtration
1552 chromatography., molecular-sieve filtration, exclusion chromatography, and gel-permeation
1553 chromatography. This technique is primarily limited to substances such as biomolecules with
1554 molecular weights greater than 10,000 Daltons. In similar types of solutions (similar solutes and
1555 similar concentrations), the molecules or ions have a similar shape and molecular weight that is
1556 approximately proportional to the hydrodynamic diameter (size) of the molecule or ion. The solid
1557 phase consists of a small-grain inert resin that contains microscopic pores in its matrix that will
1558 allow molecules and ions up to a certain diameter, called included particles, to enter the resin.
1559 Larger particles are excluded. Of the included particles, the smaller ones spend more time in the
1560 matrices. Separation of the molecules or ions is based on the fact that those substances that are
1561 excluded are separated in a batch from the included substances, while those that are included are
1562 separated by size. The log of the molecular weight of the included molecules or ions is
1563 approximately inversely proportional to the time the particles spend in the matrix.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 14-59 DO NOT CITE OR QUOTE
-------
Separation Techniques
1564 14.7.7 Chromatographic Laboratory Methods
1565 Chromatographic separations are achieved using a variety of laboratory techniques. Some are
1566 actually quite simple to perform, while others require sophisticated instrumentation. Paper
1567 chromatography employs a solid-liquid phase system that separates molecules and ions with
1568 filter paper or similar material in contact with a developing solvent. The analytical mixture in
1569 solution is spotted at the bottom of the paper and allowed to dry, leaving the analytes on the
1570 paper. The paper is suspended so that a small part of the bottom section is in a solvent, but not so
1571 deep that the dry spots enter the solvent. By capillary action, the solvent travels up the paper. As
1572 the solvent front moves up, the chromatogram is produced with the components of the mixture
1573 partitioning between the liquid phase and the paper. Thin-layer chromatography-is similar, but
1574 the paper is replaced by a thin solid phase of separately material (silica gel, alumina, cellulose,
1575 etc.) coated on an inert support, such as plastic or glass.
1576 Column chromatographycan accommodate a larger quantity of both phases and can, therefore,
1577 separate greater quantities of material by accepting larger loads or provide more separating power
1578 with an increased quantity of solid phase. In the procedure, a solid phase is packed in a glass or
1579 metal column and a liquid phase is passed through the column under pressure supplied by gravity
1580 or low-pressure pumping action. For this reason, gravity flow (or pumping the liquid phase under
1581 pressures similar to those generated by gravity flow) is often referred to as low-pressure
1582 chromatography. The liquid phase is usually referred to as the eluentand the column is eluted
1583 with the liquid. Column chromatography is the common method used in ion-exchange chroma-
1584 tography. With column chromatography, separation depends on: (1) type of ion-exchange resin
1585 used (i.e., cationic, anionic, strong, or weak); (2) eluting solution (its polarity affects ion
1586 solubility, ionic strength affects displacement of separating ions, and pH affects net charge of
1587 exchange groups or their degree of ionization in solution); (3) flow rate, grain size, and
1588 temperature, which affect how closely equilibrium is approached (generally, low flow rate, small
1589 grain size, and high temperature aid the approach to equilibrium and, therefore, increase the
1590 degree of separation); and (4) column dimensions (larger diameter increases column capacity,
1591 while increased length increases separation efficiency by increasing distance between ion bands
1592 as they travel through the column) (Wahl and Bonner, 1951, pp. 137-139).
1593 Metal columns can withstand considerably more pressure than glass columns. High-pressure
1594 liquid chromatography(FIPLC) employs stainless steel columns and solid phases designed to
1595 withstand high pressures without collapsing. The method is noted for its rapid separation times
1596 because of relatively high flow rates under high pressures (up to 2,000 lbs/in2). For this reason,
1597 the acronym FIPLC alternatively represents high-performance liquid chromatography. FIPLC is
1598 often performed with a liquid-partition technique between an aqueous phase and organic phase,
MARLAP JULY 2001
DO NOT CITE OR QUOTE 14-60 DRAFT FOR PUBLIC COMMENT
-------
Separation Techniques
1599 but gel filtration, ion exchange, and adsorption methods are also employed. In the case of liquid-
1600 partition separations, either a stationary aqueous phase or stationary organic phase is selected.
1601 The former system is referred to as normal phase chromatography and the latter as reversed phase
1602 chromatography, a holdover from the first applications of the technique that employed a
1603 stationary aqueous phase. The aqueous phase is made stationary by adsorption onto a solid
1604 support, commonly silica gel, cellulose powder, or polyacrylamide. An organic stationary phase
1605 is made from particles of a polymer such as polyvinyl chloride or Teflon®. Reversed phase HPLC
1606 has been used to separate individual elements of the lanthanides and actinides and
1607 macroquantities of actinides (Choppin et al., 1995, p. 248).
1608 Gas/liquid phase systems are also used. During gas-liquid phase chromatography (GLPC) [or
1609 simply, gas chromatography (GC)], the gas phase flows over the liquid phase (coated onto an
1610 inert solid) as an inert carrier gas—commonly helium (He) or nitrogen (N2)—flows through the
1611 system at low pressure. The carrier gas is supplied from a tank of the stored gas.
1612 14.7.8 Advantages and Disadvantages of Chromatographic Systems
1613 Ion-exchange chromatography is by far the predominant chromatographic method used for the
1614 separation of radionuclides. Its advantages and disadvantages is presented exclusively in this
1615 section.
1616 Advantages Disadvantages
1617 • Highly selective. • May require high volume of eluent.
1618 • Highly efficient as a preconcentration method. • Usually a relatively slow process, but rapid
1619 • Works as well with carrier-free tracer quantities selective elution processes are known.
1620 as with weighable amounts. • Requires narrow pH control.
1621 • Produces a high yield (recovery).
1622 • Can separate radionuclides from interfering
1623 counter-ions.
1624 • Simple process requiring simple equipment.
1625 • Wide scope of applications.
1626 • Can handle high volumes of sample.
1627 14.8 Precipitation and Coprecipitation
1628 14.8.1 Introduction
1629 Two of the most common and oldest methods for the separation and purification of ions in
1630 radioanalytical chemistry are precipitation and coprecipitation. Precipitation is used to isolate
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 14-61 DO NOT CITE OR QUOTE
-------
Separation Techniques
1631 and collect a specific radionuclide from other (foreign) ions in solution by forming an insoluble
1632 compound. Either the radionuclide is precipitated from solution itself, or the foreign ions are
1633 precipitated, leaving the radionuclide in solution. Sometimes a radionuclide is present in solution
1634 at sub-micro concentrations, i.e., levels so low that the radionuclide will not form an insoluble
1635 compound upon addition of a counter-ion. In these cases, the radionuclide can often be brought
1636 down from solution by coprecipitation, associating it with an insoluble substance that precipitates
1637 from solution. This phenomenon is especially important in gravimetric analysis and
1638 radiochemistry. In gravimetric analysis, carrying down of impurities is a problem. For
1639 radiochemists, coprecipitation is a valuable tool.
1640 14.8.2 Solutions
1641 Precipitation and coprecipitation provide an analytical method that is applied to ions in solution.
1642 Solutions are simply homogeneous mixtures (a physical combination of substances), which can
1643 be solids, liquids, or gases. The components of a solution consist of a solute and a solvent. The
1644 solute is generally defined as the substance that is dissolved, and the solvent is the substance that
1645 dissolves the solute. In an alternative definition, particularly suitable for liquid components when
1646 it is not clear what is being dissolved or doing the dissolving, the solute is the minor constituent
1647 and the solvent is the major constituent. In any event, the solute and solvent can consist of any
1648 combinations of substances, so long as they are soluble in each other. However, in this chapter,
1649 we are generally referring to aqueous solutions in which a solute is dissolved in water. The terms
1650 below further describe solutions:
1651 • Solubility is defined as the concentration of solute in solution that exists in equilibrium with
1652 an excess of solute; it represents the maximum amount of solute that can dissolve in a given
1653 amount of the solvent. The general solubilities of many of the major compounds of concern
1654 are described in Table 14.10.
1655 • An unsaturated solution is one in which the concentration of the solute is less than the
1656 solubility. When additional solute is added to an unsaturated solution, it dissolves.
1657 • A saturated solution is one that is in equilibrium with an excess of the solute. The
1658 concentration of a saturated solution is equal to the solubility of the solute. When solute is
1659 added to the saturated solution, no more solute dissolves.
1660 • A supersaturated solution is a solution in which the concentration of solute is temporarily
1661 greater than its solubility—an unstable condition. Therefore, when additional solute is added
MARLAP JULY 2001
DO NOT CITE OR QUOTE 14-62 DRAFT FOR PUBLIC COMMENT
-------
Separation Techniques
1662
1663
to a supersaturated solution, solute comes out of solution as solid until the concentration
decreases to that of the saturated solution.
1664
1665
1666
1667
1668
1669
1670
1671
1672
1673
1674
1675
1676
1677
1678
1679
1680
1681
1682
1683
1684
1685
1686
1687
1688
1689
1690
TABLE 14.10 — General solubility behavior of some cations of interest
(i)
The Common Cations
Na+1, K+1, NH4+1, Mg+2, Ca+2, Sr+2, Ba+2, Al+3, Cr+3, Mn+2, Fe+2, Fe+3,
Co+2, Ni+2, Cu+2, Zn+2, Ag+1, Cd+2, Sn+2, Hg2+2, Hg+2, and Pb+2
There are general rules of solubilities for the common cations found in most basic chemistry
texts (e.g., Pauling, 1970, p. 453).
Under the class of mainly soluble substances:
• All nitrates (NO3") are soluble.
• All acetates (C2H3O2") are soluble.
• All chlorides (Cl"), bromides (Br"), and iodides (I") are soluble, except for those of silver,
mercury, and lead. PbCl2 and PbBr2 are sparingly soluble in cold water, and more soluble
in hot water.
• All sulfates (SO4"2) are soluble, except those of barium, strontium, and lead. CaSO4,
Ag2SO4, and Hg2SO4 are sparingly soluble.
• Most salts of sodium (Na), potassium (K), and ammonium (NH4+) are soluble. Notable
exceptions are NaSb(OH)6, K3Co(NO2)6, K2PtCl6, (NH4)2PtCL6, and (NH4)3Co(NO2)6.
Under the class of mainly insoluble substances:
• All hydroxides (OH"1) are insoluble, except those of the alkali metals (Li, Na, K, Rb, and
Cs), ammonium, and barium (Ba). Ca(OH)2 and Sr(OH)2 are sparingly soluble.
• All normal carbonates (CO3"2) and phosphates (PO4"3) are insoluble, except those of the
alkali metals and ammonium. Many hydrogen carbonates and phosphates are soluble, i.e.,
Ca(HCO3)2, Ca(H2PO4)2.
• All sulfides (S"2), except those of the alkali metals, ammonium, and the alkaline-earth
metals (Be, Mg, Ca, Sr, Ba, and Ra), are insoluble. Both aluminum- and chromium sulfide
are hydrolyzed by water, resulting in the precipitation of A1(OH)3 and Cr(OH)3.
• Some cations, such as Ba+2, Pb+2, and Ag+1, form insoluble chromates (CrO4"2), which can
be used as a basis for separation.
JULY 2001
DRAFT FOR PUBLIC COMMENT
14-63
MARLAP
DO NOT CITE OR QUOTE
-------
Separation Techniques
1691
1692
1693
1694
1695
1696
1714
Actinide Elements
The solubility properties of the actinide M ions are similar to those of the tripositive
lanthanide ions, while the behavior of the actinide M+4 ions closely resembles that of Ce+4.
• The fluorides (F), oxalates (C2O42), hydroxides (OH), and phosphates are insoluble.
• The nitrates, halides (except fluorides), sulfates, perchlorates (ClO^, and sulfides are all
soluble.
1697 (1) Solubility data for specific compounds can be found in the CRC Handbook of Chemistry and Physics (CRC,
1698 1999) and in the NAS-NS monographs.
1699 14.8.3 Precipitation
1700 Precipitation is accomplished by combining a selected ion(s) in solution with a suitable counter-
1701 ion in sufficient concentrations to exceed the solubility of the resulting compound and produce a
1702 supersaturated solution. Nucleation occurs and growth of the crystalline substance then proceeds
1703 in an orderly manner to produce the precipitate (see Section 14.8.3.1, "Solubility and the
1704 Solubility Product Constant, Ksp"). The precipitate is collected from the solvent by a physical
1705 method, such as filtration or centrifugation. A cation (such as Sr+2, for example) will precipitate
1706 from an aqueous solution in the presence of a carbonate anion, forming the insoluble compound,
1707 strontium carbonate (SrCO3), when sufficient concentrations of each ion are present in solution
1708 to exceed the solubility of SrCO3. The method is used to isolate and collect strontium from water
1709 for radioanalysis (EPA, 1984, pp. Sr-04-1-19).
1710 A precipitation process should satisfy three main requirements:
1711 • The targeted species should be precipitated quantitatively.
1712 • The resulting precipitate should be in a form suitable for subsequent handling; it should be
1713 easily filterable and should not creep.
• If it is used as part of a quantitative scheme, the precipitate should be pure or of known purity
at the time of weighing for gravimetric analysis.
1715 at the time of weighing for gravimetric analysis.
1716 Precipitation processes are useful in several different kinds of laboratory operations, particularly
1717 gravimetric yield determinations—as a separation technique and for preconcentration—to
1718 eliminate interfering ions, or for coprecipitation.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 14-64 DRAFT FOR PUBLIC COMMENT
-------
Separation Techniques
1719 14.8.3. 1 Solubility and the Solubility Product Constant, Ksp
1720 Chemists routinely face challenges in the laboratory as a result of the phenomenon of solubility.
1721 Examples include keeping a dissolved component in solution and coprecipitating a trace-level
1722 analyte from solution.
1723 Solubility equilibrium refers to the equilibrium that describes a solid dissolving in solution, such
1724 as strontium carbonate dissolving in water, for example:
1725 SrCO3(s) - Sr+2(soln) + CO3'2(soln)
1726 or, alternately, a solid forming from solution, with the carbonate precipitating:
1727 Sr+2(soln) + CO3'2(soln) - SrCO3(s)
1728 The solubility product constant, K^ is the equilibrium constant for the former process, a solid
1729 dissolving and forming ions in solution. Leussing explains Ksp in general terms as follows:
1730 "For an electrolyte, MJVn, which dissolves and dissociates according to the equation:
173 1 MJVn(s) ^ MmNn(so\n) ^ mM+n (soln.) + nNm (soln.)
1732 "The equilibrium conditions exists that:
1733 aMmNn(s) = aMmNn(soln) = ^M+n^oln) ' ^ N-m(soln.)
1734 "[The value a is the activity of the ions in solution, a measure of the molar concentration
1735 (moles/L) of an ion in solution under ideal conditions of infinite dilution.] (Also see Section
1736 14.6.1, Principles of Electrodeposition, for a discussion of activity as applied to the Nernst
1737 equation.) [This equation] results in the familiar solubility product expression since the
1738 activity of a solid under given conditions is a constant. Expressing the activities in terms of
1739 the product of molar concentrations and activity coefficients, j [a measure of the extent the
1740 ion deviates from ideal behavior in solution; thus a = y • c where y < 1], [this] equation
1741 becomes...
1742 [AFT [tfT fM+nfN-n, = a constant = Ksp
1743 (Leussing, 1959, pp. 689-690).
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 1 4-65 DO NOT CITE OR QUOTE
-------
Separation Techniques
1744 For dilute solutions of electrolytes (< 10"2 molar), the activity coefficient is approximately one
1745 (y~ 1; it approaches one as the solution becomes more dilute, becoming one under the ideal
1746 conditions of infinite dilution). Then, the solubility product constant is expressed in terms of the
1747 concentrations of ions in solution, the typical form in which the equation is found in most
1748 chemistry textbooks:
1749 Ksp=[M+n]m [N'm]n
1750 For strontium carbonate, Ksp is defined in terms of the concentrations of Sr+2 and CO3"2:
1751 Ksp = [Sr+2][CO3-2] = 1.6 x 10-9
1752 In order for the carbonate to precipitate, the product of the concentration of the ions in solution
1753 representing the ions in the equilibrium expression, the common ions, must exceed the value of
1754 the Ksp. The concentration of each common ion does not have to be equal. For example, if [Sr+2]
1755 is 1 x 10"6 molar, then the carbonate ion concentration must be greater than 0.0016 molar for
1756 precipitation to occur because (1 x 10"6) x (.0016) = 1.6 x 10"9.
1757 At higher concentrations (> 10"2 molar), where the ions in solution deviate from ideal behavior,
1758 the value of the activity coefficient decreases, and the concentrations of the ions do not
1759 approximate their activities. Under these conditions, the concentrations do not reflect the
1760 behavior of the dissolution equilibrium, and the equation cannot be used for precipitation or
1761 solubility calculations. More complex estimations of activity coefficients must be made and
1762 applied to the general equation (Birkett et al., 1988, pp. 2.6-1 to 2.6-24). Generally, radiochemi-
1763 cal separations use an excess of a precipitating agent. The exact solution concentrations do not
1764 need to be known but they should be high to ensure compete reaction. Practical radiochemical
1765 separations performed based on solubility (either Ksp or coprecipitation phenomenon) are best
1766 described by M.L. Salutsky (1959, pp. 744-755).
1767 Analysts often need to know if a precipitate will form when two solutions are mixed. For
1768 example:
1769 "If a chemist mixes 100 mL of 0.0050 MNaCl with 200 mL of 0.020 M Pb(NO3)2, will lead
1770 chloride precipitate? The ion product, Q, must be calculated and compared to Ksp for the
1771 process:
1772 PbCl2(s) - Pb+2(soln) + 2Cr(soln)
MARLAP JULY 2001
DO NOT CITE OR QUOTE 14-66 DRAFT FOR PUBLIC COMMENT
-------
Separation Techniques
1773 After the two solutions are mixed, [Pb+2] = 1.3 x 10'2 M (0.2 L x 2.0 x 10'2 M/0.3 L), and [Cl"
1774 ] = 1.7 x 10'3 M (0.1 L x 5.0 x 10'3 M/0.3 L). The value for the ion product is calculated from
1775 the expression
1776 Q = [Pb+2][Clf or [1.3 x 1Q-2][1.7 x 10'3]2
1777 Q = 3.8xlQ-8
1778 The numerical value for Ksp is 1.6 x 10"5. Because the ion product Q is less than Ksp, no
1779 precipitate will form. Only when the ion product is greater than Ksp will a precipitate form."
1780 Conditions in the solution phase can affect solubility. For example, the solubility of an ion is
1781 lower in an aqueous solution containing a common ion, one of the ions comprising the
1782 compound, than in pure water because a precipitate will form if the Ksp is exceeded. This
1783 phenomenon is known as the common ion effect and is consistent with LeChatelier's Principle.
1784 For example, the presence of soluble sodium carbonate (Na2CO3) in solution with strontium ions
1785 can cause the precipitation of strontium carbonate, because carbonate ions from the sodium salt
1786 contribute to their overall concentration in solution and tend to reverse the solubility equilibrium
1787 of the "insoluble" strontium carbonate:
1788 Na2CO3(s) - 2 Na+1(soln) + CCV2(soln)
1789 SrCO3(s) - Sr+2(soln) + CCV2(soln)
1790 Alternatively, if a complexing agent or ligand is available that can react with the cation of a
1791 precipitate, the solubility of the compound can be markedly enhanced. An example from Section
1792 14.3.4.3, "Formation and Dissolution of Precipitates," provides an illustration of this
1793 phenomenon. In the determination of 90Sr, Sr+2 is separated from the bulk of the solution by direct
1794 precipitation of the sulfate (SrSO4). The precipitate is redissolved by forming a complex ion with
1795 EDTA, Sr(EDTA)'2, to separate it from lanthanides and actinides (DOE, 1994, Method RP520):
1796 SrSO4(s) - Sr+2(soln) + SO4'2(soln)
1797 Sr+2(soln) + EDTA'4 - Sr(EDTA)-2(soln)
1798 Additionally, many metal ions are weakly acidic and hydrolyze in solution. Hydrolysis of the
1799 ferric ion (Fe+3) a classical example of this phenomenon:
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 14-67 DO NOT CITE OR QUOTE
-------
Separation Techniques
1800 Fe+3 + H2O - Fe(OH)+2 + H+1
1801 When these metal ions hydrolyze, producing a less soluble complex, the solubility of the salt is a
1802 function of the pH of the solution, increasing as the pH decreases. The minimum solubility is
1803 found under acidic conditions when the concentrations of the hydrolyzed species become
1804 negligible. As demonstrated by Leussing, the solubility of a salt also depends upon the activity of
1805 the solid phase. There are a number of factors that affect the activity of the solid phase (Leussing,
1806 1959, pp. 690-692):
1807 • Polymorphism is the existence of a chemical substance in two or more crystalline forms. For
1808 example, calcium carbonate can have several different forms; only one form of a crystal is
1809 stable at a given temperature. At ordinary pressures and temperatures, calcite with a solubility
1810 of 0.028 g/L, is the stable form. Aragonite, another common form of calcium carbonate
1811 (CaCO3), has a solubility of 0.041g/L at these conditions. It is not necessarily calcite that
1812 precipitates when solutions of sodium carbonate and calcium nitrate are mixed. Extremely
1813 low concentrations of large cations, such as strontium, barium, or lead, promote the
1814 precipitation of aragonite over calcite (Wray and Daniels, 1957). On aging, the more soluble
1815 aragonite converts to calcite.
1816 • Various possible hydrates of a solid have different solubilities. For instance, at 25 °C, the
1817 molar solubility of gypsum (CaSO42H2O) is 0.206 and that of anhydrite (CaSO4) is 0.271.
1818 • The solid phase can undergo a reaction with a salt in solution.
1819 • Particle size of a solid can affect its solubility, because it has been demonstrated that the
1820 solubility of smaller particles is greater than that of larger particles.
1821 • Age of a precipitate can affect solubility. For example, Biederman and Schindler (1957) have
1822 demonstrated that the solubility of precipitated ferric hydroxide [Fe(OH)3] undergoes a four-
1823 fold decrease to a steady state after 200 hours.
1824 • Exchange of ions at the surface of the crystal with ions in the solution can affect the solubility
1825 of a solid. This effect is a function of the amount of surface available for exchange and is,
1826 therefore, greater for a finely divided solid. For example, Kolthoff and Sandell (1933)
1827 observed that calcium oxalate (CaC2O4) can exchange with either sulfate or barium ions:
1828 CaC2O4(s) + S
-------
Separation Techniques
1829 CaC2O4(s) + Ba+2(soln) - BaC2O4(s) + Ca+2(soln)
1830 The excess of common ions that appears on the right-hand side of the equations represses
1831 the solubility of calcium oxalate according to the laws of mass action.
1832 Ideally, separation of common ions from foreign ions in solution by precipitation will result in a
1833 pure solid that is easy to filter. This method should ensure the production of a precipitate to meet
1834 these criteria as closely as possible. The physical process of the formation of a precipitate is quite
1835 complex, and involves both nucleation and crystal growth. Nucleation is the formation within a
1836 supersaturated solution of the smallest particles of a precipitate (nuclei) capable of spontaneous
1837 growth. The importance of nucleation is summarized by Salutsky (1959, p. 734):
1838 "The nucleation processes govern the nature and purity of the resulting precipitates. If the
1839 precipitation is carried out in such a manner as to produce numerous nuclei, precipitation will
1840 be rapid, individual crystals will be small, filtration and washing difficult, and purity low. On
1841 the other hand, if precipitation is carried out so that only a few nuclei are formed, precipita-
1842 tion will be slower, crystals larger, filtration easier, and purity higher. Hence, control of
1843 nucleation processes is of considerable significance in analytical chemistry."
1844 Once the crystal nuclei are formed, crystal growth proceeds through diffusion of the ions to the
1845 surface of the growing crystal and deposition of those ions on the surface. This crystal growth
1846 continues until supersaturation of the precipitating material is eliminated and equilibrium
1847 solubility is attained.
1848 Thus, the goal is to produce fewer nuclei during precipitation so that the process will occur
1849 slowly, within reasonable limits, and larger crystals will be formed. Impurities result from three
1850 mechanisms: (1) inclusion, either by isomorphous replacement (isomorphic inclusion),
1851 replacement of a common ion in the crystal structure by foreign ions of similar size and charge to
1852 form a mixed crystal, or by solid solution formation (nonisomorphic inclusion), simultaneous
1853 crystallization of two or more solids mixed together; (2) surface absorption of foreign ions; and
1854 (3) occlusion, the subsequent entrapment of adsorbed ions as the crystal grows. Slow growth
1855 gives the isomorphous ion time to be replaced by a common ion that fits the crystal structure
1856 perfectly, producing a more stable crystal. It also promotes establishment of equilibrium
1857 conditions for the formation of the crystal structure so that adsorbed impurities are more likely to
1858 desorb and be replaced by a common ion rather than becoming entrapped. In addition, for a given
1859 weight of the solid that is forming, a small number of large crystals present an overall smaller
1860 surface area than a large number of small crystals. The large crystals provide less surface area for
1861 impurities to adsorb.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 14-69 DO NOT CITE OR QUOTE
-------
Separation Techniques
1862 14.8.3.2 Factors Affecting Precipitation
1863 Several factors affect the nature and purity of the crystals formed during precipitation. A
1864 knowledge of these factors permits the selection and application of laboratory procedures that
1865 increase the effectiveness of precipitation as a technique for the separation and purification of
1866 ions, and for the formation of precipitates that are easily isolated. These factors, summarized
1867 from Berg (1963, pp. 251-284) and Salutsky (1959, pp. 736-742), include the following:
1868 • Rate of precipitation. Formation of large, well-shaped crystals is encouraged through slow
1869 precipitation because fewer nuclei form and they have time to grow into larger crystals to the
1870 detriment of smaller crystals present. Solubility of the larger crystals is less than that of
1871 smaller crystals because smaller crystals expose more surface area to the solution. Larger
1872 crystals also provide less surface area for the absorption of foreign ions. Slow precipitation
1873 can be accomplished by adding a very dilute solution of the precipitant gradually, with
1874 stirring, to a medium in which the resulting precipitate initially has a moderate solubility.
1875 • Concentration of Ions and Solubility of Solids. The rate of precipitation depends on the
1876 concentration of ions in solution and the solubility of the solids formed during the
1877 equilibrium process. A solution containing a low concentration of ions, but sufficient
1878 concentration to form a precipitate, will slow the process, resulting in larger crystal
1879 formation. At the same time, increasing the solubility of the solid, either by selecting the
1880 counter-ion for precipitation or by altering the precipitating conditions, will also slow
1881 precipitation. Many radionuclides form insoluble solids with a variety of ions, and the choice
1882 of precipitating agent will affect the solubility of the precipitate. For example, radium sulfate
1883 (RaSO4) is the most insoluble radium compound known. Radium carbonate (RaCO3) is also
1884 insoluble, but its Ksp is greater than that of radium sulfate (Kirby and Salutsky, 1964, p. 9).
1885 • Temperature. Precipitation at higher temperature slows nucleation and crystal growth
1886 because of the increased thermal motion of the particles in solution. Therefore, larger crystals
1887 form, reducing the amount of adsorption and occlusion. However, most solids are more
1888 soluble at elevated temperatures, effectively reducing precipitate yield; an optimum
1889 temperature balances these opposing factors.
1890 • Digestion. Extremely small particles, with a radius on the order of one micron, are more
1891 soluble than larger particles because of their larger surface area compared to their volume
1892 (weight). Therefore, when a precipitate is heated over time (digestion) the small crystals
1893 dissolve and larger crystals grow (Ostwaldripening). Effectively, the small crystals are
1894 recrystallized, allowing the escape of impurities (occluded ions) and growth of larger crystals.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 14-70 DRAFT FOR PUBLIC COMMENT
-------
Separation Techniques
1895 This process reduces the surface area for adsorption of foreign ions and, at the same time,
1896 replaces the impurities with common ions that properly "fit" the crystal lattice. Recrystal-
1897 lization perfects the crystal lattice, producing a purer precipitate (see Reprecipitation below).
1898 Digestion is used in an 131I determination to increase the purity of the lead iodide (PbI2)
1899 crystals (EPA, 1984, pp. 1-01-1-9).
1900 • Degree of Supersaturation. A relatively high degree of supersaturation is required for
1901 spontaneous nucleation, and degree of supersaturation is the main factor in determining the
1902 physical character of a precipitate. Generally, the higher the supersaturation required, the
1903 more likely a curdy, flocculated colloid will precipitate because more nuclei form under
1904 conditions of higher supersaturation and crystal growth is faster. In contrast, the lower the
1905 supersaturation required, the more likely a crystalline precipitate will form because fewer
1906 nuclei form under these conditions and crystal growth is slower. Most perfect crystals are
1907 formed, therefore, from supersaturated solutions that require lower ion concentrations to
1908 reach the necessary degree of supersaturation and, as a result, inhibit the rate of nucleation
1909 and crystal growth. Degree of supersaturation ultimately depends on physical properties of
1910 the solid that affect its formation. Choice of counter-ion will determine the type of solid
1911 formed from a radionuclide, which, in turn, determines the degree of saturation required for
1912 precipitation. Many radionuclides form insoluble solids with a variety of ions, and the choice
1913 of precipitating agent will affect the nature of the precipitate.
1914 • Solvent. The nature of the solvent affects the solubility of an ionic solid (precipitate) in the
1915 solvent. The polarity of water can be reduced by the addition of other miscible solvents such
1916 as alcohols, thereby reducing the solubility of precipitates. Strontium chromate (SrCrO4) is
1917 soluble in water, but it is insoluble in a methyl alcohol (CH3OH)-water mixture and can be
1918 effectively precipitated from the solution (Berg, 1963, p. 364). In some procedures,
1919 precipitation is achieved by adding alcohol to an aqueous solution, but the dilution effect
1920 might reduce the yield because it lowers the concentration of ions in solution.
1921 • Ion Concentration. The common-ion effect causes precipitation to occur when the
1922 concentration of ions exceeds the solubility-product constant. In some cases, however, excess
1923 presence of common ions increases the solubility of the precipitate by decreasing the activity
1924 of the ions in solution, as they become more concentrated in solution and deviate from ideal
1925 behavior. An increase in concentration of the ions is necessary to reach the activity of ions
1926 necessary for precipitate formation.
1927 • Stirring. Stirring the solution during precipitation increases the motion of particles in solution
1928 and decreases the localized buildup of concentration of ions by keeping the solution
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 14-71 DO NOT CITE OR QUOTE
-------
Separation Techniques
1929 thoroughly mixed. Both of these properties slow nucleation and crystal growth, thus
1930 promoting larger and purer crystals. This approach also promotes recrystallization because
1931 the smaller crystals, with their net larger surface area, are more soluble under these
1932 conditions. Virtually all radiochemical laboratories employ stirring with a magnetic stirrer
1933 during precipitation reactions.
1934 • Complex-Ion Formation. Formation of complex ions can be used to hold back impurities
1935 from precipitating by producing a more soluble form of a solid. The classical example of this
1936 phenomenon is the precipitation of lead (Pb+2) in the presence of silver ions (Ag+1). Chloride
1937 ion (Cl"1) is the precipitating agent that produces insoluble lead chloride (PbCl2). In an excess
1938 of the agent, silver chloride (AgCl) is not formed because a soluble salt containing the
1939 complex ion, AgCl^1 is formed. Complex -ion formation is also used to form precipitates (see
1940 Section 14.3, "Complexation").
1941 • pH Effect. Altering the pH of aqueous solutions will alter the concentration of ions in the
1942 precipitation equilibrium by the common-ion effect, if the hydrogen ion (H+1) or hydroxide
1943 ion (OH"1) is common to the equilibrium. For example, calcium oxalate (CaC2O4) can be
1944 precipitated or dissolved, depending on the pH of the solution, as follows:
1945 Ca+2 + C2
-------
Separation Techniques
1957 HS'1 - H+
1958 Pb+2 +S'2 - PbS
1959 The pH can also influence selective formation of precipitates. Barium chromate will
I960 precipitate in the presence of strontium at pH 4 to 8, leaving strontium in solution. Sodium
1961 carbonate is added and strontium precipitates after ammonia (NH3) is added to make the
1962 solution more alkaline. This procedure is the basis for the separation of radium from
1963 strontium in the radioanalysis of strontium in drinking water (EPA, 1980, p. 63).
1964 • Precipitation from Homogeneous Solution. Addition of a precipitating agent to a solution of
1965 ions causes a localized excess of the reagent (higher concentrations) to form in the mixture.
1966 The excess reagent is conducive to rapid formation of a large number of small crystals,
1967 producing a precipitate of imperfect crystals that contains excessive impurities. The
1968 precipitate formed under these conditions is sometimes voluminous and difficult to filter.
1969 Localized excesses can also cause precipitation of more soluble solids than the expected
1970 precipitate.
1971 These problems largely can be avoided if the solution is homogenous in all stages of
1972 precipitate formation, and if the concentration of precipitating agent is increased, as slowly as
1973 practical, to cause precipitation from the most dilute solution possible. This increase in
1974 concentration is accomplished, not by adding the precipitating agent directly to the solution,
1975 but rather by generating the agent throughout the solution, starting with a very small
1976 concentration and slowly increasing the concentration while stirring. The precipitating agent
1977 is generated indirectly as the result of a chemical change of a reagent that produces the
1978 precipitating agent internally and homogeneously throughout the solution. The degree of
1979 supersaturation is low because the concentration of precipitating agent in solution is always
1980 uniformly low enough for nucleation only. This method produces larger crystals with fewer
1981 impurities.
1982 Table 14.11 (Salutsky, 1959, p. 741) summarizes methods used for precipitate formation
1983 from homogeneous solution. Descriptions of these methods can be found in Gordon et al.
1984 (1959).
1985 Some agents are generated by decomposition of a compound in solution. Hydrogen sulfide,
1986 for example, is produced from thioacetamide:
1987 CH3CSNH2 + 2 H2O - CH3COO4 + H2S + NH4+1
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 14-73 DO NOT CITE OR QUOTE
-------
Separation Techniques
1988
1989
1990
1991
Copper sulfide (CuS) coprecipitates technetium from a homogeneous medium by the
generation of hydrogen sulfide by this method (EPA, 1973, pp. 67-72). Other agents alter the
pH of the solution (see "pH Effect" above). Hydrolysis of urea, for example, produces
ammonia, which raises the pH of a solution:
1992
1993
1994
1995
1996
1997
1998
1999
H2NCONH2 +H2O
CO2 + 2 NH3
2000
2001
2002
2003
2004
2005
2006
Precipitant
Hydroxide
Phosphate
Oxalate
Sulfate
Sulfide
lodate
Carbonate
Chromate
Periodate
Chloride
Arsenate
TABLE 14.11 — Summary of methods for utilizing precipitation
from homogeneous solution(1)
Reagent
Element Precipitated
Urea
Acetamide
Hexamethylenetetraamine
Metal Chelate and H2O2
Triethyl Phosphate
Trimethyl Phosphate
Metaphosphoric Acid
Urea
Dimethyl Oxalate
Diethyl Oxalate
Urea and an Oxalate
Dimethyl Sulfate
Sulfamic Acid
Potassium Methyl Sulfate
Ammonium Persulfate
Metal Chelate and Persulfate
Thiocetamide
Iodine and Chlorate
Periodate and Ethylene Diacetate
(or B-Hydroxy Acetate)
Ce(III) and Bromate
Trichloroacetate
Urea and Dichromate
Potassium Cyanate and Dichromate
Cr(III) and Bromate
Acetamide
Silver Ammonia Complex
and B-Hydroxyethyl Acetate
Arsenite and Nitrite
Al, Ga, Th, Fe(III), Sn, and Zr
Ti
Th
Fe(III)
ZrandHf
Zr
Zr
Mg
Th, Ca, Am, Ac, and Rare Earths
Mg, Zn, and Ca
Ca
Ba, Ca, Sr, and Pb
Ba, Pb, and Ra
Ba, Pb, and Ra
Ba
Ba
Pb, Sb, Bi, Mo, Cu, and As, Cd, Sn, Hg,
andMn
Th and Zr
ThandFe(III)
Ce(IV)
Rare Earths, Ba, and Ra
Ba and Ra
Ba, Ra
Pb
Pb
Ag
Zr
MARLAP
DO NOT CITE OR QUOTE
14-74
JULY 2001
DRAFT FOR PUBLIC COMMENT
-------
Separation Techniques
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
2018
2019
2020
2021
2022
2023
Precipitant
Tetrachlorophthalate
Dimethylgloxime
8-Hydroxyquinoline
Fluoride
Reagent
Tetrachlorophthalic Acid
Urea and Metal Chelate
Urea and Metal Chelate
Fluoroboric Acid
Element Precipitated
Th
Ni
Al
La
(1) Salutsky, 1959, p. 741
• Reprecipitation. This approach increases the purity of precipitates. During the initial
precipitation, crystals collected contain only a small amount of foreign ions relative to the
common ions of the crystal. When the precipitate is redissolved in pure solvent, the foreign
ions are released into solution, producing a concentration of impurities much lower than that
in the original precipitating solution. On reprecipitation, a small fraction of impurities is
carried down with the precipitate, but the relative amount is much less than the original
because their concentration in solution is less. Nevertheless, foreign ions are not eliminated
because absorption is greater at lower, rather than at higher, concentrations. On balance,
reprecipitation increases the purity of the crystals. Reprecipitation is used in the procedure to
determine americium (Am) in soil (DOE, 1990 and 1997, Method Am-01). After americium
is coprecipitated with calcium oxalate (CaC2O4), the precipitate is reprecipitated to purify the
solid.
2024 14.8.3.3 Optimum Precipitation Conditions
2025 There is no single, fixed rule to eliminate all impurities during precipitation (as discussed in the
2026 section above), but over the years, a number of conditions have been identified from practical
2027 experience and theoretical considerations that limit these impurities (Table 14.12). Precipitations
2028 are generally carried out from dilute solutions adding the precipitant slowly with some form of
2029 agitation to a hot solution. Normally, the precipitant is then allowed to age before it is removed
2030 by filtration and washed. Reprecipitation is then commonly performed. Reprecipitation is one of
2031 the most powerful techniques available to the analyst because it increases purity, regardless of the
2032 form of the impurity.
2033 Table 14.12 highlights the optimum precipitation conditions to eliminate impurities.
JULY 2001
DRAFT FOR PUBLIC COMMENT
14-75
MARLAP
DO NOT CITE OR QUOTE
-------
Separation Techniques
2034
2035
2036
2037
2038
2039
2040
2041
2042
2043
2044
2045
2046
2047
2048
2049
2050
2051
2052
2053
2054
2055
2056
2057
2058
2059
2060
2061
2062
2063
2064
2065
2066
TABLE 14.12 — Influence of precipitation conditions on the purity of precipitates
0,2)
Condition
Dilute solutions
Slow precipitation
Prolonged digestion
High temperature
Agitation
Washing the precipitate
Reprecipitation
Form of Impurity
Mixed
Crystals
o
+
-
-
+
o
+
Surface
Adsorption
+
+
+
+
+
+
+
Occlusion and
Inclusion
+
+
+
+
+
o
+
Post-
precipitation
o
-
-
-
o
o
0
(1) +, increased purity; -, decreased purity; o, little or no change in purity
(2) Salutsky, 1959, p. 764
14.8.4 Coprecipitation
In many solutions, especially those of environmental samples, the concentration of the
radionuclide of interest is too low to cause precipitation, even in the presence of high
concentrations of its counter-ion, because the product of the concentrations does not exceed the
solubility product. Radium in most environmental samples, for example, is not present in
sufficient concentration to cause its very insoluble sulfate (RaSO4) to precipitate. The
radionuclide can often be brought down selectively and quantitatively from solution during
precipitation of an alternate insoluble compound by a process called coprecipitation. The
insoluble compound commonly used to coprecipitate radium isotopes in many radioanalytical
procedures is another insoluble sulfate, barium sulfate (BaSO4) (EPA, 1984, Method Ra-01;
EPA, 1980, Method 900.1). The salt is formed with barium, also a member of the alkaline earth
family of elements with chemical properties very similar to those of radium. Alternatively, a
different salt that is soluble for the radionuclide can be used to cause coprecipitation. Radium can
be coprecipitated with lanthanum fluoride, even though radium fluoride is soluble itself. For
trace amounts of some radionuclides, other isotopic forms of the element are available that can
be added to the solution to bring the total concentration of all forms of the element to the level
that will result in precipitation. For trace quantities of 90Sr, inactive strontium (85Sr), which will
not interfere with the radioanalysis of 90Sr, is added to permit the precipitation of strontium
carbonate in the presence of carbonate ions. The added ion that is present in sufficient
concentration to cause a precipitate to form is called a carrier (Section 14.9, "Carriers and
Tracers"). Barium, lanthanum, and stable strontium, respectively, are carriers in these examples
(DOE, 1995, Method RP5001; DOE, 1990 and 1997, Method Sr-02; EPA, 1984, Sr-04). The
MARLAP
DO NOT CITE OR QUOTE
14-76
JULY 2001
DRAFT FOR PUBLIC COMMENT
-------
Separation Techniques
2067 term carrier is also used to designate the insoluble compound that causes coprecipitation. Barium
2068 sulfate, lanthanum fluoride (LaF3), and strontium carbonate are sometimes referred to as the
2069 carrier in these coprecipitation procedures. See Wahl and Bonner (1951, p. 403) for additional
2070 examples of tracers and their carriers used for coprecipitation.
2071 The common definition of coprecipitation is, "the contamination of a precipitate by substances
2072 that are normally soluble under the conditions of precipitation" (Salutsky, 1959, p. 748). In a very
2073 broad sense, coprecipitation is alternately defined as the precipitation of one compound
2074 simultaneously with one or more other compounds to form mixed crystals (Berg, 1963, p. 296).
2075 Each is present in macro concentrations (i.e., sufficient concentrations to exceed the solubility
2076 product of each). As the term is used in radiochemistry, coprecipitation is the simultaneous
2077 precipitation of one compound that is normally soluble under the conditions of precipitation with
2078 one or more other compounds that form a precipitate under the same conditions. Coprecipitation
2079 of two or more rare earths as oxalates, barium and radium as sulfates, or zirconium (Zr) and
2080 hafnium (Hf) as phosphates are examples of this broader definition (Salutsky, 1959, p. 748). By
2081 either definition, coprecipitation introduces foreign ions into a precipitate as impurities that
2082 would normally be expected to remain in solution; and precipitation techniques, described in the
2083 previous section, are normally used to maximize this effect while minimizing the introduction of
2084 true impurities. As a method to separate and collect radionuclides present in solution at very low
2085 concentration, coprecipitation is performed in a controlled process to associate the ion of choice
2086 selectively with a precipitate, while excluding other foreign ions that would interfere with the
2087 analytical procedure.
2088 14.8.4.1 Coprecipitation Processes
2089 In order to choose the best conditions to coprecipitate an ion selectively, two processes should be
2090 considered. First is precipitation itself and the appropriate techniques employed to minimize
2091 association of impurities (see Section 14.8.3). Second is coprecipitation mechanisms and the
2092 controlling factors associated with each. Three processes (described above in Section 14.8.3.1,
2093 "Solubility and the Solubility Product Constant") are responsible for coprecipitation, although
2094 the distinction between these processes is not always clear (Hermann and Suttle, 1961, p. 1369).
2095 They consist of: (1) inclusion, i.e., uptake from solution of an ion similar in size and charge to
2096 the solid forming the precipitate in order to form a mixed crystal or solid solution; (2) surface
2097 adsorption; and (3) occlusion (mechanical entrapment).
2098 Inclusion. If coprecipitation is accomplished from a homogeneous solution allowing the crystals
2099 to form slowly in an orderly manner, then inclusion contributes to the coprecipitation process.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 14-77 DO NOT CITE OR QUOTE
-------
Separation Techniques
2100 Under these conditions, the logarithmic distribution law applies, which represents the most
2101 efficient coprecipitation method that involves mixed crystals (Salutsky, 1959, p. 750):
2102 log(VIf) = A log(P;/Pf)
2103 In the equation, \ is the concentration of impurity in solution at the start of crystallization and If
2104 is the concentration at the end. P represents the corresponding concentration of the primary ion in
2105 solution. Lambda, A, is the logarithmic distribution coefficient and is a constant. Values of A for
2106 some tracers distributed in solid carriers can be found in Wahl and Bonner (1951, p. 393).
2107 Lambda values greater than one represent removal of a foreign ion by inclusion during
2108 coprecipitation. The larger the value of lambda, the more effective and selective for a specific ion
2109 the process is. Lambda is also inversely proportional to the rate of precipitation. Slow
2110 precipitation, as accomplished by homogeneous precipitation, results in larger values and more
2111 efficient coprecipitation. For example, "Actinium [Ac] has been selectively removed from
2112 solutions containing iron and aluminum [Al] through slow oxalate precipitation by the controlled
2113 hydrolysis of dimethyl oxalate" (Hermann and Suttle, 1961, p. 1376). Also, as described in
2114 Section 14.8.3.2, "Factors Affecting Precipitation," technetium is coprecipitated with copper
2115 sulfide (CuS) carrier produced by the slow generation of hydrogen sulfide (H2S) as thioacetamide
2116 is hydrolyzed in water (EPA, 1973, pp. 67-72).
2117 Generally, A decreases as the temperature increases; thus, coprecipitation by inclusion is favored
2118 by lower temperature.
2119 Digestion of the precipitate at elevated temperature over lengthy time periods—a process that
2120 promotes recrystallization and purer crystals—will often cause mixed crystals to form by an
2121 alternate mechanism (i.e., homogeneous distribution) that is not as efficient, but which is often as
2122 successful as logarithmic distribution. The equilibrium distribution lawis represented by
2123 (Salutsky, 1959, p. 749):
2124 (I/P)ppt = D(I/P)soln.
2125 where I represents the amount of impurity and P the amount of primary substance forming the
2126 precipitate. The symbol .Dis the homogeneous distribution coefficient. Values of D greater than
2127 one represent removal of a foreign ion by inclusion during coprecipitation. Some values of D can
2128 be found in Wahl and Bonner (1951, p. 393). According to Hermann and Suttle:
2129 "Homogeneous distribution is conveniently obtained at ordinary temperatures by rapid
2130 crystallization from supersaturated solutions with vigorous stirring. Under such conditions
MARLAP JULY 2001
DO NOT CITE OR QUOTE 14-78 DRAFT FOR PUBLIC COMMENT
-------
Separation Techniques
2131 the precipitate first formed is very finely divided, the recrystallization of the minute crystals
2132 is rapid, and each molecule (sic) passes many times between solution and precipitate. If this
2133 process is repeated often enough, an equilibrium between solid and solution is obtained, and
2134 all the resulting crystals grow from a solution of constant composition" (Hermann and Suttle,
2135 1961, pp. 1473-1474).
2136 In either case, optimal results are obtained through inclusion when the precipitate contains an ion
2137 with chemical properties similar to those of the foreign ion, although it is not necessary for the
2138 similarity to exist in every successful coprecipitation. Barium sulfate is very successful in
2139 coprecipitating Ra+2, primarily because radium is in the same chemical family as barium, and has
2140 the same charge and a similar ionic radius. For best results, the radius of the foreign ion should
2141 be within approximately 15 percent of that of one of the common ions in the precipitate
2142 (Hermann and Suttle, 1961, p. 1479).
2143 Surface Adsorption. During surface adsorption, ions are adsorbed from solution onto the surfaces
2144 of precipitated particles. The conditions leading to surface adsorption are described by Salutsky:
2145 "The surface of a precipitate is particularly active. Ions at the surface of a crystal (unlike
2146 those within the crystal) are incompletely coordinated and, hence are free to attract other ions
2147 of opposite charge from solution" (Salutsky, 1959, p. 754).
2148 Adsorption involves a primary adsorption layer that is held very tightly, and a counter-ion layer
2149 held more loosely. Ions common to the precipitate are adsorbed most strongly at the surface to
2150 continue growth of the crystal. During precipitation of BaSO4, barium ions (Ba+2) and sulfate ions
2151 (SO4"2) are the primary ions adsorbed. If only one of the common ions remains in solution, then
2152 foreign ions of the opposite charge are adsorbed to maintain electrical neutrality. When barium
2153 sulfate is precipitated from a solution containing excess barium ions, for example, foreign ions
2154 such as Cl"1, if present, are adsorbed after sulfate ions are depleted in the precipitation process.
2155 Foreign ions of the same charge, such as Na+1, are repelled from the surface. Surface adsorption
2156 can be controlled, therefore, by controlling the concentration of ions during precipitation or by
2157 the addition of ions to alter the concentration. A precipitate of silver chloride (AgCl) in excess
2158 Ag+1 repels 212Pb+2, but in a solution containing an equal quantity of the common silver and
2159 chloride ions, approximately 2 percent of 212Pb is adsorbed (Salutsky, 1959, pp. 754-755). In
2160 contrast, almost 86 percent of 212Pb is adsorbed if an iodide solution is added to precipitate the
2161 silver ions as silver iodide (Agl), thereby reducing the concentration of silver ions and making
2162 the chloride ion in excess in the solution. According to the Paneth-Fajans-Hahn adsorption rule,
2163 the ion most adsorbed will be the one that forms the least soluble compound with an ion of the
2164 precipitate. For example, barium sulfate in contact with a solution containing excess sulfate ions
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 14-79 DO NOT CITE OR QUOTE
-------
Separation Techniques
2165 will adsorb ions of Pb > Ca > K > Na, which reflects the order of solubility of the respective
2166 sulfates: thus, PbSO4 < CaSO4 < K2SO4 < Na2SO4 (Salutsky, 1959, p. 755).
2167 "Because adsorption is a surface phenomenon, the larger the surface area of a precipitate, the
2168 greater the adsorption of impurities" (Salutsky, 1959, p. 755). For that reason, colloidal crystals
2169 exhibit a high degree of nonspecific adsorption. When a colloid is flocculated by the addition of
2170 an electrolyte, the electrolyte can be adsorbed as an impurity. This interference largely can be
2171 eliminated by aging the precipitate, thereby growing larger crystals and reducing the surface area.
2172 Additionally, nonvolatile impurities can be replaced on the particle by washing the colloidal
2173 precipitate with a dilute acid or ammonium salt solution. Well-formed large crystals exhibit
2174 much less adsorption, and adsorption is not a significant factor in coprecipitation with these
2175 solids. The tendency for a particular ion to be adsorbed depends on, among other factors, charge
2176 and ionic size (Berg, 1963, p. 299). Large ions with a high charge exhibit high adsorption
2177 characteristics: a high ionic charge increases the electrostatic attraction to the charged surface,
2178 and an ion with a large radius is less hydrated by the solution and not as attracted to the solution
2179 phase.
2180 "The amount of adsorption is also affected by prolonged standing of the precipitate in contact
2181 with the solution. The fraction adsorbed is higher for some tracer ions, while the fraction is lower
2182 for others. Recrystallization occurring during standing decreases the surface area so that the
2183 fraction of tracer carried will decrease unless the tracer is trapped in the growing crystals ... in
2184 which case the fraction carried may increase." (Wahl, 1951, p. 117).
2185 Adsorption also depends on the concentration of an ion in solution (Berg, 1963, p. 299). A high
2186 concentration of impurity increases the probability of solute interaction at the solid surface and
2187 favors adsorption. Salutsky comments on the percent adsorption:
2188 "Generally, the percent adsorption is much greater at low concentrations than at high
2189 concentrations. At very high concentrations of impurity, adsorption reaches a maximum
2190 value, i.e., the adsorption is saturated" (Salutsky, 1959, pp. 755-756).
2191 Occlusion. Occlusion of an impurity within a precipitate results when the impurity is trapped
2192 mechanically by subsequent crystal layers. For that reason, occluded impurities cannot be
2193 physically removed by washing. Occlusion is more prevalent with colloidal precipitates than with
2194 large crystals because of the greater surface area of colloidal solids. Freshly prepared hydroxides
2195 and sulfides commonly contain occluded impurities, but most of them are released upon aging of
2196 the precipitate.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 14-80 DRAFT FOR PUBLIC COMMENT
-------
Separation Techniques
2197 Mechanical entrapment occurs particularly when the precipitating agent is added directly to a
2198 solution. Because of the localized high concentrations of precipitant, impurities are precipitated
2199 that become occluded by the subsequent precipitation of the primary substance. The speed of the
2200 precipitation process also affects the extent of occlusion. Occlusion can be reduced, therefore, by
2201 homogeneous precipitation. Coprecipitation of strontium by barium sulfate, for example, is
2202 accomplished by the homogeneous generation of sulfate by the hydrolysis of dimethyl sulfate,
2203 (CH3)2SO4 (Hermann and Suttle, 1961, p. 1480). Digestion also eliminates occluded particles as
2204 the solid is recrystallized. Considerable occlusion occurs during nucleation, and, therefore,
2205 reducing the precipitation rate by lowering the temperature and reducing the number of nuclei
2206 formed reduces the initial coprecipitation by occlusion.
2207 This type of coprecipitation is not limited to solid impurities. Sometimes the solvent and other
2208 impurities dissolved in the solvent become trapped between layers of crystals. This liquid
2209 occlusion is common in numbers of minerals such as quartz and gypsum.
2210 14.8.4.2 Water as an Impurity
2211 In addition to other impurities, all precipitates formed from aqueous solutions contain water
2212 (Salutsky, 1959, pp. 761-763). This water might be essential water, present as an essential part of
2213 the chemical composition (e.g., MgNH4PO4 • 6H2O, Na2CO3 • H2O), or it might be nonessential
2214 water. Nonessential water can be present in the precipitate as hygroscopic water, surface water,
2215 or included water. Hygroscopic w/aterrefers to the water that a solid adsorbs from the surroun-
2216 ding atmosphere. Many colloidal precipitates are highly hygroscopic because of their large
2217 surface areas. Moreover, water can be adsorbed to the surface of the precipitate or included
2218 within the crystal matrix, as described previously.
2219 14.8.4.3 Postprecipitation
2220 Postprecipitation results when a solution contains two ions, one that is rapidly precipitated and
2221 another that is slowly precipitated by the precipitating agent (Kolthoff et al., 1969, p. 245). The
2222 first precipitate is usually contaminated by the second one. For example, calcium oxalate is a
2223 moderately insoluble compound that can be precipitated quantitatively with time. Because the
2224 precipitation tends to be slow, the precipitate is allowed to remain in contact with the solution for
2225 some time before filtering. Magnesium oxalate is too soluble to precipitate on its own under
2226 normal conditions. As long as the solution contains a predominance of calcium ions, very little
2227 magnesium precipitates. However, as the precipitation of calcium approaches quantitative levels,
2228 the competition of calcium and magnesium ions for adsorption at the surface becomes more
2229 intense. As time progresses, the magnesium oxalate adsorbed on the surface acts as seed to
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 14-81 DO NOT CITE OR QUOTE
-------
Separation Techniques
2230 induce the post-precipitation of a second solid phase of magnesium oxalate (MgC2O4). Once
2231 precipitated, the magnesium oxalate is only slightly soluble and does not redissolve.
2232 14.8.4.4 Coprecipitation Methods
2233 Selective coprecipitation of a radionuclide with an insoluble compound is primarily
2234 accomplished by the judicious selection of the compound that forms the precipitate and the
2235 concentration of solutions used in the precipitate's formation. Using good precipitation technique
2236 minimizes the coprecipitation of impurities. The compound, then, should maximize
2237 coprecipitation of the select radionuclide while providing a well-formed solid that attracts a
2238 minimum of other foreign ions as impurities. In general, conditions that favor precipitation of a
2239 substance in macroamounts also favor the coprecipitation of the same material from tracer
2240 concentrations (i.e., too low for precipitate formation) with a foreign substance (Friedlander
2241 et al., 1981, p. 294). Wahl and Bonner provide a useful summary for coprecipitation of a tracer
2242 by a carrier:
2243 "In general a tracer is efficiently carried by an ionic precipitate if: (1) the tracer ion is
2244 isomorphously incorporated into the precipitate, or (2) the tracer ion forms a slightly soluble
2245 or slightly dissociated compound with the oppositely charged lattice ion and if the precipitate
2246 has a large surface with charge opposite to that of the tracer ion (i.e., presence of excess of
2247 the oppositely charged lattice ion)" (Wahl and Bonner, 1951, p. 105).
2248 Considering the principles of precipitation and coprecipitation, radium is coprecipitated
2249 quantitatively with barium sulfate using excess sulfate in solution because: (1) radium forms the
2250 least soluble sulfate of the other elements in the alkaline earth family (Paneth-Fajans-Hahn
2251 adsorption rule); (2) the radium ion carries the same charge as the barium ion and is very similar
2252 in size (inclusion); and 3) an excess of sulfate preferentially creates a common-ion layer on the
2253 crystalline solid of sulfate ions that attracts barium ions and similar ions such as radium
2254 (absorption). For example, in a procedure to determine 226Ra in water samples, radium is
2255 coprecipitated as barium sulfate using 0.36 moles of sulfate with 0.0043 moles of barium, a large
2256 excess of sulfate (EPA, 1984, Method Ra-03).
2257 The isolation of microquantities of tracers often occurs in two steps: first the tracer is separated
2258 by coprecipitation with a carrier, and then it is separated from the carrier (Hermann and Suttle,
2259 1961, p. 1486). Use of carriers that can be easily separated from the tracer is helpful, therefore,
2260 coprecipitation by inclusion is not generally used. Coprecipitation by surface adsorption on
2261 unspecific carriers is the most common method employed. Manganese dioxide MnO2, sulfides
2262 (MnS), and hydroxides [Mn(OH)2] are important nonspecific carries because of their high surface
MARLAP JULY 2001
DO NOT CITE OR QUOTE 14-82 DRAFT FOR PUBLIC COMMENT
-------
Separation Techniques
2263
2264
2265
2266
2267
2268
2269
2270
2271
2272
2273
2274
2275
2276
2277
2278
2279
2280
2281
2282
2283
2284
2285
areas. Ferric hydroxide [Fe(OH)3] is very useful for adsorbing cations, because it forms a very
finely divided precipitate with a negative charge in excess hydroxide ion. Ferric hydroxide is
used, for example, to collect plutonium in solution after it has been isolated from tissue (DOE,
1990 and 1997, Method Pu-04). Tracers can be separated by dissolving the solid in acid and
extracting the iron in ether.
"The amount of ion adsorbed depends on its ability to compete with other ions in solution.
Ions capable of displacing the ions of the radioelements are referred to as holdback carriers
(see Section 14.9.2.4, Holdback Carriers). Highly charged ions, chemical homologs, and ions
isotopic with the radioelement are among the most efficient displacers. Thus, the addition of
a little inactive strontium makes it possible to precipitate radiochemically pure radiobarium
as the nitrate or chloride in the presence of radiostrontium" (Hermann and Suttle, 1961, p.
1487)
Tables 14.13 and 14.14 provide more details about common coprecipitating agents for
radionuclides.
TABLE 14.13 — Common coprecipitating agents for radionuclides(1)
Radionuclide
Am
Cs
Co
Fe
I
Ni
Nb
Oxidation
State
13
+1
+2
+3
-1
+2
+5
Coprecipitate
hydroxide
iodate
fluoride, oxalate, phosphate,
hydroxide
oxalate
acetate
fluoride, sulfate
acetate
phosphomolybdate,
chloroplatinate,
bismuthnitrate,
silicomolybdate
hydroxide
potassiumcobaltnitrate
1 -nitroso-2-napthol
sulfide
hydroxide
ammoniumpyrouranate
iodide
dimethylgloxime hydroxide
hydroxide, phosphate
Carrier(2)
Am+3, Fe+3
Ce+4, Th+4, Zr+4
La+3, Ce+3, Nd+3, Bi+3
Ca+2
Am+4
La+3
U02+2
Cs+1
Co+2
Co+2
Co+2
Co+2
Fe+3
Fe+3
Pb+2, Ag+1, Pd+2
Ni+2
Nb+5
Notes
JULY 2001
DRAFT FOR PUBLIC COMMENT
14-83
MARLAP
DO NOT CITE OR QUOTE
-------
Separation Techniques
2286
2287
2288
2289
2290
2291
2292
Radionuclide
Po
Pu
Ra
Sr
Tc
Th
U
Oxidation
State
+4
+3
+4
+6
+2
+2
+4
+7
+4
+4
+5
+6
Coprecipitate
tellurium
tellurate
selenium
dioxide
hydroxide
sulfide
fluoride
sulfate
fluoride
oxalate, iodate
phosphate
sodium uranylacetate
hydroxide
sulfate, chromate, chloride,
bromide
oxalate, phosphate
fluoride
carbonate
nitrate
chromate
sulfate
phosphate
hydroxide
hydroxide
chlorate, iodate,
perruthenate,
tetrafluoroborate
sulfide
hydroxide
fluoride
iodate
phosphate, peroxide
sulfate
oxalate
cupferron, pyrophosphate,
phosphate, iodate, sulfate,
oxalate
fluoride
phosphate
sulfate
cupferron
pyrouranate
Carrier®
Te
Pb+2
Se or Se2
Mn+4
Fe+3, Al+3, La+3
Cu+2, Bi+2, Pb+2
La+3, Nd+3, Ce+3, Ca+2
La+3(K+1)
La+3, Nd+3, Ce+3
Th+4
Zr+2, Bi+3
U02+2
Fe+3
Ba+2
Th+4, Ca+2, Ba+2
La+3
Sr+2, Ba+2, Ca+2
Sr+2, Ba+2
Ba+2
Sr+2, Ca+2, Pb+2
Sr+2
Fe+3
Tc+4, Fe+3, Mn+2
(Phenyl)4As+1
Tc+7, Re+7, Cu+2, Cd+2
Th+4, La+3, Fe+3, Zr+3,
Ac+3, Zn+2
Th+4, La+3, Nd+3, Ce+3
Th+4, Zr+3
Th+4, Bi+3
Ba+2
Ca+2
u+4
La+3, Nd+3
Zr+3
Ca+2
u+6
u+6
Notes
tellurate reduced with SnQ2
alkaline pH
neutral solution
from aqueous NH3, many
ions stay in solution as NH3
complex
MARLAP
DO NOT CITE OR QUOTE
14-84
JULY 2001
DRAFT FOR PUBLIC COMMENT
-------
Separation Techniques
2293
2294
2295
2296
2297
2298
2299
2300
2301
2302
2303
2304
2305
2306
2307
2308
2309
2310
2311
2312
2313
2314
2315
2316
2317
2318
2319
2320
2321
2322
2323
2324
2325
2326
2327
Radionuclide
Zr+4
Oxidation
State
Coprecipitate
phosphate
peroxide
hydroxide
fluoride
Carrier®
U+6, Al+3
U+6
Fe+3
Th+4
Notes
Th+4, Zr+3 also coprecipitate
without carbonate
(1) Compiled from: Anders, 1960; Booman and Rein, 1962; Cobble, 1964; EPA, 1973; 1980; 1984; DOE, 1990,
1995, 1997; Finston and Kinsley, 1961, Grimaldi, 1961; Grindler, 1962; Hyde, 1960; Kallmann, 1961;
Kallmann, 1964; Kirby and Salutsky, 1964; Metz and Waterbury, 1962; Sedlet, 1964; Sundermann and
Townley, 1960; and Turekian and Bolter, 1966.
(2) If the radionuclide itself is listed, alternate isotopic forms are sometimes used as carriers to form the precipitate.
TABLE 14.14 — Coprecipitation behavior of plutonium and neptunium
Carrier Compound
Hydroxides
Calcium fluoride
Lanthanum fluoride
Barium sulfate
Phosphates:
Calcium phosphate
Bismuth phosphate
Zirconium phosphate
Thorium pyrophosphate
Thorium hypophosphate
U(IV) hypophosphate
Oxalates:
Lanthanum oxalate
Bismuth oxalate
Thorium oxalate
U(IV) oxalate
lodates:
Zirconium iodate
Ceric iodate
Thorium iodate
Sodium uranyl acetate
Zirconium phenylarsenate
Thorium peroxide
Bismuth arsenate
Pu(ill)
C
C
C
C
C
C
NC
NC
C
C
C
C
NC
NC
Pu(IV)
C
C
C
C
C
C
C
C
C
C
C
C
C
C
C
C
C
NC
C
C
C
Pu(VI)
C
NC
NC
NC
NC
NC
NC
NC
NC
NC
NC
NC
NC
NC
C
NC
NC
Np(IV)
C
C
C
C
C
C
C
NC
C
C
C
C
NC
C
C
C
Np(V)
C
C
NC
NC
NC
Poor
Poor
Np(VI)
C
NC
NC
NC
NC
NC
C
NC
"C" indicates nearly quantitative coprecipitation under proper conditions; "NC" indicates that
coprecipitation can be made less than 1-2 percent under proper conditions. [Data compiled from
Seaborg and Katz, Korkisch (1969), and the NAS-NS 3050, 3058 and 3060 monographs.]
JULY 2001
DRAFT FOR PUBLIC COMMENT
14-85
MARLAP
DO NOT CITE OR QUOTE
-------
Separation Techniques
2328
2329
2330
2331
2332
2333
2334
2335
2336
2337
2338
2339
2340
2341
2342
2343
2344
2345
2346
2347
2348
2349
2350
2351
14.8.5 Colloidal Precipitates
Many precipitates exhibit colloidal properties, especially when freshly formed (Salutsky, 1959,
p. 744). The term "colloid state" refers to the dispersion of one phase that has colloidal
dimensions (less than one micrometer, but greater than one nanometer) within a second phase. A
colloidal solution is a colloid in which the second phase is a liquid (also known as a sol).
However, in radiochemistry, a colloid refers to the dispersion of solid particles in the solution
phase. The mixture is not a true solution: particles of the dispersed phase are larger than typical
ions and molecules, and can often be viewed by a light microscope. Colloidal precipitates are
usually avoided in analytical procedures because they are difficult to filter and to wash.
Moreover, the purity of the precipitate is controlled by the tremendously large surface area of the
precipitate and by the localized electrical character of the colloidal surface.
The stability of colloidal solutions and suspensions is governed by two major forces, one of
attraction between the particles (van der Waals) and one of repulsion (electrical double layer)
(Salutsky, 1959, p. 745). This repulsive force is a result of the adsorptive capacity of the colloidal
particles for their own ions. For instance, when silver chloride is precipitated in the presence of
excess silver ions, the particles adsorb silver ions and become positively charged. Then counter-
ions of opposite charge (in this case, nitrate ions) tend to adsorb to the particles to form a second
electrical layer, as illustrated in Figure 14.5.
Counter ions
Adsorbed Layer
(Primary Layer)
Ions in surface
FIGURE 14.5 — The electrical double layer: A schematic representation of adsorption of nitrate
counter-ions onto a primary adsorbed layer of silver ions at the surface of a silver chloride crystal
(Peters et al., 1974).
In a similar fashion, in the presence of a slight excess of alkali chloride, the silver chloride
particles would adsorb chloride ions and become negatively charged. Therefore, precipitates
brought down in the presence of an excess of one of the lattice ions tend to be contaminated with
ions of the opposite charge. Moreover, because all of the particles have the same charge, they
repel each other. If these repulsive forces exceed the attractive van der Waals' forces, a stable
colloid results, and the tightness with which the counter-ions are held in and with the water layer,
MARLAP
DO NOT CITE OR QUOTE
14-86
JULY 2001
DRAFT FOR PUBLIC COMMENT
-------
Separation Techniques
2352 or the completeness with which they cover the primary adsorbed ion layer, determines the
2353 stability of the colloid.
2354 Such adsorption of ions upon the surface of solids in solution is largely, but not entirely, based
2355 upon electrical attraction, otherwise adsorption would not be selective. Recall that there are four
2356 other factors, in addition to magnitude of charge, that affect the preferential adsorption by a
2357 colloid (see Surface Adsorption in Section 14.8.4.1, "Coprecipitation Process").
2358 • The Paneth-Faj ans-Hahn Law dictates that when two or more types of ions are available for
2359 adsorption, the ion that forms the least soluble compound with one of the lattice ions will be
2360 adsorbed preferentially.
2361 • The ion present in the greater concentration will be adsorbed preferentially.
2362 • Ions with a large radius will be adsorbed more readily than ions with a smaller radius because
2363 the larger ion is less hydrated by the solution and not as attracted to the solution phase.
2364 • The ion that is closer to the same size as the lattice ion will be adsorbed preferentially. For
2365 example, radium ions are adsorbed tightly onto barium sulfate, but not onto calcium sulfate;
2366 radium ions are close in size to barium ions, but are much larger than calcium ions.
2367 If an excess of electrolyte is added to the colloidal solution, the electrical double layer is
2368 destroyed and the particles can agglomerate to form larger particles that can settle to the bottom
2369 of the container, a process known as flocculation (or coagulation). For example, Smith et al.
2370 (1995) used polyethylene glycol to remove colloidal silica from a dissolved-soil solution before
2371 the addition of the sample to an ion-exchange resin. Alternatively, the process whereby
2372 coagulated particles pass back into the colloidal state is known as deflocculation., (or peptiza-
2373 tion). Special precautions should be taken during the washing of coagulated precipitates to assure
2374 that deflocculation does not occur. When coagulation is accomplished through charge
2375 neutralization, deflocculation would occur if the precipitate was washed with water. A solution
2376 containing a volatile electrolyte such as nitric acid should be used instead.
2377 There are two types of colloidal solutions (Salutsky, 1959, p. 744):
2378 • Hydrophobic colloids show little or no attraction for water. These solutions have a low
2379 viscosity, can be easily flocculated by the addition of an appropriate electrolyte, and yield
2380 precipitates that are readily filterable.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 14-87 DO NOT CITE OR QUOTE
-------
Separation Techniques
2381 • Hydrophilic colloids have a high affinity for water and are often highly viscous. They are
2382 more difficult to flocculate than hydrophobic colloids, and relatively large amounts of
2383 electrolytes are necessary to cause precipitation. The flocculate keeps water strongly adsorbed
2384 and tends to form jellylike masses that are difficult to filter.
2385 Colloidal precipitations can be a useful separation technique. Because of their great adsorption
2386 capacity, colloidal precipitates are excellent scavengers (collectors) for concentrating trace
2387 substances (Salutsky, 1959, p. 747). Unspecific carriers such as manganese dioxide, sulfides and
2388 hydrated oxides are frequently used as scavengers. For example, protactinium can be efficiently
2389 scavenged and concentrated on manganese dioxide that is precipitated by adding a manganous
2390 salt to a solution containing permanganate. Ferric hydroxide is commonly used to scavenge
2391 cations (Section 14.8.4.4, "Coprecipitation Methods"). Moreover, scavenging precipitations can
2392 sometimes be used to remove interferences. For example, a radionuclide that is capable of
2393 existing in two oxidation states can be effectively purified by precipitation in one oxidation state,
2394 followed by scavenging precipitations for impurities, while the element of interest is in another
2395 oxidation state. A useful procedure for cerium purification involves repeated cycles of eerie
2396 iodate precipitation, reduction to Ce+3, zirconium iodate (ZrIO3) precipitation to remove
2397 impurities (with Ce+3 staying in solution), and reoxidation to Ce+4.
2398 14.8.6 Filterability of Precipitates
2399 The physical nature of a precipitate not only affects the purity of the precipitate, but also the
2400 filterability of the precipitate. Large, well-formed crystals are desirable because they tend to
2401 contain fewer impurities, and are also easier to filter and wash. Many coagulated colloidal
2402 precipitates, such as hydrous oxides or sulfides, tend to form slimy aggregates and to clog the
2403 filter during filtration. There are several approaches that can be taken to improve the physical
2404 form of the precipitate (Salutsky, 1959, pp. 758-761):
2405 • A trace quantity of a hydrophilic colloid can be added to produce complete and rapid
2406 flocculation. For example, gelatin has been used as a sensitizer'm the precipitation of zinc
2407 sulfide, hydrous silica, and various other hydrous oxides, as well-coagulated, filterable
2408 precipitates (Salutsky, 1959, p. 759).
2409 • The slow precipitation techniques described in Section 14.8.3.2, "Factors Affecting
2410 Precipitation," can be used to produce good precipitates.
2411 • Aging the precipitate can result in a precipitate more amenable to filtration. During aging,
2412 small particles with a larger solubility go into solution, and larger particles grow at the cost of
MARLAP JULY 2001
DO NOT CITE OR QUOTE 14-88 DRAFT FOR PUBLIC COMMENT
-------
Separation Techniques
2413
2414
2415
2416
2417
2418
2419
2420
2421
2422
2423
2424
2425
2426
2427
2428
2429
2430
2431
2432
the smaller ones (see "Digestion" under Section 14.8.3.2, "Factors Affecting Precipitation").
Ostwald ripening results in a decrease in the number of particles and, therefore, a decrease in
surface area. The speed of aging generally increases with temperature and with the increasing
solubility of the precipitate in the aging media. Shaking can sometimes promote aging,
perhaps by allowing particles to come into contact and to cement together.
Table 14.15 summarizes general properties of common filters used in analytical procedures.
TABLE 14.15 — General properties of common filter papers (1)
Whatman
Grade
Particle
Retention
(jim)
Porosity
Ash
(%)
Filter
speed
Applications
Qualitative filter papers
1
2
3
4
5
6
>11
>8
>6
> 20-25
>2.5
>3
Medium
Medium-
fine
Medium-
fine
Coarse
Fine
Fine
0.06
0.06
0.06
0.06
0.06
0.2
Medium-
fast
Medium
Medium
Fast
Slow
Slow
Medium crystalline precipitates. A general purpose
filter used in routine laboratory applications, air
pollution monitoring, and soil chemical assays.
Crystalline precipitates. More retentive and
adsorbent than Grade 1, but with increased filtering
time.
Double the thickness of Grade 1 for high precipitate
capacity and increased wet strength. Suitable for
suction filtration.
Coarse and gelatinous precipitates. Used in air
pollution monitoring and for routine cleanup of
biological fluids.
Fine crystalline precipitates. Most retentive of the
series, and is useful for clarifying cloudy
suspensions and water analysis.
Fine crystalline precipitates. Twice as fast as Grade
5. Often specified in boiler water analyses.
Quantitative - ashless
40
41
42
>8
> 20-25
>2.5
Medium
Coarse
Medium
0.010
0.010
0.010
Medium
fast
Slow
Medium crystalline precipitates: calcium oxalate,
well-digested barium. Widely used, general
purpose.
Gelatinous precipitates and coarse particles: iron
and aluminum hydroxides. Also used in quantitative
air analyses.
Highly retentive for fine particles. The laboratory
standard for critical gravimetric analysis.
JULY 2001
DRAFT FOR PUBLIC COMMENT
14-89
MARLAP
DO NOT CITE OR QUOTE
-------
Separation Techniques
2433
2434
2435
2436
2437
2438
2439
2440
2441
2442
2443
2444
2445
2446
2447
2448
2449
2450
2451
2452
2453
2454
Whatman
Grade
43
44
Particle
Retention
(jim)
>16
>3
Porosity
Medium
Fine
Ash
(%)
0.010
0.010
Filter
speed
Medium-
fast
Slow
Applications
Medium crystalline precipitates. Foodstuff and soil
analyses, particle collection in air pollution
monitoring by XRF.
Fine crystalline precipitates. Somewhat thinner and
faster than Grade 42.
Quantitative - hardened low ash
50
52
54
>2.7
>7
> 20-25
Fine
Medium
Coarse
0.025
0.025
0.025
Slow
Medium
Fast
Hardened papers are designed for use in Buchner
funnels, possess great wet strength and lintless
surfaces, and will withstand scraping. Grade 50:
fine crystalline precipitates. Grade 52: crystalline
precipitates, general purpose filtration. Grade 54:
gelatinous precipitates and coarse particles.
Quantitative - hardened ashless
540
541
542
>8
> 20-25
>2.7
Medium
Coarse
Fine
0.008
0.008
0.008
Medium
Fast
Slow
Crystalline precipitates. Gravimetric analysis of
metals in acid/alkali solutions. Collecting
hydroxides after precipitation from strong alkali
solutions.
Coarse gelatinous precipitates. Used for strongly
acidic or alkaline conditions.
Fine crystalline precipitates from under demanding
acidic/alkali conditions.
(1) Fisher (2000-01)
14.8.7 Advantages and Disadvantages of Precipitation and Coprecipitation
Advantages
Provides the only practical method of separation
or concentration in some cases.
Can be highly selective and virtually quantitative.
High degree of concentration is possible.
Provides a large range of scale (mg to industrial).
Convenient, simple process.
Carrier can be removed and procedure continued
with tracer amounts of material (e.g., carrier iron
separated by solvent extraction).
Disadvantages
Can be time consuming to digest, filter, and/or
wash the precipitate.
Precipitate can be contaminated by carrying of
ions or postprecipitation.
Sarge amounts of carrier might interfere with
subsequent separation procedures.
Coprecipitating agent might contain isotopic
impurities of the analyte radionuclide.
Scavenger precipitates are not as selective and
are more sensitive to changes in separation
procedures.
MARLAP
DO NOT CITE OR QUOTE
14-90
JULY 2001
DRAFT FOR PUBLIC COMMENT
-------
Separation Techniques
2455
14.9 Carriers and Tracers
2456
14.9.1 Introduction
2457
2458
2459
2460
2461
2462
2463
2464
2465
2466
2467
2468
2469
2470
2471
2472
2473
2474
2475
2476
2477
2478
Radiochemical analysis frequently requires the radiochemist to separate and determine
radionuclides that are present at extremely small quantities. The amount can be in the picomole
range or less, at concentrations in the order of 10"15 to 10"11 molar. Analysis of radionuclides
using counting techniques, such as alpha spectrometry, liquid scintillation, proportional counting,
or gamma spectrometry, allows activities of radionuclides of 10,000 disintegrations per minute
(dpm) to be determined easily, even though the number of atoms (and mass percent) of these
materials is vanishingly small. Table 14.16 identifies the number of atoms and mass present in
several radionuclides, based on an activity of 500 dpm.
TABLE 14.16 — Atoms and mass of select radionuclides equivalent to 500 dpm(
Radionuclide
Radium-226
Polonium-210
Lead-212
Thallium-208
Half-life
1590 y
140 d
10.6 h
3.1m
Number of Atoms
6.0x10"
1.5 x 108
4.5 x 105
2.3 x 103
Mass (g)
2.3 x 10-10
5.0 xlO'14
1.6 xlO'16
8.0 xlO'19
(1) Based on Wahl andBonner, 1951, p. 102
Considering the minute masses of these analytes and their subsequently low concentration in
solution, it is obvious why conventional techniques of analysis, such as gravimetry, spectro-
photometry, titrimetry, and electrochemistry, cannot be used for their quantitation. However, it is
not immediately obvious why these small quantities might present other analytical difficulties.
As described below, the behavior of such small quantities of materials can be seriously affected
by macro constituents in an analytical mixture in a way that may be unexpected chemically.
14.9.2 Carriers
2479 The key to radiochemical analysis of samples with multiple radionuclides is effective separation
2480 of the different analytes. Separations are most easily accomplished when performed on a macro
2481 scale. As described above, however, the analytes are frequently at levels that challenge the
2482 analyst and the conventional methods to perform the separations. The use of a material that is
2483 different in isotopic make-up to the analyte and that raises the effective concentration of the
2484 material to the macro level is referred to as a carrier. In many cases, the carrier is a non-
2485 radioactive isotope of the analyte. Some carriers are stable isotopes of chemically similar
2486 elements.
JULY 2001
DRAFT FOR PUBLIC COMMENT
14-91
MARLAP
DO NOT CITE OR QUOTE
-------
Separation Techniques
2487 A distinction exists between traditional and radiochemical analyses when referring to macro
2488 amounts. Generally, carriers are present in quantities from a few tenths to several hundred
2489 milligrams of material during the progress of the radiochemical separation.
2490 14.9.2.1 Isotopic Carriers
2491 An isotopic carrier is usually a stable isotope of the analyte. Stable strontium (consisting of
2492 naturally occurring 84Sr, 86Sr, 87Sr, and 88Sr) is frequently used as the carrier in the analysis of 89Sr
2493 and 90Sr. Regardless of the stability of the isotope, the number of protons in the nucleus
2494 ultimately governs the chemical properties of the isotope. Thus, all nuclei that have 38 protons
2495 are strontium and react as strontium classically does.
2496 The purpose of adding a carrier is to raise the chemical concentration of the analyte to the point
2497 where it can be separated using conventional techniques, but for the carrier to perform properly,
2498 it must have the same oxidation state and chemical form as the analyte. It is important then to add
2499 the carrier to the sample as early as possible in chemical process. For example, in the determina-
2500 tion of 131I in milk, the radioiodine might be present as I"1, KV1, CH3I, or I2. The analyst should
2501 assume that all states are present, and treat the sample so that all atoms are brought to a common
2502 oxidation state and chemical form during some step in the procedure, before any separation takes
2503 place. If the final step is precipitation of Agl and the carrier is in the IC^"1 form, no precipitate
2504 will form since AgIO3 that forms when Ag+1 is added is relatively soluble compared to Agl.
2505 Furthermore, if separations of other radioisotopes are performed before this step, there is the
2506 possibility that quantities of the radioiodine could be trapped in the precipitate with other
2507 separated analytes. When concentrations of these materials are very small, even small losses are
2508 significant. The carrier also functions to prevent losses of the analyte during the separation of
2509 other radionuclides or interfering macro-contaminants. This is another reason that it is essential
2510 to add the carrier prior to any chemical treatment of the sample.
2511 The laws of equilibrium for precipitation, distillation, complexation, and oxidation-reduction will
2512 apply to the entire chemical form of analyte in solution, both carrier and radioisotope. If, for
2513 example, 99.995 percent of all strontium is determined to be precipitated during a radiochemical
2514 procedure, then the amount of stable strontium remaining in solution will be 0.005 percent,
2515 which means that 0.005 percent of the radiostrontium still remains in the solution as well. Losses
2516 such as this occur during any chemical process. Frequently then, carriers are used in radiochemi-
2517 cal analyses not only to raise the chemical concentration of the element, but also to determine the
2518 yield of the process. In order to determine the exact amount of radionuclide that was originally
2519 present in the sample, the yield (sometimes called the recovery) of the radionuclide collected at
2520 the end of the procedure should be known. However, since the amount of analyte at the start of
MARLAP JULY 2001
DO NOT CITE OR QUOTE 14-92 DRAFT FOR PUBLIC COMMENT
-------
Separation Techniques
2521 the procedure is the unknown, the yield should be determined by an alternate method. The mass
2522 of the radioanalyte is insignificant in comparison to the carrier, and measuring the yield of the
2523 carrier (gravimetrically, for example) will allow the calculation of the yield of the analyte.
2524 14.9.2.2 Nonisotopic Carriers
2525 Non-isotopic carriers are materials that are similar in chemical properties to the analyte being
2526 separated, but do not have the same number of protons in their nucleus. Usually these carriers
2527 will be elements in the same family in the periodic table. In the classical separation of radium by
2528 the Curies, the slight difference in solubility of radium chloride versus barium chloride allowed
2529 the tedious fractional crystallization of radium chloride to take place (Hampel, 1968, p. 586).
2530 When barium is present in macro-quantities and the radium in femtogram quantities, however,
2531 the two may be easily precipitated together as a sulfate.
2532 For several elements, non-isotopic carriers are chosen from a different family of elements, but
2533 they have the same ionic charge or similar crystalline morphology as the analyte. Lanthanum and
2534 neodymium as +3 ions are frequently used as nonisotopic carriers for U(IV) and Pu(IV) in their
2535 final separation as insoluble fluorides by the process of coprecipitation (Metz and Waterbury,
2536 1962, p. 254) (see also Section 14.8, "Precipitation and Coprecipitation"). The chemical form of
2537 the uranium and plutonium is particularly important for this process; the +4 oxidation state will
2538 coprecipitate, but the +6 in the MO2+2 form, will not. Uranium(IV) is present in solution as UO2+2
2539 and, therefore, will not be coprecipitated with lanthanum fluoride. However, it is very important
2540 to note that even though the precipitation of LaF3 may be quantitative (i.e., >99.995 percent may
2541 be precipitated), there is no measure of how much uranium will also be coprecipitated. Since
2542 uranium and plutonium are not chemically equivalent, the laws of solubility product constant for
2543 lanthanum cannot be applied to uranium. For these types of processes, separate methods should
2544 be used to determine the chemical yield of the process.
2545 For alpha counting rare earths, fluorides (such as NdF3) are frequently used to coprecipitate
2546 elements (Hindman, 1983 and 1986; Sill and Williams, 1981).
2547 Another group of non-isotopic carriers can be described as general scavengers. Substances with
2548 high surface areas, or the ability to occlude contaminants in their floe, can be used to effect gross
2549 separation of all radionuclides from macro quantities of interfering ions. Ferric hydroxide,
2550 manganese dioxide (MnO2) and sulfides (MnS), and hydrated oxides [Mn(OH)x] are examples of
2551 these nonspecific carriers that have been used in many radiochemical separations to eliminate
2552 gross quantities of interfering substances.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 14-93 DO NOT CITE OR QUOTE
-------
Separation Techniques
2553 14.9.2.3 Common Carriers
2554 Carriers for specific analytes are discussed below.
2555 Alkaline Earths
2556 STRONTIUM AND BARIUM. Carrier-free strontium (Sr2) and barium (Ba+2) will coprecipitate with
2557 ferric hydroxide [Fe(OH)3], while calcium (Ca+2) exhibits the opposite behavior and does not
2558 coprecipitate with ferric hydroxide. Lead sulfate (PbSO4) will also carry strontium and barium.
2559 Frequently, inactive strontium and barium are used as carriers for the radionuclides in order to
2560 facilitate separation from other matrix constituents and from calcium. The precipitates used most
2561 frequently in radiochemical procedures are the chromates (CrO4"2), nitrates (NC^"1), oxalates
2562 (C2O4"2), sulfates (SO4-2), and barium chloride (BaCl2). Several different methods of separation
2563 are identified here:
2564 • Chromate precipitation is used in the classical separation of the alkaline earths. Barium
2565 chromate (BaCrO4) is precipitated from a hot solution buffered to a pH of 4 to minimize
2566 strontium and calcium contamination of the barium precipitate. Ammonium ion (NH4+1) is
2567 then added to the solution, and strontium chromate (SrCrO4) is precipitated.
2568 • Barium and strontium can be separated from calcium as the nitrates. Fuming nitric acid is
2569 used to increase the nitric acid concentration to 60 percent, conditions at which barium and
2570 strontium nitrate [Ba(NO3)2 and Sr(NO3)2] precipitate and calcium does not.
2571 • Oxalate precipitation does not separate one alkaline earth from another, but it is usually used
2572 to produce a weighable and reproducible form suitable for radioassay. The precipitation is
2573 accomplished from a basic solution with ammonium oxalate [(NH4)2C2O4].
2574 • Barium sulfate (BaSO4) precipitation is generally not used in separation procedures. It is
2575 more common as a final step to produce a precipitate that can be readily dried, weighed, and
2576 mounted for counting. Barium is readily precipitated by slowly adding dilute sulfuric acid
2577 (H2SO4) to a hot barium solution and digesting the precipitate. For the precipitation of
2578 strontium or calcium sulfate (SrSO4 and CaSO4), a reagent such as alcohol should be added to
2579 lower the solubility, and the precipitant must be coagulated by heat.
2580 • Insolubility of barium chloride (BaCl2) in strong hydrochloric acid solution (HC1) is the basis
2581 of the method to separate barium from calcium, strontium, and other elements. The
MARLAP JULY 2001
DO NOT CITE OR QUOTE 14-94 DRAFT FOR PUBLIC COMMENT
-------
Separation Techniques
2582 precipitation is performed either by adding an ether-hydrochloric acid solution or by bubbling
2583 dry hydrogen chloride gas into the aqueous solution.
2584 RADIUM. Radium (Ra) yields the same types of insoluble compounds as barium: sulfates,
2585 chromates, carbonates (CO3"2), phosphates (PO4"3), oxalates, and sulfites (SO3"2). Hence, radium
2586 coprecipitates with all barium compounds and, to a lesser extent, with most strontium and lead
2587 compounds. Barium sulfate and barium chromate are most frequently used to carry radium. Other
2588 compounds that are good carriers for radium include ferric hydroxide when precipitated at
2589 moderately high pH with sodium hydroxide (NaOH), barium chloride when precipitated from a
2590 cold mixed solvent of water and alcohol saturated with hydrochloric acid, barium iodate (BaIO3)
2591 and various insoluble phosphates, fluorides (F"1) and oxalates (e.g., thorium phosphate
2592 [Th3(PO4)], lanthanum fluoride (LaF3), and thorium oxalate [Th(C2O4)].
2593 Rare Earths. Scandium. Yttrium, and Actinium
2594 Ferric hydroxide and calcium oxalate (CaC2O4) will coprecipitate carrier-free rare earths without
2595 difficulty.
2596 The rare earths will coprecipitate one with another in almost all of their reactions; one rare earth
2597 can always be used to coprecipitate another. The rare earth hydroxides, fluorides, oxalates, and 8-
2598 hydroxyquinolates in ammoniacal solution are insoluble. Conversely, the rare earth hydroxides
2599 will carry a number of elements that are insoluble in basic solution; the rare earth oxalate will
2600 coprecipitate calcium; and the rare earth fluorides tend to carry barium and zirconium (Zr). In the
2601 absence of macro quantities of rare earths, actinium will carry on barium sulfate and lead sulfate
2602 (PbSO4).
2603 Lead
2604 Ferric hydroxide and aluminum hydroxide [A1(OH)3] carry lead very effectively from ammonium
2605 solutions under a variety of conditions. Lead is carried by barium or radium chloride, but not
2606 carried by barium or radium bromide (BaBr2 or RaBr2). This behavior has been used to separate
2607 radiolead isotopes from radium salts. Lead is also carried by barium carbonate (BaCO3), barium
2608 sulfate, radium sulfate, radium chloride, lanthanum carbonate [La2(CO)3], barium chloride, and
2609 silver chromate (Ag2CrO4). Calcium sulfate in the presence of alcohol has also been used to
2610 coprecipitate lead.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 14-95 DO NOT CITE OR QUOTE
-------
Separation Techniques
2611 Polonium
2612 Trace quantities of polonium (Po) are carried almost quantitatively by bismuth hydroxide
2613 [Bi(OH)3] from ammoniacal solution. Ferric, lanthanum, and aluminum hydroxides have also
2614 been used as carriers for polonium in alkaline solutions. Colloidal platinum and coagulated silver
2615 hydroxide (AgOH) and ferric hydroxide sols have been used to carry polonium. Because of the
2616 high oxidation state of polonium, it is susceptible to being a contaminant in almost any
2617 precipitate. Removal of polonium by electrodeposition on nickel metal is recommended prior to
2618 final precipitation for any gross counting technique (proportional counting and liquid
2619 scintillation, for example).
2620 Actinides
2621 THORIUM. Thorium (Th) will coprecipitate with ferric, lanthanum [La(OH)3], and zirconium
2622 hydroxide [Zr(OH)4]. These hydroxide carriers are nonspecific, and therefore, will only remove
2623 thorium from a simple group of contaminants or as a group separation. The ferric hydroxide
2624 precipitation is best carried out at pH 5.5-6.
2625 Thorium will coprecipitate quantitatively with lanthanum fluoride from strongly acidic solutions,
2626 providing an effective means to remove small quantities of thorium from uranium solutions.
2627 However, the rare earths will also carry quantitatively, and zirconium and barium radioisotopes
2628 will carry unless macro quantities of these elements are added as holdback carriers (see Section
2629 14.9.2.4, "Holdback Carriers").
2630 Precipitation of thorium with barium sulfate is possible from strongly acidic solutions containing
2631 high concentrations of alkali metal sulfates; however, this coprecipitation is nonspecific. Other
2632 actinides, lead, strontium, rare earths, bismuth, scandium (Sc), and yttrium will also carry.
2633 Coprecipitation of thorium on hydrogen hypophosphate (HPO3"2) or phosphate carriers can be
2634 performed from rather strongly acidic solutions. Zirconium phosphate [Zr3(PO4)4] serves as a
2635 good carrier for trace levels of thorium. Moreover, thorium also will carry quantitatively on
2636 zirconium iodate from a strongly acidic solution. If coprecipitation is performed from a strongly
2637 acidic solution and the precipitate is washed with a solution containing iodate, the rare earths and
2638 actinium are eliminated. Ce+4 must be reduced to Ce+3 before precipitation so that it does not
2639 carry.
2640 PROTACTINIUM. Protactinium will be carried quantitatively on hydroxide, carbonate, or
2641 phosphate precipitates of tantalum (Ta), zirconium, niobium (Nb), hafnium (Hf), and titanium
MARLAP JULY 2001
DO NOT CITE OR QUOTE 14-96 DRAFT FOR PUBLIC COMMENT
-------
Separation Techniques
2642 (Ti). It is also carried by adsorption onto flocculent precipitates of calcium hydroxide [Ca(OH)2)]
2643 or ferric hydroxide, and it is carried by manganese dioxide, which is produced by addition of
2644 potassium permanganate (KMnO4) to a dilute nitric acid (HNO3) solution containing manganese
2645 nitrate. However, titanium and zirconium are also carried under these conditions.
2646 URANIUM. Trace concentrations of uranium can be coprecipitated with any of the common
2647 insoluble hydroxides. When coprecipitating U(VI) with hydroxides at pH 6-7, the ammonium
2648 used must be free of carbonate or some of the uranium will remain in solution as the stable
2649 anionic carbonate complex. Hydroxide precipitation is nonspecific, and many other metals will
2650 carry with the uranium.
2651 Uranium(IV) can be coprecipitated as the fluoride or phosphate [UF4 or U3(PO4)4] from relatively
2652 strong acid media; however, U(VI) phosphate [(UO2)3(PO4)2] is precipitated only from very weak
2653 acid solutions (pH 5-6) by the addition of carbonate-free ammonium. The rare earths, and other
2654 metals can also coprecipitate under these conditions.
2655 In general, U+4 should behave similarly to Pu+4 and Np+4, and should be carried by lanthanum
2656 fluoride, eerie and zirconium iodates [Ce(IO4)3 and Zr(IO4)4], cerium and thorium oxalates
2657 [Ce2(PO4)3], barium sulfate, zirconium phosphate [Ce2(PO4)3], and bismuth arsenate (BiAsO4).
2658 However, U(VI) does not carry with these agents as long as the concentration of either carrier or
2659 that of uranium is not too high.
2660 PLUTONIUM AND NEPTUNIUM. Classically, plutonium (Pu) and neptunium (Np) in their ter- and
2661 tetravalent oxidation states have been coprecipitated with lanthanum fluoride in the method most
2662 widely used for the isolation of femtograms of plutonium. However, large amounts of aluminum
2663 interfere with coprecipitation of plutonium, and other insoluble fluorides, such as the rare earths,
2664 calcium, and U+4, coprecipitate.
2665 AMERICIUM AND CURIUM. Bismuth phosphate (BiPO4), which historically has been used to
2666 precipitate plutonium, will also carry americium and curium from 0.1-0.3 M nitric acid.
2667 Impurities such as calcium and magnesium are not carried under these conditions.
2668 Lanthanum fluoride provides a convenient carrier for Am+3 and Cm+3. A lanthanum fluoride
2669 precipitation is not totally specific, but it can provide a preliminary isolation from the bulk of the
2670 fission products and uranium. Additionally, a lanthanum fluoride precipitation can be used to
2671 separate americium from curium. Am+3 is oxidized to Am(V) in dilute acid with persulfate, and
2672 fluoride is added to precipitate Cm+3 on lanthanum fluoride.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 14-97 DO NOT CITE OR QUOTE
-------
Separation Techniques
2673 14.9.2.4 Holdback Carriers
2674 It is often necessary to add holdback carriers to analytical mixtures to prevent unwanted
2675 radionuclides from being carried in a chemical process. Coprecipitation of a radionuclide with
2676 ferric hydroxide carries other ions in addition to the analyte, because of its tendency to adsorb
2677 other ions and occlude them in its crystal matrix. The addition of a holdback carrier, a highly-
2678 charged ion, such as Co+3, represses counter-ion exchange and adsorption to minimize the
2679 attraction of foreign ions. The amount of a given substance adsorbed onto a precipitate depends
2680 on its ability to compete with other ions in solution. Therefore, ions capable of displacing the
2681 radionuclide ions (the hold-back carrier) are added to prohibit the coprecipitation of the
2682 radionuclide. Highly charged ions, chemical homologs, and ions isotopic with the radionuclide
2683 are among the most efficient holdback carriers. Hence, the addition of inactive strontium makes
2684 it possible to precipitate radiochemically pure radiobarium as the nitrate or chloride in the
2685 presence of radiostrontium. Actinium and the rare earth elements can be separated from
2686 zirconium and radium by lanthanum fluoride coprecipitation with the addition of zirconium and
2687 barium holdback carriers. Holdback carriers are used in other processes as well. The extraction of
2688 lutetium from water employs neodymium ions (Nd+3) to avoid adsorption loses (Choppin et al.,
2689 1995, p. 262).
2690 14.9.2.5 Yield (Recovery) of Isotopic Carriers
2691 The use of an isotopic carrier to determine the chemical yield of the analyte is a critical step in
2692 the plan of a radiochemical analysis. The analytical method being used to determine the final
2693 amount of carrier will govern the method of separation. If a gravimetric method is to be used for
2694 the final yield determination, the precipitate must have all the characteristics that would be used
2695 for macro gravimetric analysis—easily dried, definite stoichiometry, non-hygroscopic, and the
2696 like.
2697 Similarly, the reagent used as source of carrier at the beginning of the analysis must be of
2698 primary-standard quality to ensure that the initial mass of carrier added can be determined very
2699 accurately. For a gravimetric yield determination, the equation would be the following:
2700 Percent Yield = (mass of carrier in final separation step) x 100%
2701 (mass of carrier added)
2702 It should be recognized that the element of interest is the only quantity used in this formula. For
2703 example, if strontium nitrate is used as the primary standard and strontium sulfate is the final
MARLAP JULY 2001
DO NOT CITE OR QUOTE 14-98 DRAFT FOR PUBLIC COMMENT
-------
Separation Techniques
2704 precipitate, both masses should be corrected, using a gravimetric factor, so that only the mass of
2705 strontium is used in the equation in both the numerator and denominator.
2706 Other methods to determine the yield of the carrier include atomic absorption spectrometry, ultra-
2707 violet/visible spectrometry, titrimetry, and potentiometry.
2708 14.9.3 Tracers
2709 The term tracer was classically used to express the concentration of any pure radionuclide in
2710 solution that had a mass too small to be measured by an analytical balance (<0.0001 to
2711 0.00001 g). More recently, the definition of a tracer has become more pragmatic. The current
2712 definition of a tracer is a known quantity of a radioisotope that is added to a solution of a
2713 chemically equivalent radioisotope of unknown concentration so that the yield of the chemical
2714 separation can be monitored. In general, a tracer is not a carrier, and a carrier is not a tracer.
2715 The analysis of 241Am in an environmental sample provides an example of a radioisotope
2716 employed in a manner consistent with the recent use of the term tracer. In the analytical
2717 procedure, no stable isotope of americium exists to act as a carrier. Femtogram quantities of
2718 243Am can be produced, however, with accurately known activities. If a known quantity of 243Am
2719 in solution is added to the unknown sample containing 241Am at the beginning of the separation
2720 procedure, and if the resulting activity of 243Am can be determined at the end of the procedure,
2721 then the yield of 241Am can be determined accurately for the process. 243Am added to the sample
2722 in this example is used as a tracer. A measurable mass of this element was not used, but a known
2723 activity was added through addition of the solution. During the course of the radiochemical
2724 separation, lanthanides may have been used to help carry the americium through analysis.
2725 However, they are not used to determine the yield in this example and would be considered,
2726 therefore, a non-isotopic carrier.
2727 When using a tracer in an analytical method, it is important to consider the availability of a
2728 suitable isotope, its chemical form, its behavior in the system, the amount of activity required, the
2729 form in which it should be counted, and any health hazards associated with it (McMillan, 1975,
2730 p. 298).
2731 Perhaps the most important property of the tracer is its half-life. It is preferable to select an
2732 isotope with a half-life that is long compared to the duration of the experiment. By doing so, one
2733 avoids the problems of having to handle high levels of activity at the beginning of the experiment
2734 and of having to make large decay corrections.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 14-99 DO NOT CITE OR QUOTE
-------
Separation Techniques
2735 Purity of the tracer is of critical importance. Radionuclide and radiochemical impurities are the
2736 two principal types of impurities encountered. Radionuclide impurity refers to the presence of
2737 radionuclides other than those desired. For instance, it is very difficult to obtain 236Pu tracer that
2738 does not contain a very small quantity of 239Pu. This impurity should be taken into account when
2739 calculating the 239Pu activity levels of samples. Radiochemical impurity refers to the nuclide of
2740 interest being in an undesired chemical form. This type of impurity has its largest effects in
2741 organic tracer studies, where the presence of a tracer in the correct chemical form is essential. For
2742 example, the presence of 32P-labeled pyrophosphate in an orthophosphate tracer could lead to
2743 erroneous results in an orthophosphate tracer study.
2744 Tracer solutions can also contain other forms of radiochemical impurities. Many tracers are
2745 actinides or other isotopes that have progeny that are radioactive. Tracer solutions are purchased
2746 with known specific activities for the isotopes listed in the solutions. However, from the time of
2747 production of the tracer, ingrowth of progeny radioisotopes occurs. 236Pu is used as a tracer for
2748 239Pu and 240Pu analysis, for example. 236Pu has a half-life of 2.9 years and decays to 232U, which
2749 has a half-life of 72 years. After solutions of 236Pu have been stored for about three years, half of
2750 the radionuclide will be converted to 232U. If the solution is then used as a tracer in a procedure
2751 for analysis of uranium and plutonium in soil, erroneously high results would be produced for the
2752 content of uranium if a gross-counting technique is used. Thus, it is important to consider
2753 chemical purification of a tracer solution prior to use to remove unwanted radioactive progeny.
2754 Tracer analysis is very dependent upon the identical behavior of the tracer and the analyte.
2755 Therefore, tracers should be added to the system as early as possible, and complete isotopic
2756 exchange should be ensured as discussed previously (see Section 14.10, "Radiochemical
2757 Equilibrium"). Obvious difficulties arise when a tracer is added to a solid sample, especially if
2758 the sample is subdivided. Unless complete dissolution and isotopic exchange is ensured, results
2759 should be interpreted carefully.
2760 Isotopes selected for tracer work should be capable of being easily measured. Gamma-emitting
2761 isotopes are ideal because they can easily be detected by gamma spectroscopy without being
2762 separated from other matrix constituents. Alpha- and beta-emitting tracers require separation
2763 before counting. Some common tracers are listed below:
2764 • 85Sr has a 514 KeV gamma ray that can be used to monitor the behavior of strontium in a
2765 system, or for yield determination in a 89Sr/90Sr procedure, as long as the gamma is accounted
2766 for in the beta-counting technique.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 14-100 DRAFT FOR PUBLIC COMMENT
-------
Separation Techniques
2767 • 99mTc with a half-life of 6.02 h and a 143 KeV gamma ray is sometimes used as a yield
2768 monitor for "Tc determinations. Samples are counted immediately to determine the chemical
2769 recovery, then the 99mTc is allowed to decay before analysis of the "Tc.
2770 • 152Eu and 145Sm are frequently used in the development of a new method to estimate the
2771 behavior of the +3 actinides and lanthanides.
2772 • 3H, 14C, 32P, and 36C1 are frequently used in biological studies. In some of these studies, the
2773 radionuclide is covalently bonded to a molecule. As a result, the chemical behavior of the
2774 radionuclide will follow that of the molecule, not the element.
2775 • 229Th is used for Th determinations, both in alpha spectroscopy and inductively coupled
2776 plasma-mass spectroscopy (ICP-MS).
2777 . 232u is commonly used as a tracer in alpha spectroscopy, whereas 236U is used for ICP-MS
2778 determinations. It should be noted that 232U decays to 228Th and therefore needs to be taken
2779 into account if determining Th isotopes in the same sample.
2780 • 242Pu and 236Pu are both used as tracers in Pu analyses. However, 236Pu decays to 232U, which
2781 needs to be taken into account when analyzing both Pu and U in the same sample aliquant.
2782 • 243Am is employed in the analysis of 241Am and Cm by alpha spectroscopy. It is assumed that
2783 Am and Cm are displaying similar chemical behavior.
2784 14.9.3.1 Characteristics of Tracers
2785 The behavior of tracers is often different from that of elements in normal concentrations. The
2786 chemical form of a radionuclide predominant at normal concentrations, for example, might not
2787 be the primary form at tracer concentrations. Alternatively, a shift in the equilibrium that is partly
2788 responsible for a radionuclide's chemical behavior might increase or reduce its concentration as a
2789 result of the low tracer concentration. Hydrolysis reactions are influenced particularly by changes
2790 in concentration because water is one of the species in the equilibrium. For example, hydrolysis
2791 of the uranyl ion is represented by (Choppin et al., 1995, p.243):
2792 m • UO2+2 + p- H2O - (UO2)m(OH) 2m-p + p- H+1
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 14-101 DO NOT CITE OR QUOTE
-------
Separation Techniques
2793 At tracer quantities, the equilibrium will shift to the left as the amount of the uranyl ion
2794 decreases. At 10"3 molar (pH 6), the uranyl ion is 50 percent polymerized; at 10"6 molar, there is
2795 negligible polymerization.
2796 Interactions of radionuclides with impurities present special problems at low concentration.
2797 Difficulties include adsorption onto impurities such as dust, silica, or colloidal or suspended
2798 material, or adsorption onto the walls of the container. Generally, 10"8 to 10"7 moles are needed to
2799 cover a container's walls; but at tracer concentrations, much less is present (Choppin et al., 1995,
2800 p. 242). Adsorption depends on (see Surface Adsorption within Section 14.8.4.1, "Coprecipita-
2801 tion Processes"):
2802 • Concentration. A larger percentage is adsorbed at lower tracer concentrations than at higher
2803 concentrations, because a larger surface area is available compared to the amount of tracer
2804 present. Dilution with carrier decreases the amount of tracer adsorbed because the carrier is
2805 competing for adsorption, and the relative amount of tracer interacting with the walls is much
2806 less.
2807 • Chemical State. Adsorption increases with charge on the ion.
2808 • Nature of the Surface Material. Surfaces that have a negative charge or that contain hydroxyl
2809 groups can interact with cations through electrostatic attraction and hydrogen bonding,
2810 respectively.
2811 • pH. Generally, adsorption decreases with a lower pH (higher hydrogen ion concentration)
2812 because the ions interact with negatively-charged surfaces, and hydrogen bonding decreases
2813 their ability to interact with metal ions.
2814 All these processes will reduce the quantity of analyte available for radiochemical procedures
2815 and, therefore, the yield of a procedure. The amount measured by the detection process will be
2816 correspondingly lower, introducing additional error and uncertainty that would go undetected at
2817 normal concentrations.
2818 Adsorption can be useful, however. For example, carrier-free yttrium (Y+3) is quantitatively
2819 adsorbed onto filter paper from basic strontium solutions at concentrations at which yttrium
2820 hydroxide, Y(OH)3, will not precipitate. Also, carrier-free niobium (Nb) has been adsorbed on
2821 glass fiber filters for a fast specific separation technique (Friedlander et al., 1981, p. 296).
MARLAP JULY 2001
DO NOT CITE OR QUOTE 14-102 DRAFT FOR PUBLIC COMMENT
-------
Separation Techniques
2822 Specific behavior characteristics of compounds in separation techniques are further described
2823 below. Additional discussion can also be found in the respective sections found earlier in this
2824 document that describe each separation technique.
2825 14.9.3.2 Coprecipitation
2826 Often, the concentration of tracer is so low that precipitation will not occur in the presence of a
2827 counter-ion that, at normal concentrations, would produce an insoluble salt. Under these
2828 conditions, carriers are used to coprecipitate the tracer. (Coprecipitation is described in
2829 Section 14.8)
2830 14.9.3.3 Deposition on Nonmetallic Solids
2831 Radionuclides can be deposited onto preformed ionic solids, charcoal, and ion-exchange resins
2832 (Wahl and Bonner, 1951, p. 124). The mechanisms of adsorption onto preformed ionic solids are
2833 similar to those responsible for Coprecipitation: counter-ion exchange and isomorphous exchange
2834 (Section 14.8, "Precipitation and Coprecipitation"). Adsorption is favored by a large surface area,
2835 charge of the solid and radionuclide, solubility of compound formed between the solid and the
2836 radionuclide, and time of contact; however, it depends, to a large extent, on whether or not the
2837 radionuclide ion can fit into the crystal lattice of the precipitate. Similarly, adsorption onto
2838 charcoal depends on the amount of charcoal and its surface area, time of contact, and nature of
2839 the surface, because it can be modified by the presence of other ions or molecules.
2840 Adsorption of radionuclides, with and without carriers (Friedlander et al., 1981, p. 297), onto
2841 ion-exchange resins, followed by selective elution, has been developed into a very efficient
2842 separation technique (Wahl and Bonner, 1951, p. 145) (see Section 14.6.4, "Ion-Exchange
2843 Chromatography"). Friedlander et al. (1981), illustrates this phenomenon:
2844 "Ion-exchange separations generally work as well with carrier-free tracers as with weighable
2845 amounts of ionic species. A remarkable example was the original isolation of mendelevium at
2846 the level of a few atoms (p. 298)...The transuranium elements in the solution were ...
2847 separated from one another by elution ...through a cation-exchange column" (p. 450).
2848 14.9.3.4 Radiocolloid Formation
2849 At the tracer level, a radionuclide solution is not necessarily truly homogeneous, but can be a
2850 microparticle (colloid) of variable size or aggregation (Adolff and Guillaumont, 1993, p. 196).
2851 Carrier-free tracers can become colloidal by two mechanisms:
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 14-103 DO NOT CITE OR QUOTE
-------
Separation Techniques
2852 1. Sorption onto a preexisting colloidal impurity (approximately 0.001 |i to 0.5 |i), such as
2853 dust, cellulose fibers, glass fragments, organic material, and polymeric metal hydrolysis
2854 products (Choppin et al., 1995, p. 243; Adolff and Guillaumont, 1993, p. 196)
2855 2. Polycondensation of a monomeric species consisting of aggregates of 103 to 107
2856 radioactive atoms (Adolff and Guillaumont, 1993, p. 197)
2857 The presence of radiocolloids in solution can be detected by one or more of the following
2858 characteristics of the solution, which is not typical behavior of a true solution (Adolff and
2859 Guillaumont, 1993, p. 196):
2860 • The radionuclide can be separated from solution by a physical method such as ultrafiltration
2861 or ultracentrifugation.
2862 • The radionuclide does not follow the laws of a true solution when a chemical gradient
2863 (diffusion, dialysis, isotopic exchange) or electrical gradient (electrophoresis, electrolysis,
2864 electrodialysis) is applied.
2865 • Adsorption on solid surfaces and spontaneous deposition differ from those effects observed
2866 for radionuclides in true solution.
2867 • Autoradiography reveals the formation of aggregates of radioactive atoms.
2868 Several factors affect the formation of radiocolloids (Wahl and Bonner, 1951, pp. 145-148):
2869 • Solubility of the Tracer. The tendency of the tracer radionuclide to hydrolyze and form an
2870 insoluble species with another component of the solution favors radiocolloid formation,
2871 while the presence of ligands that form soluble complexes hinders formation; low pH tends
2872 to minimize hydrolysis of metallic radionuclides.
2873 • Foreign Particles. The presence of foreign particles provides sites for the tracer to adsorb
2874 onto their surfaces; solutions containing ultrapure water prepared with micropore filters
2875 reduce their presence, although the preparation of water completely free of suspended
2876 particles is difficult.
2877 • Electrolytes. Electrolytes affect the nature (species) of the tracer ions in solution (see Section
2878 14.10, Radiochemical Equilibrium)., as well as the charge on both the radiocolloid and the
2879 foreign particle from which the colloid might have been derived.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 14-104 DRAFT FOR PUBLIC COMMENT
-------
Separation Techniques
2880 • Solvent. Polar and nonpolar solvents can favor the formation of radiocolloids, depending on
2881 the specific radiocolloid itself.
2882 • Time. The amount of radiocolloidal formation generally increases with the age of solution.
2883 14.9.3.5 Distribution (Partition) Behavior
2884 Distribution (partition) coefficients, which reflect the behavior of solutes during solvent
2885 extraction procedures (Section 14.4, "Solvent Extraction"), are virtually independent of
2886 concentration down to tracer concentrations (Friedlander et al., 1981, p. 299). Whenever the
2887 radioactive substance itself changes into a different form, however, the coefficient naturally
2888 changes, affecting the distribution between phases during extraction or any distribution
2889 phenomena, such as ion-exchange or gas-liquid chromatography (Section 14.7, "Chromatog-
2890 raphy"). Several properties of tracer solutions can alter the physical or chemical form of the
2891 radionuclide in solution and alter its distribution behavior (Wahl and Bonner, 1951, pp. 149-
2892 151):
2893 • Radiocolloid formation might concentrate the radionuclide in the alternate phase or at the
2894 interface between the phases.
2895 • Shift in equilibrium during complex-ion formation or hydrolysis reactions can alter the
2896 concentration of multiple radionuclide species in solution (Section 14.9.3.1, "Characteristics
2897 of Tracers").
2898 14.9.3.6 Vaporization
2899 Radioisotope concentrations that challenge the minimum detectible concentration (MDC) can be
2900 vaporized from solid surfaces or solution (Section 14.5, "Volatilization and Distillation"). Most
2901 volatilization methods of these trace quantities of radionuclides can be performed without
2902 specific carriers, but some nonisotopic carrier gas might be required (Friedlander et al., 1981,
2903 p. 300).
2904 Vaporization of these amounts of materials from solid surfaces differs from the usual process of
2905 vaporization of macroamounts of material, because the surface of the solid is usually not
2906 completely covered with the radionuclide (Wahl and Bonner, 1951, pp. 151-158). Carrier-free
2907 radionuclides at the surface are bonded with the surface particles instead of with themselves, and
2908 the bonds broken during the process are between the solid and the radioisotope, rather than
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 14-105 DO NOT CITE OR QUOTE
-------
Separation Techniques
2909 between the radioisotope particles themselves. Additionally, the nature of the radioisotope can be
2910 altered by small trace quantities of gases such as oxygen and water present in the vacuum.
2911 Therefore, the identity of the radionuclide species vaporizing might be uncertain, and the data
2912 from the procedure can be hard to interpret. The rate of vaporization of radioisotopes also
2913 decreases with time, because the number of radioisotope particles available on the solid surface
2914 decreases with time.
2915 Radioisotopes near the MDC and macroquantities of radionuclide solutes should behave very
2916 similarly in vaporization experiments from solution, however, because both are present as a
2917 small fraction of the solution. They are, therefore, surrounded and bonded to solvent molecules
2918 rather than to other solute particles (Wahl and Bonner, 1951, p. 156). The nature of the solvent,
2919 the pH, and the presence of electrolytes generally affect the solubility of the solute and its
2920 vaporization behavior.
2921 14.9.3.7 Oxidation and Reduction
2922 Some radionuclides exist in only one oxidation state in solution, but others can exist in several
2923 stable states (Tables 14.1 and 14.2). If multiple states are possible, it might be difficult to
2924 ascertain in which state the radionuclide actually exists because the presence of trace amounts of
2925 oxidation or reduction (redox) impurities might convert the radionuclide to a state other than the
2926 one in which it was prepared (Wahl and Bonner, 1951, pp. 158-159). Excess redox reagents can
2927 often be added to the solution to convert the forms to a fixed ratio and keep the ratio constant
2928 during subsequent procedures.
2929 For a redox equilibrium such as:
2930 PuO2+2 + 4 H+1 + Hg - Pu+4 + Hg+2 + 2 H2O
2931 the Nernst equation is used to calculate the redox potential, E, from the standard potential, E°:
2932 E = E° - kT ln([Pu+4] [Hg+2]/[PuO2+2] [FT1]4)
2933 where k is a constant for the reaction (R/2F, containing the ideal gas constant, R, and Faraday's
2934 constant, F) and T is the absolute temperature. Water and metallic mercury (Hg) do not appear i
2935 the equation, because their activity is one for a pure substance. Minute concentrations of ions ir
2936 solution exhibit the same redox potential as macroquantities of ions because, E depends on the
2937 ratio of ion concentrations and not their total concentration.
in
in
MARLAP JULY 2001
DO NOT CITE OR QUOTE 14-106 DRAFT FOR PUBLIC COMMENT
-------
Separation Techniques
2938 Electrolysis of some solutions is used for electrodeposition of a carrier-free metal on an electrode
2939 (Choppin et al., 1995, p. 246) or other substance, leaving the impurities in solution (Friedlander
2940 et al., 1981, p. 301). The selectivity and efficiency, characteristic of deposition of macro-
2941 quantities of ions at a controlled potential, is not observed, however, for these metals. The
2942 activity of the ion is not known, even if the concentration is, because the activity coefficient is
2943 dependent on the behavior of the mixed electrolytic system. In addition, the concentration of the
2944 metal in solution might not be known because losses may occur through adsorption or
2945 complexation with impurities. Electrolytic deposits are usually extremely thin—a property that
2946 makes them useful for counting measurements (Wahl and Bonner, 1951, p. 162).
2947 Deposition by chemical displacement is sometimes used for the separation of tracer from bulk
2948 impurities (Friedlander et al., 1981, p. 301). Polonium and lead spontaneously deposit from a
2949 solution of hydrochloric acid onto a nickel disk at 85 °C (Blanchard, 1966). Alpha and beta
2950 counting is then used to determine 210Po and 210Pb. The same technique is frequently used in low-
2951 level analysis of transuranic elements to remove lead and polonium so that they do not interfere
2952 with the subsequent alpha analysis of the elements. Wahl and Bonner (1951, pp. 460-465)
2953 contains a helpful table (6F) of electrochemical methods used for the oxidation and reduction of
2954 carrier-free tracers.
2955 14.10 Radiochemical Equilibrium
2956 14.10.1 Basic Principles of Equilibrium
2957 Radiochemical analysis is based on the assumption that an element reacts the same chemically,
2958 whether or not it is radioactive. This assumption is valid when the element (analyte) and the
2959 carrier/tracer are in the same oxidation state, complex, or compound. The atomic weight of most
2960 elements is great enough that the difference in atomic weight between the radionuclide of interest
2961 and the carrier or tracer will not result in any chemical separation of the isotopes. This
2962 assumption might not be valid for the very lightest elements (e.g., H, Li, Be, and B) when mass
2963 fractionation or measuring techniques are used.
2964 Most radiochemical procedures involve the addition of one of the following:
2965 • A carrier of natural isotopic composition (i.e., the addition of stable strontium carrier to
2966 determine 89Sr/90Sr; EPA, 1980, Method 905.0).
2967 • A stable isotope tracer (i.e., enriched 180,15N, and 14C, are frequently used in mass
2968 spectroscopy studies).
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 14-107 DO NOT CITE OR QUOTE
-------
Separation Techniques
2969 • A radionuclide tracer (i.e., the addition of a known quantity of 236Pu tracer to determine 239Pu
2970 by alpha spectroscopy; DOE, 1990 and 1997, Method Pu-02).
2971 To achieve quantitative yields, there must be complete equilibration (isotopic exchange) between
2972 the added isotope and all the analyte species present. In the first example, isotopic exchange of
2973 the carrier with the radiostrontium is achieved and a weighable, stoichiometric compound of the
2974 carrier and radionuclide are produced. The chemical recovery from the separation technique is
2975 determined gravimetrically. Alternatively, a known quantity of an isocesiumtope is used and
2976 determined independently by mass analysis, pulse-height analysis, or another counting technique.
2977 Carriers and tracers are added as soon in the sample preparation process as possible, usually after
2978 the bulk sample is dried and homogenized, but before sample decomposition to ensure that the
2979 chemistry of the carriers or tracers is truly representative of the radioisotope of interest. Thus,
2980 losses occurring during sample preparation steps, before decomposition, are not quantified and
2981 might not be detected, although losses during these earlier steps are usually minimized. Having
2982 the carriers and tracers present during the sample decomposition provides an opportunity to
2983 equilibrate the carrier or tracer with the sample so that the carrier, tracer, and analyte are in the
2984 identical chemical form. While this can initially appear to be rather easy, in some cases it is
2985 extremely difficult. The presence of multiple valence states and the formation of chemical
2986 complexes are two conditions that introduce a host of equilibration problems (Section 14.2.2,
2987 "Oxidation-Reduction Reactions"; Section 14.2.3, "Common Oxidation States"; and Section
2988 14.2.4, "Oxidation State in Solution"). Crouthamel and Heinrich (1971, pp. 5473-5474) has an
2989 excellent discussion of the intricacies and challenges associated with attaining true isotopic
2990 exchange:
2991 "Fortunately, there are many reactions which have high exchange rates. This applies even
2992 to many heterogeneous systems, as in the heterogeneous catalysis of certain electron
2993 transfer reactions. In 1920, Hevesy, using ThB (212Pb), demonstrated the rapid exchange
2994 between active lead nitrate and inactive lead chloride by the recrystallization of lead
2995 chloride from the homogeneously mixed salts. The ionization of these salts leads to the
2996 chemically identical lead ions, and a rapid isotopic exchange is expected. Similar
2997 reversible reactions account for the majority of the rapid exchange reactions observed at
2998 ordinary temperatures. Whenever possible, the analyst should conduct the isotope
2999 exchange reaction through a known reversible reaction in a homogeneous system. The
3000 true homogeneity of a system is not always obvious, particularly when dealing with the
3001 very low concentrations of the carrier-free isotopes. Even the usually well-behaved alkali-
3002 metal ions in carrier-free solutions will adsorb on the surfaces of their containment
MARLAP JULY 2001
DO NOT CITE OR QUOTE 14-108 DRAFT FOR PUBLIC COMMENT
-------
Separation Techniques
3003 vessels or on colloidal and insoluble material in the solution. This is true especially in the
3004 heavier alkali metals, rubidium and cesium. Cesium ions in aqueous solution have been
3005 observed to absorb appreciably to the walls of glass vessels when the concentrations were
3006 below 10-6g/mL"
3007 The reaction described above can be written as follows:
3008 212Pb(NO3)2(s) + PbCl2(s) - Pb(NO3)2 + 212PbCl2
3009 Any of the following techniques may be employed to achieve both chemical and isotopic
3010 equilibration:
3011 • Careful adding, mixing, stirring, shaking, etc., to assure a homogeneous solution and prevent
3012 layering.
3013
3014
3015
3016
• Introducing the carrier or tracer in several different chemical forms or oxidation states,
followed by oxidation or reduction to a single state.
• Treating the carrier or tracer and sample initially with strong oxidizing or reducing agents
during decomposition (e.g., wet ashing or fusion).
3017 • Carrying out repeated series of oxidation-reduction reactions.
3018 • Requiring that, at some point during the sample decomposition, all the species be together in
3019 a clear solution.
3020 Once a true equilibration between carrier or tracer and sample occurs, the radiochemistry
3021 problem shifts from one of equilibration to that of separation from other elements, and ultimately
3022 a good recovery of the radionuclide of interest.
3023 Crouthamel and Heinrich summarizes the introduction to equilibration (isotopic exchange)
3024 (Crouthamel and Heinrich, 1971, pp. 5475-5476):
3025 "Probably the best way to give the reader a feeling for the ways in which isotopic
3026 exchange is achieved in practice is to note some specific examples from radiochemical
3027 procedures. The elements which show strong tendencies to form radiocolloids in many
3028 instances may be stabilized almost quantitatively as a particular complex species and
3029 exchange effected. Zirconium, for example, is usually exchanged in strong nitric acid-
JULY2001 MARLAP
DRAFT FOR PUBLIC COMMENT 14-109 DO NOT CITE OR QUOTE
-------
Separation Techniques
3030 hydrofluoric acid solution. In this medium, virtually all the zirconium forms a ZrF62"
3031 complex. Niobium exchange is usually made in an oxalate or fluoride acid medium. The
3032 exchange of ruthenium is accomplished through its maximum oxidation state, Ru(VIII)
3033 which can be stabilized in a homogeneous solution and distilled as RuO4. Exchange may
3034 also be achieved by cycling the carrier through oxidation and reduction steps in the
3035 presence of the radioactive isotope. An iodine carrier with possible valence states of -1 to
3036 +7 is usually cycled through its full oxidation-reduction range to ensure complete
3037 exchange. In a large number of cases, isotopic exchange is not a difficult problem;
3038 however, the analyst cannot afford to relax his attention to this important step. He must
3039 consider in each analysis the possibility of both the slow exchange of certain chemical
3040 species in homogenous solution and the possible very slow exchange in heterogeneous
3041 systems. In the latter case, this may consist simply of examining the solutions for
3042 insoluble matter and taking the necessary steps to either dissolve or filter it and to assay
3043 for possible radioactive content."
3044 Also see the discussion of equilibration of specific radionuclides in Section 14.10.9, "Review of
3045 Specific Radionuclides."
3046 14.10.2 Oxidation State
3047 Some radionuclides exist in solution in one oxidation state that does not change, regardless of the
3048 kind of chemical treatment used for analysis. Cesium (Cs), radium, strontium, tritium (3H), and
3049 thorium are in the +1, +2, +2, +1, and +4 oxidation states, respectively, during all phases of
3050 chemical treatment. However, several radionuclides can exist in more than one state, and some
3051 are notable for their tendency to exist in multiple states simultaneously, depending on the other
3052 components present in the mixture. Among the former are cobalt, iron, iodine, and technetium,
3053 and among the latter are americium, plutonium, and uranium. To ensure identical chemical
3054 behavior during the analytical procedure, the radionuclide of interest and its carriers and/or
3055 tracers in solution must be converted to identical oxidation states. The sample mixture containing
3056 the carriers and/or tracer is treated with redox agents to convert each state initially present to the
3057 same state, or to a mixture with the same ratio of states. Table 6E in Wahl and Bonner (1951, pp.
3058 450-459) provides a list of traditional agents for the oxidation and reduction of carrier-free
3059 tracers that is a useful first guide to the selection of conditions for these radioequilibrium
3060 processes.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 14-110 DRAFT FOR PUBLIC COMMENT
-------
Separation Techniques
3061 14.10.3 Hydrolysis
3062 All metal ions (cations) in aqueous solution interact extensively with water, and, to a greater or
3063 lesser extent, they exist as solvated cations (Katz et al., 1986, p. 1141):
3064 Ra+2 + x-H2O - Ra(H2O)x+2
3065 The more charged the cation, the greater is its interaction with water. Solvated cations, especially
3066 those with +4, +3, and small +2 ions, tend to act as acids by hydrolyzing in solution. Simply
3067 stated, hydrolysis is complexation where the ligand is the hydroxyl ion. To some extent, all metal
3068 cations in solution undergo hydrolysis and exist as hydrated species. The hydrolysis reaction for a
3069 metal ion is represented simply as (Choppin et al., 1995, p.650):
3070 M+n + m- H2O - M(OH)m+(n-m) + m- H+1
3071 Hydrolysis of the ferric ion (Fe+3) is a classical example:
3072 Fe+3 + H2O - Fe(OH)+2 + H+1
3073 Considering the hydrated form of the cation, hydrolysis is represented by:
3074 M(OH2)x+n - M(OH2)x4(OH)(n-1)+ + H+1
3075 In the latter equation, the hydrated complex ion associated with the hydroxide ion, is known as
3076 the aquo-hydroxo species (Birkett et al., 1988, p. 2.7-3). As each equation indicates, hydrolysis
3077 increases the acidity of the solution, and the concentration of the hydrogen ion (pH) affects the
3078 position of equilibrium. An increase in acidity (increase in H+1 concentration; decrease in pH)
3079 shifts the position of equilibrium to the left, decreasing hydrolysis, while a decrease in acidity
3080 shifts it to the right, increasing hydrolysis. The extent of hydrolysis, therefore, depends on the pH
3081 of the solution containing the radionuclide. The extent of hydrolysis is also influenced by the
3082 radius and charge of the cation (charge/radius ratio). Generally, a high ratio increases the
3083 tendency of a cation to hydrolyze. A ratio that promotes hydrolysis is generally found in small
3084 cations with a charge greater than one (Be+2, for example). The thorium cation, Th+4, with a
3085 radius three times the size of the beryllium ion but a +4 charge, is hydrolyzed extensively, even at
3086 a pH of four (Baes and Mesmer, 1976, p. 158). It is not surprising, therefore, that hydrolysis is an
3087 especially important factor in the behavior of several metallic radionuclides in solution, and is
3088 observed in the transition, lanthanide, and actinide groups. For the actinide series, the +4 cations
3089 have the greatest charge/radius ratio and undergo hydrolysis most readily. Below pH 3, the
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 14-111 DO NOT CITE OR QUOTE
-------
Separation Techniques
3090 hydrolysis of Th4+ is negligible, but at higher pH, extensive hydrolysis occurs. Uranium(IV)
3091 undergoes hydrolysis in solution at a pH above 2.9 with U(OH)3+being the predominant
3092 hydrolyzed species. Neptunium ions undergo hydrolysis in dilute acid conditions with evidence
3093 of polymer formation in acidic solutions less than 0.3 M. The hydrolysis of plutonium is the most
3094 severe, often leading to polymerization (see Section 14.10.4, "Polymerization"). In summary, the
3095 overall tendency of actinides to hydrolyze decreases in the order (Katz et al., 1986, p. 1145):
3096 An+4 > AnO2+2 > An+3 > AnO2+1
3097 where "An" represents the general chemical symbol for an actinide.
3098 For some cations, hydrolysis continues past the first reaction with water, increasing the number
3099 of hydroxide ions (OH"1) associated with the cation in the aquo-hydroxo species:
3100 U+4 + H2O - U(OH)+3 + H+1
3101 U(OH)+3 +H2O - U(OH)2+2 + H+1
3102 This process can, in some cases, conclude with the precipitation of an insoluble hydroxide, such
3103 as ferric hydroxide. "Soluble hydrolysis products are especially important in systems where the
3104 cation concentrations are relatively low, and hence the range of pH relatively wide over which
3105 such species can be present and can profoundly affect the chemical behavior of the metal" (Baes
3106 and Mesmer, 1976, p. 3).
3107 Solutions containing trace concentrations of metallic radionuclides qualify as an example of
3108 these systems. The form of hydrolysis products present can control important aspects of chemical
3109 behavior such as (Baes and Mesmer, 1976, p. 3):
3110 • Adsorption of the radionuclide on surfaces, especially on mineral and soil particles.
3111 • Tendency to coagulate colloidal particles.
3112 • Solubility of the hydroxide or metal oxide.
3113 • Extent of complex formation in solution.
3114 • Extent of extraction from solution by various reagents.
3115 • Ability to oxidize or reduce the radionuclide to another oxidation state.
3116 Thus, a knowledge of the identity and stability of radionuclide ion hydrolysis products is
3117 important in understanding or predicting the chemical behavior of trace quantities of
3118 radionuclides in solution (Baes and Mesmer, 1976, p. 3). As the equilibrium equation indicates,
MARLAP JULY 2001
DO NOT CITE OR QUOTE 14-112 DRAFT FOR PUBLIC COMMENT
-------
Separation Techniques
3119 H+1 is produced as cations hydrolyze. Undesirable consequences of hydrolysis can, therefore, be
3120 minimized or eliminated by the addition of acid to the analytical mixture to reverse hydrolysis or
3121 prevent it from occurring. Numerous steps in radioanalytical procedures are performed at low pH
3122 to eliminate hydrolytic effects. It is also important to know the major and minor constituents of
3123 any sample, since hydrolysis effects are a function of pH and metal concentration. Thus,
3124 maintaining the pH of a high iron-content soil sample below pH 3.0 is important, even if iron is
3125 not the analyte.
3126 14.10.4 Polymerization
3127 The hydrolysis products of radionuclide cations described in the preceding section are
3128 monomeric—containing only one metal ion. Some of these monomers can spontaneously form
3129 polymeric metal hydroxo polymers in solution, represented by formation of the dimer (Birkett
3130 etal., 1988, p. 2.8-1):
3131 2 M(H2O)x.1(OH)+(n-1) - [(H2O)x.2M(OH)2M(H2O)x.2]+2(n-1) + 2 H2O
3132 The polymers contain -OH-bridges between the metal ions that, under high temperature,
3133 prolonged aging, and/or high pH, can convert to -O-bridges, leading eventually to precipitation of
3134 hydrated metal oxides. Birkett et al. (1988) states that:
3135 "Formation of polymeric hydroxo species has been reported for most metals, although in
3136 some cases, the predominant species in solution is the monomer. Some metals form only
3137 dimers or trimers, while a few form much larger, higher-molecular-weight polymeric species.
3138 "Increasing the pH of a metal ion solution, by shifting the position of hydrolysis
3139 equilibrium ..., results in an increased concentration of hydrolyzed species ..., which in turn
3140 causes increased formation of polymeric species .... Diluting a solution has two opposing
3141 effects on the formation of polymeric species:
3142 "(1) Because dilution of acidic solutions causes a decrease in FT1 concentration (i.e.,
3143 an increase in pH), it causes a shift in the hydrolyzed equilibrium toward
3144 formation of hydrolyzed species.
3145 "(2) On the other hand, dilution decreases the ratio of polymeric to monomeric
3146 complexes in solution. For metals that form both monomeric and polymeric
3147 complexes, this means that monomeric species predominate beyond a certain level
3148 of dilution" (Birkett et al., 1988, p. 2.8-2).
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 14-113 DO NOT CITE OR QUOTE
-------
Separation Techniques
3149 Because this type of polymerization begins with hydrolysis of a cation, minimizing or
3150 eliminating polymerization can be achieved by the addition of acid to lower the pH of the
3151 analytical solution to prevent hydrolysis (Section 14.10.3, "Hydrolysis").
3152 14.10.5 Complexation
3153 Radionuclides exist as metal ions in solution, and many have a tendency to form stable complex
3154 ions with molecules or anions present as analytical reagents or impurities. The tendency to form
3155 complex ions is, to a considerable extent, an expression of the same properties that lead to
3156 hydrolysis; high positive charge on a +3 or +4 ion provides a strong driving force for the
3157 interaction with ligands (Katz et al., 1986, p. 1146) (Section 14.3, "Complexation").
3158 Complex-ion formation by a radionuclide alters its form, introducing in solution additional
3159 species of the radionuclide whose concentrations depend on the magnitude of the formation
3160 constant(s). Alternate forms have different physical and chemical properties, and behave
3161 differently in separation techniques, such as extraction or partition chromatography. The behavior
3162 of alternate forms of radionuclides can present problems in the separation scheme that should be
3163 avoided if possible or addressed in the protocol. Some separation schemes, however, take
3164 advantage of the behavior of alternate radionuclide species formed by Complexation, which can
3165 alter the solubility of the radionuclides in a solvent or their bonding to an ion-exchange resin
3166 (Section 14.3.4.2, "Separation by Solvent Extraction and Ion-Exchange Chromatography").
3167 14.10.6 Radiocolloid Interference
3168 The tendency of some radionuclides in solution, particularly tracer levels of radionuclides, to
3169 form radiocolloids, alters the physical and chemical behavior of those radionuclides (see Section
3170 14.9.3.4, "Radiocolloid Formation"). Radioanalytical separations will not perform as expected in
3171 solutions containing radiocolloids, particularly as the solubility of the radionuclide species
3172 decreases.
3173 Solutions containing large molecules, such as polymeric metal hydrolysis products, are more
3174 likely to form radiocolloids (Choppin et al., 1995, p. 243). "If the solution is kept at sufficiently
3175 low pH and extremely free of foreign particles, sorption and radiocolloid formation are usually
3176 avoided as major problems" (Choppin et al. 1995, p. 243). If tracer levels of radionuclides are
3177 present, trace impurities become especially significant in the radiochemical procedure, and
3178 should be minimized or avoided whenever possible (Crouthamel and Heinrich, 1971, p. 5493).
MARLAP JULY 2001
DO NOT CITE OR QUOTE 14-114 DRAFT FOR PUBLIC COMMENT
-------
Separation Techniques
3179 Crouthamel and Heinrich provide some specific insight into radiocolloidal interference in the
3180 equilibration problem:
3181 "The transition metals tend to form radiocolloids in solution, and in these heterogeneous
3182 systems the isotopic exchange reaction between a radiocolloid and inactive carrier added to
3183 the solution is sometimes slow and, more often, incomplete. Elements which show a strong
3184 tendency to form radiocolloids, even in macro concentrations and acid solutions, are titanium,
3185 zirconium, hafnium, niobium, tantalum, thorium, and protactinium, and, to a lesser degree,
3186 the rare earths. Other metals also may form radiocolloids, but generally offer a wider choice
3187 of valence states which may be stabilized in aqueous solutions" (Crouthamel and Heinrich,
3188 1971, p. 5474).
3189 14.10.7 Isotope Dilution Analysis
3190 The basic concept of isotope dilution analysis is to measure the changes in specific activity of a
3191 substance upon its incorporation into a system containing an unknown amount of that substance.
3192 Friedlander et al. (1981), define specific activity.
3193 "Specific activity is defined as the ratio of the number of radioactive atoms to the total
3194 number of atoms of a given element in the sample (N*/N). In many cases where only the
3195 ratios of specific activities are needed, quantities proportional to N*/N, such as activity/mole,
3196 are referred to as specific activity" (Friedlander et al., 1981, p. 432).
3197 For example, isotope dilution can be used to determine the amount of some inactive material A
3198 in a system (Wang et al., 1975). To the system containing x grams of an unknown weight of the
3199 inactive form of A, y grams of active material A* of known activity D is added. The specific
3200 activity of the added active material, Sl3 is given by:
3201 Sj = D/y
3202 After ensuring isotopic exchange, the mixture of A and A* is isolated, but not necessarily
3203 quantitatively, and purified. The specific activity, S2, is measured. Owing to the conservation of
3204 matter,
3205 S2 = D/(x + y)
3206 and by substituting for SjV for D and rearranging, the amount x of inactive A is given as
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 14-115 DO NOT CITE OR QUOTE
-------
Separation Techniques
3207 x = y(S1/S2-l)
3208 However, this equation is valid only if complete isotopic exchange has occurred, a task not
3209 always easy to achieve.
3210 14.10.8 Masking and Demasking
3211 Masking is the prevention of reactions that are normally expected to occur through the presence
3212 or addition of a masking reagent. Masking reactions can be represented by the general reversible
3213 equation:
3214 A + Ms-A-Ms
3215 where A is the normal reacting molecule or ion, and Ms is the masking agent. The decreased
3216 concentration of A at equilibrium determines the efficiency of masking. An excess of masking
3217 agent favors the completeness of masking, as expected from LeChatelier's Principle. Feigl (1936,
3218 p. 409) has described masking reagent and the masking of a reaction as follows: "... the
3219 concentration of a given ion in a solution can be so diminished by the addition of substances
3220 which unite with the ion to form complex salts that an ion product sufficient to form a precipitate
3221 or cause a color reaction is no longer obtained. Thus we speak of the masking of a reaction and
3222 call the reagent responsible for the disappearance of the ions necessary for the reaction, the
3223 masking reagent." The concepts of masking and demasking are discussed further in Perrin (1979.
3224 pp. 600-643) and in Dean (1995, pp. 2.9-2.15).
3225 Masking techniques are frequently used in analytical chemistry because they often provide
3226 convenient and efficient methods to avoid the effects of unwanted components of a system
3227 without having to separate the interferant physically. Therefore, the selectivity of many analytical
3228 techniques can be increased through masking techniques. For example, copper can be prohibited
3229 from carrying on ferric hydroxide at pH 7 by the addition of ammonium ions to complex the
3230 copper ions. Fe3+ and A13+ both interfere with the extraction of the +3 actinides and lanthanides in
3231 some systems, but Fe3+ can be easily masked through reduction with ascorbic acid, and A13+ can
3232 be masked through complexation with fluoride ion (Horwitz et al., 1993 and 1994). In another
3233 example, uranium can be isolated on a U/TEVA column (Eichrom Industries, Inc., Darien, IL)
3234 from nitric acid solutions by masking the tetravalent actinides with oxalic acid; the tetravalent
3235 actinides are complexed and pass through the column, whereas uranium is extracted (SpecNews,
3236 1993). Strontium and barium can be isolated from other metals by cation exchange from a
3237 solution of water, pyridine, acetic acid and glycolic acid. The other metals form neutral or
3238 negative complexes and pass through the cation column, while strontium and barium are retained
MARLAP JULY 2001
DO NOT CITE OR QUOTE 14-116 DRAFT FOR PUBLIC COMMENT
-------
Separation Techniques
3239
3240
3241
3242
3243
3244
3245
3246
3247
3248
3249
3250
3251
3252
3253
3254
3255
3256
3257
3258
3259
3260
3261
3262
3263
3264
3265
(Orlandini, 1972). Masking phenomena are present in natural systems as well. It has been
demonstrated that humic and fulvic acids can complex heavy metals such that they are no longer
bioavailable and are, therefore, not taken up by plants. Tables 14.17 and 14.18 list common
masking agents.
TABLE 14.17 — Masking agents for ions of various metals
(i)
Metal Masking Agent
Ag Br, citrate, Cl", CN", I", NH3, SCN" S2 O3"2, thiourea, thioglycolic acid, diethyldithiocarbamate,
thiosemicarbazide, bis(2-hydroxyethyl)dithiocarbamate
Al Acetate, acetylacetone, BF4", citrate, C2O4"2, EDTA, F", formate, 8-hydroxyquinoline-5-sulfonic acid,
mannitol, 2,3-mercaptopropanol, OH", salicylate, sulfosalicylate, tartrate, triethanolamine, tiron
As Citrate, 2,3-dimercaptopropanol, NH2OHHC1, OH", S2"2, tartrate
Au Br, CN-, NH3, SCN", S2O3"2, thiourea
Ba Citrate, cyclohexanediaminetetraacetic acid, AfTV-dihydroxyethylglycine, EDTA, F", SO4"2, tartrate
Be Acetylacetone, citrate, EDTA, F", sulfosalicylate, tartrate
Bi Citrate, Cl', 2,3-dimercaptopropanol, dithizone, EDTA, I', OH", Na5P3O10, SCN', tartrate, thiosulfate,
thiourea, triethanolamine
Ca BF4", citrate, 7V,yV-dihydroxyethylglycine, EDTA, F", polyphosphates, tartrate
Cd Citrate, CN", 2,3-dimercaptopropanol, dimercaptosuccinic acid, dithizone, EDTA, glycine, I", malonate,
NH3, 1,10-phenanthroline, SCN', S2O3"2, tartrate
Ce Citrate, AfjV-dihydroxyethylglycine, EDTA, F", PO4"3, reducing agents (ascorbic acid), tartrate, tiron
Co Citrate, CN", diethyldithiocarbamate, 2,3-dimercaptopropanol, dimethylglyoxime, ethylenediamine, EDTA,
F', glycine, H2O2, NH3, NO2, 1,10-phenanthroline, Na5P3O10, SCN", S2O3"2, tartrate
Cr Acetate, (reduction with) ascorbic acid + KI, citrate, AfjV-dihydroxyethylglycine, EDTA, F", formate,
NaOH + H2O2, oxidation to CrO4"2,Na5P3O10, sulfosalicylate, tartrate, triethylamine, tiron
Cu Ascorbic acid + KI, citrate, CN", diethyldithiocarbamate, 2,3-dimercaptopropanol, ethylenediamine, EDTA
glycine, hexacyanocobalt(III)(3-), hydrazine, I", NaH2PO2, NH2OHHC1, NH3, NO"2, 1,10-phenanthroline,
S"2, SCN" + SO3"2, sulfosalicylate, tartrate, thioglycolic acid, thiosemicarbazide, thiocarbohydrazide,
thiourea
Fe Acetylacetone, (reduction with) ascorbic acid, C2O4"2, citrate, CN" 2,3-dimercaptopropanol, EDTA, F",
NH3, NH2OHHC1, OH", oxine 1,10-phenanthroline, 2,2'-bipyridyl, PO4"3, P2O7"4, S"2, SCN", SnCl2, S2O3"2,
sulfamic acid, sulfosalicylate, tartrate, thioglycolic acid, thiourea, tiron, triethanolamine, trithiocarbonate
Ga Citrate, Cl", EDTA, OH", oxalate, sulfosalicylate, tartrate
Ge F", oxalate, tartrate
Hf See Zr
Hg Acetone, (reduction with) ascorbic acid, citrate, Cl", CN", 2,3-dimercaptopropan-l-ol, EDTA, formate, I",
SCN", SO3"2, tartrate, thiosemicarbazide, thiourea, triethanolamine
In Cl", EDTA, F", SCN", tartrate thiourea, triethanolamine
Ir Citrate, CN", SCN", tartrate, thiourea
La Citrate, EDTA, F", oxalate, tartrate, tiron
JULY 2001
DRAFT FOR PUBLIC COMMENT
14-117
MARLAP
DO NOT CITE OR QUOTE
-------
Separation Techniques
VIetal Masking Agent
Mg
Mn
Vlo
Nb
Nd
NH4+
Ni
Np
Os
a
Pb
Rare
Re
Rh
Ru
Sb
Sc
Se
Sn
Ta
Te
Th
Ti
n
LJ
V
Citrate, C2O4"2, cyclohexane-l,2-diaminetetraacetic acid, yV,yV-dihydroxyethylglycine, EDTA, F", glycol,
hexametaphosphate, OH", P2O7"4, triethanolamine
Citrate, CN", C2O42, 2,3-dimercaptopropanol, EDTA, F, Na5P3O10, oxidation to MnO4", P2O7"4, reduction to
Mn(II) with NH2OH HC1 or hydrazine, sulfosalicylate, tartrate, triethanolamine, triphosphate, tiron
Acetylacetone, ascorbic acid, citrate, C2O4"2, EDTA, F", H2O2, hydrazine, mannitol, Na5P3O10, NH2OHHC1,
oxidation to molybdate, SCN", tartrate, tiron, triphosphate
Citrate, C2O4 2, F, H2O2, OH', tartrate
EDTA
HCHO
Citrate, CN", A^A^-dihydroxyethylglycine, dimethylglyoxime, EDTA, F", glycine, malonate, Na5P3O10,
NH3 1,10-phenanthroline, SCN", sulfosalicylate, thioglycolic acid, triethanolamine, tartrate
F
CN-, SCN-, thiourea
H202
Acetate, (C6H5)4AsCl, citrate, 2,3-dimercaptopropanol, EDTA, I", Na5P3O10, SO4"2, S2O3"2, tartrate, tiron,
tetraphenylarsonium chloride, triethanolamine, thioglycolic acid
Acetylacetone, citrate, CN", EDTA, I", NH3, NO2, SCN", S2O3"2, tartrate, triethanol-amine
Citrate, CN", EDTA, I", NH3, NO2", SCN", S2O3'2, tartrate, urea
Reduction to Pu(IV) with sulfamic acid
C2O4"2, citrate, EDTA, F, tartrate
Earths
Oxidation to perrhenate
Citrate, tartrate, thiourea
CN", thiourea
Citrate, 2,3-dimercaptopropanol, EDTA, I", OH", oxalate, S"2, S2"2, S2O3"2, tartrate, triethanolamine
Cyclohexane-l,2-diaminetetraacetic acid, F", tartrate
Citrate, F", I", reducing agents, S"2, SO3"2, tartrate
Citrate, C2O3"2, 2,3-dimercaptopropanol, EDTA, F", I", OH", oxidation with bromine water, PO4"3, tartrate,
triethanolamine, thioglycolic acid
Citrate, F", H2O2, OH", oxalate, tartrate
Citrate, F", I", reducing agents, S"2, sulfite, tartrate
Acetate, acetylacetone, citrate, EDTA, F", SO4"2, 4-sulfobenzenearsonic acid, sulfosalicylic acid, tartrate,
triethanolamine
Ascorbic acid, citrate, F", gluconate, H2O2, mannitol, Na5P3O10, OH", SO4"2, sulfosalicylic, acid, tartrate,
triethanolamine, tiron
Citrate, Cl", CN", EDTA, HCHO, hydrazine, NH2OHHC1, oxalate, tartrate, triethanolamine
Citrate, (NH4)2CO3, C2O4"2, EDTA, F", H2O2, hydrazine + triethanolamine, PO4"3, tartrate
(reduction with) Ascorbic acid, hydrazine, or NH2OH HC1, CN", EDTA, H2O2, mannitol, oxidation to
vanadate, triethanolamine, tiron
MARLAP
DO NOT CITE OR QUOTE
14-118
JULY 2001
DRAFT FOR PUBLIC COMMENT
-------
Separation Techniques
VIetal Masking Agent
W Citrate, F~, H2O2, hydrazine, Na5P3O10, NH2OHHC1, oxalate, SCN", tartrate, tiron, triphosphate, oxidation
to tungstate
Cyclohexane-l,2-diaminetetraacetic acid, F"
Citrate, CN", A^A^-dihydroxyethylglycine, 2,3-dimercaptopropanol, dithizone, EDTA, F", glycerol, glycol,
hexacyanoferrate(II)(4-), Na5P3O10, NH3, OH", SCN", tartrate, triethanolamine
Arsenazo, carbonate, citrate, C2O"2, cyclohexane-l,2-diaminetetraacetic acid, EDTA, F", H2O2, PO4"3, P2O7"
4, pyrogallol, quinalizarinesulfonic acid, salicylate, SO4"2 + H2O2, sulfosalicylate, tartrate, triethanolamine
Cnmnilefl frnm Pen-in C1979 nn 609-611 ^ and Dean (1995 nn 7)
TABLE 14.18 — Masking agents for anions and neutral molecules
Anion or Neutral
Molecule Masking Agent
Boric Acid
Bf
Br2
BrO3"
Chromate(VI)
Citrate
cr
C12
cio3-
cio4-
CN-
EDTA
F-
Fe(CN)3-3
Germanic Acid
r
I2
io3-
io4-
MnO4-
MoO4'2
NO2
Oxalate
Phosphate
S
s-2
Sulfate
Sulfite
so6-2
Se and its anions
F", glycol, mannitol, tartrate, and other hydroxy acids
Hg(H)
Phenol, sulfosalicylic acid
Reduction with arsenate(III), hydrazine, sulfite, or thiosulfate
Reduction with arsenate(III), ascorbic acid, hydrazine, hydroxylamine, sulfite, or thiosulfate
Ca(II)
Hg(II), Sb(III)
Sulfite
Thiosulfate
Hydrazine, sulfite
HCHO, Hg(II), transition-metal ions
Cu(II)
Al (III), Be(II), boric acid, Fe(III), Th(IV), Ti(IV), Zr(IV)
Arsenate(III), ascorbic acid, hydrazine, hydroxylamine, thiosulfate
Glucose, glycerol, mannitol
Hg(H)
Thiosulfate
Hydrazine, sulfite, thiosulfate
Arsenate(III), hydrazine, molybdate(VI), sulfite, thiosurfate
Reduction with arsenate(III), ascorbic acid, azide, hydrazine, hydroxylamine, oxalic acid,
sulfite, or thiosulfate
Citrate, F, H2O2, oxalate, thiocyanate + Sn(II)
Co(II), sulfamic acid, sulfanilic acid, urea
Molybdate(VI), permanganate
Fe(III), tartrate
CN-, S2', suffite
Permanganate + sulfuric acid, surfur
Cr(III) + heat
HCHO, Hg(II), permanganate + sulfuric acid
Ascorbic acid, hydroxylamine, thiosulfate
JULY 2001
DRAFT FOR PUBLIC COMMENT
14-119
MARLAP
DO NOT CITE OR QUOTE
-------
Separation Techniques
3333
3334
3335
Anion or Neutral
Molecule Masking Agent
Tel" Diaminobenzidine, sulfide, sulfite
Tungstate Citrate, tartrate
Vanadate Tartrate
3336 (1) Compiled from Perrin( 1979, p. 612) and Dean (1995)
3337 Demasking refers to any procedure that eliminates the effect of a masking agent already present
3338 in solution. There are a variety of methods for demasking, including changing the pH of the
3339 solution and physically removing, destroying, or displacing the masking agent. The stability of
3340 most metal complexes depends on pH, so simply raising or lowering the pH is frequently
3341 sufficient for demasking. Another approach to demasking involves the formation of new
3342 complexes or compounds that are more stable than the masked species. For example, boric acid
3343 commonly is used to demask the fluoride complexes of Sn4+ or Mo6+, and hydroxide is used to
3344 demask the thiocyanate complexes of Fe3+. In addition, it might be possible to destroy the
3345 masking agent in solution through a chemical reaction (i.e., via the oxidation of EDTA in acidic
3346 solutions by permanganate or another strong oxidizing agent).
3347 14.10.9 Review of Specific Radionuclides
3348 14.10.9.1 Americium
3349 Americium is a metal of the actinide series which is produced synthetically by neutron activation
3350 of uranium or plutonium followed by beta decay.
3351 Isotopes
3352 Twenty isotopes of americium are known, 232Am through 248Am, including three metastable
3353 states. All isotopes are radioactive. 243Am and 241Am, alpha emitters, are the longest lived with a
3354 half-lives of 7,380 years and 432.7 years, respectively. 241Am and 243Am also undergo
3355 spontaneous fission. 242mAm has a half-life of 141 years, and the half-lives of the remaining
3356 isotopes are measured in hours, minutes, or seconds. 241Am is the most common isotope of
3357 environmental concern.
3358 Occurrence
3359 None of the isotopes of americium occur naturally. It is produced synthetically by neutron
3360 bombardment of 238U or 239Pu followed by beta decay of the unstable intermediates. 241Am is
MARLAP JULY 2001
DO NOT CITE OR QUOTE 14-120 DRAFT FOR PUBLIC COMMENT
-------
Separation Techniques
3361 found in military wastes and can be extracted from reactor wastes. Some industrial ionization
3362 sources also contain americium. Decay of 241Pu injected in the atmosphere during weapons
3363 testing contributes to the presence of 241Am.
3364 The silver metal is prepared by reduction of americium fluoride (AmF3) or americium oxide
3365 (AmO2) with active metals at high temperatures and is purified by fractional distillation, taking
3366 advantage of its exceptionally high vapor pressure compared to other transuranium elements.
3367 Kilogram quantities of 241Am are available, but only 10 to 100 g quantities of 243Am are prepared.
3368 Soft gamma emission from 241Am is used to measure the thickness of metal sheets and metal
3369 coatings, the degree of soil compaction, sediment concentration in streams, and to induce X-ray
3370 fluorescence in chemical analysis. As an alpha emitter, it is mixed with beryllium to produce a
3371 neutron source for oil-well logging and to measure water content in soils and industrial process
3372 streams. The alpha source is also used to eliminate static electricity and as an ionization source in
3373 smoke detectors.
3374 Solubility of Compounds
3375 Among the soluble salts are the nitrate, halides, sulfate, and chlorate of americium(in). The
3376 fluoride, hydroxide, and oxalate are insoluble. The phosphate and iodate are moderately soluble
3377 in acid solution. Americium(VI) is precipitated with sodium acetate to produce the hydrate,
3378 NaAmO2(C2H3O2)3-xH2O.
3379 Review of Properties
3380 The study of the properties of americium is very difficult because of the intense alpha radiation
3381 emitted by 241Am and 243Am, but some properties are known. Americium metal is very ductile
3382 and malleable but highly reactive and unstable in air, forming the oxide. It is considered to be a
3383 slightly more active metal than plutonium and is highly reactive combing directly with oxygen,
3384 hydrogen, and halides to form the respective compounds, AmO2, AmH3, and AX3. Alloys of
3385 americium with platinum, palladium, and iridium have been prepared by hydrogen reduction of
3386 americium oxide in the presence of the finely divided metals.
3387 Unless the transuranium elements are associated with high-level gamma emission, the principal
3388 toxicological problems associated with the radionuclides are the result of internal exposure after
3389 inhalation or ingestion. When inhaled or ingested, they are about equally distributed between
3390 bone tissue and the liver. At high doses transuranics lead to malignant tumors years later. In
3391 addition, large quantities of 241Am could conceivably lead to criticality problems, producing
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 14-121 DO NOT CITE OR QUOTE
-------
Separation Techniques
3392 external radiation hazards or neutron exposure from (a,n) reactions. 241Am is also a gamma
3393 emitter.
3394 Americium is generally thought to be absorbed by all common rocks at pH values found in the
3395 environment. Complexation of Am(ni) by naturally occurring ligands, however, would be
3396 expected to strongly reduce its adsorption.
3397 Solution Chemistry
3398 Americium can exist in solution in the +3, +4, +5, and +6 oxidation states. Simple aqueous ions
3399 of Am+3 and AmO2+2 (VI oxidation state) are stable in dilute acid, but Am+3 is the predominant
3400 oxidation state. Free radicals produced by radiolysis of water by alpha particles reduce the higher
3401 states spontaneously to Am+3. The +3 oxidation state exists as Am(OH)3 in alkaline solution.
3402 Simple tetravalent americium is unstable in mineral acid solutions, disproportionating rapidly to
3403 produce Am+3 and AmO2+1 [Am(V)] in nitric and perchloric acid solutions. Conversely,
3404 dissociation of Am(OH)4 or AmO2 [both Am(IV)] in sulfuric acid solutions produces solutions
3405 containing Am+3 and AmO2+2. Stability is provided by complexation with fluoride ions and
3406 oxygen-containing ligands such as carbonate and phosphate ions. The AmO2+1 ion also
3407 disproportionates in acid solutions to yield Am+3 and AmO2+2, but the process for 241Am is so
3408 slow that radiation-induced reduction dominates. Evidence exists for the presence of Am+7 in
3409 alkaline solutions from the oxidation of AmO2+2.
3410 OXIDATION-REDUCTION BEHAVIOR. Although disproportionation reactions convert the +4 and +5
3411 oxidation states into the +3 and +6 states, radiolysis eventually converts the higher oxidation
3412 state into Am+3. Redox processes are used, however, to produce solutions of alternate oxidation
3413 states and to equilibrate the forms of americium into a common state, usually +3, but sometimes
3414 +6.
3415 The +4 state is reduced to Am+3 by iodide. In dilute, non-reducing solutions, peroxydisulfate
3416 (S2O8"2) oxidizes both the +3 and +5 states to the +6 state. Ce+4 and ozone (O3) oxidizes the +5
3417 state to +6 in perchloric acid solution. Electrolytic oxidation of Am+3 to AmO2+2 occurs in
3418 phosphoric, nitric, and perchloric acid solutions and solutions of sodium bicarbonate (Na2CO3).
3419 The latter ion is reduced to Am+3 by iodide, hydrogen peroxide, and the nitrite ion (NO/).
3420 COMPLEXATION. The +3 oxidation state forms complexes in the following order of strength (in
3421 aqueous solution): F" > H2PO4" > SCN" > NO3" > Cl". Both americium (+3) and (IV) form
3422 complexes with organic chelants. These are stable in aqueous and organic solvents. Americium
MARLAP JULY 2001
DO NOT CITE OR QUOTE 14-122 DRAFT FOR PUBLIC COMMENT
-------
Separation Techniques
3423 (IV) however can be easily reduced unless special oxidizing conditions are maintained." The
3424 AmO2+2 ion also forms significant complex ions with nitrate, sulfate, and fluoride ions.
3425 HYDROLYSIS. The actinide elements are known for their tendency to hydrolyze and, in many
3426 cases, form insoluble polymers. In the predominant +3 oxidation state in solution, americium,
3427 with its large radius, has the least tendency of the +3 actinides to hydrolyze; yet, hydrolysis is
3428 expected to occur with some polymerization. Hydrolysis that does occur is complicated and
3429 depends on the nature of the cations present and may start at pH values as low as 0.5-1.0. In
3430 contrast, the AmO2+2, like all actinyl ions, undergoes hydrolysis to an appreciable extent. The
3431 tendency to form polymers of colloidal dimensions, however, appears to be small relative to
3432 other actinide ions in the +6 oxidation state. Precipitation occurs early on after relatively small
3433 polymeric aggregates form in solution. The strong tendency to form insoluble precipitates after a
3434 small amount of hydrolysis makes characterization of the water-soluble polymers a difficult
3435 problem.
3436 RADIOCOLLOIDS. At trace concentrations, a colloidal form of Am+2 can easily be prepared, so
3437 steps should be taken to avoid its formation during analytical procedures. At high pH ranges,
3438 colloids form from the Am(OH)3, and at lower pH ranges through adsorption of Am+3 onto
3439 foreign particles. Their formation depends on storage time, pH, and ionic strength of the solution.
3440 Dissolution of Samples
3441 Americium is generally dissolved from irradiated reactor fuels, research compounds, and soil,
3442 vegetation, and biological samples. Spent fuel elements may be difficult to dissolve but
3443 eventually yield to digestion with hydrofluoric acid, nitric acid, or sulfuric acid. Aqua regia is
3444 used if platinum is present, and hydrochloric acid with an oxidizing agent such as sodium
3445 chlorate. Perchloric acid, while a good solvent for uranium, reacts too vigorously. Sodium
3446 hydroxide-peroxide is a good basic solvent. Research compounds, usually salts, yield to hot
3447 concentrated nitric or sulfuric acid. Soil samples are digested with concentrated nitric acid,
3448 hydrofluoric acid, or hydrochloric acid. Vegetation and biological samples are commonly wet
3449 ashed, and the residue is treated with nitric acid.
3450 Separation Methods
3451 The separation of americium, particularly from other transuranics, is facilitated by the
3452 exceptional stability of Am(in) compared to the trivalent ions of other actinides, which more
3453 readily convert to higher oxidation states under conditions that americium remains trivalent.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 14-123 DO NOT CITE OR QUOTE
-------
Separation Techniques
3454 PRECIPITATION AND COPRECIPITATION. Coprecipitation with lanthanum fluoride (LaF3) is
3455 achieved after reduction of higher oxidation states to Am(in). Select oxidation of other
3456 transuranic elements such as neptunium and plutonium to the IV or VI oxidation states
3457 solubilizes these radionuclides leaving americium in the insoluble form. Although coprecipita-
3458 tion with rare earths as fluorides or hydroxides from a bicarbonate solution of americium(VI), is
3459 used to purify americium, it is not as effective as ion-exchange procedures. Other coprecipitating
3460 agents for americium(in) include thorium oxalate [Th(C2O4)2], calcium oxalate (CaC2O4), ferric
3461 hydroxide [Fe(OH)3), and lanthanum potassium sulfate [LaK(SO4)2]. Americium(IV) is also
3462 coprecipitated with these reagents as well as with zirconium phosphate [Zr3(PO4)2].
3463 Americium(VI) is not coprecipitated with any of these reagents but with sodium uranyl acetate
3464 [NaUO2(C2H3O2)2].
3465 SOLVENT EXTRACTION. Organic solvents and chelating agents are available for separating
3466 americium from other radionuclides by selectively extracting either americium or the alternate
3467 radionuclide from aqueous solutions into an organic phase. Tributyl phosphate (TBP) in kerosene
3468 or thenoyltrifluoroacetone (TTA) in xylene removes most oxidation states of neptunium and
3469 plutonium from americium(in) in the presence of dilute nitric acid. The addition of sodium
3470 nitrate (6 M) tends to reverse the trend making americium more soluble in TBP than uranium,
3471 neptunium, or plutonium radionuclides. Di(2-ethylhexyl)phosphoric acid (HDEHP) in toluene is
3472 highly effective in extracting americium(ni) and is used in sample preparation for alpha
3473 spectroscopic analysis.
3474 Recently, solvent extraction chromatography has offered an efficient, easy technique for rapidly
3475 separating americium and other transuranic elements. A process using octylphenyl-N,N-
3476 diisobutyl carbamoylphoshpine oxide (CMPO) in dissolved TBP and fixed on an inert polymeric
3477 resin matrix has been used to isolate americium(in). The column is loaded with 2 M nitric acid,
3478 and americium is eluted with 4 M hydrochloric acid. It is important to note that iron, found in
3479 most environmental samples, does not effect the americium isolation if the iron is kept in the +2
3480 oxidation state. The ferric ion (Fe+3) is detrimental to the separation.
3481 ION EXCHANGE. Separation of americium can be achieved by cation-exchange chromatography.
3482 Any of its oxidation states absorb on a cation resin in dilute acid solution, but the higher
3483 oxidation states are not important in cation-exchange separations because they are unstable
3484 toward reduction to the +3 state. Generally, americium(in) is the last tripositive ion among the
3485 actinides eluted from a cation-exchange matrix, although the order may not be maintained under
3486 all conditions. Many eluting agents are available for specific separations. Concentrated
3487 hydrochloric acid, for example, has been used for separating actinides such as americium from
3488 the lanthanides. Anion-exchange chromatography has been widely used for separating
MARLAP JULY 2001
DO NOT CITE OR QUOTE 14-124 DRAFT FOR PUBLIC COMMENT
-------
Separation Techniques
3489 americium. Anionic complexes of americium(ni) form at high chloride concentrations, providing
3490 a chemical form that is easily exchanged on an anion-exchange column. The column can be
3491 eluted using dilute hydrochloric acid or a dilute hydrochloric acid/ammonium thiocyanate
3492 solution. Anion-exchange separations of americium are also realized with columns prepared with
3493 concentrated nitric acid solutions. The sequential separation of the actinides is accomplished
3494 readily using anion-exchange chromatography. Americium, plutonium, neptunium, thorium,
3495 protactinium, curium, and uranium can all be separated by the proper application of select acid or
3496 salt solutions to the column.
3497 ELECTRODEPOSITION. Americium can be electrodeposited for alpha spectrometry measurement
3498 on a highly-polished platinum cathode. The sample is dissolved in a dilute hydrochloric acid
3499 solution that has been adjusted to a pH of about six with ammonium hydroxide solution using
3500 methyl red indicator. The process runs for one hour at 1.2 amps.
3501 Methods of Analysis
3502 241Am is detected and quantified by either alpha counting or gamma spectroscopy. Trace
3503 quantities of 241Am are analyzed by alpha counting, after separation from interfering
3504 radionuclides by solvent extraction, coprecipitation, or ion-exchange chromatography. The
3505 isolated radionuclide is collected by coprecipitation, filtered, and mounted on a planchet or
3506 electroplated onto a platinum electrode for counting by alpha spectrometry. 243Am is added to the
3507 analytical solution as a tracer to measure chemical recovery. 241Am in bulk soil samples can be
3508 determined by gamma spectroscopy.
3509 Compiled from: Ahrland, 1986; Baes and Mesmer, 1976; Choppin et al., 1995; Considine
3510 and Considine, 1983; Cotton and Wilkinson, 1988; DOE, 1990 and 1997, 1995; 1997;
3511 Ehmann and Vance, 1991; Greenwood and Earnshaw, 1984; Haissinsky and Adolff, 1965;
3512 Katz et al., 1986; Lindsay, 1988; Metz and Waterbury, 1962; NEA, 1982; SCA, 2001;
3513 Penneman, 1994; Penneman and Keenan, 1960; Schulz and Penneman, 1986; Seaborg and
3514 Loveland, 1990; Horwitzetal, 1993.
3515 14.10.9.2 Cesium
3516 Cesium is the last member of the naturally occurring alkali metals in group IA of the periodic
3517 table with an atomic number of 55. As such, its radiochemistry is simplified because the Group
3518 IA metals form only +1 ions. Elemental cesium is a very soft, silver-white metallic solid in the
3519 pure state with a melting point of only 28.5 °C. It tarnishes quickly to a golden-yellow color
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 14-125 DO NOT CITE OR QUOTE
-------
Separation Techniques
3520 when exposed to small amounts of air. In larger amounts of air it ignites spontaneously. It is
3521 normally stored under xylente/toluene to prevent contact with air.
3522 Isotopes
3523 Cesium isotopes of mass number 112 to 148 have been identified. 133Cs is the only stable isotope.
3524 134Cs and 137Cs are the only two isotopes of significance from an environmental perspective. Both
3525 are formed from the nuclear fission process. Their half-lives are 2.06 and 30.17 years,
3526 respectively.
3527 Occurrence
3528 Cesium is widely distributed in the Earth's crust with other alkali metals. In granite and
3529 sedimentary rocks the concentration is less than 7 ppm. In seawater it is about 0.002 ppm, but in
3530 mineral springs the concentration may be greater than 9 mg/L. Cesium is found in complex
3531 minerals such as carnallite, a potassium and magnesium chloride mineral that contains small
3532 percentages of cesium compounds; lepidolite ores, a lithium aluminum silicate; and pollucite, a
3533 cesium-rich ore of the oxides of cesium, aluminum, and silicon. 137Cs is produced in nuclear
3534 fission and occurs in atmospheric debris from weapons tests and accidents. It is a very important
3535 component of radioactive fallout; and because of its moderately long half-life and high solubility,
3536 it is a major source of long-lived external gamma radiation from fallout. It accounts for 30
3537 percent of the gamma activity of fission products stored for one year, 70 percent in two years,
3538 and 100 percent after five years.
3539 Cesium metal is not produced on a commercial scale. It is isolated from its minerals, however, by
3540 acid extraction, fusion with alkaline fluxes, or direct reduction of an ore to metallic cesium.
3541 Extraction and fusion yield a cesium salt, which is treated by oxidation-reduction processes to
3542 make the pure metal. The salt is either roasted with carbon, heated with calcium or lithium, or
3543 electrolyzed as a melt to reduce the cation to pure cesium. Special equipment should be used in
3544 these processes because of the very reactive chemical nature of the metal.
3545 Metallic cesium is used in photoelectric cells, spectrographic instruments, scintillation counters,
3546 and other optical and detecting devices, sometimes alloyed with calcium, strontium, or barium to
3547 facilitate handling. Its most recognized use is in the atomic clock that serves to define the second.
3548 Cesium has been considered as a fuel in ion-propulsion engines for deep space travel and as a
3549 heat-transfer medium for some applications. 137Cs has replaced 60Co in the treatment of cancer
3550 and has been used in industrial radiography for the control of welds. Cesium compounds are used
3551 in glass and ceramic production, as an absorbent in carbon dioxide production plants, and in the
MARLAP JULY 2001
DO NOT CITE OR QUOTE 14-126 DRAFT FOR PUBLIC COMMENT
-------
Separation Techniques
3552 preparation of density gradients for the separation of macromolecules by centrifugation. 37Cs is
3553 also used commercially as a sealed source in liquid scintillation spectrometers. The 661 keV
3554 gamma ray it emits is used to create an electron (Compton effect) distribution which allows the
3555 degree of sample quench to be determined.
3556 Solubility of Compounds
3557 Most cesium salts are very soluble in water and dilute acids. Among the salts of common anions,
3558 the notable exceptions are cesium perchlorate and periodate (CsClO4 and CsIO4). Several cesium
3559 compounds of large anions are insoluble. Examples include the following: silicotungstate
3560 [Cs8SiW12O42], permanganate (CsMnO4), chloroplatinate (Cs2PtCl6), tetraphenylborate
3561 [CsB(C6H5)4], alum [CsAl(SO4)2], and cobaltnitrate complex [Cs3Co(NO3)6].
3562 Review of Properties
3563 Cesium is the most active and electropositive of all the metals. It forms compounds with most
3564 inorganic and organic anions; it readily forms alums with all the trivalent cations that are found
3565 in alums. The metal readily ionizes, and in ammonia solutions and it is a powerful reducing
3566 agent. When exposed to moist air, it tarnishes initially forming oxides and a nitride and then
3567 quickly melts or bursts into flame. With water the reaction is violent. Cesium reacts vigorously
3568 with halogens and oxygen, and it is exceptional among the alkali metals in that it can form stable
3569 polyhalides such as CsI3. Reaction with oxygen forms a mixture of oxides: cesium oxide (Cs2O),
3570 cesium peroxide (Cs2O2), and cesium superoxide (CsO2). The toxicity of cesium compounds is
3571 generally not important unless combined with another toxic ion.
3572 137Cs, introduced into the water environment as cations, is attached to soil particles and can be
3573 removed by erosion and runoff. However, soil sediment particles act as sinks for 137Cs, and the
3574 radionuclide is almost irreversible bound to mica and clay minerals in freshwater environments.
3575 It is unlikely that 137Cs will be removed from these sediments under typical environmental
3576 conditions. Solutions of high ionic strength as occur in estuarine environments might provide
3577 sufficient exchange character to cause cesium to become mobile in the ecosphere.
3578 Solution Chemistry
3579 The cesium ion exists in only the +1 oxidation state, and its solution chemistry is not complicated
3580 by oxidation-reduction reactions. As a result, it undergoes complete, rapid exchange with carriers
3581 in solution. The cesium ion is colorless in solution and is probably hydrated as a hexaaquo
3582 complex.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 14-127 DO NOT CITE OR QUOTE
-------
Separation Techniques
3583 COMPLEXATION. Cesium ions form very few complex ions in solution. The few that form are
3584 primarily with nitrogen-donor ligands or beta-diketones. Anhydrous beta-diketones are insoluble
3585 in water, but in the presence of additional coordinating agents, including water, they become
3586 soluble in hydrocarbons. One solvent-extraction procedure from aqueous solutions is based on
3587 chelation of cesium with 1,1,1 -trifluoro-3-(2 '-thenoyl)acetone (TTA) in a hydrocarbon solvents.
3588 Cesium is sandwiched between crown ligands, associated with the oxygen atoms of the ether, in
3589 [Cs9(18-C-6)14]+9.
3590 HYDROLYSIS. With the small charge and large radius of the cesium ion, hydrolysis reactions are
3591 inconsequential.
3592 ADSORPTION. When cesium is present in extremely low concentrations, even in the presence of 2
3593 M acid, adsorption on the walls of glass and plastic containers leads to complications for the
3594 radioanalyst. Half the activity of cesium radionuclides, for example, can be lost from acid
3595 solutions stored for one month in these containers. Experiments indicate that addition of 1 jig
3596 cesium carrier per mL of solution is sufficient to stabilize acid solutions for six months.
3597 Dissolution of Samples
3598 Radiochemists generally dissolve cesium samples from irradiated nuclear fuel, activated cesium
3599 salts, natural water, organic material, agriculture material, and soils. Nuclear fuel samples are
3600 generally dissolved in HC1, HNO3, HF, or a combination of these acids. Care should be taken to
3601 ensure that the sample is representative if 137Cs has been used as a burn-up monitor. Precautions
3602 should also be taken with these samples to prevent loss of cesium because of leaching or
3603 incomplete sample dissolution. Most cesium salts dissolve readily in water and acid solutions. In
3604 water samples, the cesium might require concentration, preferably by ion exchange, or by
3605 precipitation or coprecipitation if interfering ions are present. Organic materials are either
3606 decomposed by HNO3 or dry ashed, and the cesium is extracted with hot water or hot acid
3607 solution. Extraction and leaching procedure have been use to assess exchangeable or teachable
3608 cesium using ammonium acetate solutions or acid solutions, but soils are generally completely
3609 solubilized in HNO3, HC1, HF, H2SO4, or a mixture of these acids in order to account for all the
3610 cesium in a soil sample.
3611 Separation Methods
3612 PRECIPITATION AND COPRECIPITATION. Cesium is separated and purified by several precipitation
3613 and coprecipitation methods using salts of large anions. Gravimetric procedures rely on
3614 precipitation to collect cesium for weighing, and several radiochemical techniques isolate cesium
MARLAP JULY 2001
DO NOT CITE OR QUOTE 14-128 DRAFT FOR PUBLIC COMMENT
-------
Separation Techniques
3615 radionuclides for counting by precipitation or coprecipitation. Cesium can be precipitated, or
3616 coprecipitated in the presence of cesium carrier, by the chlorate, cobaltinitrate, platinate, and
3617 tetraphenylborate ions. Other alkali metals interfere and should be removed before a pure
3618 insoluble compound can be collected. Cesium can be isolated from other alkali metals by
3619 precipitation as the silicotungstate. The precipitate can be dissolved in 6 M sodium hydroxide,
3620 and cesium can be further processed by other separation procedures. The tetraphenylb orate
3621 procedure first removes other interfering ions by a carbonate and hydroxide precipitation in the
3622 presence of iron, barium, lanthanum, and zirconium carriers. Cesium is subsequently precipitated
3623 by the addition of sodium tetraphenylb orate to the acidified supernatant. Alum also precipitates
3624 cesium from water samples in the presence of macro quantities of the alkali metals. Trace
3625 quantities of cesium radionuclides are precipitated using stable cesium as a carrier.
3626 ION EXCHANGE. The cesium cation is not retained by anion-exchange resins and does not form a
3627 suitable anion for anion-exchange chromatography. The process is used, however, to separate
3628 cesium from interfering ions that form anionic complexes. Cesium elutes first in these
3629 procedures. Cesium is retained by cation-exchange resins. Because the cesium ion has the largest
3630 ionic radius and has a +1 charge, it is less hydrated than most other cations. Therefore, cesium
3631 has a small hydrated radius and can approach the cation exchange site to form a strong
3632 electrostatic association with the ion-exchange resin. Binding of alkali metal ion to cation
3633 exchange resins follows the order: Cs+1>Rb+1>K+1>Na+1>Li+1. Cesium is generally the last alkali
3634 metal ion to elute in cation-exchange procedures. In some procedures, the process is not
3635 quantitative after extensive elution.
3636 SOLVENT EXTRACTION. Cesium does not form many complex ions, and solvent extraction is not
3637 a common procedure for its separation. One solvent-extraction procedure, however, is based on
3638 chelation of cesium with l,l,l-trifluoro-3-(2'-thenoyl)acetone (TTA) in a solvent of methyl
3639 nitrate/hydrocarbons. Cesium can also be extracted from fission product solutions with sodium
3640 tetraphenylb orate in amyl acetate. It can be stripped from the organic phase by 3 M HC1.
3641 Methods of Analysis
3642 Macroscopic quantities of cesium have been determined by gravimetric procedures using one of
3643 the precipitating agents described above. Spectrochemical procedures for macroscopic quantities
3644 include flame photometry, emission spectroscopy, and X-ray emission.
3645 Gamma ray spectrometry allows detection of 134Cs, 136Cs, and 137Cs down to very low levels. The
3646 gamma ray measured for 137Cs (661 Kev) actually is emitted from it progeny 136mBa. However,
3647 since the half-life of the barium isotope is so short (2.5 min) it is quickly equilibrated with its
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 14-129 DO NOT CITE OR QUOTE
-------
Separation Techniques
3648 parent cesium isotope (i.e., secular equilibrium). 137Cs is used as part of a group of nuclides in a
3649 mixed radioactivity source for calibration of gamma ray spectrometers.
3650 Compiled from: Choppin et al., 1995; Considine and Considine, 1983; Cotton and
3651 Wilkinson, 1988; Emsley, 1989; EPA, 1973; EPA, 1973; EPA, 1980; Finston and Kinsley,
3652 1961; Friedlander et al., 1981; Hampel, 1968; Hassinsky and Adolff, 1965; Kallmann, 1964;
3653 Lindsay, 1988; Sittig, 1994.
3654 14.10.9.3 Cobalt
3655 Cobalt, atomic number 27, is a silvery-grey, brittle metal found in the first row of the transition
3656 elements in the periodic table, between iron and nickel. Although it is in the same family of
3657 elements as rhodium and iridium, it resembles iron and nickel in its free and combined states.
3658 Isotopes
3659 59Co is the only naturally occurring isotope of the element. The other twenty-two isotopes and
3660 their metastable states, ranging from mass numbers 50 to 67, are radioactive. Isotopes with mass
3661 numbers less than 59 decay by positron emission or electron capture. Isotopes with mass
3662 numbers greater than 59 decay by beta and gamma emission. Except for 60Co, the most important
3663 radionuclide, their half-lives range from milliseconds to days. The principle isotopes of cobalt
3664 (with their half-lives) are "Co (272 d), 58Co (71 d), and 60Co (5.27 y). Isotopes 57 and 58 can be
3665 determined by X-ray as well as gamma spectrometry. Isotope 60 is easily determined by gamma
3666 spectrometry.
3667 Occurrence and Uses
3668 The cobalt content of the crust of the earth is about 30 ppm, but the element is widely distributed
3669 in nature, found in soils, water, plants and animals, meteorites, stars, and lunar rocks. Over 200
3670 cobalt minerals are known. Commercially, the most important are the arsenides, oxides, and
3671 sulfides. Important commercial sources also include ores of iron, nickel, copper, silver,
3672 manganese, and zinc. 60Co is produced by neutron activation of stable 59Co. 56Co and "Co are
3673 prepared by bombardment of iron or nickel with protons or deuterons.
3674 Some of the metallic cobalt is isolated from its minerals, but much of the metal is produced
3675 primarily as a byproduct of copper, nickel, or lead extraction. The processes are varied and
3676 complicated because of the similar chemical nature of cobalt and the associated metals.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 14-130 DRAFT FOR PUBLIC COMMENT
-------
Separation Techniques
3677 Since ancient times cobalt ores has been used to produce the blue color in pottery, glass, and
3678 ceramics. Cobalt compounds are similarly used as artist pigments, inks, cotton dyes, and to speed
3679 the drying of paints and inks. They also serves as catalysts in the chemical industry and for
3680 oxidation of carbon monoxide in catalytic converters. One of the major uses of cobalt is the
3681 preparation of high-temperature or magnetic alloys. Jet engines and gas turbines are
3682 manufactured from metals with a high content of cobalt (up to 65 percent) alloyed with nickel,
3683 chromium, molybdenum, tungsten, and other metals.
3684 Little use if made of pure cobalt except as a source of radioactivity from 60Co. The radionuclide
3685 is used in cancer radiotherapy, as a high-energy gamma source for the radiography of metallic
3686 objects, fluids, and other solids, or as an injectable radionuclide for the measurement of flow
3687 rates in pipes.
3688 Solubility of Compounds
3689 Most simple cobalt compounds contain cobalt (U), but cobalt (U) and cobalt(UI) display varied
3690 solubilities in water. To some extent, their solubilities depend on the oxidation state of the metal.
3691 For example, all the halides of cobalt (II) are soluble but the only stable halide of cobalt (HI), the
3692 fluoride, is insoluble. The sulfates of both oxidation states are soluble in water. The acetate of
3693 cobalt (II) is soluble, but that of cobalt (HI) hydrolyses in water. The bromate, chlorate, and
3694 perchlorate of cobalt (II) are also soluble. Insoluble compounds include all the oxides of both
3695 oxidation states, cobalt (U) sulfide, cyanide, oxalate, chromate, and carbonate. The hydroxides
3696 are slightly soluble. Several thousand complex compounds of cobalt are known. Almost all are
3697 cobalt (in) complexes and many are soluble in water.
3698 Review of Properties
3699 Metallic cobalt is less reactive than iron and is unreactive with water or oxygen in air unless
3700 heated, although the finely divided metal is pyrophoric in air. On heating in air it forms the
3701 oxides, cobalt (II) oxide (CoO) below 200 °C and above 900 °C and cobalt (II)-cobalt (in) oxide
3702 (Co3O4) between the temperatures. It reacts with common mineral acids and slowly with
3703 hydrofluoric and phosphoric acids to form cobalt (H) salts and with sodium and ammonium
3704 hydroxides. On heating, it reacts with halogens and other nonmetals such as boron, carbon,
3705 phosphorus, arsenic, antimony, and sulfur.
3706 Cobalt exists in all oxidation states from -1 to +4. The most common are the +2 and +3 oxidation
3707 states. The +1 state is found in a several complex compounds, primarily the nitrosyl and carbonyl
3708 complexes and certain organic complexes. The +4 state exist in some fluoride complexes.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 14-131 DO NOT CITE OR QUOTE
-------
Separation Techniques
3709 Cobalt(II) is more stable in simple compounds and is not easily hydrolyzed. Few simple
3710 compounds are known for the +3 state, but cobalt is unique in the numerous stable complex
3711 compounds it forms.
3712 The toxicity of cobalt is not comparable to metals such as mercury, cadmium, or lead. Inhalation
3713 of fine metallic dust can cause irritation of the respiratory system, and cobalt salts can cause
3714 benign dermatosis. 60Co is made available in various forms, in sealed aluminum or monel
3715 cylinders for industrial applications, as wires or needles for medical treatment, and in various
3716 solid and solution forms for industry and research. Extreme care is required in handling any of
3717 these forms of cobalt because of the high-energy gamma radiation from the source.
3718 Solution Chemistry
3719 In aqueous solution and in the absence of complexing agents, cobalt (n) is the only stable
3720 oxidation state, existing in water as the pink-red hexaaquo complex ion, Co(H2O)6+2. Simple
3721 cobalt ions in the +3 oxidation state decompose water in an oxidization-reduction process that
3722 generates cobalt (II):
3723 4 Co+3 + 2 H2O - 4 Co+2 + O2 + 4 H+1
3724 Complexation of cobalt (HI) decreases its oxidizing power and most complex ions of the +3
3725 oxidation state are stable in solution.
3726 COMPLEXATION. Several thousand complexes of cobalt have been prepared and extensively
3727 studied, including neutral structures and those containing complex cations and/or anions. Among
3728 these, the cobalt (ID) complexes are the strongest and represent one of the largest groups of
3729 complex compounds. The most common cobalt (IE) compounds contain six ligands bonded to
3730 the metal atom or cation (coordination number six) in an octahedral arrangement. It forms many
3731 complex ions with nitrogen-compounds such as ammonia and amines ([Co(NH3)6]Cl3) by
3732 coordinating through the nitrogen atom, and with those containing carbon (K3[Co(CN)6]), oxygen
3733 and sulfur ([Co(H2O)6]Cl3), and halides (Na3[CoF6]). Complex compounds with mixed ligands
3734 are common: [Co(NH3)5(H2O)]Cl3 and [Co(NH3)3Cl3].
3735 The +2 oxidation state forms complexes with a coordination of four or six, and in aqueous
3736 solution, [Co(H2O)6]+2 is in equilibrium with some [Co(H2O)4]+2. In alkaline solution Co+2
3737 precipitates as Co(OH)2, but the ion is amphoteric; and in concentrated hydroxide solutions, the
3738 precipitate dissolves forming [Co(OH)4]"2. Many complexes of the form [Co(X)4]4 exist with
3739 monodentate anionic ligands such as Cl"1, Br"1,1"1, SCN"1, N34, and OH"1. Many aquo-halo
MARLAP JULY 2001
DO NOT CITE OR QUOTE 14-132 DRAFT FOR PUBLIC COMMENT
-------
Separation Techniques
3740 complexes are known; they are various shades of red and blue. The aquo complex, [Co(H2O)6]+2,
3741 is pink.
3742 Chelate complexes are well-known and are used to extract cobalt from solutions of other ions.
3743 Acetylacetone (acac) is used, for example, in a procedure to separate cobalt from nickel. Co+2 and
3744 Ni+2 do not form chelates with the acac, Co+3 does, however, and can be easily extracted.
3745 OXIDATION-REDUCTION BEHAVIOR. Most simple cobalt +3 compounds are unstable because the
3746 +3 state is a strong oxidizing agent. It is very unstable in aqueous media, rapidly reducing to the
3747 +2 state at room temperature. The aqueous ion of cobalt(II), [Co(H2O)6]+2, can be oxidized,
3748 however, to the +3 state either by electrolysis or by ozone (O3) in cold perchloric acid (HC1O4);
3749 solutions at 0 °C have a half-life of about one week. Compounds of the cobalt(in) complex ions
3750 are formed by oxidizing the +2 ion in solution with oxygen or hydrogen peroxide (H2O2) in the
3751 presence of ligands. The cobalt(in) hexamine complex forms according to:
3752 4 CoX2 + 4 NH4X + 20 NH3 + O2 ^ 4 [Co(NH3)6]X3 + 2 H2O
3753 HYDROLYSIS. The hydrolysis of the +2 oxidation state of cobalt is not significant in aqueous
3754 media below pH 7. At pH 7, hydrolysis of 0.001 M solution of the cation begins and is
3755 significant at a pH above 9. The hydrolysis of the +3 oxidation state is reminiscent of the
3756 hydrolysis of iron (HI), but it is not as extensive. Hydrolysis of cobalt (in) is significant at pH 5.
3757 In contrast, the hydrolysis of iron (ID) becomes significant at a pH of about 3.
3758 Dissolution of Samples
3759 Cobalt minerals, ores, metals, and alloys can be dissolved by treatment first with hydrochloric
3760 acid, followed by nitric acid. The insoluble residue remaining after application of this process is
3761 fused with potassium pyrosulfate and sodium carbonate. In extreme cases, sodium peroxide
3762 fusion is used. Biological samples are dissolved by wet ashing, digesting with heating in a
3763 sulfuric-perchloric-nitric acid mixture.
3764 Separation Methods
3765 PRECIPITATION AND COPRECIPITATION. Cobalt can be precipitated by hydrogen sulfide (H2S),
3766 ammonium sulfide (NH4S), basic acetate (C2H3O2"1/HO"1), barium carbonate (BaCO3), zinc oxide
3767 (ZnO), potassium hydroxide and bromine (KOH/Br2), ether and hydrochloric acid [(C2H5)2O and
3768 HC1], and cupferron. Cobalt sulfide (CoS) is coprecipitated with stannic sulfide (SnS2) when
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 14-133 DO NOT CITE OR QUOTE
-------
Separation Techniques
3769 low-solubility sulfides are precipitated in mineral acids. Care should be taken to avoid
3770 coprecipitation of zinc sulfide (ZnS).
3771 Cobalt can be separated from other metals by hydroxide precipitation using pH control to
3772 selectively precipitate metals such as chromium, zinc, uranium, aluminum, tin, iron (+3),
3773 zirconium, and titanium at low pH. Cobalt precipitates at pH 6.8, and magnesium, mercury,
3774 manganese, and silver at a pH greater than 7. Cobalt is not be separated from metals such as iron,
3775 aluminum, titanium, zirconium, thorium, copper, and nickel using ammonium hydroxide
3776 (NH4OH) solutions (aqueous ammonia), because an appreciable amount of cobalt is retained by
3777 the hydroxide precipitates of these metals produced using this precipitating agent. Various
3778 precipitating agents can be used to remove interfering ions prior to precipitating cobalt: iron by
3779 precipitating with sodium phosphate (Na3PO4) or iron, aluminum, titanium, and zirconium with
3780 zinc oxide.
3781 The separation of cobalt from interfering ions can be achieved by the quantitative precipitation of
3782 cobalt with excess potassium nitrite (KNO2) to produce K3[Co(NO2)6] (caution — unstable to
3783 heating after standing for some time). Ignition can be used to collect the cobalt as its mixed oxide
3784 (Co3O4). Cobalt can also be precipitated with a-nitroso-p-napthol (l-nitroso-2-napthol) to
3785 separate it form interfering metals. Nickel can interfere with this precipitation, but can be
3786 removed with dimethylglyoxime. Precipitation as mercury tetracyanatocobaltate (II)
3787 (Hg[Co(SCN)4]} also is used, particularly for gravimetric analysis, and precipitation with
3788 pyridine in thiocyanate solution is a quick gravimetric product, [Co(C5H5N)4](SCN)2.
3789 SOLVENT EXTRACTION. Various ions or chelates have been used in solvent extraction systems to
3790 isolate cobalt from other metals. Separation has been achieved by extracting either cobalt itself
3791 or, conversely, extracting contaminating ions into an organic solvent in the presence of
3792 hydrofluoric acid (HF), hydrochloric acid, and calcium chloride (HCl/CaCl2), hydrobromic acid
3793 (HBr), hydroiodic acid (HI), or ammonium thiocyanate (NH4SCN). For example, cobalt (II) has
3794 been separated from nickel (II) by extracting a hydrochloric acid solution containing calcium
3795 chloride with 2-octanol. The ion is not extracted by diethyl ether from hydrobromic acid
3796 solutions, but it is extracted from ammonium thiocyanate solutions by oxygen-containing organic
3797 solvents in the presence of iron (HI) by first masking the iron with citrate.
3798 Several chelate compounds have been used to extract cobalt from aqueous solutions.
3799 Acetylacetone (acac) forms a chelate with cobalt (IE), but not cobalt (II), that is soluble in
3800 chloroform at pH 6 to 9, permitting separation from several metals including nickel. Cobalt (II)
3801 can be oxidized to cobalt (in) with hydrogen peroxide (H2O2) prior to extraction, a-nitroso-p-
3802 napthol has also been used as a chelating agent in the separation of cobalt (in) by solvent
MARLAP JULY 2001
DO NOT CITE OR QUOTE 14-134 DRAFT FOR PUBLIC COMMENT
-------
Separation Techniques
3803 extraction. Diphenylthiocarbazone (dithiozone) has been used at pH 8 to extract cobalt into
3804 carbon tetrachloride and chloroform after metals that form dithiozonates in acid solution (pH 3-
3805 4) have been removed. 8-quinolinol has been used in a similar manner at pH up to 10. Masking
3806 agents added to the system impede the extraction of iron, copper, and nickel.
3807 ION-EXCHANGE CHROMATOGRAPHY. Anion-exchange resins have been used extensively to
3808 separate cobalt from other metals. The chloro-metal complexes, prepared and added to columns
3809 in molar hydrochloric acid solutions, are eluted at varying concentrations of hydrochloric acid.
3810 Trace amounts of 59Fe, 60Co, and 65Zn and their respective carriers have been separated from
3811 neutron-irradiated biological tissue ash with a chloride system. 60Co has been eluted carrier-free
3812 from similar samples and columns prepared with hydrobromic acid. Cobalt and contaminated
3813 metals in nitric-acid systems behave in a manner similar to hydrochloric-acid systems. Cobalt
3814 (n)-cyanide and cyanate complexes have been used to separate cobalt from nickel. The basic
3815 form of quaternary amine resins (the neutral amine form) has been used in the column
3816 chromatography of cobalt. Both chloride- and nitrate-ion systems have resulted in the association
3817 of cobalt as a complex containing chloride or nitrate ligands as well as the neutral (basic)
3818 nitrogen atom of the amine resin. Resins incorporating chelates in their matrix system have been
3819 used to isolate cobalt. 8-quinolinol resins are very effective in separating cobalt from copper.
3820 ABSORBENT CHROMATOGRAPHY. Several inorganic adsorbents such as alumina, clays, and silica
3821 are used to separate cobalt. Complex ions of cobaltamines separate on alumina as well as cobalt
3822 (II) complexes of tartaric acid and dioxane. A complex of nitroso-R-salts are absorbed onto an
3823 alumina column while other metals pass through the column. Cobalt is eluted with sulfuric acid.
3824 Cobalt dithizonates absorb on alumina from carbon tetrachloride solutions. Cobalt is eluted with
3825 acetone. The separation of cobalt from iron and copper has been achieved on aluminum
3826 hydroxide [A1(OH)3]. Clay materials, kalolinite, benotite, and montmorilloite, separate cobalt (n)
3827 from copper (n). Copper (II) absorbs and cobalt (n) elutes with water. Silica gel and activated
3828 silica have both been used as adsorbents in cobalt chromatography.
3829 Organic adsorbents such as 8-hydroxyquinoline and dimethylgloxime have been used in cobalt-
3830 absorbent chromatographic systems. Powdered 8-hydroxyquinoline separates cobalt (II) from
3831 other cations and anions, for example, and dimethylglyoxime separates cobalt from nickel.
3832 Cobalt-cyano complexes absorb on activated charcoal, and cobalt is eluted from the column
3833 while the anionic complexes of metals such as iron, mercury, copper, and cadmium remain on
3834 the column.
3835 Numerous paper chromatograph systems employing inorganic or chelating ligands in water or
3836 organic solvents are available to separate cobalt from other metals. In one system, carrier-free
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 14-135 DO NOT CITE OR QUOTE
-------
Separation Techniques
3837 60Co and 59Fe from an irradiated manganese target were separated with an acetone-hydrochloric
3838 solvent.
3839 ELECTRODEPOSTION. Most electroanalytical methods for cobalt are preceded by isolating the
3840 cobalt from interfering ions by precipitation or ion exchange. The electrolyte is usually an
3841 ammonia solution that produces the hexamine complex of cobalt (II), Co(NH3)6+2 in solution.
3842 Reducing agents such as hydrazine sulfate are added to prevent anodic deposits of cobalt and the
3843 oxidation of the cobalt (Il)-amine ion. Cobalt and nickel can be separated electrolytically by
3844 using an aqueous solution of pyridine with hydrazine to depolarize the platinum anode. The
3845 nickel is deposited first, and the voltage is increased to deposit cobalt.
3846 Methods of Analysis
3847 57Co, 58Co, and 60Co maybe concentrated from solution by coprecipitation and determined by
3848 gamma-ray spectrometry. 60Co is most commonly produced by the neutron activation of 59Co, in
3849 a reactor or an accelerator. 58Co is most commonly produced from the following reaction in
3850 nuclear reactors, 58Ni(n,p)58Co, due to the presence of nickel bearing alloys which undergo
3851 corrosion and are transported through the reactor core. 58Co is the most significant contributor to
3852 the gamma ray induced radiation fields in these facilities. "Co can be produced by either of the
3853 following, 58Ni(n,d)57Co [reactor] or 56Fe(d,n)57Co [accelerator], 57Co and 60Co are frequently
3854 used as part of a mixed radionuclide source for calibration of gamma ray spectrometers.
3855 Compiled from: Baes and Mesmer, 1976; Bate and Leddicotte, 1961; Cotton and Wilkinson,
3856 1988; Dale and Banks, 1962; EPA, 1973; Greenwood and Earnshaw, 1984; Haissinsky and
3857 Adloff, 1965; Hillebrand et al., 1980; Larsen, 1965; Latimer, 1952; Lingane, 1966.
3858 14.10.9.4 Iodine
3859 Iodine is a nonmetal, the last naturally occurring member of the halogen series, with an atomic
3860 number of 53. In the elemental form it is a diatomic molecule, I2, but it commonly exists in one
3861 of four nonzero oxidation states: -1 with metal ions or hydrogen; and +1, +5, and +7 with other
3862 nonmetals, often oxygen. Numerous inorganic and organic compounds of iodine exist, exhibiting
3863 the multiple oxidation states and wide range of physical and chemical properties of the element
3864 and its compounds. Existence of multiple oxidation states and the relative ease of changing
3865 between the -1, 0, and +5 state allows readily available methods for separation and purification of
3866 radionuclides of iodine in radiochemical procedures.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 14-136 DRAFT FOR PUBLIC COMMENT
-------
Separation Techniques
3867 Isotopes
3868 There are 42 known isotopes of iodine, including seven metastable states. The mass numbers
3869 range from 108 to 142. The only stable isotope is naturally occurring 127I. The half-lives of the
3870 radionuclides range from milliseconds to days with the single exception of long-lived 129I
3871 (t1/2=l-57 x 107 y). Iodine radionuclides with lower mass numbers decay primarily by electron
3872 capture. The higher mass number are, for the most part, beta emitters. The significant
3873 radionuclides are 125I (t1/2=60.Id, electron capture), 129I (beta), and 131I (t1/2=8.0d, beta).
3874 Occurrence and Uses
3875 Iodine is widely distributed, but never found in the elemental form. The average concentration in
3876 the earth's crust is about 0.3 ppm. In seawater, iodine concentration, in the form of sodium or
3877 potassium iodide, is low (about 50 ppb), but it is concentrated in certain seaweed, especially kelp.
3878 It is also found in brackish waters from oil and salt wells. The sources are saltpeter and nitrate-
3879 bearing earth in the form of calcium iodate, well brine, and seaweed. Iodine is produced from
3880 calcium iodate by extraction of the iodate from the source with water and reduction of the iodate
3881 with sodium bisulfite to iodine. Iodine is precipitated by mixing with the original iodate liquor to
3882 cause precipitation. Iodine can also be obtained from well brine, where the iodide ion is oxidized
3883 with chlorine, and then the volatile iodine is blown out with a stream of air. Sodium or potassium
3884 iodide in seaweed is calcined to an ash with sulfuric acid, which oxidizes the iodide to iodine.
3885 Iodine from any of these processes can be purified by sublimation.
3886 Isotopes of iodine of mass > 128 may all be formed as a result of fission of uranium and
3887 plutonium. Nuclear reactors and bomb tests are the most significant sources of these radioiso-
3888 topes with the exception of 131I. That isotope is routinely produced for use in medical imaging
3889 and diagnosis. The isotopes released from the other sources represent a short term environmental
3890 health hazard should there be an abnormal release from reactors or if bomb testing or use were to
3891 occur.
3892 This was the case in both 1979 and 1986 when the power reactor events at Three Mile Island and
3893 Chernobyl, caused releases of radioiodines. During the former event a ban on milk distribution in
3894 the downwind corridor was enforced as a purely preventative measure. In the latter case, signifi-
3895 cant releases of iodines and other isotopes caused more drastic, long term measures for food
3896 quarantine.
3897 Deposits on the surface of plants could provide a quick source of exposure if consumed directly
3898 from fruits and vegetables or indirectly from cow's milk. It would readily accumulate in the
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 14-137 DO NOT CITE OR QUOTE
-------
Separation Techniques
3899 thyroid gland, causing a short-term exposure of concern. It represent the greatest short-term
3900 exposure after a nuclear detonation and has been released in power plant accidents. 129I, with of a
3901 half-life of more than 15 million years, represent a long-term environmental hazard. In addition
3902 to its long half-life, the environmental forms of iodine in the environment are highly soluble in
3903 groundwater and are poorly sorbed by soil components. It is not absorbed at all by granite, and
3904 studies at a salt repository indicate that 129I would be only one of few radionuclides that would
3905 reach the surface before it decayed. Therefore, research on the fate of 129I that might be released
3906 suggests that the radionuclide would be highly disseminated in the ecosystem.
3907 131I is routinely analyzed for in milk, soil and water. 129I is a low energy beta and gamma emitter,
3908 which has a very long half-life (1.47 x 107 years). The most significant concern for this isotope is
3909 in radioactive waste, and its potential for migration due to the chemistry of iodine in the
3910 environment. 131I is produced for medical purposes by neutron reaction as follows: 130Te(n,y)131Te
3911 - beta decay - 131I (half-life = 8 days).
3912 The major use of iodine, iodine radionuclides, and iodine compounds is in medical diagnosis and
3913 treatment. 123I, 125I, and 131I are use for diagnostic imaging of the thyroid gland and the kidneys.
3914 131I is used to treat hyperthyroidism and thyroid cancer. Stable iodine in the form of potassium
3915 iodide is added to commercial salt to prevent enlargement of the thyroid (goiter). Iodine in the
3916 form of the hormone thyroxine is also used for thyroid and cardiac treatment and hormone
3917 replacement therapy in iodine deficiency. Iodine radionuclides are used as a tracer in the
3918 laboratory and industry to study chemistry mechanisms and processes and to study biological
3919 activity and processes. Iodine is a bactericide and is used as an antiseptic and sterilization of
3920 drinking water. It is used as a catalyst in chemical processes and as silver iodide in film
3921 emulsions.
3922 Solubility of Compounds
3923 Molecular iodine is only very slightly soluble in water (0.33 g/L), but it is soluble in solutions of
3924 iodide ion, forming I3~l. It is appreciably soluble in organic solvents. Carbon tetrachloride (CC14)
3925 or chloroform (CHC13) are commonly used to extract iodine from aqueous solutions after
3926 alternate forms of the element, typically I"1 and IO34, are converted to I2. The solutions have a
3927 violet color in organic solvents, and iodine dimerizes to some extent in these solutions:
3928 2 I2 ^ I4
3929 Numerous compounds of iodine are soluble in water. All metallic iodides are soluble in water
3930 except those of silver, mercury, lead, cupurous ion, thallium, and palladium. Antimony, bismuth,
MARLAP JULY 2001
DO NOT CITE OR QUOTE 14-138 DRAFT FOR PUBLIC COMMENT
-------
Separation Techniques
3931 and tin iodides require a small amount of acid to keep them in solution. Most of the iodates and
3932 periodates are insoluble. The iodates of sodium, potassium, rubidium, and the ammonium ion are
3933 soluble in water. Those of cesium, cobaltous ion, magnesium, strontium, and barium are slightly
3934 soluble in water but soluble in hot water. Most other metallic iodates are insoluble.
3935 Review of Properties
3936 Elemental iodine (I2) is a purple-black, lustrous solid at room temperature with a density of 4.9
3937 g/cm3. The brittle crystals have a slightly metallic appearance. Iodine readily sublimes and stored
3938 in a closed clear, colorless container, it produces a violet vapor with an irritating odor. Iodine has
3939 a melting point of 114 °C and a boiling point of 184 °C.
3940 The chemical reactivity of iodine is similar to the other halogens, but it is the least electro-
3941 negative member of the family of elements and the least reactive. It readily reduces to iodide, and
3942 is displaced from its iodides by the other halogens and many oxidizing agents. Iodine combines
3943 directly with most elements to form a large number of ionic and covalent compounds. The
3944 exceptions are the noble gases, carbon, nitrogen, and some noble metals.
3945 The inorganic compounds of iodine can be classified into three groups: (1) iodides, (2)
3946 interhalogen, and (3) oxides. Iodine forms iodides that range from ionic compounds such as
3947 potassium iodide (KI) to covalent compounds such as titanium tetraiodide (TiI4) and phosphorus
3948 triiodide (PI3), depending on the identity of the combining element. More electropositive (less
3949 electronegative) metals (on the left side of the periodic table, such as alkali metals and alkaline
3950 earths) form ionic compounds. Less electropositive metals and more electronegative nonmetals
3951 tend to form covalent compounds. Interhalogen compounds include the binary halides, such as
3952 iodine chloride (IC1), iodine trichloride (IC13), and iodine pentafluoride (IF5), or contain
3953 interhalogen cations and anions, such as IC12+1, IF6+1.1+3, Clffir"1, IC\4~l, and I6"2. Oxygen
3954 compounds constitute the oxides, I2O5 and I4O9 (containing one I+3 cation and three IO34 anions),
3955 for example; the oxyacids, such as hypoiodous acid (HIO) and iodic acid (HIO3); and compounds
3956 containing oxyanions, iodates (IO3 J) and periodates (IO44) are the common ones.
3957 Organoiodides include two categories: (1) iodides and (2) iodide derivatives with iodine in a
3958 positive oxidation state because iodine is covalently bonded to another, more electronegative
3959 element. Organoiodides contain a carbon iodide bond. They are relatively dense and volatile and
3960 more reactive than the other organohalides. They include the iodoalkanes such as ethyl iodide
3961 (C2H5I) and iodobenzene (C6H5I). Dimethyliodonium (HI) hexafluoroantimonate
3962 [(CH3)2I+3SbF6"3], a powerful methylating agent, is an example of the second category.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 14-139 DO NOT CITE OR QUOTE
-------
Separation Techniques
3963
3964
3965
3966
3967
3968
3969
3970
3971
3972
3973
3974
3975
3976
3977
3978
3979
3980
3981
3982
3983
3984
3985
3986
3987
3988
3989
3990
The toxicity of molecular iodine is primarily related to its vapor state and to solutions. Iodine
vapor is an eye and nasal irritant, potentially causing damage to the eyes and serious respiratory
damage. Solid iodine is not a serious problem unless confined to the skin where it causes
discoloration and eventually burns. Solutions of iodine are toxic if taken internally. The
radionuclides of iodine are radiotoxic, primarily because of their concentration in the thyroid
gland. Radiotoxicity of 129I, if released, is a concern because of its extremely long half-life. 131I,
with a half-life of eight days, is a short-term concern. The whole-body effective biological half-
lives of 129I and 131I are 140 d and 7.6 d, respectively.
Solution Chemistry
OXIDATION-REDUCTION BEHAVIOR. Iodine can exist in multiple oxidation states in solution, but
the radiochemist can control the states by selection of appropriate oxidizing and reducing agents.
In acid and alkaline solutions, the common forms of iodine are: I"1,12, and IO34. Hypoiodous acid
(HIO) and the hypoiodite ion (IO"1) can form in solution, but they rapidly disproportionate:
5 HIO ^ 2 I2 + KV1 + H+1 + 2 H2O
3 IO4 ^ 2 I'1 + IO34
Iodine itself is not a powerful oxidizing agent, less than that of the other halogens (F2, C12, and
Br2), but its action is generally rapid. Several oxidizing and reducing agents are used to convert
iodine into desired oxidation states during radiochemical procedures. These agents are used to
promote radiochemical equilibrium between the analyte and the carrier or tracer or to produce a
specific oxidation state before separation: I2 before extraction in an organic solvent or I"1 before
precipitation, as examples. Table 14.19 presents oxidizing and reducing agents commonly used
in radiochemical procedures:
Table 14.19 — Common radiochemical oxidizing and reducing agents for iodine
Redox Process
r1 - 12
r1 - 103-'
i2-r
r1 - icv1
Redox Reagent
HNO2 (NaNO2 in acid)
MnO2 in acid
6 M HNO3
NaHSO3 and NaHSO4 (in acid)
Na2SO3 and Na2S2O3
Fe2(SO4)3 (in acid)
KMnO4
50%CrO3inl8NH2SO4
Notes
Does not affect other halides
Well suited for laboratory work
MARLAP
DO NOT CITE OR QUOTE
14-140
JULY 2001
DRAFT FOR PUBLIC COMMENT
-------
Separation Techniques
3991
3992
3993
3994
3995
3996
3997
3998
3999
4000
4001
4002
4003
4004
4005
4006
Redox Process
r1 - KV
KV1 - 12
KV1 - I2
KV - r1
i, -r1
Redox Reagent
NaClOinbase
NH2OH-HC1
NH2OH-HC1
H2C2O4inl8NH2SO4
NaHSO3 in acid
SO2 gas
NaHSO3 and (NH4)2SO3
Notes
Radiochemical exchange between I2 and I"1 in solution is complete within time of mixing and
before separation. In contrast, exchange between I2 and IO34 or KV1 in acid solution and between
IO34 and IO44 in acid or alkaline solution is slow. For radiochemical analysis of iodine,
experimental evidence indicates that the complete and rapid exchange of radioiodine with carrier
iodine can be accomplished by the addition of the latter as I"1 and subsequent oxidation to IO44 by
NaCIO in alkaline solution, addition of IO44 and reduction to I"1 with NaHSO3, or addition of one
followed by redox reactions first to one oxidation sate and then back to the original state.
COMPLEXATION. As a nonmetal, iodine is generally not the central atom of a complex, but it can
act as a ligand to form complexes such as Sil6'2 and CoI6"3. An important characteristic of
molecular iodine is its ability to combine with the iodide ion to form polyiodide anions. The
brown triioide is the most stable:
4007 I9
4008 The equilibrium constant for the reaction in aqueous solution at 25 °C is 725, so appreciable
4009 concentrations of the anion can exist in solution, and the reaction is responsible for the solubility
4010 of iodine in iodide solutions.
4011 HYDROLYSIS. Iodine hydrolyzes in water through a disproportionation reaction:
4012 I2 + H2O ^ H+1 + r1 + HIO
4013 Because of the low solubility of iodine in water and the small equilibrium constant (k=2.0 x
4014 10"13), hydrolysis produces negligible amounts of the products (6.4 x 10"6 M) even when the
4015 solution is saturated with iodine. Disproportionation of HIO produces a corresponding minute
4016 quantity of IO34 (see the reaction above). In contrast, in alkaline solution, I2 produces I"1 and IO"1:
4017 I, + 2 OH4 ^ F1 + IO'1 + H9O
JULY 2001
DRAFT FOR PUBLIC COMMENT
14-141
MARLAP
DO NOT CITE OR QUOTE
-------
Separation Techniques
4018 The equilibrium constant favors the products (k=30), but the actual composition of the solution is
4019 complicated by the disproportionation of IO"1 (illustrated above), giving I"1 and IO3~1. The
4020 equilibrium constant for the reaction of IO"1 with hydroxide ion is very large (IO20), and the rate
4021 of the reaction is very fast at all temperatures. Therefore, the actual products obtained by
4022 dissolving iodine in an alkaline solution are indeed I"1 and KV1, quantitatively, and IO"1 does not
4023 exist in the solution.
4024 Dissolution of Samples
4025 Iodine compounds in rocks are often in the form of iodides that are soluble in either water or
4026 dilute nitric acid when the finely divided ores are treated with one of these agents. Those that are
4027 insoluble under these conditions are solubilized with alkali fusion with sodium carbonate or
4028 potassium hydroxide, followed by extraction of the residue with water. Insoluble periodiates can
4029 be decomposed by cautious ignition, converting them to soluble iodides.
4030 Metals containing iodine compounds are dissolved in varying concentrations of nitric, sulfuric, or
4031 hydrochloric acids. Dissolution can often be accomplished at room temperature or might require
4032 moderation in an ice bath.
4033 Organoiodides are decomposed with a sodium peroxide, calcium oxide, or potassium hydroxide
4034 by burning in oxygen in a sealed bomb. Wet oxidation with mixtures of sulfuric and chromic
4035 acids or with aqueous hydroxide is also used.
4036 Separation Methods
4037 PRECIPITATION. The availability of stable iodine as a carrier and the relative ease of producing
4038 the iodide ion make precipitation a simple method of concentrating and recovering iodine
4039 radionuclides. The two common precipitating agents are silver (Ag+1) and palladium (II) (Pd+2)
4040 cations, which form silver iodide (Agl) and palladium iodide (PdI2), respectively. Silver iodide
4041 can be solubilized with a 30 percent solution of potassium iodide. Palladium precipitates iodide
4042 in the presence of chloride and bromide, allowing the separation of iodide from these halides.
4043 The precipitating agent should be free of palladium (IV), which will precipitate chloride. If
4044 palladium (II) iodide is dried, precaution should be taken as the solid slowly looses iodine if
4045 heated at 100 °C. lodate can be precipitated as silver iodate, and periodate as lead periodate.
4046 SOLVENT EXTRACTION. One solvent extraction method is commonly used to isolate iodine. After
4047 preliminary oxidation-reduction steps to insure equilibrium of all iodine in solution, molecular
4048 iodine (I2) is extracted from aqueous solutions by a nonpolar solvent, usually carbon tetrachloride
MARLAP JULY 2001
DO NOT CITE OR QUOTE 14-142 DRAFT FOR PUBLIC COMMENT
-------
Separation Techniques
4049 or chloroform. It is not uncommon to add small quantities of the oxidizing or reducing agent to
4050 the extraction solution to ensure and maintain all iodine in the molecular form. Hydroxylamine is
4051 added, for example, if iodate is the immediate precursor of iodine before extraction.
4052 ION-EXCHANGE CHROMATOGRAPHY. Both cation and anion exchange procedures are used to
4053 separate iodine from contaminants. Cation-exchange chromatography has been used to remove
4054 interfering cations. To remove 137Cs activity, an iodine sample in the iodide form is absorbed on a
4055 cation-exchange resin and eluted with ammonium sulfite [(NH4)2SO3], to ensure maintenance of
4056 the iodide form. Cesium cations remain the resin. Bulk resin is also used, and iodide is washed
4057 free of the resin as the periodate with sodium hypochlorite (NaCIO) as the oxidizing agent.
4058 Anion-exchange resins provide absorption of the iodide ion. The halides have been separated
4059 from each other on an anion-exchange column prepared in the nitrate form by eluting with 1 M
4060 sodium nitrate. Iodide can also be separated from contaminants by addition to an anion
4061 exchanger and elution as periodate with sodium hypochlorite. The larger periodate anion is not as
4062 strongly attracted to the resin as the iodide ion. 131I separation, collection, and analysis is
4063 performed by absorbing the radionuclide on an anion-exchange resin and gamma counting it on
4064 the sealed column after eluting the contaminants.
4065 DISTILLATION. Molecular iodine is a relatively volatile substance. Compared to many
4066 contaminating substances, particularly metal ions in solution, its boiling point of 184 °C is very
4067 low, and the volatility of iodine provides a method for its separation from other substances. After
4068 appropriate oxidation-reductions steps to convert all forms of iodine into the molecular form,
4069 iodine is distilled from aqueous solution into sodium hydroxide and collected by another
4070 separation process, typically solvent extraction. In hydroxide solution, molecular iodine is
4071 converted to a mixture of iodide and hypoiodite ions and then into iodide and periodate ions, and
4072 suitable treatment is required to convert all forms into a single species for additional procedures.
4073 Methods of Analysis
4074 Macroquantities of iodine can be determined gravimetrically by precipitation as silver iodide or
4075 palladium iodide. The latter substance is often used to determine the chemical recovery in
4076 radiochemical analyses. Microquantities of 129I and 131I are coprecipitated with palladium iodide
4077 using stable iodide as a carrier and counted for quantification. 129I usually is beta counted in a
4078 liquid-scintillation system, but it can also be determined by gamma-ray spectrometry. 131I is
4079 determined by gamma-ray emission.
4080 Compiled from: Adams, 1995; APHA, 1998; Armstrong et al., 1961; Bailar et al., 1984;
4081 Choppin et al., 1995; Considine and Considine, 1983; Cotton and Wilkinson, 1988; DOE,
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 14-143 DO NOT CITE OR QUOTE
-------
Separation Techniques
4082 1990 and 1997, 1997; EPA, 1973; EPA, 1980; Ehmann and Vance, 1991; Greenwood and
4083 Earnshaw, 1984; Haissinsky and Adloff, 1965; Kleinberg and Cowan, 1960; Latimer, 1952;
4084 Lindsay, 1988.
4085 14.10.9.5 Plutonium
4086 Plutonium, with an atomic number of 94 is an actinide and the second element in the transuranic
4087 series. Essentially all plutonium is an artifact, most produced by neutron bombardment of 238U
4088 followed by two sequential beta emissions, but trace quantities of plutonium compounds can be
4089 found in the natural environment. Plutonium radiochemistry is complicated by the five possible
4090 oxidation states that can exist; four can be present in solution at one time.
4091 Isotopes
4092 Plutonium has 18 isotopes with mass numbers ranging from 232 to 247, and all isotopes are
4093 radioactive. Some have a long half-life: the isotope of greatest importance, 239Pu, has a half-life
4094 of 24,110 years, but 242Pu and 244Pu have a half-lives of 376,000 and 76,000,000 years,
4095 respectively. 238Pu, 240Pu, and 241Pu have a half-lives of 87.74, 6,537, and 14.4 years, respectively.
4096 Four of these isotopes decay by alpha emission accompanied by weak gamma rays: 238Pu, 239Pu,
4097 240Pu, and 242Pu. In contrast, 241Pu decays by beta emission with weak gamma rays but its progeny
4098 is 241Am, an intense gamma emitter. 239Pu and 241Pu are fissile materials—they can be split by
4099 both fast and slow neutrons. 240Pu, and 242Pu are fissionable but have very small neutron fission
4100 cross-sections. 240Pu partly decays by spontaneous fission, although a small amount of
4101 spontaneous fission occurs in most plutonium isotopes.
4102 Occurrence and Uses
4103 There are minute quantities of plutonium compounds in the natural environment as the result of
4104 thermal neutron capture and subsequent beta decay of naturally occurring 238U. All plutonium of
4105 concern is an artifact, the result of neutron bombardment of uranium in a nuclear reactor.
4106 Virtually all nuclear power-plants of all sizes and the waste from the plants contain plutonium
4107 because 238U is the main component of fuel used in nuclear reactors. It is also associated with the
4108 nuclear weapons industry and its waste. Virtually all the plutonium in environmental samples is
4109 found in air samples as the results of atmospheric weapons testing. Plutonium in plant and crop
4110 samples is essentially caused by surface absorption.
4111 Plutonium is produced in nuclear reactors from 238U that absorbs neutrons emitted by the fission
4112 of 235U, which is a naturally occurring uranium isotope found with 238U. 239U is formed and emits
MARLAP JULY 2001
DO NOT CITE OR QUOTE 14-144 DRAFT FOR PUBLIC COMMENT
-------
Separation Techniques
4113 a beta particle to form 239Np that decays by beta emission to form 239Pu. Once started, the process
4114 is spontaneous until the uranium fuel rods become a specific uranium-plutonium mixture. The
4115 rods are dissolved in acid, and plutonium is separated primarily by solvent extraction, finally
4116 producing a concentrated plutonium solution. Pure plutonium metal can be prepared by
4117 precipitating plutonium peroxide or oxalate, igniting the precipitate to PuO2, converting the oxide
4118 to PuF3, and reducing Pu(Tfl) to the metal in an ignited mixture containing metallic calcium.
4119 Large quantities of 239Pu have been used as the fissile agent in nuclear weapons and as a reactor
4120 fuel when mixed with uranium. It is also used to produce radioactive isotopes for research,
4121 including the study of breeder reactors, and 238Pu is used as a heat source to power instruments
4122 for space exploration and implanted heart pacemakers.
4123 Solubility of Compounds
4124 General solubility characteristics include the insolubility of the hydroxides, fluorides, iodates,
4125 phosphates, carbonates, and oxalates of Pu(Tfl) and Pu(IV). Some of these can be dissolved in
4126 acid solution, however. The corresponding compounds of PuO2+1 and PuO2+2 are soluble, with the
4127 exception of the hydroxides. The binary compounds represented by the carbides, silicides,
4128 sulfides, and selenides are of particular interest because of their refractory nature. One of the
4129 complicating factors of plutonium chemistry is the formation of a polymeric material by
4130 hydrolysis in dilute acid or neutral solutions. The polymeric material can be a complicating factor
4131 in radiochemical procedures and be quite unyielding in attempts to destroy it.
4132 Review of Properties
4133 Plutonium metal has some unique physical properties: a large piece is warm to the touch because
4134 of the energy produced by alpha decay, and it exists in six allotropic forms below its melting
4135 point at atmospheric pressure. Each form has unusual thermal expansion characteristics that
4136 prevents the use of unalloyed plutonium metal as a reactor fuel. The delta phase, however, can be
4137 stabilized by the addition of aluminum or gallium and be used in reactors. Chemically, plutonium
4138 can exist in five oxidation states: in, IV, V, VI, and VII. The first four states can be observed in
4139 solution, and solid compounds of all five states have been prepared. The metal is a silver-grey
4140 solid that tarnishes in air to form a yellow oxide coating. It is chemically reactive combining
4141 directly with the halogens, carbon, nitrogen, and silicon.
4142 Plutonium is a very toxic substance, but outside the body, it does not represent great danger from
4143 it low penetrating alpha emission or emission of its low intensity beta, gamma, or neutron
4144 radiation. Ingested plutonium is not readily absorbed into the body, but passes through the
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 14-145 DO NOT CITE OR QUOTE
-------
Separation Techniques
4145 digestive tract and expelled before it can cause significant harm. Inhaled plutonium presents a
4146 significant danger. Particularly, inhalation of particles smaller than one micron would be a
4147 serious threat due to the alpha-emitting radionuclide being in direct contact with lung tissue.
4148 Plutonium would also be very dangerous if it were to enter the blood stream through an open
4149 wound, because it would concentrate in the liver and bones, leading to damage to the bone
4150 marrow and subsequent related problems. For these reasons, plutonium is handled in gloveboxes
4151 with associated precautions taken to protect the worker from direct contact with the material.
4152 When working with plutonium in any form, precautions should also be taken to prevent the
4153 accumulation of quantities of fissionable plutonium that would achieve a critical mass,
4154 particularly in solution where it is more likely to become critical than solid plutonium.
4155 Most of the plutonium in the environment is the result of weapons testing. More than 99 percent
4156 of the plutonium from these activities was released during atmospheric tests, but a small portion
4157 was also released during ground tests. An even smaller quantity is released by nuclear fuel
4158 reprocessing plants, some in the ocean, and by nuclear waste repositories. Part of the atmospheric
4159 plutonium, originally part of the weapons, settled to the earth as an insoluble oxide, locating in
4160 the bottom sediments of lakes, rivers, and oceans or becoming incorporated in sub-surface soils.
4161 The majority of environmental plutonium isotopes are the result of atmospheric nuclear bomb
4162 tests. If the bomb material is made from uranium, the oxide is enriched to high percentages of
4163 235U, the fissile isotope. The 238U isotope does not fission, but absorbs 1-2 neutrons during the
4164 explosion forming isotopes of 239U and 240U. These isotopes beta decay within hours to their
4165 neptunium progeny, which in turn decay to 239Pu and 240Pu. Bombs from plutonium would yield
4166 higher fractions of 24°.241.242pu.
4167 Plutonium formed as a result of atmospheric tests is most likely to be in the form of a fine
4168 particulate oxide. If as in the case of a low altitude or underground test, there is a soil component,
4169 the plutonium will be fused with siliceous minerals. The behavior of the soluble form of
4170 plutonium would be similar to that released from fuel reprocessing plants and from nuclear waste
4171 sites. Like the insoluble oxide, most of the soluble form is found in sediments and soils, but a
4172 small percentage is associated with suspended particles in water. Both the soluble form of
4173 plutonium and the form suspended on particulate matter are responsible for plutonium transporta-
4174 tion in the environment. Plutonium in soil is found where the humic acid content is high. In non-
4175 humic, carbonate-rich soils, plutonium migrates downward. Migration in the former soil is slow
4176 (<0.1 cm/y) and in the latter it is relatively fast (1-10 cm/y). In subsurface oxic soil, plutonium is
4177 relatively mobile, transported primarily by colloids. In wet anoxic soils, most of the plutonium is
4178 quickly immobilized, although a small fraction remains mobile. The average time plutonium
4179 remains in water is proportional to the amount of suspended material. For this reason, more than
MARLAP JULY 2001
DO NOT CITE OR QUOTE 14-146 DRAFT FOR PUBLIC COMMENT
-------
Separation Techniques
4180
4181
4182
4183
4184
4185
4186
4187
4188
4189
4190
4191
4192
4193
4194
4195
4196
4197
4198
4199
4200
4201
4202
4203
4204
4205
4206
4207
90 percent of plutonium is removed from costal water, while the residence time in mid-ocean
water where particulate matter is less is much longer.
Solution Chemistry
The equilibration problems of plutonium are among the most complex encountered in
radiochemistry. Plutonium can form five oxidation states in solution, +3, +4, +5, +6, and +7. The
first four are present in solution as Pu+3, Pu+4, PuO2+1, PuO2+2. They coexist in dilute acid
solution, and sometimes all four are present in substantial quantities. Problems of disproportiona-
tion and auto-oxidation in freshly prepared solutions also complicate the chemistry of plutonium.
The +7 state can form in alkaline solutions, and it has been suggested that the ion in solution is
PuO5"3. Plutonium ions tend to hydrolyze and form complex ions in solution. The +4 ion can
form long chain polymers that do not exhibit the usual chemical behavior of the +4 oxidation
state. Finally, the different oxidation states exhibit radically different chemical behavior. As a
result of these effects, it is possible to mix a plutonium sample with plutonium tracer, subject the
mixture to a relatively severe chemical treatment using hot acids or similar reagents, and still
selectively recover portions of either the tracer or the sample. This characteristic explains the
challenge in achieving reproducible radiochemical results for plutonium.
OXIDATION-REDUCTION BEHAVIOR. Numerous redox agents are available to oxidize and reduce
any of the five states of plutonium to alternate oxidation states. The following table provides a
convenient method of preparation of each state and illustrates the use of redox reagents in
plutonium chemistry:
Table 14.20 — Redox agents in plutonium chemistry
Oxidation State
III
IV
IV
V
VI
VII
Form
Pu+3
Pu+4
PuO2-nH2O
(polymer)
Pu02+1
Pu02+2
Pu05-3 (?)
Method of Preparation
Dissolve Pu metal in HC1 and reduce Pu+4 with NH2OH, N2H4,
SO2, or by cathodic reduction
Oxidize Pu+3 with hot HNO3; treat Pu+3 or PuO2+2 with NO/1
Heat Pu+4 in very dilute acid; peptize Pu(OH)4
Reduce PuO2+2 with stoichiometric amount of I"1 or ascorbic acid;
electrolytic reduction of PuO2+2
Oxidize Pu+4 with hot dilute HNO3 or AgO; ozonize Pu+4 in cold
dilute HNO3 with Ce+3 or Ag+1 catalyst
Oxidize PuO2+2 in alkali with O3, S2O8"2 or radiation
JULY 2001
DRAFT FOR PUBLIC COMMENT
14-147
MARLAP
DO NOT CITE OR QUOTE
-------
Separation Techniques
4208 Unlike uranium, the +3 oxidation state is stable enough in solution to be useful in separation
4209 chemistry. Disproportionation reactions convert Pu+4 to Pu+3 and PuO2+2 releasing H+1. The
4210 presence of acid in the solution or complexing agents represses the process. Similarly, PuO2+1
4211 disproportionates producing the same products but with the consumption of H+1. For this reason,
4212 PuO2+1 is not predominant in acid solutions. These disproportionation reactions can be involved
4213 in redox reactions by other reagents. Instead of direct oxidation or reduction, the disproportiona-
4214 tion reaction can occur first, followed by direct oxidation or reduction of one of the products.
4215 It is possible to prepare stable aqueous solutions in which appreciable concentrations of the first
4216 four oxidation states exist simultaneously: the +3, +4, +5, and +6 states. The relative proportions
4217 of the different oxidation states depend on the acid, the acid concentration, the method of
4218 preparation of the solution, and the initial concentrations of each of the oxidation states. These
4219 relative concentrations will change over time and ultimately establish an equilibrium specific to
4220 the solution. In 0.5 M HC1 at 25 °C, for example, the equilibrium percentages of the four
4221 oxidation states prepared from initially pure Pu+4 are +3 (27.2%), +4 (58.4%), +5 (0.7%), and +6
4222 (13.6%). Freshly prepared plutonium samples are frequently in the +4 state, while an appreciable
4223 amount of the +3 and +6 oxidation states will be present in long-standing tracer solutions.
4224 A convenient solution to this plutonium equilibration problem takes the form of a two step
4225 process:
4226 • boil the combined sample and tracer with a concentrated inorganic acid (e.g., HNO3) to
4227 destroy any +4 polymers that might have formed, and
4228 • cool and dilute the solution; then rapidly (to avoid reforming polymers) treat the solution
4229 with excess iodide ion (solution turns brown or black) to momentarily reduce all of the
4230 plutonium to the +3 oxidation state.
4231 The solution will immediately start to disproportionate in the acid medium, but the plutonium
4232 will have achieved a true equilibrium starting at a certain time from one state in the solution.
4233 Alpha particles emitted by 239Pu can decompose solutions of the radionuclide by radiolysis. The
4234 radiolysis products then oxidize or reduce the plutonium, depending on the nature of the solution
4235 and the oxidation state of the element. The nature of the anion present greatly influences the rate
4236 of the redox process. For the radiochemist it is important to recognize that for old plutonium
4237 solutions, particularly those in low acidity, the oxidation labeled states are not reliable.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 14-148 DRAFT FOR PUBLIC COMMENT
-------
Separation Techniques
4238 HYDROLYSIS AND POLYMERIZATION. Hydrolysis is most pronounced for relatively small and
4239 highly charged ions such as Pu+4, but plutonium ions in any oxidation state are more easily
4240 hydrolyzed than their larger neptunium and uranium analogues.
4241 Trivalent plutonium tends to hydrolyze more than neptunium or uranium, but the study of its
4242 hydrolysis characteristics has been hindered by precipitation, formation of Pu+4, and unknown
4243 polymerization. In strongly alkaline solutions, Pu(OH)3 precipitates; the solubility product
4244 constant is estimated to be 2 x 10"20.
4245 Plutonium(IV) exists as a hydrated ion in solutions that are more acidic than 0.3 M H+1. Below
4246 0.3 M, it undergoes much more extensive hydrolysis than any other plutonium species, or at
4247 lower acidities (0.1 M) if the plutonium concentration is lower. Thus, the start of hydrolysis
4248 depends on the acid/plutonium ratio as well as the temperature and presence of other ions. On
4249 hydrolysis, only Pu(OH)+3 is important in the initial phases, but it tends to undergo irreversible
4250 polymerization, forming polymers with molecular weights as high as 1010 and chemical
4251 properties much different from the free ion. Presence of the polymer can be detected by its bright
4252 green color. When plutonium (IV) hydroxide [Pu(OH)4] is dissolved in dilute acid, the polymer
4253 also forms. Similarly, if a solution of Pu+4 in moderately concentrated acid is poured slowly into
4254 boiling water, extensive polymerization occurs. The colloidal character of the polymer is
4255 manifested by its strong adsorption onto glass, silica, or small bits of paper or dirt. The chemical
4256 characteristics of the polymer, with regard to precipitation, ion-exchange, and solvent extraction,
4257 is markedly different than the chemistry of the common +4 oxidation state of plutonium. Care
4258 should be taken in the laboratory to avoid the formation of these polymers. For instance, these
4259 polymers can be formed by overheating solutions during evaporation. Moreover, diluting an
4260 acidic plutonium solution with water can cause polymerization because of localized areas of low
4261 acidity, even when the final concentration of the solution is too high for polymerization.
4262 Therefore, plutonium solutions should always be diluted with acid rather than water. Polymeric
4263 plutonium can also be formed if insufficient acid is used when dissolving plutonium (IV)
4264 hydroxide.
4265 Immediately after formation, these polymers are easy to decompose by acidification with
4266 practically any concentrated inorganic acid or by oxidation. Because depolymerization is slow at
4267 room temperature and moderate acid concentrations, solutions should be made at least 6 M and
4268 boiled to destroy the polymers. The polymer is rapidly destroyed under these conditions. Adding
4269 strong complexing agents such as fluoride, sulfate, or other strong complexing agents can
4270 increase the rate of depolymerization. However, if the polymers are allowed to "age," they can be
4271 very difficult to destroy.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 14-149 DO NOT CITE OR QUOTE
-------
Separation Techniques
4272 The PuO2+1 ion has only a slight tendency to hydrolyze, beginning at pH 8, but study of the extent
4273 of the process is inhibited by the rapid disproportionation of hydrolyzed plutonium(V).
4274 Hydrolysis of PuO2+2 is far more extensive than expected for a large +2 ion. Hydrolysis begins at
4275 pH of about 2.7 to 3.3, giving an orange color to the solution that yields to bright yellow by pH 5.
4276 Between pH 5 and 7, dimerizatons seem to occur, and by pH 13 several forms of plutonium
4277 hydroxide have been precipitated with solubility products of approximately 2.5 x 10"25.
4278 COMPLEXATION. Plutonium ions tend to form complex ions in the following order:
4279 Pu+4 > Pu+3 « PuO2+2 > PuO2+1
4280 Divalent anions tend to form stronger complexes, and the order for simple anions with Pu+4 is:
4281 carbonate > oxalate > sulfate > fluoride > nitrate >
4282 chloride > bromide > iodide > perchlorate
4283 Complexation is preferably through oxygen and fluorine rather than nitrogen, phosphorus, or
4284 sulfur. Plutonium also forms complexes with ligands such as phosphate, acetate, and
4285 tributylphosphate (TBP). Strong chelate complexes form with EDTA, tartrate, citrate, 2-
4286 thenoyltrifluoroacetone (TTA), acetylacetone (acac), and cupferron. Plutonium(IV) forms a
4287 strong complex with fluoride (PuF+3) that is used to solubilize plutonium oxides and keep it in
4288 the aqueous phase during extraction of other elements with organic solvents. The complex with
4289 nitrate, Pu(NO3)6"2, allows the recovery of plutonium from nuclear fuels. Carbonate and acetate
4290 complexes prevent precipitation of plutonium from solution even at relatively high pH.
4291 Dissolution of Samples
4292 Metallic plutonium dissolves in halogen acids such as hydrochloric acid, but not in nitric or
4293 concentrated sulfuric acids. The metal dissolves in hydrofluoric nitric acid mixtures. Plutonium
4294 oxide dissolves with great difficulty in usual acids when ignited. Boiling with concentrated nitric
4295 acid containing low concentrations of hydrofluoric acid or with concentrated phosphoric acid is
4296 used. Fusion methods have also been used to dissolve the oxide as well as other compounds of
4297 plutonium. Plutonium in biological samples is readily soluble, in the case of metabolized
4298 plutonium in excreted samples, or highly refractory, in the case of fallout samples. Most
4299 procedures for fallout or environmental samples involve treatment with hydrofluoric acid or
4300 fusion treatment with a base.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 14-150 DRAFT FOR PUBLIC COMMENT
-------
Separation Techniques
4301 Separation Methods
4302 Extensive work has been done on methods to separate plutonium from other elements. Both
4303 laboratory and industrial procedures has received considerable treatment. The methods described
4304 below represents only a brief approach to separation of plutonium, but they indicate the nature of
4305 the chemistry employed.
4306 PRECIPITATION AND COPRECIPITATION. Macro quantities of plutonium are readily precipitated
4307 from aqueous solution, and the methods is the basis of separating plutonium from other
4308 radionuclides in some procedures. Contamination of other metals can be a problem, however;
4309 zirconium and ruthenium give the most trouble. Plutonium is precipitated primarily as the
4310 hydroxide, fluoride, peroxide, or oxalate. Both Pu(in) and Pu(IV) are precipitated from acid
4311 solution by potassium or ammonium hydroxide as hydrated hydroxides or hydrous oxides. On
4312 redissolving in acid, Pu (IV) tends to form the polymer, and high concentration of acid is needed
4313 to prevent its formation. Pu(IV) peroxide is formed on the addition of hydrogen peroxide to
4314 Pu(ni), Pu(IV), Pu(V), and Pu(VI) because of the oxidizing nature of hydrogen peroxide. The
4315 procedure has been used to prepare highly pure plutonium compounds from americium and
4316 uranium.
4317 Coprecipitation of plutonium can be very specific with the control of its oxidation states and
4318 selection of coprecipitating reagents. Lanthanum fluoride, a classical procedure for coprecipita-
4319 tion of plutonium, will bring down Pu(ni) and Pu(IV) but not Pu(VI). Only elements with similar
4320 redox and coprecipitation behavior interfere. Separation from other elements as well as
4321 concentration from large volumes with lanthanum fluoride is also important because not many
4322 elements form acid-soluble lanthanum fluoride coprecipitates. Bismuth phosphate (BiPO4) is also
4323 used to coprecipitate Pu(in) and Pu(IV). In contrast to lanthanum fluoride and bismuth
4324 phosphate, zirconium phosphate (ZrPO4) and an organic coprecipitate, zirconium phenylarsenate
4325 [Zr(C6H5)AsO4], will coprecipitate Pu(IV) exclusively.
4326 SOLVENT EXTRACTION. A wide variety of organic extractants have been developed to separate
4327 plutonium from other radionuclides and metals by selectively extracting them from aqueous
4328 media. The extractants, among others, include organophosphorus compounds such as phosphates
4329 (organoesters of phosphoric acid), amines and their quaternary salts, alcohols, ketones, ethers,
4330 and amides. Chelating agents such as thenoyltrifluoroacetone (TTA) and cupferron have also
4331 been used. Numerous studies have been performed on the behavior of these systems. It has been
4332 found that the performance of an extracting system is primarily related to the organic solvent in
4333 which the extractant is dissolved and the concentration of the extractant in the solvent, the nature
4334 of the aqueous medium (the acid present and its concentration (pH) and the presence of salting
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 14-151 DO NOT CITE OR QUOTE
-------
Separation Techniques
4335 agents), the temperature of the system, and the presence and nature of oxidizing agents. One
4336 common system, used extensively in the laboratory and in industrial process to extract plutonium
4337 from fission products, illustrates the use of solvent extraction to separate plutonium from
4338 uranium and other metals. The PUREX process (pjutonium uranium reduction extraction) is used
4339 in most fuel reprocessing plants to separate the radionuclides. It employs TBP, tri-/>butyl
4340 phosphate [(C4H9)3PO], in a hydrocarbon solvent, as the extractant. The uranium fuel is dissolved
4341 in nitric acid as Pu(Tfl), and plutonium is oxidized to Pu(IV) and uranium to U(VI) by oxidizing
4342 agents. Plutonium and uranium are extracted into a 30 percent TBP solution, and the organic
4343 phase is scrubbed with nitric acid solution to remove impurities. The plutonium is removed by
4344 back-extracting it as Pu(ni) with a nitric acid solution containing a reducing agent.
4345 Solvent extraction chromatography has provided an efficient, easy technique for rapidly
4346 separating plutonium and other transuranic elements. A process using octylphenyl-N,N-
4347 diisobutyl carbamoylphoshpine oxide (CMPO) in TBP and fixed on an inert polymeric resin
4348 matrix has been used to isolate plutonium (IV). All plutonium in the analyte is adjusted to
4349 plutonium (IV), and the column is loaded from 2 M nitric acid. Plutonium is eluted with 4 M
4350 hydrochloric acid and 0.1 M hydroquinone or 0.1 M ammonium hydrogen oxalate (NH4HC2O4).
4351 It is important to note that iron, found in most environmental samples, does not effect the
4352 separation if the element is kept in the +2 oxidation state as ferrous ions. This is commonly
4353 achieved using ascorbic acid. The ferric ion (Fe+3) is detrimental to the separation.
4354 ION-EXCHANGE CHROMATOGRAPHY. Ion-exchange chromatography has been used extensively
4355 for the radiochemical separation of plutonium. All cationic plutonium species in non-complexing
4356 acid solutions readily exchanges onto cation resins at low acid concentrations and desorb at high
4357 acid concentrations. Plutonium in all its oxidation states form neutral or anionic complexes with
4358 various anions, providing an alternate means for eluting the element. Various cation-exchange
4359 resins have been used with hydrochloric, nitric, perchloric, and sulfuric acids for separation of
4360 plutonium from metals including other actinides, but the most common use of plutonium cation-
4361 exchange chromatography is concentrating a dilute solution or separation from nonabsorbable
4362 impurities such as organic reagents, redox agents, for example.
4363 Anion-exchange chromatography is the primary ion-exchange method for the separation of
4364 plutonium from other metals and the separation of the plutonium oxidation states, and many
4365 procedures have been developed using this method. On a strong anion-exchange resin, for
4366 example, the higher oxidation states (IV, V, and VI) occurs at hydrochloric acid concentrations
4367 above 6 M, while desorption occurs at 2 M acid. Plutonium (in) does not absorb on the column,
4368 and plutonium (VI) absorbs from 2 to 3 M hydrochloric acid solution. Plutonium can be
4369 separated from other actinides and most other elements by absorbing the plutonium cations—
MARLAP JULY 2001
DO NOT CITE OR QUOTE 14-152 DRAFT FOR PUBLIC COMMENT
-------
Separation Techniques
4370 Pu(IV) and Pu (VI)—onto a strong-anion resin from 6 M hydrochloric acid, and subsequently
4371 eluting the plutonium by reducing it to plutonium ILL Anion exchange in 7 to 8 M nitric acid is
4372 also an effective method for separating plutonium. The radionuclide loads on the column as
4373 Pu(NO3)6"2 and is eluted with dilute acid or after reduction.
4374 ELECTRODEPOSTION. Separation methods based on elecrtodeposition are not common, but one
4375 method for the alpha analysis of plutonium is in use. Plutonium is electrodeposited on a stainless
4376 steel disc from an ammonium sulfate solution at 1.2 amps for one hour. The separation is used
4377 after isolating the radionuclide by extraction chromatography, and the plutonium isotopes are
4378 resolved by alpha spectroscopy.
4379 Methods of Analysis
4380 238Pu, 239Pu, 240Pu, and 241Pu are collected for analysis either by electrodepositon on a platinum or
4381 nickel disc or by microprecipitation with lanthanum fluoride (LaF3). Radionuclides of 238Pu,
4382 239Pu, and 240Pu are determined by alpha spectrometry or gas flow proportional counting. 241Pu is
4383 beta counted. 236Pu or 242Pu are used as a tracer for measuring chemical yield. They are measured
4384 by alpha spectrometry.
4385 Compiled from: Baes and Mesmer, 1976; Choppin et al., 1995; Coleman, 1965; Cotton and
4386 Wilkinson, 1988; DOE 1990, 1995, and 1997; EPA 1973 and 1980; Metz and Waterbury,
4387 1962; Seaborg and Loveland, 1990; Weigel et al., 1986.
4388 14.10.9.6 Radium
4389 Radium, with an atomic number of 88, is the heaviest (last) member of the family of alkaline
4390 earth metals, which, in addition, includes beryllium (Be), magnesium (Mg), calcium (Ca),
4391 strontium (Sr), and barium (Ba). It is the most basic and reactive of the series, and exists
4392 exclusively as +2 cations in compounds and solution. All isotopes are radioactive, and essentially
4393 all analyses are made by radioactive measurements.
4394 Isotopes
4395 There are 25 isotopes of radium from 205Ra to 234Ra; all are radioactive. The most important with
4396 respect to the environmental contamination are members of the 238U and 232Th naturally occurring
4397 decay series: 226Ra and 228Ra, respectively. 226Ra is the most abundant isotopic form with a half-
4398 life of 1,602 years. As a member of the 238U series, it is produced by alpha emission from 230Th.
4399 226Ra emits an alpha particle and, in turn, produces 222Rn, an inert gas that is also an alpha
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 14-153 DO NOT CITE OR QUOTE
-------
Separation Techniques
4400 emitter. Radium generates radon at the rate of 0.1 jiL per day per gram of radium, and its
4401 radioactivity decreases at the rate of about one percent each 25 years. 228Ra, half-life of 5.77
4402 years, is produced in the 232Th decay series by emission of an alpha particle from 232Th itself.
4403 Occurrence
4404 In nature, radium is primarily associated with uranium and thorium, particularly in the uranium
4405 ores—carnotite and pitchblende, where 226Ra is in radioactive equilibrium with 238U and its other
4406 progeny. The widespread dispersal of uranium in rocks and minerals results in a considerable
4407 distribution of radium isotopes throughout nature. It is generally found in trace amounts in most
4408 materials, therefore, the radium/uranium ratio is about 1 mg radium per 3 kg uranium (1 part in 3
4409 x 106 parts uranium). This leads to a terrestrial abundance of approximately 10"6 ppm: 10"12g/g in
4410 rocks and minerals. Building materials, such as bricks and concrete blocks for example, that
4411 contain mineral products also contain radium. With leaching from soil, the concentration is about
4412 10"13g/L in river and streams, and uptake in biological systems produces concentrations of
4413 10"14g/g in plants and 10"15g/g in animals.
4414 Uranium ores have been processed with hot mineral acids or boiling alkali carbonate to remove
4415 radium and/or uranium. Extracted radium was usually coprecipitated with barium sulfate,
4416 converted to carbonate or sulfide, and solubilized with hydrochloric acid. Separation from
4417 barium was usually accomplished by fractional crystallization of the chlorides, bromides, or
4418 hydroxides, since barium salts are usually slightly more soluble. The free metal has been
4419 prepared by electrolysis of radium chloride solutions, using a mercury cathode. The resulting
4420 amalgam is thermally decomposed in a hydrogen atmosphere to produce the pure metal. The
4421 waste streams from these industrial operations contain radium, primarily as a coprecipitate of
4422 barium sulfate. Since many other natural ores also contain uranium and radium, processing can
4423 result in uranium and its equilibrium progeny appearing in a product or byproduct. Apatite, a
4424 phosphate ore, is used to produce phosphoric acid, and the gypsum byproduct contains all the
4425 radium originally present in the ore.
4426 226Ra extracted from ores has historically been used in diverse ways as a source of radioactivity.
4427 It has been mixed with a scintillator to produce luminous paint, and at one time, the most
4428 common use for its salts was radiation therapy. As a source of gamma radiation, radium activity
4429 was enhanced by sealing a radium salt in a capsule that prevented escape of the gaseous progeny,
4430 222Rn, and allowing the radon to decay into its successive progeny. Two progeny are 214Pb and
4431 214Bi, the principal emitters of gamma radiation in the source. For the most part radium has been
4432 replaced in medical technology by other sources of radioactivity, but numerous capsules
4433 containing the dry, concentrated substances still exist.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 14-154 DRAFT FOR PUBLIC COMMENT
-------
Separation Techniques
4434 Radium salts are used in various instruments for inspecting structures such as metal castings by
4435 gamma-ray radiography, to measure the thickness of catalyst beds in petroleum cracking units,
4436 and to continuously measure and control the thickness of metals in rolling mills. Radium is also
4437 used for the preparation of standard sources of radiation, as a source of actinium and protac-
4438 tinium, and as a source of ionizing radiation in static charge eliminators. In combination with
4439 beryllium, it is a neutron source for research, in the analysis of materials by neutron activation,
4440 and radio-logging of oil wells.
4441 Radium in the environment is the result of natural equilibration and anthropogical activity such
4442 as mining and processing operations. Radium is retained by many rock and soil minerals,
4443 particularly clay minerals, and migrates only very slowly in through these materials. The decay
4444 progeny of 226Ra, gaseous 222Rn, is an important environmental pollutant and represents the most
4445 significant hazard from naturally occurring radium. Concentration of the alpha-emitting gas in
4446 some occupied structures contributes to the incidence of lung cancer in humans. During the
4447 decay of 226Ra, the recoil of the parent nucleus after it emits an alpha particle, now 222Rn, causes
4448 an increased fraction of radon to escape from its host mineral, a larger fraction than can be
4449 explained by intramineral migration or diffusion.
4450 In groundwater, radium likely encounters dissolved sulfate and/or carbonate anions, which could
4451 precipitate radium sulfate or radium carbonate. Although both salts are relatively insoluble, a
4452 sulfate concentration of 0.0001 M would still allow an equilibrium concentration of about 0.1
4453 ppm Ra+2 to exist in solution. Thus, the insolubilities of either of these salts are not likely to
4454 prevent contamination of the environment.
4455 Radium also contaminates the environment because of past disposal practices of some proces-
4456 sing, milling, and reclamation operations. Radium process tailings have been discovered in land
4457 areas as seams or pockets of insoluble radium compounds, such as barium radium sulfate, or
4458 unprocessed radium (uranium) ore, such as carnotite. Release of solid or liquid process streams
4459 and subsequent mixing with local soil has resulted in intimate contamination of soil particles,
4460 primarily as Ra+2 absorbed onto clay-sized fractions. This form of absorbed radium is tightly
4461 bound to soil but can be extracted partially by hot concentrated acid solutions.
4462 Solubility of Compounds
4463 The solubility of radium compounds can usually be inferred from the solubility of the correspon-
4464 ding barium compound and the trend in the solubilities of the corresponding alkaline earth
4465 compounds. The common water-soluble radium salts are the chloride, bromide, nitrate, and
4466 hydroxide. The fluoride, carbonate, phosphate, biphosphate (hydrogen phosphate), and oxalate
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 14-155 DO NOT CITE OR QUOTE
-------
Separation Techniques
4467 are only slightly soluble. Radium sulfate is the least soluble radium compound known, insoluble
4468 in water and dilute acids, but it is soluble in concentrated sulfuric acid, forming a complex ion
4469 with sulfate anions, Ra(SO4)2"2.
4470 Radium compounds are essentially insoluble in organic solvents. In most separation procedures
4471 based on extraction, other elements, not radium, are extracted into the organic phase. Exceptions
4472 are known (see "Separation," below), and crown ethers have been developed recently that
4473 selectively remove radium from an aqueous environment.
4474 Review of Properties
4475 Radium is highly toxic exclusively because of its radioactive emissions: gamma radiation of the
4476 element itself and beta particles emitted by some of its decay progeny. It concentrates in bones
4477 replacing calcium and causing anemia and cancerous growths. Its immediate progeny, gaseous
4478 radon, is an alpha emitter that is a health threat when inhaled.
4479 Metallic radium is brilliant white and reacts rapidly with air, forming a white oxide and black
4480 nitride. It is an active metal that reacts with cold water to produce radium hydroxide, hydrogen,
4481 and other products. The radium ion in solution is colorless. Its compounds also are colorless
4482 when freshly prepared but darken and decompose on standing because of the intense alpha
4483 radiation. The original color returns when the compound is recrystallized. Alpha emissions also
4484 cause all radium compounds to emit a blue glow in air when sufficient quantities are available.
4485 Radium compounds also are about 1.5 °C higher in temperature than their surroundings because
4486 of the heat released when alpha particles loose energy on absorbance by the compound. Glass
4487 containers turn purple or brown in contact with radium compounds and eventually the glass
4488 crystallizes and becomes crazed.
4489 Like all alkaline earths, radium contains two valence electrons (7s2) and forms only +2 ions in its
4490 compounds and in solution. The ionic radius of radium in crystalline materials is 152 pm (0.152
4491 nm or 1.52 A), the largest crystalline radius of the alkaline earth cations (Ra+2 > Ba+2 > Sr+2 >
4492 Ca+2 >Mg+2 > Be+2). In contrast, the hydrated ion radius in solution is the smallest of the alkaline
4493 earth cations, 398 pm (Be+2 > Mg+2 > Ca+2 > Sr+2 > Ba+2 > Ra+2). With the smallest charge-to-
4494 crystal-radius ratio among the alkaline earths of 1.32 (+2/1.52), the smallest hydrated radius of
4495 radium is expected, because the ratio represents the least attractive potential for water molecules
4496 in solution.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 14-156 DRAFT FOR PUBLIC COMMENT
-------
Separation Techniques
4497 Solution Chemistry
4498 Existing exclusively in one oxidation state (+2), the chemistry of radium is uncomplicated by
4499 oxidation-reduction reactions that could produce alternate states in solution. It is made even less
4500 complicated by its weak tendency to form complex ions or hydrolyze in solution. These
4501 properties are a reflection of the small charge-to-crystal-radius ratio of 1.32, described above. In
4502 general, radiochemical equilibrium is established with carriers by stirring, followed by either
4503 standing or digesting in the cold for several minutes. Adsorption of trace amounts of radium on
4504 surfaces, however, is an important consideration in its radiochemistry.
4505 COMPLEXATION. Radium, like other alkaline-earth cations, forms few complexes in acid
4506 solution. Under alkaline conditions, however, several one-to-one chelates are formed with
4507 organic ligands: among others, with EDTA, diethylenetriaminepentaacetate (DTPA),
4508 ethyleneglycol bis(2-aminoethylether)-tetraacetate (EGTA), nitrilotriacetate (NTA or NTTA),
4509 and citrate. The most stable complex ion forms with DTPA. The tendency to form complexes
4510 decreases as their crystalline size increases and their charge-crystal-radius ratio decreases. Since
4511 crystalline sizes of the cations are in the order: Ra+2 > Ba+2 > Sr+2 > Ca+2, radium has the least
4512 tendency to form complex ions, and few significant complexes of radium with inorganic anions
4513 are known. One notable exception is observed in concentrated sulfuric acid, which dissolves
4514 highly insoluble radium sulfate (RaSO4) by forming Ra(SO4)2"2.
4515 Complex-ion chemistry is not used in most radium radiochemical procedures. Complexing
4516 agents are primarily employed as elution agents in cation exchange, in separations from barium
4517 ions by fractional precipitation, and in titration procedures. Alkaline citrate solutions have been
4518 used to prevent precipitation of radium in the presence of lead and barium carriers until complete
4519 isotopic exchange has been accomplished.
4520 HYDROLYSIS. Similar to their behavior complex-ion formation, alkaline earths show less and less
4521 tendency to hydrolyze with increasing size of the ions, and the tendency decreases with
4522 increasing ionic strength of the solution. Therefore, hydrolysis of radium is an insignificant factor
4523 in their solution chemistry.
4524 ADSORPTION. The adsorption of trace amounts of radium on surfaces is an important considera-
4525 tion in its radiochemistry. Although not as significant with radium as with some ions with higher
4526 charges, serious losses from solution can occur under certain conditions. Adsorption on glass is a
4527 particular problem, and adsorption on polyethylene has been reported. Adsorption gradually
4528 increases with increasing pH and depends strongly on the nature of the surface. In the extreme,
4529 up to 50 percent radium has been observed to adsorb onto glass from neutral solution in 20 days,
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 14-157 DO NOT CITE OR QUOTE
-------
Separation Techniques
4530 and 30 percent from 0.13 M hydrochloric acid (HC1). Fortunately, adsorbed radium can be
4531 removed from glass with strong acid.
4532 The presence of insoluble impurities, such as traces of dust or silica, increases adsorption, but
4533 adsorption is negligible from very pure solutions at low pH values. Tracer radium solutions,
4534 therefore, should be free from insoluble impurities, and radium should be completely in solution
4535 before analysis. The solutions should also be maintained in at least 1 M mineral acid or contain
4536 chelating agents. Addition of barium ion as a carrier for radium will probably decrease the
4537 amount of radium adsorption. Radium residues from solubilization of samples that contain silica
4538 or lead or barium sulfates and those that result in two or more separate solutions should be
4539 avoided since the radium might divide unequally between the fractions. Destruction of silica with
4540 HF, reduction of sulfates to sulfides with zinc dust, and subsequent dissolution of the residue
4541 with nitric acid are procedures used to avoid this problem.
4542 Dissolution of Samples
4543 Soil, mineral, ore samples, and other inorganic solids are dissolved by conventional treatment
4544 with mineral acids and by fusion with sodium carbonate (Na2CO3). Hydrofluoric acid (FTP) or
4545 potassium fluoride (KF) is used to remove silica. Up to 95 percent radium removal has been
4546 leached from some samples with hot nitric acid (F£NO3), but such simple treatment will not
4547 completely dissolve all the radium in soil, rock, and mineral samples. Biological samples are wet
4548 ashed first with mineral acids or decomposed by heating to remove organic material. The residue
4549 is taken up in mineral acids or treated to remove silica. Any dissolution method that results in
4550 two or more separate fractions should be avoided, since the adsorption characteristics of trace
4551 quantities of radium may cause it to divide between the fractions.
4552 Barium sulfate (BaSO4), often used to coprecipitate radium from solution, can be dissolved
4553 directly into alkaline EDTA solutions. Radium can be repeatedly reprecipitated and dissolved by
4554 alternate acidification with acetic acid and dissolution with the EDTA solution.
4555 Solutions resulting from dissolution of solid samples should be made at least 1 M with mineral
4556 acid before storage to prevent radium from absorbing onto the surface of glass containers.
4557 Separation Methods
4558 COPRECIPITATION. Radium is almost always present in solution in trace amounts, and even the
4559 most insoluble radium compound, radium sulfate, can not be used to separate and isolate radium
4560 from solution by direct precipitation. Therefore, the cation is commonly removed from solution
MARLAP JULY 2001
DO NOT CITE OR QUOTE 14-158 DRAFT FOR PUBLIC COMMENT
-------
Separation Techniques
4561 in virtually quantitative amounts by coprecipitation. Since radium forms the same types of
4562 insoluble compounds as barium: sulfates (SO4"2), chromates (CrO4"2), carbonates (CO3"2),
4563 phosphates (PO4"3), oxalates (C2O4"2), and sulfites (SO4"2), it coprecipitates with all insoluble
4564 barium compounds, and to a lesser extent with most insoluble strontium and lead compounds.
4565 Barium sulfate and barium chromate are most frequently used to carry radium during coprecipita-
4566 tion. Other compounds that are good carriers for radium include: ferric hydroxide when
4567 precipitated at moderately high pH with sodium hydroxide (NaOH) or ammonium hydroxide
4568 (NH4OH), barium chloride (BaCl2) when precipitated from a cold mixed solvent of water and
4569 alcohol saturated with hydrochloric acid, barium iodate (BaIO3), and various insoluble
4570 phosphates, fluorides (F"1), and oxalates (e.g., thorium phosphate [Th3(PO4)], lanthanum fluoride
4571 (LaF3), and thorium oxalate [Th(C2O4)]. Lead sulfate (PbSO4) can be used if a carrier-free radium
4572 preparation is required, since quantitative lead-radium separations are possible while quantitative
4573 barium-radium separations are very difficult.
4574 ION EXCHANGE. Radium has been separated from other metals on both cation- and anion-
4575 exchange resins. Barium and other alkaline earths are separated on cation-exchange columns
4576 under acidic conditions. In dilute hydrochloric acid solutions (3 M), the affinity of the cation for
4577 the exchange site is dominated by ion-dipole interactions between the water molecules of the
4578 hydrated ion and the resin. Ions of smaller hydrated radius (smaller charge-to-crystal-radius ratio)
4579 tend to displace ions of larger hydrated radius. The affinity series is Ra+2>Ba+2>Sr+2>Ca+2, and
4580 radium elutes last. Increasing the acid concentration to 12 M effectively reverses the order of
4581 affinity, since the strong acid tends to dehydrate the ion, and ion-resin affinity is dominated more
4582 by ionic interactions, increasing in the order of increasing crystal radius: Ca+2>Sr+2>Ba+2>Ra+2,
4583 and calcium elutes last. Radium has also been separated from tri- and tetravalent ions since these
4584 ions have a much stronger affinity for the cation-exchange resin. Radium with its +2 charge is
4585 only partially absorbed, while trivalent actinium and tetravalent thorium, for example, will be
4586 completely absorbed. Tracer quantities of radium also has been separated from alkaline earths by
4587 eluting a cation-exchange column with chelating agents such as lactate, citrate, and EDTA;
4588 radium typically elutes last, since it forms weaker interactions with the ligands.
4589 Anion-exchange resins have been used to separate radium from other metal ions in solutions of
4590 chelating agents that form anionic complexes with the cations. The affinity for the columns
4591 decreases in the order Ca > Sr > Ba > Ra, reflecting the ability of the metal ions to form stable
4592 complex anions with the chelating agents. The difficult separation of barium from radium has
4593 been accomplished by this procedure. Radium is also separated from metals such as uranium,
4594 polonium, bismuth, lead, and protactinium that form polychloro complex anions. Since radium
4595 does not form a chlorocomplex, it does not absorbs on the anion exchanger (carrying a positive
4596 charge), and remains quantitatively in the effluent solution.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 14-159 DO NOT CITE OR QUOTE
-------
Separation Techniques
4597 Ion-exchange methods are not easily adapted for the separation of macro-scale quantities of
4598 radium, because the intense radiation degrades the synthetic resin and insoluble radium
4599 compounds usually form in the ion-exchange column.
4600 SOLVENT EXTRACTION. Radium compounds have very low solubilities in organic solvents. In
4601 most extraction procedures, other organic-soluble complexes of elements, not radium, are
4602 extracted into the non-aqueous phase, leaving radium in the water. Radium is separated from
4603 actinium, thorium, polonium, lead, bismuth, and thallium, for example, by extracting these
4604 elements as 2-thenoyltrifluoroacetone (TTA) complexes. Radium does not form the complex
4605 except at very high pH, and is not extracted. One notable exception to this generality is the
4606 extraction of radium tetraphenylborate by nitrobenzene from an alkaline solution. The presence
4607 of EDTA inhibits formation of the tetraphenylborate, however, and radium is not extracted in the
4608 presence of EDTA either.
4609 More recent developments have employed crown ethers to selectively extract radium as a
4610 complex ion from water samples for analysis. Radium-selective extraction membranes have also
4611 been used to isolate radium from solutions.
4612 Methods of Analysis
4613 Radium is detected and quantified by counting either alpha or gamma emissions of the
4614 radionuclide or its progeny. Gamma-ray spectroscopy can be used on macro 226Ra samples
4615 (approximately 50 g or more) without pretreatment unless 235U, even in very small quantities, is
4616 present to interfere with the measured peak. The most sensitive method for the analysis of 226Ra
4617 is de-emanation of 222Rn from the radium source, complete removal, followed by alpha counting
4618 the 222Rn and its progeny. The procedure is lengthy and expensive, however. The radium in a
4619 liquid sample is placed in a sealed tube for a specified time to allow the ingrowth of 222Rn. The
4620 radon is collected in a scintillation cell and stored for several hours to allow for ingrowth of
4621 successive progeny products. The alpha radiation is then counted in the scintillation cell called a
4622 Lucas cell. The primary alpha emissions are from 222Rn, 218Po, and 214Po. Complete retention of
4623 radon can also be accomplished by sealing the radium sample hermetically in a container and
4624 alpha- or gamma-counting.
4625 228Ra can also be determined directly by gamma spectroscopy, using the gamma-rays of its
4626 progeny, 228Ac, without concern for interference; however, a lower detection limit is obtained if
4627 the 228Ac is measured by beta counting. In the beta-counting procedure, 228Ra is separated, time is
4628 allowed for actinium ingrowth, the 228Ac is removed by solvent extraction, ion-exchange, or
4629 coprecipitation, and then measured by beta counting.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 14-160 DRAFT FOR PUBLIC COMMENT
-------
Separation Techniques
4630 224Ra can be determined by chemically isolating the 212Pb, which is in equilibrium with the 224Ra.
4631 After an appropriate ingrowth period, 212Pb is determined by alpha counting its progeny, 212Bi and
4632 212Po.
4633 Compiled from: Baes and Mesmer, 1976; Choppin et al., 1995; Considine and Considine,
4634 1983; DOE, 1990 and 1997, 1997; EPA, 1984; Friedlander et al., 1981; Green and Earnshaw,
4635 1984; Hassinsky and Asloff, 1965; Kirby and Salutsky, 1964; Lindsay, 1988; Salutsky, 1997;
4636 Sedlet, 1966; Shoesmith, 1964; Sunderman and Townley, 1960; Turekian and Bolter, 1966;
4637 Vdovenko and Dubasov, 1975.
4638 14.10.9.7 Strontium
4639 Strontium, atomic number 38, is the fourth member of the alkaline-earth metals, which includes
4640 beryllium (Be), magnesium (Mg), calcium (Ca), strontium (Sr), barium (Ba), and radium (Ra).
4641 Like radium, it exist exclusively in the +2 oxidation state in both compounds and in solution,
4642 making its chemistry simpler than many of the radionuclides reviewed in this section.
4643 Isotopes
4644 Strontium exists in 29 isotopic forms, including three metastable states, ranging in mass number
4645 from 77 to 102. Natural strontium is a mixture of four stable isotopes: 84Sr, 86Sr, 87Sr, and 88Sr.
4646 The lower mass number isotopes decay by electron capture, and the isotopes with higher mass
4647 numbers are primarily beta emitters. The half-lives of most isotopes are short, measured in
4648 milliseconds, seconds, minutes, hours, or days. The exception is 90Sr, a beta emitter with a half-
4649 life of 29.1 years.
4650 Occurrence and Uses
4651 Strontium is found in nature in two main ores, celestite (SrSO4) and strontianite (SrCO3), widely
4652 distributed in small concentrations. Small amounts are found associated with calcium and barium
4653 minerals. The earth's crust contains 0.042 percent strontium, ranking twenty-first among the
4654 elements occurring in rock and making it as abundant as chlorine and sulfur. The element ranks
4655 11th in abundance in sea water, about 8-10 ppm. The only naturally occurring radioactive isotopes
4656 of strontium are the result of spontaneous fission of uranium in rocks. Other nuclear reactions
4657 and fallout from nuclear weapons test are additional sources of fission products. 90Sr is a fission
4658 product of 235U, along with 89Sr, and short-lived isotopes, 91Sr to 102Sr. 85Sr can be produced by
4659 irradiation of 85Rb with accelerated protons or deuterons.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 14-161 DO NOT CITE OR QUOTE
-------
Separation Techniques
4660 Stable strontium is produced from its ores. The sulfate ore is leached with hydrochloric acid
4661 solution to remove impurities and shaken with sodium carbonate for several hours to produce
4662 strontium carbonate. Washing this product or the carbonate ore with hot water and several
4663 reprecipitation steps produce a fine grade of strontium carbonate. The metal is produced by
4664 converting the carbonate to strontium chloride with hydrochloric acid or to strontium oxide by
4665 heating. Strontium chloride in a melt with potassium chloride is electrolyzed or the oxide is
4666 reduced by heating with aluminum in a vacuum to distill off the metal. An alternate method
4667 electrolyzes an aqueous solution of the chloride with a mercury cathode. The resultant mercury
4668 amalgam is heated in hydrogen to drive off the mercury.
4669 The major use of strontium is in glass production for color television picture tubes. Strontium is
4670 used in producing ferrite magnets, in refining zinc, to produce hardness and durability in alloys of
4671 tin and lead, as a deoxidizer in copper and bronze, and "getter" in electron tubes. Strontium
4672 hydroxide forms soaps and greases with numerous organic acids that are stable, resistant to
4673 oxidation and decomposition over a wide temperature range, and resistant to decomposition by
4674 water and the leaching action of hydrocarbons. The beta emission of 90Sr and its progeny, 90Y
4675 (ti/2=64 h), has found applications in industry, medicine, and research. The radionuclides are in
4676 equilibrium in about 25 days. The radiation of 90Y is more penetrating than that of strontium. It is
4677 used with zinc sulfide in some luminescent paints. Implants of 90Sr provide radiation therapy for
4678 the treatment of the pituitary gland and breast and nerve tissue. The radiation from strontium has
4679 been used in thickness gauges, level measurements, automatic control processes, diffusion
4680 studies of seawater, and a source of electrical power. Since 90Sr is one of the long-lived and most
4681 energetic beta emitters, it might prove to be a good source of power in space vehicles, remote
4682 weather stations, navigational buoys, and similar long-life, remote devices. Both 89Sr and 90Sr
4683 have been used in physical chemistry experiments and in biology as tags and tracers. 90Sr to 87Sr
4684 ratios are used in geological dating, because 87Sr is formed by decay of long-lived 87Rb.
4685 Solubility of Compounds
4686 Several simple salts of strontium are soluble in water. Among these are the acetate, chloride,
4687 bromide, iodide, nitrate, nitrite, permanganate, sulfide, chlorate, bromate, and perchlorate.
4688 Strontium hydroxide is slightly soluble and is precipitated only from concentrated solutions.
4689 Review of Properties
4690 Strontium is a low-density (2.54 g/cm3) silver-white metal. It is as soft as lead and is malleable
4691 and ductile. Three allotropic forms exit with transition temperatures of 235 and 540 °C. Freshly
MARLAP JULY 2001
DO NOT CITE OR QUOTE 14-162 DRAFT FOR PUBLIC COMMENT
-------
Separation Techniques
4692 cut strontium is silver in appearance, but it rapidly turns a yellowish color on formation of the
4693 oxide in the air. It is stored under mineral oil to prevent oxidation.
4694 The metal decomposes water, producing strontium hydroxide [Sr(OH)2] and hydrogen, and the
4695 finely divided metal ignites spontaneously in the air. The hydroxide forms strontium peroxide
4696 (SrO2) when treated with hydrogen peroxide in the cold. Strontium is a strong reducing agent and
4697 combines directly with hydrogen, halogens, oxygen, and sulfur to form, respectively, the simple
4698 binary compounds: hydride (SrH2), halogens (SrX2), oxide (SrO), and sulfide (SrS). The metal
4699 reacts with nitrogen to form the nitride (Sr3N2) only on heating to 380 °C. It also reacts
4700 vigorously with most acids to form Sr+2 salts and hydrogen. With nitric acid the reaction is fast,
4701 producing nitrogen dioxide. In contrast, reaction with sulfuric acid is slow because of the
4702 formation of the insoluble sulfate [Sr(SO4)2].
4703 Strontium isotopes are some of the principal constituents of radioactive fallout following
4704 detonation of nuclear weapons, and they are released in insignificant amounts during normal
4705 operations of reactors and fuel reprocessing operations. Their toxicity is higher, however, than
4706 that of other fission products, and 90Sr represent a particular hazard because of its long half-life,
4707 energetic beta emission, tendency to contaminate food, especially milk, and high retention in
4708 bone structure. Strontium in bone is difficult to eliminate and has a biological half-life of
4709 approximately eleven years (4,000 d).
4710 Strontium occurring in groundwater is primarily in the form of strontium carbonate. Its solubility
4711 under oxidizing and reducing conditions is approximately 0.001 M (0.15 g/L or 150 g/m3).
4712 Solution Chemistry
4713 Strontium exists exclusively in the +2 oxidation state in solution, so the chemistry of strontium is
4714 uncomplicated by oxidation-reduction reactions that could produce alternate states in solution.
4715 COMPLEXATION. Strontium has little tendency to form complexes. Of the few complexing agents
4716 for strontium, the significant agents in radiochemistry to date are EDTA, oxalate, citrate,
4717 ammoniatriacetate, methylanine-N,N-diacetate, 8-quinolinol, and an insoluble chelate with
4718 picrolonate. The most stable complex ion forms with EDTA. Coordination compounds of
4719 strontium are not common. These chelating agents are used primarily in ion-exchange
4720 procedures. Amine chelates of strontium are unstable, and the p-diketones and alcohol chelates
4721 are poorly characterized. In contrast, cyclic crown ethers and cryptates form stronger chelates
4722 with strontium than with calcium, the stronger chelating metal with EDTA and more traditional
4723 chelating agents. Cryptates are a macrocyclic chelate of the type, N[(CH2CH2O)2CH2CH2]3N, an
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 14-163 DO NOT CITE OR QUOTE
-------
Separation Techniques
4724 octadentate ligand containing six oxygen atoms and two nitrogen atoms as ligand bonding sites
4725 that encapsulates the cation. It might find use in the extraction chemistry of strontium.
4726 HYDROLYSIS. The tendency of the alkaline-earth cations to hydrolyze decreases as their atomic
4727 number increases. The tendency is greater than that of the corresponding alkali metals, but
4728 hydrolysis of potassium, for example, is insignificant. An indication of the tendency of a cation
4729 to hydrolyze is the solubility of their hydroxides, and the solubility of the alkaline earths become
4730 more soluble with increasing atomic number. Strontium hydroxide is slightly soluble in water (8
4731 g/L at 20 °C). In comparison, the hydroxide of beryllium, the first element in the alkaline earth
4732 series, has a solubility of approximately 3 x 10"4 g/L.
4733 Dissolution of Samples
4734 Dissolution of samples for the analysis of strontium is generally simple. Water is used to dissolve
4735 soluble compounds: acetate, bromide, chloride, iodide, chlorate, perchlorate, nitrate, nitrite, and
4736 permangenate. Hydrochloric or nitric acid dissolves the fluoride, carbonate, oxalate, chromate,
4737 phosphate, sulfate, and oxide. Strontium in limestone, cement, soil, bone, and other biological
4738 material can be dissolved from some samples in hot hydrochloric acid. Insoluble silica, if present,
4739 can be filtered or centrifuged. In some cases, soil can be leached to remove strontium. As much
4740 as 99.5 percent of the strontium in some crushed soil samples has been leached with 1 M nitric
4741 acid by three extractions. Soil samples have also been suspended overnight in ammonium acetate
4742 at pH 7. If leaching is not successful, soil samples can be dissolved by alkali fusion of the ground
4743 powder with potassium hydroxide, nitrate, or carbonate. Strontium is taken up from the residue
4744 in nitric acid. Biological materials such as plant material or dairy products are solubilized by
4745 ashing at 600 °C and taking up milk residue in hot, concentrated hydrochloric acid and plant
4746 residue in aqua regia. Wet ashing can be used by treating the sample with nitric acid followed by
4747 an equal-volume mixture of nitric and perchloric acids. Human and animal bone samples are
4748 ashed at 900 °C and the residue dissolved in concentrated hydrochloric acid.
4749 Separation Methods
4750 PRECIPITATION AND COPRECIPITATION. The common insoluble salts of strontium are the fluoride,
4751 carbonate, oxalate, chromate, and sulfate. Most are suitable for radiochemical procedures, and
4752 strontium separation have the advantage of stable forms of strontium that can be used as a carrier
4753 and are readily available. Precipitation of strontium nitrate in 80 percent nitric acid has been used
4754 to separate stable strontium carrier and 90Sr from its progeny, 90Y, and other soluble nitrates
4755 (calcium, for example). The solubility of strontium chloride in concentrated hydrochloric
4756 solution has been used to separate strontium from barium—barium chloride is insoluble in the
MARLAP JULY 2001
DO NOT CITE OR QUOTE 14-164 DRAFT FOR PUBLIC COMMENT
-------
Separation Techniques
4757 acid. Barium and radium (as coprecipitant) have been removed from strontium by precipitating
4758 barium as the chromate at a carefully controlled pH of 5.5. Strontium chromate will not
4759 precipitate unless the pH is raised. Strontium can also be separated from yttrium by precipitation
4760 of the much less soluble yttrium hydroxide by raising an acid solution of the cations to a pH of
4761 about 8 with ammonium hydroxide. Strontium hydroxide is slightly soluble and will not
4762 precipitate without high concentrations of hydroxide or strontium or both. Carrier-free strontium
4763 is coprecipitated with ferric hydroxide, and lead sulfate is also used.
4764 SOLVENT EXTRACTION. The application of organic solvents for separation of strontium from
4765 other metals has not been extensive. Thenoyltrifluoroacetone (TTA) has been used to extract
4766 carrier-free strontium at a pH > 10. At pH 5,90Y is extracted with TTA from strontium, which
4767 remains in aqueous solution. 8-hydroxyquinolinol in chloroform has also been used to extract
4768 strontium. The few procedures that have been available are mainly used to separate the alkaline
4769 earths from each other. A 1:1 mixture of ethyl alcohol and diethy ether extracts calcium from
4770 strontium.
4771 In recent years, extraction procedures have been developed based on the complexation of
4772 strontium cations with crown ethers in 1-octanol. Strontium can be extracted with these mixture
4773 from 1 M to 7 M nitric acid solutions. The most advantageous application of strontium extraction
4774 procedures has been found in extraction chromatography. An extraction resin consisting of
4775 4,4'(5')-bis(Y-butylcyclohexano)-18-crown-6 (DtBuCH18C6) in 1-octanol on an inert polymeric
4776 matrix is highly selective for strontium nitrate and will separate the cation from many other
4777 metals including calcium, barium, and yttrium. This column is used to separate strontium from
4778 potassium, cerium, plutonium, and neptunium (K+1, Ce+4, Pu+4, Np+4, respectively). The column
4779 is prepared and loaded from 8 M nitric acid. The ions listed above are eluted with 3 M nitric acid
4780 containing oxalic acid. Strontium is eluted with 0.05 M nitric acid.
4781 ION-EXCHANGE CHROMATOGRAPHY. Ion-exchange chromatography is used to separate trace
4782 quantities of strontium, but separation of macro quantities is very time consuming. Strontium is
4783 absorbed on cation-exchange resins, and elution is often based on the formation of a stable
4784 complex. Carrier-free strontium is separated from fission products, including barium, on a
4785 cation-exchange resin and eluted with citrate. In a similar process, strontium was also separated
4786 form other alkaline earths, magnesium, calcium, barium, and radium, eluting with ammonium
4787 lactate at pH 7 and 78 °C. Good separations were also obtained with hydrochloric solutions and
4788 ammonium citrate. 90Sr and 90Y are separated on a cation-exchange column, eluting yttrium with
4789 ammonium citrate at pH 3.8 and strontium at pH 6.0. Strontium and calcium have also been
4790 separated in EDTA solutions at pH 5.3. Strontium is retained on the column, and calcium elutes
4791 as the calcium-EDTA complex. Strontium elutes with 3 M hydrochloric acid.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 14-165 DO NOT CITE OR QUOTE
-------
Separation Techniques
4792 Not many procedures use anion-exchange chromatography for separation of strontium. 90Sr has
4793 been separated from 90Y on an anion-exchange resin pretreated with hydroxide. Strontium is
4794 eluted form the column with water, and yttrium is eluted with 1 M hydrochloric acid. The
4795 alkaline earths have been separated by anion-exchange column pretreated with dilute ammonium
4796 citrate, loading the column with the chloride form of the metals, and eluting with ammonium
4797 citrate at pH 7.5.
4798 Methods of Analysis
4799 Macroquantities of strontium are determined by gravimetric methods and atomic absorption
4800 spectrometry, and emission spectrometry. Strontium is precipitated as strontium carbonate or
4801 sulfate in gravimetric procedures. For atomic absorption analysis, the separated sample is ashed,
4802 and the product is dissolved in hydrochloric acid. Lanthanum is added to the solution to
4803 precipitate interfering anions, phosphate, sulfate, or aluminate, that would occur in the flame.
4804 89Sr and 90Sr are determined by analysis of their beta emissions. With a short half-life of 53 d,
4805 89Sr is only found in fresh fission products. 90Sr is a beta emitter with a half-life of 27.7 y. Its
4806 progeny is 90Y, which emits beta particles with a half-life of 64.0 h, producing stable 90Zr.
4807 Neither 90Sr nor 90Y is a gamma emitter. 90Sr is determined directly from its beta emission, before
4808 90Y grows in, by beta counting immediately (three to four hours) after it is collected by
4809 precipitation. The chemical yield can be determined gravimetrically by the addition of stable
4810 strontium, after the separation of calcium. Alternatively, 90Sr can be measured from the beta
4811 emission of 90Y while it reaches secular equilibrium (two to three weeks). The 90Y is separated by
4812 solvent extraction and evaporated to dryness or by precipitation, then beta counted. The chemical
4813 yield of the yttrium procedure can be determined by adding stable yttrium and determining the
4814 yttrium gravimetrically. 89Sr has a half-life of 52.7 d and is only present in fresh fission material.
4815 If it is present with 90Sr, it can be determined by the difference in activity of combined 89Sr and
4816 90Sr (combined or total strontium) and the activity of 90Sr. Total strontium is measured by beta
4817 counting immediately after it is collected by precipitation, and 90Sr is measured by isolating 90Y
4818 after ingrowth. 85Sr can be used as a tracer for determining the chemical yield of 90Sr (determined
4819 by isolating 90Y), but its beta emission interferes with beta counting of total strontium and must
4820 be accounted for in the final activity.
4821 An alternative method for determining 89Sr and 90Sr in the presence of each other is based on the
4822 equations for decay of strontium radionuclides and ingrowth of 90Y. Combined strontium is
4823 collected and immediately counted to determine the total strontium. During ingrowth, the
4824 mixture is recounted, and the data from the counts are used to determine the amount of 89Sr and
4825 90Sr in the original (fresh) mixture.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 14-166 DRAFT FOR PUBLIC COMMENT
-------
Separation Techniques
4826 Compiled from: Baes and Mesmer, 1976; Choppin et al., 1995; Considine and Considine,
4827 1983; CRC, 1998-99; DOE, 1990 and 1997, 1997; EPA, 1973; EPA, 1980; Greenwood and
4828 Earnshaw, 1984; Hassinsky and Adloff, 1965; Riley, 1995; Sunderman and Townley, 1960;
4829 Turekian and Bolter, 1966.
4830 14.10.9.8 Technetium
4831 Technetium, atomic number 43, was the first element to be made artificially. Techetium has no
4832 stable isotopes. Natural technetium is known to exist but only in negligibly small quantities
4833 resulting from the spontaneous fission of natural uranium. Technetium is chemically very similar
4834 to rhenium, but significant differences exist that cause them to behave quite differently under
4835 certain conditions.
4836 Isotopes
4837 Thirty-one radioisotopes and unstable isomers of technetium are known with mass numbers
4838 ranging from 86 to 113. The half-lives range from seconds to millions of years. The lower mass
4839 number isotopes decay by primarily by electron capture and the higher mass number isotopes by
4840 beta emission. The significant isotopes (with half-lives/decay modes) are 95mTc (61 d/electron
4841 capture and isomeric transition), 99mTc (6.01 h/isomeric transition by low-energy gamma), and
4842 "Tc (2.13 x 105 y/beta to stable ruthenium-99). Other, long-lived isotopes are 97Tc (2.6 x
4843 106/electron-capture) and 98Tc (4.2 x 106 y/beta emission).
4844 Occurrence and Uses
4845 The first synthesis of technetium was through the production of 99Mo by bombardment of 98Mo
4846 with neutrons and subsequent beta decay to 99Tc. Technetium is also a major constituent of
4847 nuclear reactor fission products and has been found in very small quantities in pitchblende from
4848 the spontaneous fission of naturally occurring uranium.
4849 Technetium makes up about 6 percent of uranium fission products in nuclear power plant fuels. It
4850 is recovered from these fuels by solvent extraction and ion-exchange after storage of the fuels for
4851 several years to allow the highly radioactive, short-lived products to decay. Technetium is
4852 recovered as ammonium pertechnate (NH4TcO4) after its solutions are acidified with
4853 hydrochloric acid, precipitated with sulfide, and the sulfide (Tc2S7) is reacted with hydrogen
4854 peroxide. Rhenium and molybdenum are also removed by extraction with organic solvents. The
4855 metal is obtained by reduction of ammonium pertechnate with hydrogen at 600 °C.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 14-167 DO NOT CITE OR QUOTE
-------
Separation Techniques
4856 Potassium pertechnates (KTcO4) have been used in water (55 ppm) as corrosion inhibitors for
4857 mild carbon steel in aerated distilled water, but currently there is no significant uses of elemental
4858 technetium or its compounds, although technetium and some of its alloys are a superconductor.
4859 The corrosion protection is limited to closed systems to prevent release of the radioactive isotope.
4860 99mTc, with a half-life of only 61 days, has been used in tracer work. 99mTc is used in medical
4861 diagnosis as a radioactive tracer. As a complex, the amount of 99mTc required for gamma
4862 scanning is very small, thus, it is referred to as non-invasive scanning. It is used for cardiovascu-
4863 lar and brain studies and the diagnosis of liver, spleen, and thyroid disorders. There are more than
4864 20 99mTc compounds available commercially for diagnostic purposes. With iodine isotopes, they
4865 are the most frequently used radionuclides for diagnostics. 99mTc has also been used to determine
4866 the deadtime of counting detectors.
4867 Solubility of Compounds
4868 The nature of the compounds has not been thoroughly delineated, but ammonium pertechnate
4869 is soluble in water, and technetium heptoxide forms soluble pertechnetic acid (HTcO4) when
4870 water is added.
4871 Review of Properties
4872 Technetium is a silver-grey metal that resembles platinum in appearance. It tarnishes slowly in
4873 moist air to give the oxyacid, pertechnetic acid (HTcO4). It has a density of 11.5 g/cm3. The metal
4874 reacts with oxygen at elevated temperatures to produce the volatile oxide, technetium heptoxide.
4875 Technetium dissolves in warm bromine water, nitric acid, aqua regia, and concentrated sulfuric
4876 acid, but it is insoluble in hydrochloric and hydrofluoric acids. Technetium forms the chlorides
4877 (TcCl4 and TcCl6) and fluorides (TcF5 and TcF6) by direct combination of the metal with the
4878 respective halogen. The specific halide is obtained by selecting the proper temperature and
4879 pressure for its formation.
4880 "Tc has a high specific activity. As a contamination hazard, it should be handled in a glove box.
4881 The behavior of technetium in groundwater is highly dependent on its oxidation state. Under
4882 oxidizing conditions, pertechnate is the predominant species. It is very soluble and only slightly
4883 absorbed to mineral components. For those reasons, it has a relatively high dissemination
4884 potential in natural systems. Under reducing conditions, technetium precipitates as technetium
4885 dioxide (TcO2), which is very insoluble. With the production of "Tc in fission fuels and
4886 considering its long half-life, the soluble form of the radionuclide is an environmental concern
4887 wherever the fuel is reprocessed or stored. As a consequence, "Tc would be expected to be one
MARLAP JULY 2001
DO NOT CITE OR QUOTE 14-168 DRAFT FOR PUBLIC COMMENT
-------
Separation Techniques
4888 of the principle contributors to a radioactive release to the environment, even from repositories
4889 with barriers that could retain the radionuclide up to 10,000 years. Studies of a salt repository
4890 indicate that "Tc is one of the few radionuclides that might reach the surface before it decays.
4891 Solution Chemistry
4892 All oxidation states between -1 and +7 can be expected for technetium, but the important ones in
4893 solution are +4 and +7. The +4 state exist primarily as the slightly soluble oxide, TcO2. It is
4894 soluble only in the presence of complexing ligands; TcCl6"2, for example, is stable in solutions
4895 with a chloride concentration greater than 1 M. The most important species in solution is the
4896 pertechnate ion [TcO^1 as Tc(VII)], which is readily soluble and easily formed from lower
4897 oxidation states with oxidizing agents such as nitric acid and hydrogen peroxide. There is no
4898 evidence of polymeric forms in solution as a result of hydrolysis of the metal ion.
4899 OXIDATION-REDUCTION BEHAVIOR. Most radioanalytical procedures for technetium are
4900 performed on the pertechnate ion, TcO44. The ion can be reduced by hydrochloric acid, the
4901 thiocyanate ion (SCN"1), organic impurities, anion-exchange resins, and some organic solvents.
4902 The product of reduction can be TcO2 [Tc(IV)], although a multiplicity of other products are
4903 expected in complexing media. Even though the +7 oxidation state is easy to reduce, the
4904 reduction process is sometimes slow. Unless precautions are taken to maintain the appropriate
4905 oxidation state, however, erratic results will be obtained during the radioanalytical procedure.
4906 Several examples illustrate the precaution. Dissolution should always be performed under
4907 strongly oxidizing conditions to ensure conversion of all states to the +7 oxidation state since
4908 complications because of slow exchange with carrier and other reagents are less likely to occur if
4909 this state is maintained. Technetium is extracted with various solvents in several radioanalytical
4910 procedures, but the method can be very inefficient because of reduction of the pertechnate ion by
4911 some organic solvents. The presence of an oxidizing agent such as hydrogen peroxide will
4912 prevent the unwanted reduction. In contrast, TcO^1 is easily lost on evaporation of acid solutions
4913 unless a reducing agent is present or evaporation is conducted at a relatively low temperature.
4914 COMPLEXATION. Technetium forms complex ions in solution with several simple inorganic
4915 ligands such as fluoride and chloride. The +4 oxidation state is represented by the TcX6"2 ion
4916 where X = F, Cl, Br, and I. It is formed from TcO^1 by reduction to the +4 state with iodide in
4917 HX. TcF6"2 is found in HF solutions during decomposition of samples, before further oxidation.
4918 Complex ions formed between organic ligands and technetium in the +5 oxidation state are
4919 known with the general formula, TcO3XLL, where X is a halide and L is an organic ligand. the
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 14-169 DO NOT CITE OR QUOTE
-------
Separation Techniques
4920 ligands typically bond through an oxygen or nitrogen atom. Other organic complexes of the +5
4921 state have the general formulas: TcOX2L2, TcOX/1, and TcOX5'2.
4922 Dissolution of Samples
4923 Dissolution of samples containing technetium requires two precautions: it is essential that acid
4924 solutions be heated only under reflux conditions to avoid losses by volatilization, and dissolution
4925 should be done only with strongly oxidizing conditions to ensure conversion of all lower
4926 oxidation states to Tc(VII). In addition, problems with slow carrier exchange are less likely for
4927 the VII oxidation state. Molybdenum targets are dissolved in nitric acid or aqua regia, but the
4928 excess acid interferes with many subsequent analytical steps. Dissolution in concentrated sulfuric
4929 acid followed by oxidation with hydrogen peroxide after neutralization avoids these problems of
4930 excess acid. Other technetium samples can be dissolved by fusion with sodium peroxide/sodium
4931 hydroxide (Na2O2/NaOH) fluxes.
4932 Separation Methods
4933 PRECIPITATION AND COPRECIPITATION. The various oxidation states of technetium are
4934 precipitated in different forms with different reagents. Technetium (VII) is primarily present in
4935 solution as the pertechnate anion, and macro quantities are precipitated with large cations such as
4936 thallium (Tl+1), silver (Ag+1), cesium (Cs+1), and tetraphenylarsonium [(C6H5)4As+1]. the
4937 latter ion is the most efficient if ice-bath conditions are used. Perchtechnate is coprecipitated
4938 without interference from molybdenum with these cations and perrhenate (ReO44), perchlorate
4939 (ClO^1), periodate (lO/1), and tetrafluoroborate (BF^1). The salt consisting of tetraphenylar-
4940 senium and the perrhenate froms a coprecipitate fastest, in several seconds. Technetium (VII) can
4941 be precipitated from solution as the heptasulfide (Tc2S7) by the addition of hydrogen sulfide (or
4942 hydrogen sulfide generating compounds such as thioacetamide and sodium thiosulfate) from 4 M
4943 sulfuric acid. Since many other transition metals often associated with technetium also from
4944 insoluble compounds with sulfide, the method is primarily used to concentrate technetium.
4945 Technetium (IV) is carried by ferric hydroxide. The method can be use to separate technetium
4946 from rhenium. The precipitate is solubilized and oxidized with concentrated nitric acid, and iron
4947 is removed by precipitation with aqueous ammonia. Technetium is also coprecipitated as the
4948 hexachlorotechnate (IV) (TcCl6"2) with thallium and a,a'-dipyridylhexachlororhenate (IV).
4949 Technetium (VI) (probably as TcO4.2) is carried quantitatively by molybdenum 8-hydroxyquino-
4950 late and by silver or lead molybdate. Technetium (in) is carried quantitatively by iron or zinc
4951 hydroxide and the sulfide, hydroxide, and 8-hydroxyquinolate of molybdenum.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 14-170 DRAFT FOR PUBLIC COMMENT
-------
Separation Techniques
4952 SOLVENT EXTRACTION. Technetium, primarily in the Tc (VII) state (pertechnetate) can be
4953 isolated by extraction with organic solvents, but the principal disadvantage of all extraction
4954 systems is the inevitable introduction of organic material that might reduce the pertechnetate
4955 anion and cause difficulties in subsequent analytical steps. The pertechnetate ion is extracted
4956 with pyridine from a 4 M sodium hydroxide solution, but perrhenate and permanganate ions are
4957 also extracted. The anion also extracts into chloroform in the presence of the tetraphenyl-
4958 arsonium ion as tetraphenylarsonium pertechnetate. Extraction is more favorable from neutral or
4959 basic sulfate solutions than chloride solutions. Perrhenate and perchlorate are also extracted but
4960 molybdenum does not interfere. Small amounts of hydrogen peroxide in the extraction mixture to
4961 prevent reduction of pertechnetate. Technetium is back-extracted into 0.2 M perchloric acid or 12
4962 M sulfuric acid. Other organic solvents are have also been used to extract pertechnetate from acid
4963 solutions, including alcohols, ketones, and tributyl phosphate. Ketones and cyclic amines are
4964 more effective for extraction form basic solutions. Tertiary amines and quaternary ammonium
4965 salts are more effective extracting agents than alcohols, ketones, and tributyl phosphate. Back
4966 extraction is accomplished several ways, depending on the extraction system. A change in pH,
4967 displacement by another anion such as perchlorate, nitrate, or bisulfate, or addition of a nonpolar
4968 solvent to an extraction system consisting of an oxygen-containing solvent.
4969 A recent extraction method has been used successfully for extraction chromatography and
4970 extractive filtration. A column material consisting of trioctyl and tridecyl methyl ammonium
4971 chlorides impregnated in an inert apolar polymeric matrix is used to separate "Tc by loading the
4972 radionuclide as the pertechnetate ion from a 0.1 M nitric acid solution. It is stripped off the
4973 column with 12 M nitric acid. Alternatively, the extraction material is used in a filter disc, and
4974 the samples containing "Tc are filtered from water at pH 2 and rinsed with 0.01 M nitric acid.
4975 Technetium is collected on the disc.
4976 Lower oxidation states of technetium are possible. The thiocyanate complexes of technetium (V)
4977 is soluble in alcohols, ethers, ketones, and trioctylphosphine oxide or trioctylamine hydrochloride
4978 in cyclohexane or 1,2-dichloroethane. Technetium (IV), as TcCl6"2, extracts into chloroform in
4979 the presence of high concentrations of tetraphenylarsonium ion. Pertechnate and perrhenate are
4980 both extracted from alkaline solution by hexone (methyl isobutyl ketone), but reduction of
4981 technetium to the IV state with hydrazine or hydroxylamine results in the extraction of perrhenate
4982 only.
4983 ION-EXCHANGE CHROMATOGRAPHY. Ion-exchange chromatography is primarily performed with
4984 technetium as the pertechnate anion. Technetium does not exchange on cation resins, so
4985 technetium is rapidly separated from other cations on these columns. In contrast, it is strongly
4986 absorbed on strong anion exchangers and is eluted with anions that have a greater affinity for the
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 14-171 DO NOT CITE OR QUOTE
-------
Separation Techniques
4987 resin. Technetium and molybdenum are separated using ammonium thiocyanate as the eluent. A
4988 good separation of pertechnate and molybdate has been achieved on an anion-exchange resin in
4989 the phosphate form where the molybdate is preferentially absorbed. Good separation of
4990 pertechnate and perrhenate are obtained with perchlorate as the eluent.
4991 VOLATILIZATION. The volatility of technetium heptoxide allows the co-distillation of technetium
4992 with acids. Co-distillation from perchloric acid gives good yields, but only a partial separation
4993 from rhenium is achieved. Molybdenum is also carried unless complexed by phosphoric acid.
4994 Separation from rhenium can be achieved from sulfuric acid, but yields of technetium are can be
4995 very poor because of its reduction by trace impurities in the acid. Much more reproducible results
4996 can be obtained in the presence of an oxidizing agent, but ruthenium tetroxide (RuO4) also
4997 distills under these conditions. It can be removed, however, by precipitation as ruthenium dioxide
4998 RuO2. In distillation from sulfuric acid-water mixtures, technetium distills in the low-boiling
4999 point aqueous fraction, probably as pertechnetic acid. Technetium and rhenium are separated
5000 from sulfuric-hydrochloric acid mixtures; pertechnate is reduced to non-volatile Tc (IV) and
5001 remains in the acid solution. Technetium heptoxide can be separated from molybdenum trioxide
5002 by fractional sublimation at temperatures > 300 °C.
5003 ELECTRODEPOSITION. Technetium can be electrodeposited as its dioxide (TcO2) from 2 M
5004 sodium hydroxide. The metal is partially separated from molybdenum and rhenium, but
5005 deposition only occurs from low technetium concentrations. Carrier-free 95Tc and 96Tc have been
5006 electrolyzed on a platinum electrode from dilute sulfuric acid. Optimum electroplating of
5007 technetium has been achieved at pH 5.5 in the presence of very dilute fluoride ion. Yields were
5008 better with a copper electrode instead of platinum—about 90 percent was collected in two hours.
5009 Yields of 98-99 percent were achieved for platinum electrodes at pH 2-5 when the plating time of
5010 up to 20 hours was used. In 2 M sulfuric acid containing traces of fluoride, metallic technetium
5011 instead of the dioxide is deposited on the electrode.
5012 Methods of Analysis
5013 "Tc is analyzed by ICP-MS or from its beta emission. No gamma rays are emitted by this
5014 radionuclide. For ICP-MS analysis, technetium is stripped from an extraction chromatography
5015 resin and measured by the spectral system. The results should be corrected for interference by
5016 "Ru, if present. For beta analysis, technetium can be electrodeposited on a platinum disc and beta
5017 counted. Alternatively, it is collected by extraction-chromatography techniques. The resin from a
5018 column or the disc from a filtration system is placed in a liquid scintillation vial and counted.
5019 99mjc (ti/2=g oh), measured by gamma-ray spectrometry, can be used as a tracer for measuring the
5020 chemical yield of "Tc procedures. Beta emission from the tracer should then be subtracted from
MARLAP JULY 2001
DO NOT CITE OR QUOTE 14-172 DRAFT FOR PUBLIC COMMENT
-------
Separation Techniques
5021 the total beta count when measuring "Tc. Alternatively, samples are counted immediately after
5022 isolation and concentration of technetium to determine the chemical recovery, then the 99mTc is
5023 allowed to decay before analysis of the "Tc. A widely used medical application is the technetium
5024 generator. 98Mo is neutron irradiated and chemically oxidized to "MoO42". This solution is ion
5025 exchange onto an acid-washed alumina column. After about 1.25 days, the activity of 99mTc has
5026 grown-in to its maximum concentration. The 99Tc is eluted with a 0.9% solution of NaCl, while
5027 the 99Mo remains on the column. The column may have its 99mTc removed after another 1.25
5028 days, but at a slightly smaller concentration. The 99mTc thus separated is carrier free. This process
5029 historically was referred to as "milking," and the alumina column was call the "cow."
5030 Compiled from: Anders, 1960; CRC, 1998-99; Choppin et al., 1995; Cobble, 1964;
5031 Considine and Considine, 1983; Coomber, 1975; Cotton and Wilkinson, 1988; DOE, 1990
5032 and 1997, 1995, 1997; Ehmann and Vance, 1991; Fried, 1995; Greenwood and Earnshaw,
5033 1984; Hassinsky and Adloff, 1965; Kleinberg et al., 1960; Lindsay, 1988; SCA, 2001; and
5034 Wahl and Bonner, 1951.
5035 14.10.9.9 Thorium
5036 Thorium, with an atomic number of 90, is the second member in the series of actinide elements.
5037 It is one of only three of the actinides—thorium, protactinium, and uranium—that occur in nature
5038 in quantities sufficient for practical extraction. In solution, in all minerals, and in virtually all
5039 compounds, thorium exists in the +4 oxidation state; it is the only actinide exclusively in the +4
5040 state in solution.
5041 Isotopes
5042 There are 24 isotopes of thorium ranging inclusively from 213Th to 236Th; all are radioactive.
5043 232Th, the parent nuclide in the natural decay series, represents virtually 100 percent of the
5044 thorium isotopes in nature, but there are a trace amounts of 227Th, 228Th, 230Th, 231Th, and 234Th.
5045 The remaining isotopes are artifacts. The most important environmental contaminants are 232Th
5046 and 230Th, (a member of the 238U decay series). They have half-lives of 1.41 x 1010 years and
5047 75,400 years, respectively.
5048 Occurrence and Uses
5049 Thorium is widely but sparsely dispersed in the Earth's crust. At an average concentration of
5050 approximately 10 ppm, it is over three times as abundant as uranium. In the ocean and rivers,
5051 however, its concentration is about one-thousandth that of uranium (about 10"8 g/L) because its
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 14-173 DO NOT CITE OR QUOTE
-------
Separation Techniques
5052 compounds are much less soluble under environmental conditions. There are six minerals whose
5053 essential element is thorium; thorite (uranothorite) and thorianite are common examples. Several
5054 lanthanum and zirconium minerals are also thorium-bearing minerals; examples include
5055 monazite sand and uraninite. In each mineral, thorium is present as its oxide, thorium dioxide
5056 (ThO2). Monazite sand is the most common commercial mineral, but thorite is also a source of
5057 thorium.
5058 Thorium is extracted from its minerals with hot sulfuric acid or hot concentrated alkali,
5059 converted into thorium nitrate [Th(NO3)4] (its chief commercial compound), extracted with
5060 organic solvents (commonly kerosene containing tributylphosphate), stripped from the organic
5061 phase by alkali solutions, and crystallized as thorium nitrate or precipitated with oxalate. The
5062 metal can be produced by electodeposition from the chloride or fluoride dissolved in fused alkali
5063 halides or by thermoreduction of thorium compounds by calcium (1,000-1,200 °C). Thorium can
5064 also be produced as a by-product in the production of other valuable metals such as nickel,
5065 uranium, and zirconium, in addition to the lanthanides. Unextracted minerals or partially
5066 extracted mill tailings represent some forms of thorium contaminants found in the environment.
5067 Very insoluble forms of thorium hydroxide [Th(OH)4] are other common species found.
5068 Metallic thorium has been used as an alloy in the magnesium industry and as a deoxidant for
5069 molybdenum, iron, and other metals. Because of its high density, chemical reactivity, poor
5070 mechanical properties, and relatively high cost, it is not used as a structural material. Thorium
5071 dioxide is a highly refractory material with the highest melting point among the oxides, 3,390
5072 °C. It has been used in the production of gas mantles, to prevent crystallization of tungsten in
5073 filaments, as furnace linings, in nickel alloys to improve corrosion resistance, and as a catalyst in
5074 the conversion of methanol to formaldehyde. 232Th is a fuel in breeder reactors. The radionuclide
5075 absorbs slow neutrons, and with the consecutive emission of two beta particles, it decays to 233U,
5076 a fissionable isotope of uranium with a half-life of 159,000 years.
5077 Solubility of Compounds
5078 Thorium exists in solution as a highly charged ion and undergoes extensive interaction with
5079 water and with many anions. Few of the compounds are water soluble; soluble thorium
5080 compounds include the nitrate [Th(NO3)4], sulfate [Th(SO4)2], chloride (ThCl4), and perchlorate
5081 [Th(ClO4)4]. Many compounds are insoluble in water and are used in the precipitation of thorium
5082 from solution, including the hydroxide [Th(OH)4], fluoride (ThF4), iodate [Th(IO3)4], oxalate
5083 [Th(C2O4)2], phosphate [Th3(PO4)4], sulfite [Th(SO3)2], dichromate [Th(Cr2O7)2], potassium
5084 hexafluorothorionate [K2ThF6], thorium ferrocyonide (II) [ThFe(CN)6], and thorium peroxide
5085 sulfate [Th(OO)2SO4].
MARLAP JULY 2001
DO NOT CITE OR QUOTE 14-174 DRAFT FOR PUBLIC COMMENT
-------
Separation Techniques
5086 The thorium ion forms many complex ions, chelates, and solvated species that are soluble in
5087 organic solvents. This property is the basis of many procedures for the separation and purification
5088 of thorium (see below). For example, certain ions, such as nitrate and sulfate, form large
5089 unsolvated complex ions with thorium that are soluble in organic solvents. Chelates of 1,3-
5090 diketones, such as acetylacetone (acac) and 2-thenoyltrifluoroacetone (TTA), form neutral
5091 molecular chelates with the thorium ion that are soluble. In addition, many neutral organic
5092 compounds have strong solvating properties for thorium, bonding to the thorium ion in much the
5093 same way water solvates the ion at low pH. Tributylphosphate (TBP), diethyl ether, methyl ethyl
5094 ketone, mesityl oxide, and monoalkyl and dialkyl phosphates are examples of such compounds.
5095 Review of Properties
5096 Thorium is the first member of the actinide series of elements that includes actinium (Ac),
5097 uranium, and the transuranium elements. Thorium is a bright, silver-white metal with a density
5098 above 11 g/cm3. It tarnishes in air, forming a dark gray oxide coating. The massive metal is
5099 stable, but in finely divided form and as a thin ribbon it is pyrophoric and forms thorium oxide
5100 (ThO2). Thorium metal dissolves in hydrochloric acid, is made passive by nitric acid, but is not
5101 affected by alkali. It is attacked by hot water and steam to form the oxide coating and hydrogen,
5102 but its reactions with water are complicated by the presence of oxygen. Thorium has four valence
5103 electrons (6d27s2). Under laboratory conditions, chlorides, bromides, and iodides of the bi- and
5104 trivalent state have been prepared. In aqueous solution and in most compounds, including all
5105 those found in nature, thorium exists only in the +4 oxidation state; its compounds are colorless
5106 in solution unless the anion provides a color. Thorium forms many inorganic compounds in acid
5107 solution.
5108 Solution Chemistry
5109 Because the only oxidation state of thorium in solution is the +4 state, its chemistry is not
5110 complicated by oxidation-reductions reactions that might produce alternate species in solution.
5111 With the +4 charge and corresponding charge-to-radius ratio of 4.0, however, thorium forms very
5112 stable complex ions with halides, oxygen-containing ligands, and chelating agents. Although
5113 Th+4 is large (0.99 A; 0.099 nm; 99 pm) relative to other +4 ions (Ti, Zr, Hf, Ce) and therefore
5114 more resistant to hydrolysis, as a highly charged ion, it hydrolyzes extensively in aqueous
5115 solutions above pH 3 and tends to behave more like a colloid than a true solution. The
5116 concentration of Th+4 is negligible under those conditions. Below pH 3, however, the
5117 uncomplexed ion is stable as the hydrated ion, Th(H2O)8 Or9+4.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 14-175 DO NOT CITE OR QUOTE
-------
Separation Techniques
5118 COMPLEXATION. Thorium has a strong tendency to form complex ions in solution. The presence
5119 of HF forms very stable complex ions, for example, with one, two, or three ligands:
5120 Th+4 + HF - ThF+3 + H+1
5121 ThF+3 + HF - ThF2+2 + H+1
5122 ThF2+2 + HF - ThF3+1 + H+1
5123 These complex ions represent the predominant species in solutions containing HF. Stable
5124 complex ions also form with oxygen-containing ligands such as nitrate, chlorate, sulfate,
5125 bisulfate, iodate, carbonate, phosphate, most carboxylate anions, and chelate anions. Some
5126 chelating agents such as salicylate, acetylacetonate (acac), theonyltrifluoroacetonate (TTA), and
5127 cupferron form complexes that are more soluble in organic solvents, This property is the basis of
5128 several radiochemical isolation methods for thorium. Through the formation of soluble complex
5129 ions, chelating agents found in some industrial wastewater or natural water samples will interfere
5130 to varying degrees with the isolation of thorium by ferric hydroxide [Fe(OH)3] coprecipitation.
5131 Alternative isolation methods should be used, such as coprecipitation from an acidic solution
5132 with an alternative reagent. Protonation of the anionic form of chelates with acid renders them
5133 useless as chelating agents. Other complexing agents also interfere with precipitation by the
5134 formation of soluble ions. Thorium, for example, does not precipitate with oxalate in the
5135 presence of carbonate ions. A procedure for separating thorium from rare-earth ions takes
5136 advantage of the formation of a soluble thorium-EDTA complex that inhibits thorium
5137 precipitation when the rare-earth ions are precipitated with phosphate. The presence of high
5138 concentrations of other complexing agents such as phosphate, chloride, and other anions found in
5139 some samples takes thorium into a completely exchangeable form when it is solubilized in high-
5140 concentrati on nitri c aci d.
5141 HYDROLYSIS. Beginning at pH 3, thorium ions undergo extensive hydrolysis to form monomeric
5142 and polymeric complexes in solution, leaving little (approximately 5 x 10"6 M) Th+4in a saturated
5143 solution at pH 3. Tracer solutions containing 234Th can be added at pH 2 to allow equilibration
5144 because it is not likely to occur if part of the thorium is hydrolyzed and bound in polymeric
5145 forms.
5146 The hydrolysis process is complex, depending on the pH of the solution and its ionic strength.
5147 Several species have been proposed: three are polynuclear species, Th2(OH)2+6, Th4(OH)8+8, and
5148 Th6(OH)15+9; and two are monomeric species, Th(OH)+3 and Th(OH)2+2. The monomeric species
5149 are of minor importance except in extremely dilute solutions, but they become more important as
5150 the temperature increases. The presence of chloride and nitrate ion diminishes hydrolysis,
5151 because the formation of corresponding complex ions markedly suppresses the process.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 14-176 DRAFT FOR PUBLIC COMMENT
-------
Separation Techniques
5152 Hydrolysis increases with increasing hydroxide concentration (pH), and eventually polymeriza-
5153 tion of the species begins. At a pH of about 5, irreversible hydrolysis produces an amorphous
5154 precipitate of thorium hydroxide, a polymer that might contain more than 100 thorium atoms.
5155 Just before precipitation, polymerization slows and equilibration might take weeks or months to
5156 obtain.
5157 Routine fuming of a sample containing organic material with nitric acid is recommended after
5158 addition of tracer, but before separation of thorium as a hydroxide precipitate because there is
5159 evidence for lack of exchange between added tracer and isotope already in solution. Complexing
5160 with organic substances in the initial solution or existence of thorium in solution as some
5161 polymeric ion have been suggested as the cause.
5162 ADSORPTION. The insoluble hydroxide that forms in solution above pH 3 has a tendency to
5163 coagulate with hydrated oxides such as ferric oxide. The high charge of the thorium cation (+4),
5164 high charge-to-radius ratio, and tendency to hydrolyze all contribute to the ability of thorium to
5165 adsorb on surfaces by ion-exchange mechanisms or chemical adsorption mechanisms. These
5166 adsorption properties greatly affect the interaction of thorium with ion-exchange resins and
5167 environmental media such as soil.
5168 Dissolution of Samples
5169 Thorium samples are ignited first to remove organic materials. Most compounds will decompose
5170 when sintered with sodium peroxide (Na2O2), and most thorium minerals will yield to alternate
5171 sodium peroxide sintering and potassium pyrosulfate (K2S2O7) fusion. It is often necessary to
5172 recover thorium from hydrolysis products produced by these processes. The hydrolysis products
5173 are treated with hydrofluoric acid, and thorium is recovered as the insoluble fluoride. Rock
5174 samples are often dissolved in hydrofluoric acid containing either nitric acid or perchloric acid.
5175 Monazite is dissolved by prolonged sintering or with fuming perchloric or sulfuric acid. Thorium
5176 alloys are dissolved in two steps, first with aqua regia (nitric and hydrochloric acid mixture)
5177 followed by fusion with potassium pyrosulfate. Thorium targets are dissolved in concentrated
5178 nitric acid containing hydrofluoric acid, mantles in nitric or sulfuric acid, and tungsten filaments
5179 with aqua regia or perchloric acid.
5180 Separation Methods
5181 PRECIPITATION AND COPRECIPITATION. Precipitation and coprecipitation are used to separate and
5182 collect thorium from aqueous solutions either for further treatment in an analytical scheme or for
5183 preparation of a sample for counting. Formation of insoluble salts is used to precipitate thorium
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 14-177 DO NOT CITE OR QUOTE
-------
Separation Techniques
5184 from solution; examples include the hydroxide, peroxide, fluoride, iodate, oxalate, and
5185 phosphate, among others. Tracer quantities of thorium are commonly coprecipitated with
5186 lanthanum fluoride (LaF3), neodymium fluoride (NdF3), and cerium fluoride (CeF3) in separation
5187 schemes and to prepare samples for alpha counting. Tracer quantities are also carried with
5188 calcium oxalate [Ca(C2O4)], ferric hydroxide [Fe(OH)3], zirconium iodate (ZrIO4), zirconium
5189 phosphate (Zr3PO4), and barium sulfate (BaSO4).
5190 ION EXCHANGE. The highly charged thorium cation is strongly adsorbed onto cation exchangers
5191 and is more difficult to elute than most other ions. Its strong adsorption property makes it
5192 possible to remove trace quantities of thorium from a large volume of solution onto small
5193 amounts of ion-exchange resin. Washing the resin with mineral acids of various concentrations
5194 separates thorium from less strongly bound cations that elute from the resin. For example, Th+4
5195 remains bonded at all hydrochloric concentrations, allowing other cations to be eluted at different
5196 concentrations of acid. Thorium is eluted by complexing agents such as citrate, lactate, fluoride,
5197 carbonate, sulfate, or oxalate that reduce the net charge of the absorbing species, causing reversal
5198 of the adsorption process.
5199 Anion exchangers are useful for separating thorium, but the contrasting behavior of thorium with
5200 the resin depends on whether hydrochloric or nitric acid is used as an eluent. In hydrochloric
5201 acid, several metal ions, unlike thorium, form negative complexes that can be readily removed
5202 from a thorium solution by adsorption onto the anionic exchanger. Thorium forms positively
5203 charged chlorocation complexes or neutral thorium chloride (ThCl4) in the acid and is not
5204 adsorbed by the resin at any hydrochloric acid concentration. In contrast, thorium forms anionic
5205 complexes in nitric acid solution that adsorb onto the exchanger over a wide range of nitric acid
5206 concentrations, reaching a maximum affinity near 7 M nitric acid. Behavior in nitric acid solution
5207 is the basis for a number of important radiochemical separations of thorium from rare earths,
5208 uranium, and other elements.
5209 ELECTRODEPOSITION. Thorium separated from other actinides by chemical methods can be
5210 electrodeposited for alpha counting from a dilute solution of ammonium sulfate adjusted to a pH
5211 of 2. The hydrous oxide of thorium is deposited in one hour on a highly polished platinum disc
5212 serving as the cathode of an electrolytic cell. The anode is a platinum-iridium alloy.
5213
5214 SOLVENT EXTRACTION. Many complexes and some compounds of thorium can be extracted from
5215 aqueous solutions into a variety of organic solvents. The TTA (a-theonyltrifluoroacetone)
5216 complex of metals is widely used in radiochemistry for the separation of ions. Thorium can be
5217 separated from most alkali metal, alkaline earth, and rare earth metals after the complex is
MARLAP JULY 2001
DO NOT CITE OR QUOTE 14-178 DRAFT FOR PUBLIC COMMENT
-------
Separation Techniques
5218 quantitatively extracted into benzene above pH 1. Backwashing the organic solution with dilute
5219 acid leaves the more soluble ions in benzene.
5220 Extraction of nitrates and chlorides of thorium into organic solvents from the respective acid
5221 solutions is widely used for isolation and purification of the element. One of the most common
5222 processes is the extraction of thorium nitrate from a nitric acid solution with TBP (triisobutyl-
5223 phosphate). TBP is usually diluted with an inert solvent such as ether or kerosene to reduce the
5224 viscosity of the mixture. Dilution reduces the extraction effectiveness of the mixture, but the
5225 solubility of many contaminating ions is greatly reduced, increasing the effectiveness of the
5226 separation when the thorium is recovered by backwashing.
5227 Long-chain amine salts have been very effective in carrying thorium in laboratory and industrial
5228 extraction process using kerosene. Complex sulfate anions of thorium are formed in sulfuric acid
5229 that act as the counter ion to the protonated quaternary amine cation. They accompany the
5230 organic salt into the organic phase.
5231 In recent years, solvent extraction chromatography procedures have been developed to separate
5232 thorium. These procedures use extraction chromatography resins that consist of extractant
5233 materials such as CMPO in TBP or DPPP (dipentylpentylphosphonate), also called DAAP
5234 (diamylamylphosphonate), absorbed onto an inert polymeric material. They are used in a column,
5235 rather than in the traditional batch mode, and provide a rapid efficient method of separating the
5236 radionuclide with the elimination of large volumes of organic waste.
5237 Methods of Analysis
5238 Chemical procedures are used for the analysis of macroscopic quantities of thorium in solution
5239 after it has been separated by precipitation, ion exchange, extraction, and/or extraction chroma-
5240 tography from interfering ions. Gravimetric determination generally follows precipitation as the
5241 oxalate that is calcined to the oxide (ThO2). Numerous volumetric analyses employ EDTA as the
5242 titrant. In the most common spectrometric method of analysis, thorin, a complex organoarsenic
5243 acid forms a colored complex with thorium that is measured in the visible spectrum.
5244 Trace quantities of thorium are measured by alpha spectrometry after chemical separation from
5245 interfering radionuclides. 227Th, 228Th, 230Th, and 232Th are determined by the measurement of
5246 their respective spectral peaks (energies), using 234Th as a tracer to determine the chemical yield
5247 of the procedure. The activity of the tracer is determined by beta counting in a proportional
5248 counter. 234Th also emits gamma radiation that can be detected by gamma spectrometry; however,
5249 the peak can not be measured accurately because of interfering peaks of other gamma-emitting
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 14-179 DO NOT CITE OR QUOTE
-------
Separation Techniques
5250 radionuclides. 229Th is sometimes used as a tracer to determine the chemical yield of the alpha
5251 spectrometric procedure, but it produces considerable recoil that might contaminate the detector.
5252 Compiled from: Ahrland, 1986; Baes and Mesmer, 1976; Cotton, 1991; Cotton and
5253 Wilkinson, 1988; DOE, 1990 and 1997, 1997; EPA, 1980 and 1984; Greenwood, 1984;
5254 Grimaldi, 1961; Hassinsky and Adloff, 1965; Hyde, 1960; Katzin, 1986; Lindsey, 1988.
5255 14.10.9.10 Tritium
5256 Unlike the elements reviewed in this section, tritium the only radionuclide of the element
5257 hydrogen. It contains two neutrons and is represented by the symbols 3H, 3T, or simply, T. The
5258 atom contains only one valence electron so its common oxidation state, besides zero, is +1,
5259 although it can exist in the -1 state as a metal hydride.
5260 Occurrence and Uses
5261 Tritium is found wherever stable hydrogen is found, with and without the other isotopes of the
5262 element (hydrogen and deuterium)—as molecular hydrogen (HT, DT, T2), water (HTO, DTO,
5263 T2O), and inorganic and organic compounds, hydrides and hydrocarbons, respectively, for
5264 example. About 99 percent of the radionuclide in nature from any source is in the form of HTO.
5265 Natural processes account for approximately one T atom per 1018 hydrogen atoms. The source of
5266 some natural tritium is ejection form the sun, but the primary source is from bombardment of 14N
5267 with cosmic neutrons in the upper atmosphere:
5268 147N + > - 3H + 126C
5269 Most tritium from this source appears as HTO.
5270 Tritium is produced in laboratory and industrial processes by nuclear reactions such as:
5271 \D + \D - 3J + \H
5272 For large-scale production of tritium, 6Li alloyed with magnesium or aluminum is the target of
5273 neutrons:
5274 63Li + > - 3J + 42He
MARLAP JULY 2001
DO NOT CITE OR QUOTE 14-180 DRAFT FOR PUBLIC COMMENT
-------
Separation Techniques
5275 The radionuclide is retained in the alloy until released by acid dissolution of the target. Large
5276 quantities are handled as HT or HTO. HTO is formed from HT when it is exposed to oxygen or
5277 water vapor. A convenient way to store tritium is as the hydride of uranium (UT3). It is formed by
5278 reacting the gas with finely divided uranium and is released by heating the compound above 400
5279 °C.
5280 Tritium is also produced in nuclear reactors that contain water or heavy water from the neutron
5281 bombardment of deuterium:
5282
5283 2D +10n - 3T
5284 Most tritium (>99%) in reactors is formed from the fission process as a ternary particle.
5285 The main use for tritium is in fission bombs to boost their yield and in thermonuclear weapons,
5286 the hydrogen bomb. Tritium bombarded with high-energy deuterons undergoes fusion to form
5287 helium and releasing neutrons:
5288 \H + \H -42He + >
5289 A tremendous amount of energy is released during the nuclear reaction, much more than the
5290 energy of the bombarding particle. Fusion research on controlled thermonuclear reactions should
5291 lead to an energy source for electrical generation.
5292 Tritium absorbed on metals are a source of neutrons when bombarded with deuterons. Mixed
5293 with zinc sulfide, it produces radioluminescence that is used in luminescent paint and on watch
5294 dials. Gaseous tritium in the presence of zinc sulfide produces a small, permanent light source
5295 found in rifle sights and exit signs. Tritium is also a good tracer since it does not emit gamma
5296 radiation. Hydrological studies with HTO is used to trace geological water and the movement of
5297 glaciers. It is also used as a tracer for hydrogen in chemical studies and biological research. In
5298 medicine, it is used for diagnosis and radiotreatment.
5299 Review of Properties
5300 Tritium decays with a half-life of 12.3 y by emission of a low-energy beta particle to form 3He,
5301 and no gamma radiation is released. The range of the beta particle is low, 6 mm in air and 0.005
5302 mm in water or soft tissue.
5303
5304 The physical and chemical properties of tritium are somewhat different than hydrogen or
5305 deuterium because of their mass differences (isotope effects). Tritium is approximately twice as
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 14-181 DO NOT CITE OR QUOTE
-------
Separation Techniques
5306 heavy as deuterium and three times heavier than hydrogen, and the isotope effect can be large for
5307 mass differences of these magnitudes. In its simple molecular form, tritium exists primarily as T2
5308 or DT. The oxide form is HTO, DTO, or T2O, with higher molecular weights than water (H2O).
5309 Thus molecules of tritiated water are heavier, and any process such as evaporation or distillation
5310 that produces a phase transition results in isotopic fractionation and enrichment of tritium in
5311 water. In a mixture of the oxides, various mixed isotopic water species are generally also present
5312 because of exchange reactions: in any mixture of H2O, D2O, and T2O, HTO and DTO are
5313 found.Molecules of HTO are more stable than H2O or HDO and are not as easily decomposed by
5314 electrolysis, to form hydrogen or oxygen. Electrolysis of a water sample to about five percent of
5315 its original volume, therefore, concentrates tritium by retaining approximately 80 percent of the
5316 tritium from the initial volume. Reaction rates of chemical bonds containing tritium are slower
5317 because of the isotope effect than those of hydrogen. The rates can be as small as 1:64 (T:H), and
5318 these differences should be considered when interpreting tracer studies of reaction mechanisms.
5319 Chemical isotope effects are large in some biological systems. Some algae and bacteria
5320 selectively exchange hydrogen isotopes, and the preference is tritium over deuterium over
5321 hydrogen. Enrichment of tritium can be about 2.5.
5322 Tritium can be introduced into organic compounds by exposing T2 to the compound for a few
5323 days or weeks, irradiation of the compound and a lithium salt with neutrons (recoil labeling), or it
5324 can be selectively introduced into a molecule by chemical synthesis using a molecular tritium
5325 source such as HTO. Beta radiation causes exchange reactions between hydrogen atoms in the
5326 compound and tritium and migration of the isotope within the molecule. Phenol (C6H5OH), for
5327 example, labeled with tritium on the oxygen atom (C6H5OT) will become C6H4TOH and
5328 C6H4TOT. When tritium samples are stored in containers made from organic polymers such as
5329 polyethylene, the container will adsorb tritium, resulting in a decrease in the concentration of
5330 tritium in the sample. Eventually, the tritium atoms will migrate to the outer surface of the
5331 container, and tritium will be lost to the environment. Catalytic exchange also occurs in tritiated
5332 solutions or solutions containing T2 gas. Exchange is very rapid with organic compounds when
5333 H+1 or OH"1 ions or if a hydrogen-transfer agent such as Pt or Pd is present.
5334 Tritium as HT or HTO will absorb on most metallic surfaces. Penetration at room temperature is
5335 very slow, and the radionuclide remains close to the surface. In the form of HTO, it can be
5336 removed with water, or by hydrogen gas in the form of HT. Heating aids the removal. When
5337 tritium is absorbed at elevated temperatures, it penetrates deeper into the surface. Adsorption
5338 under these conditions will result in enough penetration to cause structural damage to the metal,
5339 especially if the process continues for extended periods. Hydrogenous material such as rubber
5340 and plastics will also absorb tritium. It will penetrate into the material, and hydrogenous
5341 materials are readily contaminated deep into the material, and it is impossible to completely
MARLAP JULY 2001
DO NOT CITE OR QUOTE 14-182 DRAFT FOR PUBLIC COMMENT
-------
Separation Techniques
5342 remove the tritium. Highly contaminated metal or plastic surfaces can release some of the loosely
5343 bound tritium immediately after exposure in a process called outgassing.
5344 Pure T2O can be prepared by oxidation of tritium gas with hot copper(II) oxide or direct
5345 combination of the gas with oxygen in the presence of an electrical spark. It is never used for
5346 chemical or biological processes because one milliliter contains 2,650 curies. The liquid is self-
5347 luminescent, undergoes rapid self-radiolysis, and considerable radiation damage is done to
5348 dissolved species. For the same reason, very few compounds of pure tritium have ever been
5349 prepared or studied.
5350 The radiotoxity of tritium is rated medium. Tritium is not a hazard outside the body. Gamma
5351 radiation is not released by its decay. The beta emission is low in energy compared to most beta
5352 emitters and readily stopped by the outer layer of skin. Only ingested tritium can be a hazard.
5353 Exposure to tritium is primarily in the form of HT gas or HTO water vapor, although T2 and T2O
5354 may be present. Only about 0.005 percent of the activity of inhaled HT gas is incorporated into
5355 lung tissue, and most is exhaled. Tritiated water vapor, however, is almost 100 percent absorbed
5356 from inhalation or ingestion. In addition, tritiated water can be absorbed through the skin or
5357 wounds. Not all gloves will prevent exposure because of the ability of tritium to be absorbed by
5358 the gloves themselves. Tritium is found in tissue wherever hydrogen is found. The biological
5359 half-life is about ten days, but the value varies significantly, depending on exertion rates and
5360 fluid intake.
5361 Environmental tritium is formed in the gaseous and aqueous forms, but over 99 percent of tritium
5362 from all sources is found in the environment after exchange with hydrogen in water in the form
5363 of HTO. It is widely distributed in the surface waters of the earth and makes a minor contribution
5364 to the activity of ocean water. It can also be found in laboratories and industrial sites in the form
5365 of metal hydrides, tritiated pump oil, and tritiated gases such as methane and ammonia.
5366 Separation Methods
5367 DISTILLATION. Tritium in water samples is essentially in the form of HTO. It can be removed
5368 quantitatively from aqueous mixtures by distillation to dryness, which also separate it form other
5369 radionuclides. Volatile iodine radionuclides are precipitated as silver iodide before distillation, if
5370 they are present. The aqueous solution is usually distilled, however, from a basic solution of
5371 potassium permangenate, which will oxidize radionuclides, such as iodine and carbon, and
5372 oxidize organic compounds that might interfere with subsequent procedures, liquid scintillation
5373 counting, for example. Charcoal can also be added to the distillation mixture as an additional
5374 measure to remove organic material. Contaminating tritium in soil samples can be removed by
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 14-183 DO NOT CITE OR QUOTE
-------
Separation Techniques
5375 distillation from similar aqueous mixtures. All tritium in soil samples might not be recovered by
5376 this method, however, if the tritium is tightly bound to the soil matrix. Tritium also can be
5377 removed by distillation of an azeotrope mixture formed with toluene or cyclohexane. In some
5378 procedures, tritium is initially separated by distillation and then concentrated (enriched) by
5379 electrolysis in an acid or base solution. Recovery of tritium from the electrolytic cell for analysis
5380 is accomplished by a subsequent distillation.
5381 DECOMPOSITION. Organically bound tritium (OBT) in vegetation, food, and tissue samples can be
5382 removed by combustion. The sample is freeze dried (lyophilized), and the water from the process
5383 is collected in cold traps for tritium analysis. The remaining solid is collected as a pellet, which is
5384 burned at 700 °C in a highly purified mixture of argon and oxygen in the presence of a copper(I)
5385 oxide (CuO) catalyst, generated on a copper screen at the temperature of the process. Water from
5386 the combustion process, containing tritium from the pellet, and water from the freeze-drying
5387 process is analyzed for tritium by liquid scintillation counting.
5388 Tritium in HTO can be reduced to TH by heating with metals, such as magnesium, zinc, or
5389 calcium, and analyzed as a gas.
5390 CONVERSION TO ORGANIC COMPOUNDS. Compounds that react readily with water to produce
5391 hydrogen derivatives can be used to isolate and recover tritium that is present in the HTO form.
5392 Organic compounds containing magnesium (Grignard reagents) with relatively low molecular-
5393 weights will react spontaneously with water and produce a gaseous product containing hydrogen
5394 from the water. Tritium from HTO in a water sample will be included in the gaseous sample. It is
5395 collected after formation by condensation in a cold trap and vaporized into a gas tube for
5396 measurement. Grignard reagents formed from butane, acetylene, and methane can be used in this
5397 method. Tritiated butane is produced by the following chemical reaction:
5398 C4H9MgBr + THO - C4H9T + Mg(OH)Br
5399 Inorganic compounds can also be use to produce gaseous products:
5400 A14C3 + 3 HTO + 9 H2O - 3 CH3T + 4 A1(OH)3
5401 EXCHANGE. Methods to assess tritium in compounds take advantage of exchange reactions to
5402 collect the radionuclide in a volatile substance that can be collected in a gas tube for measure-
5403 ment. Acetone is one compound that easily exchanges tritium in an acid or base medium and is
5404 relatively volatile.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 14-184 DRAFT FOR PUBLIC COMMENT
-------
Separation Techniques
5405 Methods of Analysis
5406 Tritium is collected primarily as HTO along with water (H2 O) by distillation and then
5407 determined from its beta emission in a liquid scintillation system. No gamma rays are emitted.
5408 The distillation process is usually performed from a basic solution of potassium permangenate to
5409 oxidize radionuclides and organic compounds, preventing them from distilling over and
5410 subsequently interfering with counting. Charcoal can also be added to the distillation mixture as
5411 an additional measure to remove organic material. Volatile iodine radionuclides can be
5412 precipitated as silver iodide before distillation.
5413 Compiled from: Choppin et al., 1995; Cotton and Wilkinson, 1988; DOE, 1994; Duckworth,
5414 1995; Greenwood and Earnshaw, 1984; Hampel, 1968; Hassinky and Adloff, 1965; Kaplan,
5415 1995; Lindsay, 1988; Mitchell, 1961; Passo and Cook, 1994.
5416 14.10.9.11 Uranium
5417 Uranium, atomic number 92, is the last naturally occurring member of the actinide series and the
5418 precursor to the transuranic elements. Three isotopes are found in nature, and uranium was the
5419 active constituent in the salts whose study led to the discovery of radioactivity by Becquerel in
5420 1896.
5421 Isotopes
5422 There are 19 isotopes of uranium with mass numbers ranging from 222 to 242. All isotopes are
5423 radioactive with half-lives range ranging from microseconds to billions of years. 235U (0.72%)
5424 and 238U (99.27%) are naturally occurring as primordial uranium. 234U has a natural abundance of
5425 0.0055%, but is present as a part of the 238U decay natural decay chain. The 234U that was formed
5426 at the time the earth was formed has long since decayed. The half-lives of these principal
5427 isotopes of uranium are listed below.
Alpha Decay Spontaneous Fission
5428 Isotope Half-Life Half-Life
5429 234 2.46 x 105years 1.42 x 1016years
5430 235 7.04 x 108years 9.80 x 1018years
5431 238 4.48 x 109years 8.08 x 1015years
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 14-185 DO NOT CITE OR QUOTE
-------
Separation Techniques
5432 These isotopes have two different decay modes. Each decay mode has its own characteristic half-
5433 life. As seen above the alpha decay mode is the most significant, since it has the shortest half-life
5434 for each of these isotopes.
5435 Another isotope of uranium of significance is 232U (half-life 69.8 years). It is used as a tracer in
5436 uranium analyses and is also an alpha emitter so it can be determined concurrently with the major
5437 uranium isotopes by alpha spectrometry.
5438 235U and artificially produced 233U are fissionable material on bombardment with slow (thermal)
5439 neutrons. Other uranium radionuclides are fissionable with fast moving neutrons, charged
5440 particles, high-energy photons, or mesons. 238U and 235U are both parents of natural radioactive
5441 decay series, the uranium series of 238U that eventually decays with alpha and beta emissions to
5442 stable 206Pb and the actinium series of 235U that decays to 207Pb.
5443 Occurrence and Uses
5444 Naturally occurring uranium is believed to be concentrated in the earth's crust with an average
5445 concentration of approximately 4 ppm. Granite rocks contains up to 8 ppm or more, and ocean
5446 water contains 0.0033 ppm. Many uranium minerals have been discovered. Among the better
5447 known are uraninite, carnotite, adavidite, pitchblende, and coffinite. The latter two minerals are
5448 important commercial sources of uranium. It is also found in phosphate rock, lignite, and
5449 monazite sands and is commercially available from these sources. The artificial isotope, 233U, is
5450 produced from natural 232Th by absorption of slow neutrons to form 233Th, which decays by the
5451 emission of two beta particles to 233U.
5452 Uranium is extracted from uranium minerals, ores, rocks, and sands by numerous chemical
5453 extraction (leaching) processes. The extraction process is sometimes preceded by roasting the ore
5454 to improve the processing characteristic of the material. The extraction process uses either an
5455 acid/oxidant combination or sodium carbonate treatment, depending on the nature of the ore, to
5456 convert the metal to a soluble form of the uranyl ion. Uranium is recovered from solution by
5457 precipitating the uranate salt with ammonia or sodium hydroxide solution. Ammonium uranate is
5458 know as yellow cake. The uranate salt is solubilized to give a uranyl nitrate solution that is
5459 further purified by extraction into an organic phase to separate the salt from impurities and
5460 subsequent stripping with water. It is precipitated as a highly purified nitrate salt that is used to
5461 produce other uranium compounds—uranium trioxide (UO3) by thermal processing or uranium
5462 dioxide (UO2) on reduction of the tri oxide with hydrogen. Uranium tetrafluoride (UF4) is
5463 prepared, in turn, form the dioxide by treatment with hydrogen fluoride. The metal is recovered
MARLAP JULY 2001
DO NOT CITE OR QUOTE 14-186 DRAFT FOR PUBLIC COMMENT
-------
Separation Techniques
5464 by fused-salt electrolysis in molten sodium chloride-calcium chloride or reduction with more
5465 active metals such as calcium or magnesium (Ames Process) in an inert atmosphere at 1,000 °C.
5466 Early in the twentieth century, the only use of uranium was in the production of a brown-yellow
5467 tinted glass and glazes; it was a byproduct of the extraction of radium, which was used for
5468 medicinal and research purposes. Since the mid-twentieth century, the most important use of
5469 uranium is as a nuclear fuel, directly in the form of 233U and 235U, fissionable radionuclides, and
5470 in the form of 238U that can be converted to fissionable 239Pu by thermal neutrons in breeder
5471 reactors. Depleted uranium, uranium whose 235U content has been reduced to below about 0.2
5472 percent, the majority of waste from the uranium enrichment process, is used in shielded
5473 containers to transport radioactive materials, inertial guidance devices, gyro compasses,
5474 counterweights for aircraft control surfaces, ballast for missile reentry vehicles, fabrication of
5475 armor-piercing conventional weapons, and tank armor plating. Uranium metal is used as a X-ray
5476 target for production of high-energy X-rays, the nitrate salt as a photographic toner, and the
5477 acetate is used in analytical chemistry.
5478 Solubility of Compounds
5479 Only a small number of the numerous uranium compounds are soluble in water. Except for the
5480 fluorides, the halides of uranium (HI and IV) are soluble, as are the chloride and bromide of
5481 uranium (V) [UOX2] and the fluoride, chloride, and bromide of uranium (VI) [UO2X2]. Several
5482 of the uranyl (UO2+2) salts of polyatomic anions are also soluble in water: the sulfate,
5483 bicarbonate, acetate, thiocyanate, chromate, tungstate, and nitrate. The latter is one of the most
5484 water-soluble uranium compounds.
5485 Review of Properties
5486 Uranium is a dense, malleable and ductile metal that exists in three allotropic forms: alpha, stable
5487 to 688 °C where it forms the beta structure, which becomes the gamma structure at 776 °C. It is
5488 a poor conductor of electricity. The metal absorbs gases and is used to absorb tritium. Uranium
5489 metal tarnishes readily in an oxidation process when exposed to air. It burns when heated to 170
5490 °C, and the finely divided metal is pyrophoric. Uranium slowly decomposes water at room
5491 temperature, but rapidly at 100 °C. Under a flux of neutrons and other accelerated particles,
5492 atoms of uranium are displaced from their equilibrium position in its metallic lattice. With high
5493 temperatures and an accumulation of fission products, the metal deforms and swells, becoming
5494 twisted, porous, and brittle. The problem can be avoided by using some of its alloys, particularly
5495 alloys of molybdenum and aluminum.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 14-187 DO NOT CITE OR QUOTE
-------
Separation Techniques
5496 Uranium forms a large number of binary and ternary alloys with most metals. It also form
5497 compounds with many metals: aluminum, bismuth, cadmium, cobalt, gallium, germanium, gold,
5498 indium, iron, lead, magnesium, mercury, nickel, tin, titanium, zinc, and zirconium. Many binary
5499 compounds of the nonmetals are also known: hydrides, borides, carbides, nitrides, silicides,
5500 phosphides, halides, and oxides. Although other oxides are known, the common oxides are UO2,
5501 UO3, and U3O8. Uranium reacts with acids to form the +4 salts and hydrogen. It is very reactive
5502 as a strong reducing agent.
5503 Uranium compounds are toxic at high concentrations. The physiological damage occurs to
5504 internal organs, especially the kidneys. The radioactivity of natural uranium radionuclides is not
5505 of great concern, although it is high for some artificial isotopes. Natural uranium in the
5506 environment is considered a relatively low hazard, however, because of its very long half-life and
5507 low toxicity at minute concentrations.
5508 Uranium in nature is almost entirely in the IV and VI oxidation states. It occurs as the oxides,
5509 UO2 and U3O8, in the solid state. In ground water under oxic conditions it exists as UO2+2 or
5510 complexes of carbonate such as UO2(CO3)3"4. Complex formation increases its solubility under
5511 all conditions in normal groundwater and even under fairly strong reducing conditions. The
5512 amount associated with paniculate matter is small in natural oxic waters. In some waters,
5513 solubility may be limited, however, by formation of an uranyl silicate species. Uranium in
5514 general is poorly absorbed on geologic media under oxic conditions, especially at moderate and
5515 high concentrations and in the presence of high carbonate concentrations. A significant
5516 adsorption occurs at pH above about 5 or 6 because of formation of hydrolytic complexes.
5517 Reduction to the IV oxidation state would increase uptake in the environmental pH range.
5518 Solution Chemistry
5519 The radiochemistry of uranium is complicated because of the multiple oxidation states that can
5520 exist in solution and the extensive complexation and hydrolytic reactions the ions are capable of
5521 undergoing in solution. Four oxidation states are possible: +3, +4, +5, and +6; the latter two exist
5522 as oxycations: UO2+1 and UO2+2, respectively. Their stabilities vary considerably, and the +4 and
5523 +6 states are stable in solution under certain conditions; oxidation-reduction reagents are used to
5524 form and maintain these ions in solution. Each ion has different chemical properties, and those of
5525 the +4 and +6 states have been particularly exploited to stabilize, solubilize, separate, and collect
5526 uranium. The multiple possibilities of oxidation state, complexation, and hydrolysis should be
5527 carefully considered when planning any radiochemical procedures.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 14-188 DRAFT FOR PUBLIC COMMENT
-------
Separation Techniques
5528 OXIDATION-REDUCTION BEHAVIOR. The multiple oxidation states can be exploited during
5529 separation procedures by taking advantage of their different chemical properties. Thorium can be
5530 separated from uranium, for example, by oxidizing uranium in solution to the +6 oxidation state
5531 with 30 percent hydrogen peroxide (H2O2) and precipitating thorium as the hydroxide; in the +6
5532 state, uranium is not precipitated.
5533 The U+3 ion is an unstable form of uranium, produced in perchlorate or chloride solutions by
5534 reduction of UO2+2 electrochemically or with zinc amalgam. It is a powerful reducing agent, and
5535 is oxidized to U+4 by chlorine or bromine. U+3 is slowly oxidized by water with the release of
5536 hydrogen, and oxygen from air causes rapid oxidation. Aqueous solutions are red-brown and are
5537 stable for several days in 1 M hydrochloric acid, especially if kept cold; rapid oxidation occurs in
5538 more concentrated acid solutions.
5539 The tetrapositive uranous ion, U+4, is produced by dissolving water-soluble salts of the ion in
5540 solution, dissolving uranium metal with sulfuric or phosphoric acid, reduction of UO2+1 during its
5541 disproportionation reaction, reduction of UO2+2 by Cr+2 or Ti+3, or oxidation of U+3. The tetraposi-
5542 tive ion is green in solution. The ion is stable, but slowly oxidizes by oxygen from air to the +6
5543 state.
5544 The UO2+1 ion (+5 state) is extremely unstable in solution and exist only as a transient species,
5545 disproportionating rapidly to U+4 and UO2+2 according to the following reaction in the absence of
5546 complicating factors (k=1.7 x 106):
5547 2 UO+1 + 4 H+1 * UO+2 + U+4 + 2 H9O
2
5548 Maximum stability is observed in the pH range 2-4 where the reaction is considerably slower.
5549 Solutions of UO2+1 are prepared by the dissolution of UC15 or reduction of UO2+2 ions
5550 electrochemically or with U+4 ions, hydrogen, or zinc amalgam.
5551 The +6 oxidation state of uranium is generally agreed to be in the form of the dioxo or uranyl ion,
5552 UO2+2. As the only oxidation state stable in contact with air, it is very stable in solution and
5553 difficult to reduce. Because of its exceptional stability, the uranyl ion plays a central role in the
5554 radiochemistry of uranium. It is prepared in solution by the dissolution of certain water-soluble
5555 salts: nitrate, halides, sulfate, acetate, and carboxylates; by dissolution of uranium +6
5556 compounds; and oxidation of lower-oxidation state ions already in solution, U+4 with nitric acid
5557 for example. Its solutions are yellow in color.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 14-189 DO NOT CITE OR QUOTE
-------
Separation Techniques
5558 COMPLEXATION. Uranium ions form numerous complex ions, and the solution chemistry of
5559 uranium is particularly sensitive to complexing agents present. Complex-ion chemistry is very
5560 important, therefore, to the radiochemical separation and determination of uranium.
5561 Complexation, for example, provides a method to prevent the removal of uranium ions or its
5562 contaminants from solution and can influence the stability of ions in solution.
5563 Among the oxidation states exhibited in solution, the tendency for formation of anionic
5564 complexes is:
5565 U+4 > UO2+2 > U+3 > UO2+1,
5566 while the order of stability of the anionic complexes is represented by:
5567 fluoride > nitrate > chloride > bromide > iodide > perchlorate > carbonate > oxalate > sulfate.
5568 Numerous organic complexes form, including citrate, tartrate, and EDTA, especially with UO2+2.
5569 There is evidence for only a few complexes of U+3, cupferron and chloride for example. In
5570 contrast, tetrapositive uranium, U+4, forms complexes with a wide variety of anions, and many
5571 are stable: halides—including fluoride (up to eight ligands, UF8"4)—chloride, and bromide;
5572 thiocyanate; and oxygen-donors, nitrate, sulfates, phosphates, carbonate, perchlorate, and
5573 numerous carboxylates: acetate, oxalate, tartrate, citrate, and lactate. The low charge on UO2+1
5574 precludes the formation of very stable complexes. Fluoride (from hydrogen fluoride) is notable,
5575 however, in its ability to displace oxygen from the ion, forming UFg"1—which inhibits
5576 disproportionation—and precipitating the complex ion from aqueous solution. The uranyl ion,
5577 UO2+2, readily forms stable complexes with a large variety of inorganic and carboxylate anions
5578 very similar to those that complex with U+4. In addition, numerous organic ligands besides
5579 carboxylates are known that contain both oxygen and nitrogen as donor atoms. Complex-ion
5580 formation must be considered, therefore, during precipitation procedures. Precipitation of
5581 uranium ions is inhibited, for example, in solutions containing carbonate, tartrate, malate, citrate,
5582 hydroxylamine, while impurities are precipitated as hydroxides, sulfides, or phosphates.
5583 Conversely, uranium is precipitated with ammonia, while other ions are kept in solution as
5584 complexes of EDTA.
5585 HYDROLYSIS. Some uranium ions undergo extensive hydrolysis in aqueous solution. The
5586 reactions can lead to formation of polymeric products, which form precipitates under certain
5587 conditions. The tendency of the various oxidation states toward hydrolysis, a specific case of
5588 complexation, is, therefore, in the same order as that of complex-ion formation (above).
MARLAP JULY 2001
DO NOT CITE OR QUOTE 14-190 DRAFT FOR PUBLIC COMMENT
-------
Separation Techniques
5589 Little data are available on the hydrolysis of U+3 ion because it is so unstable in solution.
5590 Qualitative evidence indicates, however, that hydrolysis is about that to be expected for a +3 ion
5591 of its size, that is, as a much weaker acid than most other metals ions of this charge. The U+4 ion
5592 is readily hydrolyzed in solution, but exist as the unhydrolyzed, hydrated ion in strongly acidic
5593 solutions. Hydrolysis begins at pH
-------
Separation Techniques
5625 alkaline carbonates, hydroxides, and peroxides. Liquid biological samples may also be extracted
5626 to remove uranium, or the solid sample can be ashed by a wet or dry process and dissolved in
5627 acid solution. Wet ashing is carries out with nitric acid and completed with perchloric acid, but
5628 extreme caution should be used when using perchloric acid in the presence of organic material.
5629 Separation Methods
5630 PRECIPITATION AND COPRECIPITATION. There are a large number of reagents that will precipitate
5631 uranium over a wide pH range. The number of reagents available coupled with the two possible
5632 oxidation states of uranium in solution and the complexing properties of the ions provide many
5633 opportunities to separate uranium from other cations and the two oxidation states from each
5634 other. Precipitation can be inhibited, for example, by the presence of complexing agents that
5635 form soluble complexes. Complexes that form weak complexes with uranium and strong
5636 complexes with other cations allow the separation of uranium by its precipitation while the
5637 complexed cations remain in solution. EDTA has been used in this manner to separate uranium
5638 form many of the transition metals and alkaline earths. In contrast, uranium forms a very strong
5639 soluble complex with carbonate, and this property has been used to keep uranium in solution
5640 while ammonium hydroxide precipitates iron, titanium, zirconium, and aluminum. In a similar
5641 manner, uranium is separated from other cations as they are precipitated as sulfides or
5642 phosphates. Common precipitating reagents and used for separation include: ammonium
5643 hydroxide, precipitates uranium quantitatively at pH > 4; carbonate, which will form soluble
5644 anionic complexes with uranium (VI) at pH 5 to 11 while many other metals form insoluble
5645 hydroxides; peroxide; oxalic acid, completely precipitate uranium (IV) while uranium (VI) forms
5646 a soluble complex; iodide; iodate; phosphate for uranium (VI) over a wide pH range; sulfate;
5647 cupferron, precipitates uranium (IV) from an acidic solution but uranium (VI) from a neutral
5648 solution; and 8-hydroxyquinoline, which forms a quantitatively precipitate with uranium(VI)
5649 only.
5650 Coprecipitation of uranium is accomplished with several carriers. In the absence of carbonate, it
5651 is quantitatively coprecipitated with ferric hydroxide at pH from 5 to 8. Aluminum and calcium
5652 hydroxide are also employed to coprecipitate uranium. Uranium (VI), however, is only partially
5653 carried by metal hydroxides in the presence of carbonate, and the amount carried decreases as the
5654 concentration of carbonate increases. Small amounts of uranium (VI) coprecipatate with eerie
5655 and thorium fluoride, calcium, zirconium, and aluminum phosphate, barium carbonate, thorium
5656 hexametaphosphate, magnesium oxide, and thorium peroxide. Uranium (IV) is carried on eerie
5657 sulfate, the phosphates of zirconium, bismuth, and thorium, lanthanum and neodymium fluoride,
5658 eerie and zirconium iodates, barium sulfate, zirconium phosphate, and bismuth arsenate.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 14-192 DRAFT FOR PUBLIC COMMENT
-------
Separation Techniques
5659 SOLVENT EXTRACTION. Liquid-liquid extraction is the most common method for the separation
5660 of uranium in radioanalytical procedures. Extraction provides a high-recovery, one-batch process
5661 that is more reproducible than other methods. With the development of extraction chromatog-
5662 raphy, solvent extraction has become a very efficient process for uranium separation. Many and
5663 varied procedures are used to extract uranium from aqueous solutions, but the conditions can be
5664 summarized as: (1) composition of the aqueous phase (form of uranium, type of acid present, and
5665 presence of common cations and anions and of foreign anions); (2) nature of organic phase (type
5666 and concentration of solvent and diluent); (3) temperature; and (4) time of equilibrium.
5667 Extraction processes can be conveniently divided into three systems: those based on (1) oxygen
5668 bonding, (2) chelate formation, and (3) extraction of anionic complexes.
5669 Oxygen-bonding systems are more specific than those based on chelate formation. The employ
5670 organic acids, ethers, ketones, esters, alcohols, organophosphates (phosphoesters), and
5671 nitroalkanes. Ethers are effective for the extraction of uranyl nitrate from nitric acid solutions.
5672 Cyclic ethers are especially effective, and salting agents such as calcium nitrate increase the
5673 effectiveness. Methyl isobutyl ketone (MIBK or hexone) also effectively extracts uranium as the
5674 nitrate complex. It has been used extensively by industry in the Redox process for extracting
5675 uranium and plutonium from nuclear fuels. Aluminum hydroxy nitrate [A1OH(NO3)2] is an
5676 excellent salting agent for the process and the extraction efficiency is increased by the presence
5677 of the tetrapropylammonium cation [(C3H7)4N+1]. Another common system, used extensively in
5678 the laboratory and in industrial process to extract uranium and plutonium from fission products,
5679 known as the PUREX process (pjutonium uranium reduction extraction), is used in most fuel
5680 reprocessing plants to separate the radionuclides. It employs TBP, tri-/>butyl phosphate
5681 [(C4H9)3PO], in a hydrocarbon solvent, commonly kerosene, as the extractant. The uranium fuel
5682 is dissolved in nitric acid, and uranium and plutonium are extracted into a 30 percent TBP
5683 solution, forming a neutral complex, UO2(TBP)2. The organic phase is scrubbed with nitric acid
5684 solution to remove impurities, plutonium is removed by back-extracting it as Pu(ni) with a nitric
5685 acid solution containing a reducing agent, and uranium is removed with dilute nitric acid. A
5686 complexing agent can also be used as a stripping agent. Trioctylphosphine oxide is 100,000 times
5687 more efficient in extracting uranium (VI). In both cases, nitric acid is used both to form the
5688 uranium extracting species, uranyl nitrate, and as the salting agent. Salting with aluminum nitrate
5689 produces a higher extraction efficiency but less specificity for uranium. Specificity depends the
5690 salt used and it concentration and the diluent concentration.
5691 Uranium is also extracted with select chelate forming agents. One of the most common systems
5692 used for uranium is cupferron in diethyl ether or chloroform. Uranium (VI) is not extracted from
5693 acidic media, so impurities soluble in the mixture under acidic conditions can be extracted first.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 14-193 DO NOT CITE OR QUOTE
-------
Separation Techniques
5694 Uranium (VI) can be reduced to uranium (IV) for subsequent extraction. Other chelating agents
5695 used to extract uranium include 8-hydroxyquinoline or acetylacetone in hexone or chloroform.
5696 Amines with molecular weights in the 250 to 500 range are used to extract anionic complexes of
5697 uranium (VI) from acidic solutions. The amine forms a salt in the acidic medium consisting of a
5698 ammonium cation and complex anion, (C10H21)3NH+1 UO2(NO3)"1, for example. Selectivity of the
5699 amines for uranium (VI) is in the order: tertiary > secondary > primary. An anionic extracting
5700 system use extensively in laboratories and industry consists of triisooctyl amine (TIOA) in
5701 kerosene. Uranium is stripped with sodium sulfate or sodium carbonate solution. A number of
5702 mineral and organic acids have been used with the system: hydrochloric, sulfuric, nitric,
5703 phosphoric, hydrofluoric, acetic oxalic, formic, and maleic acid. Stripping is accomplished with
5704 dilute acid solutions.
5705 Extraction chromatography is a simple and relatively quick method for the separation of uranium
5706 on a highly selective, efficient column system. One separation column consist of a triamyl-
5707 phosphate [(C5HnO)3PO] and diamylamylphosphonate (DAAP) [C5H11O)2(C5H11)PO] mixture in
5708 an apolar polymeric matrix. In nitric acid, uranyl nitrate forms a complex with DAAP that is
5709 soluble in triamylphosphate. Uranium can be separated in this system from many other metal
5710 ions, including thorium and the transuranium ions plutonium, americium, and neptunium. It is
5711 eluted from the column with the addition of oxalate to the eluent. Another extraction chromatog-
5712 raphy column uses octylphenyl-N,N-diisobutyl carbamoylphoshpine oxide (CMPO) dissolved in
5713 TBP and fixed on the resin matrix for isolation of uranium in nitric acid. Elution occurs with the
5714 addition of oxalic acid to the eluent.
5715 ION-EXCHANGE CHROMATOGRAPHY. Both cation- and anion-exchange chromatography have
5716 been used to separate uranium from other metal ions. Both stable forms of uranium, uranium (IV
5717 and VI) are absorbed on cation-exchange resins. Uranium (IV) is more strongly absorbed, and
5718 separation of uranium (VI) (UO2+2) is limited. On some cation-exchange columns, the ion also
5719 tends to tail into other ion fractions during elution. Absorption increases with temperature,
5720 however, and increasing the pH also increases absorption up to the beginning of formation of
5721 hydrolytic precipitates at pH 3.8. In strong acid solutions, uranium (VI) is weakly absorbed
5722 compared to uranium (in and IV) cations. Use of complexing agents increases specificity either
5723 by elution of uranium (VI) with common complexes-forming anions such as chloride, fluoride,
5724 nitrate, carbonate, and sulfate or by forming EDTA, oxalate, acetate, or sulfate complexes with
5725 cations in the analyte, producing a more pronounced difference in absorption of the ions on the
5726 exchange resin. A general procedure for separating uranium (VI) from other metals using the first
5727 method is to absorb uranium (VI) at pH of 1.5 to 2 and elute the metal with acetate solution.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 14-194 DRAFT FOR PUBLIC COMMENT
-------
Separation Techniques
5728 Anion-exchange chromatography of uranium takes advantage of the stable anionic complexes
5729 formed by the various oxidation states of uranium, especially uranium (VI), with many common
5730 anions. Uranium (VI) forms both anionic or neutral complexes with acetate, chloride, fluoride,
5731 carbonate, nitrate, sulfate, and phosphate. Strong anion-exchange resins are more selective and
5732 have a greater capacity than weak exchangers whose use is more limited. Factors that affect the
5733 separations include uranium oxidation state and concentration; type of anion and concentration;
5734 presence and concentration of other metallic ions and foreign ions; temperature, resin, size,
5735 porosity, and cross-linking. The various oxidation states of uranium and other metal ions,
5736 particularly the actinides, and the effect of pH on formation of complexes and net charge of the
5737 column provide two controllable variable to control the separation process.
5738 A number of chromatographic systems are available for uranium separation on anion-exchange
5739 resins. In hydrochloric acid uranium is often absorbed and other cations are not. Uranium (VI)
5740 can be absorbed from concentrated hydrochloric acid while alkali metals, alkaline earths, rare
5741 earths, aluminum, yttrium, actinium, and thorium are washed off the column. In contrast,
5742 uranium, molybdenum, bismuth, tin, technetium, polonium, plutonium and many transition
5743 metals are absorbed on the column, and uranium is eluted exclusively with dilute hydrochloric
5744 acid. Various oxidation states provide another method of separation. Uranium (IV) is separated
5745 from praseodymium (IV), and thorium (IV) with 8 M hydrochloric acid. Thorium, plutonium,
5746 zirconium, neptunium, and uranium can be separated individually by absorbing all the ions
5747 except thorium from concentrated hydrochloric acid Plutonium (IE) elutes with concentrated
5748 acid, zirconium at 7.5 M, neptunium (IV) with 6 M hydrochloric acid and 5 percent
5749 hydroxylamine hydrochloride, and uranium at 0.1 M acid. Uranium (IV) can be separated from
5750 uranium (VI) because both strongly absorb from concentrated hydrochloric acid, but they
5751 separate at 6 M acid because uranium (IV) is not absorbed at that concentration. Uranium (VI)
5752 absorbs strongly on an anion-exchange resin in dilute hydrofluoric acid, and the absorption
5753 decreases with increasing acid concentration. Nitric acid provides an excellent method to purify
5754 uranium, because uranium is more strongly absorbed from a nitric acid/nitrate solution. More
5755 selectivity is achieved when acid concentration is low and nitrate concentrations high.
5756 Absorbance is greatest when aluminum nitrate is use as the source of nitrate. Ethyl alcohol
5757 increases absorbance significantly.
5758 ELECTRODEPOSITION. Electrochemical procedures have been used to separate metal ions from
5759 uranium in solution by depositing them on a mercury cathode from a sulfuric acid solution, using
5760 5 amps for one hour. Uranium is deposited at a cathode from acetate, carbonate, oxalate, formate,
5761 phosphate, fluoride, and chloride solutions to produce a thin, uniform film for alpha and fission
5762 counting. This is the primary use of electrodeposition of uranium in analytical work. In another
5763 procedure, uranium (VI) is electroplated on a platinum electrode from the basic solution adjacent
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 14-195 DO NOT CITE OR QUOTE
-------
Separation Techniques
5764 to the cathode that exist in a slightly acidic bulk solution. The conditions of the process should be
5765 carefully controlled to obtain high yields and adherent coatings on the electrode.
5766 VOLATILIZATION. Several halides of uranium and the uranyl ion are volatile and have the
5767 potential for separation by sublimation or fractional distillation. Practically, however, their
5768 volatility is not used to separate uranium in analytical procedures because of technical problems
5769 or the high temperatures that are required for some procedures, but volatilization has been used
5770 in industrial processes. Uranium hexafluoride and uranyl hexafluoride are volatile, and the
5771 property is used to separate 235U from 238U in natural uranium isotope mixtures. Uranium tetra-
5772 chloride and hexachloride are also volatile, and uranium has been isolated from phosphate rock
5773 by heating with a mixture of chlorine and carbon monoxide at 800 °C and collecting the
5774 tetrachl oride.
5775 Methods of Analysis
5776 Macroquantities of uranium, essentially 238U, are determined by fluorimetry. During the
5777 separation and purification process, the sample is eventually fused at 625 °C in a flux mixture
5778 containing potassium carbonate, sodium carbonate, and sodium fluoride. The residue is exposed
5779 to light and its fluorescence is measured. Total uranium or individual radionuclides of uranium,
5780 234U, 235U, and 238U, can be determined from their alpha particle emissions. Uranium radionuc-
5781 lides are collected by evaporating the sample to dryness on a stainless steel planchet, by micro-
5782 precipitation with a carrier, such as lanthanum or cerium fluoride, or electrodepositon on a
5783 platinum disc. Total alpha activity is determined with a gas-flow proportional counter or an alpha
5784 liquid scintillation system. Individual radionuclides are measured by alpha spectrometry. Alpha
5785 emissions from 232U are used as a tracer to determine chemical recovery.
5786 Compiled from: Allard et al., 1984; Ahrland, 1986; Baes and Mesmer, 1976; Bard, 1985;
5787 Booman and Rein, 1962; Choppin et al., 1995; Considine and Considine, 1983; Cotton and
5788 Wilkinson, 1988; CRC, 1998-99; DOE, 1990, 1995, and 1997; EPA, 1973; Ehmann and
5789 Vance, 1991; Fritz and Weigel, 1995; Greenwood and Earnshaw, 1984; Grindler, 1962;
5790 Hampel, 1968; Hassinsky and Adloff, 1965; Katz et al., 1986; Katzin, 1986; SCA, 2001;
5791 Weigel, 1986.
5792 14.10.9.12 Zirconium
5793 Zirconium, atomic number 40, is a member of the second-row transition elements. It exhibits
5794 oxidation states of+2, +3, and +4, and the +4 state is the most common in both the solid state
5795 and in solution. It is immediately above hafnium in the periodic table, and both elements have
MARLAP JULY 2001
DO NOT CITE OR QUOTE 14-196 DRAFT FOR PUBLIC COMMENT
-------
Separation Techniques
5796 very similar chemical properties, more so than any other two elements in the periodic table. It is
5797 very difficult, but not impossible, to prepare a sample of zirconium without the presence of
5798 hafnium.
5799 Isotopes
5800 There are twenty-nine isotopes of zirconium, including five metastable states, with mass numbers
5801 from 81 through 104. Five are naturally occurring, 90Zr, 91Zr, 92Zr, 94Zr, and 96Zr, although the
5802 least abundant, 96Zr, is radioactive with an exceptionally long half-life of 3.56 x 1017 y. The
5803 remaining isotopes have a half-life of milliseconds to days. The lower mass number isotopes
5804 decay primarily by electron capture and the upper mass number isotopes are beta emitters. 95Zr
5805 (t1/2=64.0 d) and 97Zr (t1/2=16.9 h) are fission products and are beta emitters. 93Zr (t1/2=1.53 x 106y)
5806 is a rare fission product, and 98Zr, and "Zr are short-lived products with half-lives of 30.7 s and
5807 2.1 s, respectively. All are beta emitters.
5808 Occurrence and Uses
5809 Zirconium is one of the most abundant and widely distributed metals found in the earth's crust. It
5810 is so reactive that it is found only in the combined state, principally in two minerals, zircon,
5811 zircon orthosilicate (ZrSiO4), and baddeleyite, mostly zirconium dioxide (ZrO2). Zirkite is a
5812 commercial ore that consists of both minerals. Hafnium is a minor constituent of all zirconium
5813 minerals.
5814 In the production of zirconium metal, zirconium sands, primarily zirconium dioxide, is passed
5815 through an electrostatic separator to remove titanium minerals, a magnetic separator to remove
5816 iron, ileminite, and garnet, and a gravity separator to remove the less dense silica. The recovered
5817 zircon is heated with carbon in an arc furnace to form zirconium cyanonitride, an interstitial
5818 solution of carbon, nitrogen, and oxygen (mostly carbon) in the metal. Silicon evaporates as
5819 silicon monoxide (SiO), becoming silicon dioxide (SiO2) at the mouth of the furnace. The hot
5820 zirconium cyanonitride is treated with chlorine forming volatile zirconium tetrachloride (ZrCl4),
5821 which is purified by sublimation to remove, among other impurities, contaminating oxides. The
5822 chloride is reduced in the Kroll process, in turn, with liquid magnesium under conditions that
5823 produce a metal sponge. The byproduct, magnesium chloride (MgCl2), is then removed by
5824 melting the chloride, draining it off, and removing its residues by vacuum distillation. The
5825 zirconium sponge is crushed, melted into bars, arc-melted in an inert atmosphere, and formed
5826 into ingots. For additional purification, the van Arkel-de Boer process removes all nitrogen and
5827 oxygen. Crude zirconium is heated to 200 °C in an evacuated container containing a small
5828 amount of iodine to form volatile zirconium tetraiodide (ZrI4). A tungsten filament is electrically
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 14-197 DO NOT CITE OR QUOTE
-------
Separation Techniques
5829 heated to 1,300 °C, decomposing the iodide and depositing zirconium on the filament. The
5830 commercial grade of zirconium still contains up to three percent hafnium. To be used in nuclear
5831 reactors, however, hafnium should be removed. Separation is usually accomplished by solvent
5832 extraction of zirconium from an aqueous solution of zirconium tetrachloride as a complex ion
5833 (phosphine oxide, for example), by ion-exchange, fractional crystallization of complex fluoride
5834 salts, distillation of complexes of zirconium tetrachloride with phosphorus pentachloride or
5835 phosphorus oxychloride, or differential reduction of the mixed tetrachlorides (zirconium
5836 tetrachloride is more easily reduced to the nonvolatile trichloride than hafnium tetrachloride.
5837 95Zr and 97Zr are fission products and are also produced by bombardment of naturally occurring
5838 94Zr and 96Zr, respectively, with thermal neutrons. Stable 90Zr is a product of the 90Sr decay chain:
5839 9038Sr-9°39Y + p-9040Zr + p + Y
5840 Zirconium metal and its alloys are highly corrosion resistant and withstands streams of heated
5841 water under high pressure. These properties, along with their low cross section for thermal
5842 neutrons, make them an important material for cladding uranium fuel elements and as core armor
5843 material in nuclear reactors. It is also used for making corrosive resistant chemical equipment
5844 and surgical instruments and making superconducting magnets. Zirconium compounds are also
5845 used in the ceramics industry as refractories, glazes, and enamels, in cores for foundry molds,
5846 abrasive grits, and components of electrical ceramics. Crystals of zircon are cut and polished to
5847 use in jewelry as simulated diamonds. They are also used in pyrotechnics, lamp filaments, in arc
5848 lamps, cross-linking agents for polymers, components of catalysts, as bonding agents between
5849 metal and ceramics and between ceramics and ceramics, as tanning agents, ion exchangers, and
5850 in pharmaceutical agents as deodorants and antidotes for poison ivy. 95Zr is used to follow
5851 homogenization of oil products.
5852 Solubility of Compounds
5853 The solution properties of zirconium in water are very complex, mainly because of the formation
5854 of colloids and the extensive hydrolysis and polymerization of the zirconium ion. hydrolysis and
5855 polymerization are strongly dependent on the pH of the solution, concentration of the ion, and
5856 temperature. The nitrate, chloride, bromide, iodide, perchlorate, and sulfate of zirconium are
5857 soluble in acid solution, however.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 14-198 DRAFT FOR PUBLIC COMMENT
-------
Separation Techniques
5858 Review of Properties
5859 Pure zirconium is a grey-white (silvery) lustrous metal with a density of 6.49 g/cm3. It exist in
5860 two allotropic forms, alpha and beta, with a transition temperature of 870 °C. The alpha form is
5861 stabilized by the common impurity oxygen. The amorphous powder is blue-black. Trace amounts
5862 of common impurities (< 1 percent), such as oxygen, nitrogen, and carbon, make the metal brittle
5863 and difficult to fabricate. The metal is not considered to be a good conductor of heat and
5864 electricity, but compared to other metals it is soft, malleable, and ductile. Zirconium forms alloys
5865 with most metals except, mercury, the alkali metals, and the alkaline earths. It can absorb up to
5866 ten percent oxygen and nitrogen. Zirconium is a superconductor at temperatures near absolute
5867 zero, but its superconducting properties improve when the metal is alloyed with niobium and
5868 zinc.
5869 Finely divided, dry zirconium (powder and chips) is pyrophoric and extremely hazardous. It is
5870 hard to handle and store and should be moistened for safe use. Note, however, that both wetted
5871 sponge and wet and dry stored scrap have been reported to spontaneously explode. Caution
5872 should also be observed with waste chips produced from machining and cleaning (new)
5873 zirconium surfaces. Both can be pyrophoric. In contrast, zirconium in the bulk form is extremely
5874 resistant to corrosion at room temperature and remains bright and shiny in air. Resistance is
5875 rendered by the formation of a dense, adherent, self-sealing oxide coating. The metal in this form
5876 is resistant to acids, alkalis, and seawater. Without the coating, zirconium dissolves in warm
5877 hydrochloric and sulfuric acids slowly; dissolution is more rapid in the presence of fluoride ions.
5878 The metal is also resistant to high-pressure water streams and high-temperature steam. It also has
5879 a low cross-section to thermal neutrons and is resistant to damage from neutron radiation. These
5880 properties give pure zirconium (without hafnium) very useful as a fabrication material for nuclear
5881 reactors. Zirconium metal alone, however, is not sufficiently resistant to hot water and steam to
5882 meet the needs for use in a nuclear reactor. Alloyed with small percentages of tin, iron, nickel, or
5883 chromium (Zircalloy), however, the metal meets the standards.
5884 The coated metal is becomes reactive when heated at high temperature (> 500 °C) with
5885 nonmetals, including hydrogen, oxygen, nitrogen, carbon, and the halogens, and forms solid
5886 solutions or compounds with many metals. It reacts slowly with hot concentrated sulfuric and
5887 hydrochloric acids, boiling phosphoric acid, and aqua regia. It is also attacked by fused potassium
5888 nitrate and potassium hydroxide, but is nonreactive with aqueous alkali solutions. It is not
5889 reactive with nitric acid. Hydrofluoric acid is the only reagent that reacts vigorously with
5890 zirconium.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 14-199 DO NOT CITE OR QUOTE
-------
Separation Techniques
5891 Zirconium and its compounds are considered to have a low order of toxicity. Most handling and
5892 testing indicate no level of toxicity, but some individual seem to be allergic to zirconium
5893 compounds. Inhalation of zirconium compound sprays and metallic zirconium dust have
5894 produced inflammatory affects.
5895 Very small quantities of 95Zr have been released to the environment from fuel reprocessing
5896 facilities, atmospheric testing, and the Chernobyl accident. With a half-life of 64 days, the
5897 contamination of the environment is not significant. Zirconium lost from a waste repository
5898 would be expected to move very slowly because of radiocolloidal attraction to surrounding soil
5899 particles. Hydrolysis and polymerization renders most zirconium insoluble in natural water, but
5900 absorption to suspended particles is expected to provide some mobility in an aqueous
5901 environment.
5902 Solution Chemistry
5903 The only important oxidation state of zirconium ions in aqueous solution is +4, making it a
5904 essentially a monovalent element. The solution chemistry of zirconium is quite complex,
5905 nevertheless, because of the easy formation of colloids and extensive hydrolysis and
5906 polymerization reactions that are strongly dependent on pH and ion concentration.
5907 COMPLEXATION. Zirconium ions forms complexes with numerous substances: fluoride,
5908 carbonate, borate, oxalate, and other dicarboxylic acids, among others. As a large, highly
5909 charged, spherical ion, it exhibits high coordination numbers. One of the important chemical
5910 properties of zirconium ions in solution is their formation of a very stable hexafluorozirconate
5911 complex, ZrF6"2. For that reason, hydrofluoric acid (HF) is an excellent solvent for the metal and
5912 insoluble zirconium compounds. Unfortunately, the fluorocomplex interferes with most
5913 separation and determination steps, and zirconium should be expelled by fuming with sulfuric or
5914 perchloric acid before proceeding. The addition of several milliliters of concentrated FTP to a cool
5915 solution of zirconium carrier and sample will produce initial equilibration; essentially all the
5916 zirconium is present in the +4 oxidation state as a fluoride complex. Note that addition of FTP to
5917 solutions above the azeotropic boiling point of the acid (120 °C) serves no useful purpose and
5918 simply evaporates the FTP.
5919 Tartrate and citrate ions form stable complexes even in alkaline solutions, and zirconium
5920 hydroxide will not precipitate in their presence (see hydrolysis below). Oxalate forms a complex
5921 that is less stable. The ion, [Zr(C2O4)3]"2, is only stable in acid solution. On addition of base, the
5922 complex is destroyed, and zirconium hydroxide precipitates. Sulfuric acid complexes in strongly
5923 acidic solutions, forming Zr(SO4)4"2. In concentrated HC1 solutions, ZrCl6"2 is present.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 14-200 DRAFT FOR PUBLIC COMMENT
-------
Separation Techniques
5924 Zirconium ions form chelate complexes with many organic compounds, usually through oxygen
5925 atoms in the compounds. Typical examples are: acetylacetone (acac), EDTA, thenoyltrifluoro-
5926 acetone (TTA), salicylic acid, mandelic acid, cupferron, and 8-hydroxyquinoline.
5927 HYDROLYSIS. Although Zr+4 has a large radius and any +4 cation is extensively hydrolyzed, Zr+4
5928 appears to exist at low ion concentrations (approximately 10"4 M) and high pH (1-2 M). As the
5929 Zr+4 concentration increases and the concentration of H+1 decreases, however, hydrolysis and
5930 polymerization occurs, and one or more polymeric species is dominate in solution. Amorphous
5931 hydrous oxides are precipitated near pH 2; they are soluble in base. Because of hydrolysis,
5932 soluble salts (nitrate, sulfate, perchlorate, acetate, and halides) form acidic solutions when they
5933 dissolve. The reaction seems to be essentially a direct conversion to the tetranuclear
5934 Zr4(OH)8(H2O)16+2 ion; there is no convincing evidence for the existence of ZrO+2, thought at one
5935 time to be present in equilibrium with numerous other hydrolysis products. It should be noted,
5936 however, that freshly prepared solutions of zirconium salts might react differently from a solution
5937 left standing for several days. Whatever the actual species in solution at any given time, the
5938 behavior of zirconium (IV) depends on the pH of the solution, temperature, anion present, and
5939 age of solution. In addition, zirconium compounds formed by precipitation from solution usually
5940 do not have a constant composition because of their ease of hydrolysis. Even under exacting
5941 conditions, it is difficult to obtain zirconium compounds of known, theoretical composition, and
5942 on aging, hydrolysis products becomes more polymeric and polydisperse.
5943 In acidic solutions, trace amounts of zirconium are strongly coprecipitated with most precipitates
5944 in the absence of complexing ions, especially F"1 and C2O4"2 that form soluble complex ions.
5945 In alkaline solutions, produced by the addition of hydroxide ions or ammonia, a white gelatinous
5946 precipitate of zirconium hydroxide forms. Since the hydroxide is not amphoteric, it does not
5947 dissolve in excess base. The precipitate is not a true hydroxide but a hydrated oxide, ZrO2-nH2O
5948 where n represents the variable nature of the water content. Freshly prepared zirconium
5949 hydroxide is soluble in acid; but as it dries, its solubility decreases. Precipitation is inhibited by
5950 tartrate or citrate ions because Zr+4 forms complexes with these organic anions even in alkaline
5951 solutions (see "Complexation," above).
5952 In preparing zirconium solutions, it is wise to acidify the solution with the corresponding acid to
5953 reduce hydrolysis and avoid precipitation of basic salts. During solubilization and radiochemical
5954 equilibrium with a carrier, the tendency of zirconium ions to hydrolyze and polymerize even at
5955 low pH should be kept in mind. Often, the formation of a strong complex with fluoride or TTA is
5956 necessary.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 14-201 DO NOT CITE OR QUOTE
-------
Separation Techniques
5957 RADIOCOLLOIDS. Radiocolloids of zirconium are adsorbed on practically any foreign matter (e.g.,
5958 dirt, glass, etc.). Their formation can cause problems with dissolution, achieving radiochemical
5959 equilibrium, and analysis. Generally, it is necessary to form a strong complex with fluoride (see
5960 caution above) or TTA.
5961 Dissolution of Samples
5962 Metallic zirconium is dissolved in hydrofluoric acid, hot aqua regia, or hot concentrated sulfuric
5963 acid. Hydrofluoric acid should be removed by fuming with sulfuric acid or perchloric acid
5964 (caution), because fluoride interferes with most separation and analytical procedures. Zirconium
5965 ores, rocks, and minerals are fused at high temperatures with sodium carbonate, potassium
5966 thiosulfate, sodium peroxide, sodium tetraborate, or potassium hydrogen fluoride (remove
5967 fluoride). The residue is dissolved in dilute acid or water and might require filtration to collect a
5968 residue of zirconia (impure ZrO2), which is dissolved in acid. As a minor constituent of natural
5969 sample or as a result of formation by nuclear reactions, zirconium typically dissolves during
5970 dissolution of the major constituents. The tendency to polymerize under low concentrations of
5971 acid and the formation of insoluble zirconium phosphates should be considered in any
5972 dissolution process. The tendency of zirconium to polymerize and form radiocolloids makes it
5973 important to insure equilibrium with any carrier added. Generally, formation of strong complexes
5974 with fluoride or TTA is necessary.
5975 Separation Methods
5976 PRECIPITATION AND COPRECIPITATION. One of the most insoluble precipitating agents is
5977 ammonium hydrogen phosphate (NH4)2HPO4) in 20 percent sulfuric acid. It has the advantage
5978 that it can be dissolved by hydrofluoric acid, forming hexafluorozirconate. This complex ion also
5979 forms insoluble barium hexafluorozirconate (BaZrF6), a precipitating agent that allows the
5980 precipitation of zirconium in the presence of niobium that is soluble as the heptafluoroniobate
5981 (NbF7"2). Other precipitating agents include the iodate (from 8 M nitric acid), cupferrate, the
5982 hydroxide, peroxide, selenate, and mandelate. Cupferron is used in sulfuric or hydrochloric acid
5983 solutions. It is one of the few precipitating agents in which fluoride does not interfere, but iron
5984 and titanium, among other cations, are also precipitated. The precipitate can be heated in a
5985 furnace at 800 °C to produce zirconium dioxide for the gravimetric determination of zirconium.
5986 The hydroxide begins to precipitate at pH 2 and is complete at pH 4, depending on the presence
5987 of zirconium complexes. It is not recommended unless other cations are absent, because it
5988 absorbs or coprecipitates almost all other ions. Peroxide is formed from a solution of hydrogen
5989 peroxide in acid. Selenious acid in dilute hydrochloric acid separates zirconium from some of the
5990 transition elements and thorium. Mandelic acid in hot dilute hydrochloric acid quantitatively and
MARLAP JULY 2001
DO NOT CITE OR QUOTE 14-202 DRAFT FOR PUBLIC COMMENT
-------
Separation Techniques
5991 specifically precipitates zirconium (and hafnium) ions. Large amounts of titanium, tin, iron, and
5992 other ions might be partially coprecipitated, but they can be eliminated by reprecipitation.
5993 Trace quantities of zirconium can be strongly coprecipitated by most precipitates from strong
5994 acid solutions that do not contain complex-forming ions. Bismuth and eerie phosphate readily
5995 carries zirconium, and in the absence of holdback carriers, it is almost quantitatively carried by
5996 rare-earth fluorides. Ferric hydroxide and thorium iodate are also effective carriers.
5997 SOLVENT EXTRACTION. Several extractants have been used to selectively remove zirconium from
5998 aqueous solutions; most are organophosphorus compounds. Di-/>butylphosphoric acid (DBFA)
5999 (di-/7-butylphosphate) is an extractant for zirconium and niobium. It is effective in extracting
6000 tracer and macro quantities of zirconium from 1 M aqueous solutions of nitric, hydrochloric,
6001 perchloric, and sulfuric acids and in separating it from many other elements. A 0.06 M solution
6002 in di-/7-butylether containing three percent hydrogen peroxide extracts more than 95 percent
6003 zirconium but less than one percent niobium. Tin and indium were also extracted by this mixture.
6004 Tri-/7-butylphosphate (TBP) is an excellent solvent for zirconium. It is used pure or with several
6005 nonpolar diluents, ethers, kerosene, or carbon tetrachloride. Extractability increases with acid
6006 strength. A 0.01 M solution of tri-/>octylphosphine oxide (TOPO) in cyclohexane has been use
6007 to separate zirconium form iron, molybdenum, vanadium, thorium, and hafnium.
6008 TTA and hexone (methyl isobutyl ketone) are two nonphosphorus extractants employed for
6009 separating zirconium. TTA is highly selective. A 0.5 M solution in xylene separates zirconium
6010 from aluminum, iron, thorium, uranium, and rare earths in a 6 M hydrochloric acid solution. At
6011 tracers levels, the reagent can separate 95Zr from all other fission products. It is also used to
6012 separate zirconium from hafnium. In the analysis of zirconium in zirconium-niobium-tantalum
6013 alloys, hexone separates zirconium from an aqueous solution that is 10 M hydrochloric acid and
6014 6 M sulfuric acid. This is one of the few methods that can be use to separate zirconium from
6015 these metals.
6016 ION-EXCHANGE CHROMATOGRAPHY. Zirconium can be separated form many other cations by
6017 both cation- and anion-exchange chromatography. The technique represents the best laboratory
6018 method for separating zirconium and hafnium. Cation-exchange columns strongly absorb
6019 zirconium ions, but macro quantities of zirconium and hafnium can be purified as aqueous
6020 colloidal solutions of their hydrous oxides on an organic cation-exchange resin. Many cations are
6021 retained on the column, but zirconium and hafnium, under these conditions, are not. The
6022 recovery can be as high as 99 percent with successive passages, but titanium and iron are not
6023 removed. Zirconium and hafnium can be separated on a sulfuric-acid column from 2 M
6024 perchloric acid. Hafnium is eluted first with 6 M hydrochloric acid. Fluoride complexes of
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 14-203 DO NOT CITE OR QUOTE
-------
Separation Techniques
6025 zirconium and hafnium can be separated from other non-complexing cations, because the
6026 negative complex ions are not absorb and the non-complexing ions are retained. Zirconium,
6027 hafnium, and niobium are eluted from rare earths and alkaline earths on cation-exchange
6028 columns with citrate. The three elements can be then be separated by the selection of appropriate
6029 citrate buffers, but the separations are not quantitative.
6030 The formation of stable zirconium complexes is the basis of anion-exchange chromatography of
6031 the metal. Separation of zirconium and hafnium from each other and form other cations can be
6032 achieved in hydrochloric-hydrofluoric acid mixtures. Separation of zirconium from hafnium,
6033 niobium, protactinium, and thorium, respectively, is accomplished by selection of the proper
6034 eluting agent. Elution of hafnium first with 9 M hydrochloric acid separates zirconium from
6035 hafnium, for example, while elution with 0.2 M hydrochloric acid/0.01M hydrofluoric acid
6036 recovers zirconium first. Elution with 6-7 M hydrochloric acid separates zirconium from
6037 niobium, in another example.
6038 Methods of Analysis
6039 95Zr decays with a half-life of 65.5 d, emitting a beta particle accompanied by gamma-ray
6040 emission. After several half-lives, it is in transient equilibrium with its progeny, 95Nb, which has
6041 a half-life of 35.0 d and is also a beta and gamma emitter. The progeny of 95Nb is stable 95Mo.
6042 Fresh samples of 95Zr are analyzed by their gamma-ray emission. Zirconium is collected by
6043 precipitation and filtration. The sample and filter are heated at 800 °C for one hour to decompose
6044 the filter and convert zirconium to its oxide. Zirconium dioxide (ZrO2) is collected by filtration,
6045 dried, and counted immediately.
6046 Compiled from: Baes and Mesmer, 1976; Choppin et al., 1995; Considine and Considine,
6047 1983; Cotton and Wilkinson, 1988; CRC, 1998-99; Ehmann and Vance, 1991; EPA, 1973;
6048 Greenwood and Earnshaw, 1984; Hahn, 1961; Hassinsky and Adi off, 1965; Latimer, 1952;
6049 Steinberg, 1960.
6050 14.11 References
6051 Adams, B.A. and Holmes, E.L. 1935. "Absorptive Properties of Synthetic Resins. Part I,"
6052 JournalSoc. Chem. Ind. (London), Vol. 54T, pp. T1-T6.
6053 Adams, C. 1995. "Iodine," McGraw-Hill Multimedia Encyclopedia of Science and Technology,
6054 McGraw-Hill, New York; Software Copyright: Online Computer Systems, Inc.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 14-204 DRAFT FOR PUBLIC COMMENT
-------
Separation Techniques
6055 Adolff, J.-P. and Guillaumont, R. 1993. Fundamentals of Radiochemistry, CRC Press, Boca
6056 Raton, Florida.
6057 Alvarez, A. and Navarro, N. 1996. "Method for Actinides and Sr-90 Determination in Urine
6058 Samples," Applied Radiation and Isotopes, Vol. 47, No. 9/10, pp. 869-873.
6059 Allard, B, Olofsson, U., and Torstenfelt, B. 1984. "Environmental Actinide Chemistry,"
6060 Inorganica Chimica Acta, Vol. 94, pp. 205-221.
6061 American Society of Testing Materials (ASTM) 1995. "Standard Test Method for Strontium-90
6062 in Water," Annual Book of ASTM Standards, Designation: D 5811-95.
6063 Ahrland, S. 1986. "Solution Chemistry And Kinetics of Ionic Reactions," in Katz, J.J., Seaborg,
6064 G.T., and Morss, L.R., Eds., The Chemistry of the Actinides, Vol. 2, Chapman and Hall,
6065 London, pp. 1480-1546.
6066 American Public Health Association (APHA). 1998. Standard Methods for the Examination of
6067 Water and Wastewater, Clesceri, L.S., Greenberg, A.E., Eaton, A.D., Eds., Franson, M.A.,
6068 Mang. Ed., American Public Healthe Association, Washington, DC.
6069 Anders, E. 1960. The Radiochemistry of Technetium, National Academy of Sciences-National
6070 Research Council (NAS-NS), NAS-NS 3021, Washington, DC.
6071 Armstrong, G.W., Gill, H.H., and Rolf, R.F. 1961. "The Halogens," in Treatise on Analytical
6072 Chemistry, Kolthoff, I.M. and Elving, P.J., Eds., Part II, Volume 7, John Wiley and Sons,
6073 New York, pp. 3 3 5 -424.
6074 Baes, C.F. and Mesmer, R.E. 1976. The Hydrolysis of Cations, John Wiley and Sons, New York.
6075 Bailar, J.C., Moeller, T., Kleinberg, J., Guss, C.O., Castellion, M.E., and Metz, C. 1984.
6076 Chemistry, Academic Press, Orlando, Florida.
6077 Banavali, A.D., Raimondi, J.M., and McCurdy, D.E. 1995. "The Determination of
6078 Technetium-99 in Low-Level Radioactive Waste," Radioactivity and Radiochemistry, 6:3,
6079 pp. 26-35.
6080 Bard, A.J., Parsons, R., and Jordan, J. 1985. Standard Potentials in Aqueous Solution, Marcel
6081 Dekker, New York.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 14-205 DO NOT CITE OR QUOTE
-------
Separation Techniques
6082 Bate, L.C. and Leddicotte, G. W. 1961. The Radiochemistry of Cobalt, National Academy of
6083 Sciences-National Research Council, (NAS-NS), NAS-NS 3041, Washington, DC.
6084 Berg, E.W. 1963. Physical and Chemical Methods of Separation, McGraw-Hill, New York.
6085 Berne, A. 1995. "Use of Eichrom's TRU Resin in the Determination of Am, Pu and U in Air
6086 Filter and Water Samples," Environmental Measurements Laboratory. EML-575. December.
6087 Biederman, G. and Schindler, P. 1957. "On the Solubility Product of Precipitated Iron (HI)
6088 Hydroxide," Acta Chemica Scandinavia, Vol. 11, p. 731-740.
6089 Birkett, J.D., Bodek, I, Glazer, A.E., Grain, C.F., Hayes, D., Lerman, A., Lindsay, D.B., Loreti,
6090 C.P., and Ong, J.H. 1988. "Description of Individual Processes," in Environmental Inorganic
6091 Chemistry, Bodek, I, Lyman, W.J., Reehl, W.F., and Rosenblatt, D.H., Eds., Pergammon,
6092 New York, pp. 2.1 -2.17-18.
6093 Blanchard, R.L. 1966. "Rapid Determination of Lead-210 and Polonium-210 in Environmental
6094 Samples by Deposition on Nickel," Analytical Chemistry, Vol. 38, p. 189-192.
6095 Blumenthal, W.B. 1995. "Zirconium," McGraw-Hill Multimedia Encyclopedia of Science and
6096 Technology, McGraw-Hill, New York; Software Copyright: Online Computer Systems, Inc.
6097 Bonnesen, P.V., Presley, D.J., Haverlock, T.J., and Moyer, B.A. 1995. "Removal of Technetium
6098 from Alkaline Nuclear-Waste Media by a Solvent-Extraction Process Using Crown Ethers,"
6099 presented at conference identified as CONF-9505101-6, Oak Ridge National Laboratory, Oak
6100 Ridge, Tennessee.
6101 Booman, G.L. and Rein, I.E. 1962. "Uranium," in Treatise on Analytical Chemistry, Kolthoff,
6102 I.M. and Elving, P.J., Eds., Part II, Volume 9, John Wiley and Sons, New York, pp. 1-188.
6103 Bunzl, K. and Kracke,W. 1994. "Efficient Radiochemical Separation for the Determination of
6104 Plutonium in Environmental Samples, Using a Supported, Highly Specific Extractant,"
6105 Journal of Radioanalytical and Nuclear Chemistry, Letters, Vol. 186, No. 5, pp. 401-413.
6106 Bunzl, K., Kracke, W., and Meyer, H. 1996. "A Simple Radiochemical Determination of 90Sr in
6107 Brines," Journal of Radioanalytical and Nuclear Chemistry, Letters, Vol. 2112, No. 2, pp.
6108 143-149.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 14-206 DRAFT FOR PUBLIC COMMENT
-------
Separation Techniques
6109 Burnett, W. and Yeh, C.C. 1995. "Separation of Protactinium from Geochemical Materials via
6110 Extraction Chromatography," Radioactivity and Radiochemistry, Vol. 6, No. 4, pp. 22-26.
6111 Cadieux, J.R., and Reboul, S.H. 1996. "Separation and Analysis of Actinides by Extraction
6112 Chromatography Coupled with Alpha-Particle Liquid Scintillation Spectrometry,"
6113 Radioactivity and Radiochemistry, Vol. 7, No. 2, pp. 30-34.
6114 Carney, K.P., and Cummings, D.G. 1995. "The Application of Micro-Column Solid Phase
6115 Extraction Techniques for the Analysis of Rare Earth Elements in Actinide Containing
6116 Matrices," Proceedings of the Third International Conference on Methods and Applications
6117 of Radioanalytical Chemistry in Journal of Radioanalytical and Nuclear Chemistry, Articles,
6118 Vol. 194, No. 1, pp. 41-49.
6119 Chabaux, F., Othman, D.B., and Brick, J.L. 1994. "A New Ra-Ba Chromatographic Separation
6120 and Its Application to Ra Mass-Spectrometric Measurement in Volcanic Rocks," Chemical
6121 Geology, Vol. 114, pp. 191-197. Erratum, 1994. Chemical Geology, Vol. 116, p. 301.
6122 Chemical Rubber Company (CRC). 1998-1999. Handbook of Chemistry and Physics, CRC
6123 Press, Boca Raton, FL.
6124 Chen, Q., Dahlgaard, H., Hansen, H.J.M., and Aarkrog, A. 1990. "Determination of "Tc in
6125 Environmental Samples by Anion Exchange and Liquid-Liquid Extraction at Controlled
6126 Valency," Analytica Chimica Acta, Vol. 228, pp. 163-167.
6127 Choppin, G., Rydberg, J., Liljenzin, J.O. 1995. Radiochemistry and Nuclear Chemistry,
6128 Butterworth-Heinemann, Oxford.
6129 Clark, D.L., Keog, D.W., Palmer, P.O., Scott, B.L., and Tait, C.D. 1998. "Synthesis and
6130 Structure of the First Transuranium Crown Inclusive Complex [NpO2([18]Crown-6)ClO4,"
6131 Angew. Chem. Int. Ed., Vol. 37, No. 1/2, pp. 164-166.
6132 Cleveland, J.M. 1970. The Chemistry of Plutonium, Gordon and Breach, Science Publishers, Inc.
6133 New York.
6134 Cobble, J.W. 1964. "Technetium," in Treatise on Analytical Chemistry, Kolthoff, I.M. and
6135 Elving, P.J., Eds., Part II, Volume 6, John Wiley and Sons, New York, pp. 404-434.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 14-207 DO NOT CITE OR QUOTE
-------
Separation Techniques
6136 Coleman, G.H. 1965. The Radiochemistry of Plutonium, National Academy of Sciences-
6137 National Research Council (NAS-NS), NAS-NS 3058, Washington, DC.
6138 Considine, D.M. and Considine, G.D., Eds. 1983. Van Nostrand's Scientific Encyclopedia, Van
6139 Nostrand Reinhold, New York.
6140 Coomber, D.I. 1975. "Separation Methods for Inorganic Species," in RadiochemicalMethods in
6141 Analysis, Coomber, D.I., Ed., Plenum Press, New York, pp. 175-218.
6142 Cotton, F.A. and Wilkinson, G. 1988. Advanced Inorganic Chemistry, John Wiley and Sons,
6143 New York.
6144 Cotton, S. 1991. Lanthanides and Actinides, Oxford University Press, New York.
6145 Crouthamel, C.E. and Heinrich, R.R. 1971. "Radiochemical Separations," in Treatise on
6146 Analytical Chemistry, Kolthoff, I.M. and Elving, P.J., Eds., Part I, Volume 9, John Wiley and
6147 Sons, New York, pp. 5467-5511.
6148 Dale, C.J., Warwick, P.E., and Croudace, I.W. 1996. "An Optimized Method for Technetium-99
6149 Determination in Low-Level Waste by Extraction Into Tri-n-octylamine," Radioactivity and
6150 Radiochemistry, Vol. 7, No. 3, pp. 23-31.
6151 Dale, J.M. and Banks, C.V. 1962. "Cobalt," in Treatise on Analytical Chemistry, Kolthoff, I.M.
6152 and Elving, P. J., Eds., Part H, Volume 2, John Wiley and Sons, New York, pp. 311-376.
6153 Dean, J.A. 1995. Analytical Chemistry Handbook, McGraw-Hill, New York.
6154 DeVoe, J.R. 1962. Application of Distillation Techniques to Radiochemical Separations,
6155 National Academy of Sciences- National Research Council (NAS-NS), NAS-NS 3108,
6156 Washington, DC.
6157 Dietz, Ml, Horowitz, E.P., Nelson, D.M., and Wahlgreen, M.A. 1991. "An Improved Method
6158 for Determining 89 Sr and 90 Sr in Urine," Health Physics, Vol. 61, No. 6, pp. 871-877.
6159 Dietz, M. and Horwitz, E.P. 1992. "Improved Chemistry for the Production of Y-90 for Medical
6160 Applications," Applied Radiation and Isotopes, Vol. 43, p. 1093-1102.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 14-208 DRAFT FOR PUBLIC COMMENT
-------
Separation Techniques
6161 Dietz, M.L. and Horwitz, E.P. 1993. "Novel Chromatographic Materials Based on Nuclear Waste
6162 Processing Chemistry," LC-GC, The Magazine ofSepartion Science, 11:6.
6163 U.S. Department of Energy (DOE) 1990 and 1997. EML Procedures Manual, Chieco, N.A.,
6164 Bogen, DC., and Knutson, E.G., Eds., HASL-300, 27th and 28th Editions, DOE
6165 Environmental Measurements Laboratory, New York.
6166 U.S. Department of Energy (DOE) 1993. DOE Methods for Evaluating Environmental and
6167 Waste Management Samples, Goheen, S.C., McCulloch, M., Thomas, B.L., Riley, R.G.,
6168 Skiarew, D.S., Mong, G.M., and Fadeff, S.K., Eds., DOE/EM-0089T, Washington, DC.
6169 U.S. Department of Energy (DOE) 1994. Primer on Tritium Safe Handling Practices, DOE-
6170 HDBK-1079-94, Washington, DC.
6171 U.S. Department of Energy (DOE) 1995. DOE Methods for Evaluating Environmental and
6172 Waste Management Samples, Goheen, S.C., McCulloch, M., Thomas, B.L., Riley, R.G.,
6173 Skiarew, D.S., Mong, G.M., and Fadeff, S.K., Eds., DOE/EM-0089T, Rev 2, Addendum 1,
6174 Washington, DC.
6175 U.S. Department of Energy (DOE) 1997. DOE Methods for Evaluating Environmental and
6176 Waste Management Samples, Goheen, S.C., McCulloch, M., Thomas, B.L., Riley, R.G.,
6177 Skiarew, D.S., Mong, G.M., and Fadeff, S.K., Eds., DOE/EM-0089T, Rev 2, Addendum 1,
6178 Washington, DC.
6179 Duckworth, H.E. 1995. "Triton," McGraw-Hill Multimedia Encyclopedia of Science and
6180 Technology, McGraw-Hill, New York; Software Copyright: Online Computer Systems, Inc.
6181 Ehmann, W.D. and Vancd, D.E. 1991. Radiochemistry and Nuclear Methods of Analysis, John
6182 Wiley, New York.
6183 U.S. Environmental Protection Agency (EPA) 1973. Procedures for Radiochemical Analysis of
6184 Nuclear Reactor Aqueous Solutions, Krieger, H.L. and Gold, S., Eds., EPA-R4-73-014, EPA
6185 National Environmental Research Center, Office of Research and Monitoring, Cincinnati,
6186 Ohio.
6187 U.S. Environmental Protection Agency (EPA) 1979. Radiochemical Analytical Procedures for
6188 Analysis of Environmental Samples, Johns, F.B., Hahn, P.B., Thome, D.J., and Bretthauer,
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 14-209 DO NOT CITE OR QUOTE
-------
Separation Techniques
6189 E.W., EMSL-LV-0539-17, Environmental Monitoring and Support Laboratory, Las Vegas,
6190 Nevada.
6191 U.S. Environmental Protection Agency (EPA) 1980. Prescribed Procedures for Measurement of
6192 Radioactivity in Drinking Water, Krieger, H.L. and Whittaker, E.L., Eds., EPA-600/4-80-
6193 032, EPA Environmental Monitoring and Support Laboratory, Cincinnati, Ohio.
6194 U.S. Environmental Protection Agency (EPA) 1984. Radiochemistry Procedures Manual,
6195 Liberman, R.E., Ed., EPA-520/5-84-006, EPA Eastern Environmental Radiation Facility,
6196 Office of Radiation Programs, Montgomery, Ala.
6197 Dorfner, K. 1972. Ion Exchangers: Properties and Applications, Ann Arbor Science Publishers,
6198 Ann Arbor, Michigan.
6199 Emsley, J. 1989. The Elements, Clarendon, Oxford.
6200 Paris, J.P. and Buchanan, RJ. 1964. USAEC Report, ANL-6811.
6201 Finston, H.L. and Kinsley, M.T. 1961. The Radiochemistry of Cesium, National Academy of
6202 Sciences-National Research Council (NAS-NS), NAS-NS 3035, Washington, DC.
6203 Fisher Catalog, 2000-01 2000. Fisher Scientific Company, Pittsburg, PA, pp. 578, 582-583.
6204 Flagg, J.F., and Bleidner, W.E. 1945. "The Electrodeposition of Element 43 and the Standard
6205 Potential for the Reaction Ma-MaO4," Journal of Chemical Physics, Vol. 13, pp. 269-276.
6206 Feigl, F. 1936. "Organic Reagents in Inorganic Analysis," Industry and Engineering Chemistry,
6207 Analytical Edition, Vol. 8, pp. 401-410.
6208 Fried, S. 1995. "Technetium," McGraw-Hill Multimedia Encyclopedia of Science and
6209 Technology, McGraw-Hill, New York; Software Copyright: Online Computer Systems, Inc.
6210 Friedlander,G., Kennedy, J.W., Macias, E.S., and Miller, J.M. 1981. Nuclear and
6211 Radiochemistry, 3rd Ed., John Wiley and Sons, New York.
6212 Gale, N.H. 1996. "A New Method for Extracting and Purifying Lead from Difficult Matrices for
6213 Isotopic Analysis," Analytica ChimicaActa, Vol. 332, pp. 15-21.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 14-210 DRAFT FOR PUBLIC COMMENT
-------
Separation Techniques
6214 Garcia-Leon, M. 1990. "Determination and Levels of "Tc in Environmental and Biological
6215 Samples," Journal of Radioanalytical and Nuclear Chemistry, Articles, Vol. 138, No. 1, pp.
6216 171-179.
6217 Gokel, G.W. 1991. Crown Ethers and Cryptands, Monographs in Supramolecular Chemistry,
6218 The Royal Society of Chemistry, London.
6219 Golchert, N.W. and Sedlet, J. 1969. "Radiochemical Determination of Technetium-99 in
6220 Environmental Water Samples," Analytical Chemistry., Vol. 41, No. 4, pp. 669-671.
6221 Gordon, L., Salutsky, M.L., and Willard, H.H. 1959. Precipitation from Homogeneous Solution,
6222 John Wiley and Sons, New York.
6223 Greenwood, N.N. andEarnshaw, A. 1984. Chemistry of the Elements, Pergamon, Oxford.
6224 Grimaldi, F.S. 1961. "Thorium," in Treatise on Analytical Chemistry, Kolthoff, I.M. and Elving,
6225 P.J., Eds., Part II, Volume 5, John Wiley and Sons, New York, pp. 142-216.
6226 Grindler, J.E. 1962. The Radiochemistry of Uranium, National Academy of Sciences-National
6227 Research Council (NAS-NS), NAS-NS 3050, Washington, DC.
6228 Hahn, P.P. 1945. "Radioactive Iron Procedures-Purification, Electroplating, and Analysis,"
6229 Industry and Engineering Chemistry, Analytical Edition, Vol. 17, pp. 45-46.
6230 Hahn, R.B. 1961. "Zirconium and Hafnium," in Treatise on Analytical Chemistry, Kolthoff, I.M.
6231 and Elving, P.J., Eds., Part U, Volume 5, John Wiley and Sons, New York, pp. 61-138.
6232 Hampel, C.A., Ed. 1968. The Encyclopedia of the Chemical Elements, Reinhold, New York.
6233 Hassinsky, M. and Adi off, J.-P. 1965. Radiochemical Survey of the Elements, Elsevier, New
6234 York.
6235 Hermann, J.A. and Suttle, J.F. 1961. "Precipitation and Crystallization," in Treatise on
6236 Analytical Chemistry, Kolthoff, I.M. and Elving, P.J., Eds., Part I, Volume 3, John Wiley and
6237 Sons, New York, pp. 1367-1410.
6238 Hillebrand, W.F., Lundell, G.E.F., Bright, H.A., and Hoffman, J.I. 1980, Applied Inorganic
6239 Analysis, Krieger Publishing, New York.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 14-211 DO NOT CITE OR QUOTE
-------
Separation Techniques
6240 Hindman, F.D. 1983. "Neodymium Fluoride Mounting for a Spectrometry Determinations of
6241 Uranium, Plutonium, and Americium," Analytical Chemistry, Vol. 55, pp. 2460-2461.
6242 Hindman, F.D. 1986. "Actinide Separations for a Spectrometry Using Neodymium Fluoride
6243 Coprecipitation," Analytical Chemistry, Vol. 58, pp. 1238-1241.
6244 Hirano, S., Matsuba, M., Kamada, H. 1989. "The Determination of "Tc in Marine Algae,"
6245 Radioisotopes, Vol. 38, pp. 186-189.
6246 Hiraoka, M. 1992. Crown Ethers and Analogous Compounds, Elsevier Science Publishers B.V.,
6247 Amsterdam.
6248 Holm, E., Rioseco, J., Garcia-Leon, M. 1984. "Determination of "Tc in Environmental
6249 Samples," Nuclear Instruments: Methods of Physics Research, Vol. 223, pp. 204-207.
6250 Horwitz, E.P., Kalina, D.G., Diamond, H., and Vandegrift, G.F. 1985. "The TRUEX Process- A
6251 Process for the Extraction of the Transuranic Elements from Nitric Acid Wastes Utilizing
6252 Modified PUREX Solvents," Solvent Extraction and Ion Exchange, Vol. 3, Nos. 1 and 2, pp.
6253 75-109.
6254 Horwitz, E.P., Diamond, H., and Martin, K.A. 1987. "The Extraction of Selected Actinides in the
6255 (HI) (IV) and (VI) Oxidation States from Hydrochloric Acid by OfD(iB)CMPO: The
6256 TRUEX-Chloride Process," Solvent Extraction and Ion Exchange, Vo. 5, No. 3, pp. 447-470.
6257 Horwitz, E.P., Dietz, M.L., Nelson, D.M., Larosa, J.J., and Fairman, W.D. 1990. "Concentration
6258 and Separation of Actinides from Urine Using a Supported Bifunctional Organophosphorus
6259 Extractant," Analytica Chimica Acta, Vol. 238, pp. 263-271.
6260 Horwitz, E.P., Dietz, M.L., and Fisher, D.E. 1991. "Separation and Preconcentration of
6261 Strontium from Biological, Environmental, and Nuclear Waste Samples by Extraction
6262 Chromatography Using a Crown Ether," Analytical Chemistry, Vol.63, pp. 522-525.
6263 Horwitz, E. P., Dietz, M.L., and Chiarizia, J. 1992. "The Application of Novel Extraction
6264 Chromatographic Materials to the Characterization of Radioactive Waste Solutions," Journal
6265 of Radioanalytical and Nuclear Chemistry, Vol. 161, pp. 575-583.
6266 Horwitz, E.P., Chiarizia, R., and Dietz, M.L., 1992. "A Novel Strontium-Selective Extraction
6267 Chromatographic Resin." Solvent Extraction and Ion Exchange, Vol. 10, No. 2, pp. 313-336.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 14-212 DRAFT FOR PUBLIC COMMENT
-------
Separation Techniques
6268 Horwitz, E.P., Chiarizia, R., Dietz, M.L., Diamond, Essling, A.M., Graczyk, D. 1992.
6269 "Separation and Preconcentration of Uranium from Acidic Media by Extraction
6270 Chromatography," Analytica Chimica Acta, Vol. 266, pp. 25-37.
6271 Horwitz, E.P., Chiarizia, R., Dietz, M.L., Diamond, H., and Nelson, D.M. 1993. "Separation and
6272 Preconcentration of Actinides from Acidic Media by Extraction Chromatography," Analytica
6273 Chimica Acta, Vol. 281, pp. 361-372.
6274 Horwitz, E.P., Dietz, M.L., Rhoads, S., Felinto, C., Gale, N.H., and Houghton, J. 1994. "A
6275 Lead-Selective Extraction Chromatographic Resin and Its Application to the Isolation of
6276 Lead from Geological Samples," Analytica Chimica Acta, Vol. 292, pp. 263-273.
6277 Hyde, E.K. 1960. The Radiochemistry of Thorium, National Academy of Sciences-National
6278 Research Council (NAS-NS), NAS-NS 3004, Washington, DC.
6279 International Atomic Energy Agency (IAEA). 1972. Analytical Chemistry of Nuclear Fuels,
6280 Vienna, Austria.
6281 Irving, H. and Williams, R.J.P. 1961. "Liquid-Liquid Extraction", in Treatise on Analytical
6282 Chemistry, Kolthoff, I.M. and Elving, P.J., Eds., Part I, Volume 3, John Wiley and Sons,
6283 New York, pp. 1309-1364.
6284 Jeter, H.W. and Grob, B. 1994. "Determination of Radiostrontium in Milk Samples Using an
6285 Extraction Chromatography Column," Radioactivity and Radiochemistry, Vol. 5, No. 3,
6286 1994, pp. 8-17.
6287 Kallmann, S. 1961. "The Alkali Metals," in Treatise on Analytical Chemistry, Kolthoff, I.M. and
6288 Elving, P.J., Eds., Part II, Volume 1, John Wiley and Sons, New York, pp. 301-446.
6289 Kallmann, S. 1964. "Niobium and Tantalum," in Treatise on Analytical Chemistry, Kolthoff,
6290 I.M. and Elving, P.J., Eds., Part II, Volume 6, John Wiley and Sons, New York, pp. 183-406.
6291 Kaplan, L. 1995. "Tritium"' McGraw-Hill Multimedia Encyclopedia of Science and Technology,
6292 McGraw-Hill, New York; Software Copyright: Online Computer Systems, Inc.
6293 Katz, J.J., Seaborg, G.T., and Morss, L.R. 1986. "Summary and Comparative Aspects of the
6294 Actinide Elements," in Katz, J.J., Seaborg, G.T., and Morss, L.R., Eds., The Chemistry of the
6295 Actinides, Vol. 2, Chapman and Hall, London, pp. 1121-1195.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 14-213 DO NOT CITE OR QUOTE
-------
Separation Techniques
6296 Katzin, L.I. 1986. "Thorium," in Katz, J.J., Seaborg, G.T., and Morss, L.R., Eds., The Chemistry
6297 of the Actinides, Vol. 1, Chapman and Hall, London, pp. 41-101.
6298 Kaye, J.H., Strebin, R.S., and Orr, R.D. 1995. "Rapid, Quantitative Analysis of Americium,
6299 Curium and Plutonium Isotopes in Hanford Samples Using Extraction Chromatography and
6300 Precipitation Plating," Journal of Radioanalytical andNuclea Chemistry, Articles, Vol. 194,
6301 No.l, pp. 191-196.
6302 Kirby, H.W. and Salutsky, M.L. 1964. The Radiochemistry of Radium, National Academy of
6303 Sciences-National Research Council (NAS-NS), NAS-NS 3057, Washington, DC.
6304 Kleinberg, J., Argersinger, W.J., and Griswald, E. 1966. Inorganic Chemistry, DC. Heath,
6305 Boston.
6306 Kleinberg, J. and Cowan, G.A. 1960. The Radiochemistry of Fluorine, Chlorine, Bromine, and
6307 Iodine, National Academy of Sciences-National Research Council (NAS-NRC), NAS-NRC
6308 3005, Washington, DC.
6309 Kolthoff, I.M. and Sandell, E.B. 1933. "Exchange Adsorption and Its Influence upon the
6310 Solubility of Precipitates with Ionic Lattices in Electrolyte Solution,"7ot/r/7a7 of the American
6311 Chemical Society, Vol. 5 5, p. 2170-2171.
6312 Kolthoff, I.M., Sandell, E.B., Meehan, E.J., and Bruckenstein, S. 1969. Quantitative Chemical
6313 Analysis, The Macmillan Company, New York.
6314 Korkisch, J. 1969. Modern Methods for the Separation of Rarer Metals, Pergamon Press, New
6315 York.
6316 Korkisch, J. 1989. Handbook of Ion Exchange Resins: Their Applicability to Inorganic
6317 Analytical Chemistry, Vol. I and II, CRC Press, Boca Raton, FL.
6318 Kraus, K.A., and Nelson, F. 1956. "Nuclear Chemistry and the Effects of Irradiation,"
6319 Proceedings of the International Conference on Peaceful Uses of Atomic Energy, 1st Geneva,
6320 Vol. 7, p. 113-167.
6321 Kressin, I.K. 1977. "Electrodeposition of Plutonium and Americium for High-Resolution Alpha
6322 Spectrometry," Analytical Chemistry, Vol.49, No. 6, pp. 842-846.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 14-214 DRAFT FOR PUBLIC COMMENT
-------
Separation Techniques
6323 Krivan, V. 1986. "Application of Radoiotracers to Methodological Studies in Trace Element
6324 Analysis, " in Treatise on Analytical Chemistry, Elving, P.P., Krivan, V., and Kolthoff, E.M.,
6325 Eds., John Wiley and Sons, New York, pp. 339-417.
6326 Kuska, Y. and Meinke, W. 1961. RapidRadiochemical Separations, National Academy of
6327 Sciences-National Research Council (NAS-NS), NAS-NS 3104, Washington, DC.
6328 Larsen, E.M. 1965. Transitional Elements, W.A. Benjamin, New York.
6329 Latimer, W.M. 1952. The Oxidation States of the Elements and Their Potentials in Aqueous
6330 Solutions, Prentice-Hall, Englewood Cliffs, NJ.
6331 Leussing, D.L. 1959. "Solubility," in Treatise on Analytical Chemistry, Kolthoff, I.M. and
6332 Elving, P.J., Eds., Part I, Volume 1, John Wiley and Sons, New York, pp. 675-732.
6333 Leyba, J.D., Vollmar, H.S., Fjeld, R.A., Devol, T.A., Brown, D.D., and Cadieux, J.R. 1995.
6334 "Evaluation of a Direct Extraction/Liquid Scintillation Counting Technique for the
6335 Measurement of Uranium in Water," Journal of Radioanalytical and Nuclear Chemistry,
6336 Articles, Vol. 194, No. 2, pp. 337-344.
6337 Lindsay, D.B. 1988. "Radionuclides," in Environmental Inorganic Chemistry, Bodek, I, Lyman,
6338 W.J., Reehl, W.F., and Rosenblatt, D.H., Eds., Pergammon, New York, pp. 9.1 - 9.3-3.
6339 Lingane, J.L. 1966. Analytical Chemistry of Selected Metallic Elements, Reinhold, New York.
6340 Lo, T.C., Baird, H.I., Hanson, C., Eds. 1983. Handbook of Solvent Extraction, Wiley-
6341 Interscience, New York.
6342 Lucas, H.F., Markun, F., and Boulenger, R. 1990. "Methods for Measuring Radium Isotopes:
6343 Emanometry," in The Environmental Behavior of Radium, Vol 1, Technical Report Series
6344 No. 310, International Atomic Energy Agency, Vienna, pp. 149-172.
6345 Martin, J.E. and Hylko, J.M. 1987. "Measurement of "Tc in Low-Level Radioactive Waste from
6346 Reactors Using 99mTc as a Tracer," Applied Radiation and Isotope, Vol. 38, No. 6, pp. 447-
6347 450.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 14-215 DO NOT CITE OR QUOTE
-------
Separation Techniques
6348 Marzilli, L. and Marzilli, P. A. 1995. "Cobalt," McGraw-Hill Multimedia Encyclopedia of
6349 Science and Technology, 1994 and 1996, McGraw-Hill, New York; Software Copyright:
6350 Online Computer Systems, Inc.
6351 McDowell, WJ. 1986. Alpha Counting and Spectrometry Using Liquid Scintillation Methods.,
6352 National Academy of Sciences-National Research Council (NAS-NRC), NAS-NRC 3116,
6353 Technical Information Center, Office of Scientific and Technical Information, U.S.
6354 Department of Energy, Washington, DC.
6355 McDowell, WJ. 1992. "Photon/Electron-Rejecting Alpha Liquid Scintillation (PERALS)
6356 Spectrometry: a Review," Radioactivity & Radiochemistry, Vol. 3, No. 2, pp. 26, 28, 30, 35-
6357 36, 38-42, 44-46, 48-50, 52-54.
6358 McMillan, J.W. 1975. "The Use of Tracers in Inorganic Analysis," in Radiochemical Methods in
6359 Analysis, Coomber, D.I., Ed., Plenum Press, pp. 297-348.
6360 Metz, C.F. and Waterbury, G.R. 1962. "The Transuranium Actinide Elements," in Treatise on
6361 Analytical Chemistry, Kolthoff, I.M. and Elving, P.J., Eds., Part II, Volume 9, John Wiley
6362 and Sons, New York, pp. 189-440.
6363 Mitchell, J. 1961. "Water," in Treatise on Analytical Chemistry, Kolthoff, I.M. and Elving, P.J.,
6364 Eds., Part U, Volume 1, John Wiley and Sons, New York, pp. 67-206.
6365 Mitchell, J. 1961. "Water," in Treatise on Analytical Chemistry, Kolthoff, I.M. and Elving, P.J.,
6366 Eds., Part H, Volume 1, John Wiley and Sons, New York, pp. 69-206.
6367 Mitchell, R.F. 1960. "Electrodeposition of Actinide Elements at Tracer Concentrations",
6368 Analytical Chemistry, Vol. 32, pp. 326-328.
6369 Moore, F.L. 1958. "Liquid-liquid Extraction of Uranium and Plutonium from Hydrochloric Acid
6370 Solution with Tri(iso-octyl)amine," Analytical Chemistry, Vol. 30, pp. 908-911.
6371 Morse, R.S., and Welford, GA. 1971. "Dietary Intake of 210Pb," Health Physics, Vol. 21, pp. 53-
6372 55.
6373 Nelson, F., Murase, T., and Kraus, K.A. 1964. "Ion Exchange Procedures I. Cation Exchange in
6374 Concentrated HC1 and HC1O4 Solutions," Journal of Chromatography, Vol. 13, pp. 503-535.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 14-216 DRAFT FOR PUBLIC COMMENT
-------
Separation Techniques
6375 Nguyen, S.N., Miller, P.E., Wild, J.F., and Hickman, D.P. 1996. "Simultaneous Determination of
6376 237Np, 232Th, and U isotopes in Urine Samples Using Extraction Chromatography, ICP-MS
6377 and Gamma-Ray Spectroscopy," Radioactivity and Radiochemistry, Vol. 7, No. 3, pp. 16-22.
6378 Nuclear Energy Agency (NEA). 1982. "The Geochemistry of Actinides, in Geological Disposal
6379 of Radioactive Waste: Geochemical Processes, Nuclear Energy Agency, Paris, pp.49-68.
6380 Orlandini, K.A. 1972. "Selective Ion Exchange for the Isolation of Certain Alkaline Earths," U.S.
6381 Patent No. 3694369, Sept. 26.
6382 Orlandini, K.A., King, J.G., and Erickson, M.D. 1998. "The Rapid Separation and Measurement
6383 of Technetium-99," submitted to Radiochimica Acta.
6384 Paducah Gaseous Diffusion Plant 1993. "99Tc Determination in Water," Method R-46.
6385 Parsa, B. 1998. "Contribution of Short-Lived Radionuclides to Alpha-Particle Radioactivity in
6386 Drinking Water and Their Impact on the Safe Drinking Water Act Regulations,"
6387 Radioactivity and Radiochemistry, Vol. 9, pp. 41-47.
6388 Passo, CJ. and Cook, G.T. 1994. Handbook of Environmental Liquid Scintillation Spectrometry:
6389 A Compilation of Theory and Methods, Packard Instrument Company, Meriden, CT, pp. 4 -
6390 1-6.
6391 Pauling, L. 1970. General Chemistry, Dover, New York.
6392 Penneman, R.A. 1994 and 1996. "Americium," McGraw-Hill Multimedia Encyclopedia of
6393 Science and Technology, McGraw-Hill, New York; Software Copyright: Online Computer
6394 Systems, Inc.
6395 Penneman, R.A. and Keenan, T.K. 1960. "The Radiochemistry of Americium and Curium,"
6396 National Academy of Sciences-National Research Council (NAS-NRC), NAS-NRC 3006,
6397 Washington, DC.
6398 Perrin, D.D. 1979. "Masking and Demasking in Analytical Chemistry," in Treatise on Analytical
6399 Chemistry, 2nd Ed., Kolthoff, I.M. and Elving, P.J., Eds., Part I, Volume 2, John Wiley and
6400 Sons, New York, pp. 599-643.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 14-217 DO NOT CITE OR QUOTE
-------
Separation Techniques
6401 Perry, E.S. and Weissberger, A. 1965. "Distillation," in Technique of Organic Chemistry. Perry,
6402 E.S. and Weissberger, A., Eds., Second edition, Volume IV, Wiley-Interscience, New York.
6403 Peters, D.G., Hayes, J.M, and Hieftje, G.M. 1974. Chemical Separations and Measurements:
6404 Theory and Practice of Analytical Chemistry, W.B. Saunders Company, New York.
6405 Pimpl, M. 1995. "89Sr/90Sr-Determination in Soils and Sediments Using Crown Ethers for Ca/Sr-
6406 Separation," Journal of Radioanalytical and Nuclear Chemistry, Articles, Vol. 194, No. 2,
6407 pp. 311-318.
6408 Pin, C. and Bassin, C. 1992. "Evaluation of a Strontium-Specific Extraction Chromatographic
6409 Method for Isotopic Analysis in Geological Materials," Analytica Chimica Acta, Vol. 269,
6410 pp. 249-255.
6411 Pin, C., Briot, D., Bassin, C., and Poitasson, F. 1994. " Concomitant Separation of Strontium and
6412 Samarium-Neodymium for Isotopic Analysis in Silicate Samples, Based on Specific
6413 Extraction Chromatography," Analytica Chimica Acta, Vol. 298, pp. 209-217.
6414 Pin, C. and Zalduequi, J.F.S. 1997. "Sequential Separation of Rare-Earth Elements, Thorium and
6415 Uranium by Miniaturized Extraction Chromatography: Application to Isotopic Analyses of
6416 Silicate Rocks," Analytica Chimica Acta, Vol. 339, pp. 79-89.
6417 Rieman, W. and Walton, H. 1970. Ion Exchange in Analytical Chemistry, Pergamon Press, New
6418 York.
6419 Riley, R.F. 1995. "Strontium." McGraw-Hill Multimedia Encyclopedia of Science and
6420 Technology, McGraw-Hill, New York; Software Copyright: Online Computer Systems, Inc.
6421 Salutsky, M.L. 1959. "Precipitates: Their Formation, Properties, and Purity," in Treatise on
6422 Analytical Chemistry, Kolthoff, I.M. and Elving, P.J., Eds., Part I, Volume 1, John Wiley and
6423 Sons, New York, pp. 733-766.
6424 Salutsky, M.L. 1995. "Radium," McGraw-Hill Multimedia Encyclopedia of Science and
6425 Technology, 1994 and 1996, McGraw-Hill, New York; Software Copyright: Online
6426 Computer Systems, Inc.
6427 Salutsky, M.L. 1997. "Radium," in McGraw-Hill Encyclopedia of Science and Technology,
6428 Parker, S.P., Ed. in Chief, Vol. 15, McGraw-Hill, New York, pp. 177-179.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 14-218 DRAFT FOR PUBLIC COMMENT
-------
Separation Techniques
6429 S. Cohen and Associates, Inc. (SCA) 2001. Laboratory Quality Assurance Plan: Standard
6430 Operating Procedures, Volume n, S. Cohen and Associates, Inc., Montgomery, Ala.
6431 Schulz, W.W. and Penneman, R.A. 1986. "Americium," in Katz, J.J., Seaborg, G.T., and Morss,
6432 L.R., Eds., The Chemistry of the Actinides, Vol. 2, Chapman and Hall, London, pp. 887-961.
6433 Seaborg, G. T. and Loveland, W.D. 1990. The Elements Beyond Uranium., John Wiley & Sons,
6434 New York.
6435 Sedlet, J. 1966. "Radon and Radium," in Treatise on Analytical Chemistry, Kolthoff, I.M. and
6436 Elving, P.J., Eds., Part II, Volume 4, John Wiley and Sons, New York, pp. 219-316.
6437 Sedlet, J. 1964. "Actinium, Astatine, Francium, Polonium, and Protactinium," in Treatise on
6438 Analytical Chemistry, Kolthoff, I.M. and Elving, P.J., Eds., Part II, Volume 6, John Wiley
6439 and Sons, New York, pp. 435-610.
6440 Showsmith, D.W. 1984. "The Behavior of Radium in Soil and Uranium Mill Tailings," AECL-
6441 7818, Whitshell Nuclear Research Establishment, Tinawa, Manitoba, Canada.
6442 Sill, C.W., Puphal, K.W., and Hindman, F.D. 1974. "Simultaneous Determination of Alpha-
6443 Emitting Nuclides of Radium Through Californium in Soil," Analytical Chemistry, Vol. 46,
6444 No. 12, pp. 1725-1737.
6445 Sill, C.W. and Williams, R.L. 1981. "Preparation of Actinides for Alpha Spectrometry without
6446 Electrodeposition," Analytical Chemistry, Vol. 53, pp. 421-415.
6447 Sittig, M. 1994 and 1996. "Cesium," McGraw-Hill Multimedia Encyclopedia of Science and
6448 Technology, McGraw-Hill, New York; Software Copyright: Online Computer Systems, Inc.
6449 Smith, LL., Crain, J.S., Yaeger, J.S., Horwitz, E.P., Diamond, H., and Chiarizia, R. 1995.
6450 "Improved Separation Method for Determining Actinides in Soil Samples," Journal of
6451 RadioanayticalNuclear Chemistry, Articles, Vol. 194, No 1, pp. 151-156.
6452 Smith, L.L., Orlandini, K.A., Alvarado, J.S., Hoffmann, K.M., Seely, DC., and Shannon, R.T.
6453 1996. "Application of Empore™ Strontium Rad Disks to the Analysis of Radiostrontium in
6454 Environmental Water Samples," Radiochimica Acta, Vol. 73, pp. 165-170.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 14-219 DO NOT CITE OR QUOTE
-------
Separation Techniques
6455 Smith, L.L., Alvarado, J.S., Markun, F.J., Hoffmann, K.M., Seely, DC., and Shannon, R.T. 1997.
6456 "An Evaluation of Radium-Specific, Solid-Phase Extraction Membranes," Radioactivity and
6457 Radiochemistry, Vol 8 No. 1, pp. 30-37.
6458 SpecNews 1993. "Product Overview," Vol. 2, Issue 3, Eichrom Industries, Inc., Darien, IL, p. 3.
6459 Steinberg, E.O. 1960. The Radiochemistry of Zirconium and Hafnium., National Academy of
6460 Sciences-National Research Council (NAS-NRC), NAS-NRC 3011, Washington, DC.
6461 Strebin, R. et al. ??et al.???. 1997. "Nickel-59 and Nickel-63 Determination in Aqueous
6462 Samples". DOE Methods Compendium. {SEE NOTE IN ABOVE REFERENCE}
6463 Sullivan, T.M., Nelson, D.M., and Thompson, E.G. 1993. "Monitoring for "Tc in Borehole
6464 Waters Using an Extraction Chromatographic Resin," Radioactivity and Radiochemistry, Vol
6465 4, No.2, pp. 14-18.
6466 Sunderman, D.N. and Townley, C.W. 1960. "The Radiochemistry of Barium, Calcium, and
6467 Strontium," National Academy of Sciences-National Research Council (NAS-NS), NAS-NS
6468 3010, Washington, DC.
6469 Talvitie, N.A. 1972. "Electrodeposition of Actinides for Alpha Spectrometric Determination,"
6470 Analytical Chemistry, Vol. 44, No. 2, pp. 280-283.
6471 Testa, C., Desideri, D., Meli, M.A., and Roselli, C. 1995. "New Radiochemical Procedures for
6472 Environmental Measurements and Data Quality Control," Journal of Radioanalytical and
6473 Nuclear Chemistry, Articles, Vol. 194, No. 1, pp. 141-149.
6474 Turekian, K.K. and Bolter, E. 1966. "Strontium and Barium," in Treatise on Analytical
6475 Chemistry, Kolthoff, I.M. and Elving, P.J., Eds., Part II, Volume 4, John Wiley and Sons,
6476 New York, pp. 153-218.
6477 Vdovenko, V.M. and Dubasov, Yu.V. 1975. Analytical Chemistry of Radium," in the series
6478 Analytical Chemistry of the Elements, Malament, D., Ed., John Wiley and Sons, New York.
6479 Wahl, A.C. and Bonner, N.A. 1951. Radioactivity Applied to Chemistry, John Wiley and Sons,
6480 New York.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 14-220 DRAFT FOR PUBLIC COMMENT
-------
Separation Techniques
6481 Wang, C.H., Willis, D.L., and Loveland, W.D. 1975. Radiotracer Methodology in the Biological,
6482 Environmental and Physical Sciences, Prentice-Hall, New York.
6483 Weigel, F. 1986. "Uranium," in Katz, J.J., Seaborg, G.T., and Morss, L.R., Eds., The Chemistry
6484 of the Actinides, Vol. 1, Chapman and Hall, London, pp. 169-442.
6485 Weigel, F., Katz, J.J., and Seaborg, G.T. 1986. "Plutonium," in Katz, J.J., Seaborg, G.T., and
6486 Morss, L.R., Eds., The Chemistry of the Actinides, Vol. 1, Chapman and Hall, London, pp.
6487 499-886.
6488 Weigel, F. 1995. "Plutonium," McGraw-Hill Multimedia Encyclopedia of Science and
6489 Technology, 1994 and 1996, McGraw-Hill, New York; Software Copyright: Online
6490 Computer Systems, Inc.
6491 Weigel, F. 1995. "Uranium," McGraw-Hill Multimedia Encyclopedia of Science and
6492 Technology, 1994 and 1996, McGraw-Hill, New York; Software Copyright: Online
6493 Computer Systems, Inc.
6494 Willard, H.H. and Rulfs, C.L. 1961. "Decomposition and Dissolution of Samples: Inorganic," in
6495 Treatise on Analytical Chemistry, Kolthoff, I.M. and Elving, P.J., Eds., Part I, Volume 2,
6496 John Wiley and Sons, New York, pp. 1027-1050.
6497 Wright, B.T. 1947. "Recoil of Silver Nuceli Due to D-Capture in Cadmium," Physical Review,
6498 Vol. 71, No. 12, pp. 839-841.
6499 Woittiez, J.R.W. and Kroon, KJ. 1995. "Fast, Selective and Sensitive Methods for the
6500 Determination of Pb-210 in Phosphogypsum and Phosphate Ore," Journal of Radioanalytical
6501 and Nuclear Chemistry, Vol. 194, No. 2, pp. 319-329.
6502 Wray, J.L. and Daniels, F. 1957. "Precipation of Calcite and Aragonite," Journal of the American
6503 Chemical Society, Vol. 79, pp. 2031 -203 4.
6504 Zolotov, Yu.A. and Kuz'man, N.M. 1990. Preconcentration of Trace Elements, Vol. XXV of
6505 Wilson and Wilson's Comprehensive Analytical Chemistry, G. Svehla, Ed., Elsevier Science
6506 Publishers, Amsterdam.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 14-221 DO NOT CITE OR QUOTE
-------
Separation Techniques
6507 14.12 Selected Bibliography
6508 14.12.1 Inorganic and Analytical Chemistry
6509 Baes, C.F. and Mesmer, R.E. 1976. The Hydrolysis of Cations, John Wiley and Sons, New York.
6510 Bard, A.J., Parsons, R., and Jordan, J. 1985. Standard Potentials in Aqueous Solution, Marcel
6511 Dekker, New York.
6512 Bodek, I, Lyman, W.J., Reehl, W.F., and Rosenblatt, D.H., Eds. 1988. Environmental Inorganic
6513 Chemistry, Pergammon, New York.
6514 Cotton, F.A. and Wilkinson, G. 1988. Advanced Inorganic Chemistry, John Wiley and Sons,
6515 New York.
6516 Dean, J.A. 1995, Analytical Chemistry Handbook, McGraw-Hill, New York.
6517 Dorfner, K. 1972. Ion Exchangers: Properties and Applications, Ann Arbor Science Publishers,
6518 Ann Arbor, Michigan.
6519 Greenwood, N.N. andEarnshaw, A. 1984. Chemistry of the Elements, Pergamon, Oxford.
6520 Karger, B.L., Snyder, L.R., and Horvath, C. 1973. An Introduction to Separation Science, John
6521 Wiley and Sons, New York.
6522 Kolthoff, I.M., Sandell, E.B., Meehan, E.J., and Bruckenstein, S. 1969. Quantitative Chemical
6523 Analysis, The Macmillan Company, New York.
6524 Latimer, W.M. 1952. The Oxidation States of the Elements and Their Potentials in Aqueous
6525 Solutions, Prentice-Hall, Englewood Cliffs, NJ.
6526 Zolotov, Yu.A. and Kuz'man, N.M. 1990. Preconcentration of Trace Elements, Vol. XXV of
6527 Wilson and Wilson's Comprehensive Analytical Chemistry, G. Svehla, Ed., Elsevier Science
6528 Publishers, Amsterdam.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 14-222 DRAFT FOR PUBLIC COMMENT
-------
Separation Techniques
6529 14.12.2 General Radiochemistry
6530 Adolff, J.-P. and Guillaumont, R. 1993. Fundamentals of Radiochemistry., CRC Press, Boca
6531 Raton, Florida.
6532 Friedlander, G., Kennedy, J.W., Macias, E.S., and Miller, J.M. 1981. Nuclear and
6533 Radiochemistry, John Wiley and Sons, New York.
6534 Choppin, G., Rydberg, J., Liljenzin, J.O. 1995. Radiochemistry and Nuclear Chemistry.,
6535 Butterworth-Heinemann, Oxford.
6536 Coomber, D.I., Ed. 1975. RadiochemicalMethods in Analysis, Plenum Press, New York.
6537 Wahl, A.C. and Bonner, N.A. 1951, Second Printing: May, 1958. Radioactivity Applied to
6538 Chemistry., John Wiley and Sons, New York.
6539 14.12.3 Radiochemical Methods of Separation
6540 Crouthamel, C.E. and Heinrich, R.R. 1971. "Radiochemical Separations," in Treatise on
6541 Analytical Chemistry, Kolthoff, I.M. and Elving, P.J., Eds., Part I, Volume 9, John Wiley and
6542 Sons, New York, pp. 5467-5511.
6543 Dietz, M.L. and Horwitz, E.P. 1993. "Novel Chromatographic Materials Based on Nuclear Waste
6544 Processing Chemistry," LC-GC, The Magazine ofSepartion Science, Vol. 11, No. 6, pp. 424-
6545 426, 428, 430, 434, 436.
6546 Horwitz, E. P., Dietz, M.L., and Chiarizia, J. 1992. "The Application of Novel Extraction
6547 Chromatographic Materials to the Characterization of Radioactive Waste Solutions," Journal
6548 of Radioanalytical and Nuclear Chemistry, Vol. 161, pp. 575-583.
6549 14.12.4 Radionuclides
6550 Anders, E. 1960. The Radiochemistry of Technetium, National Academy of Sciences-National
6551 Research Council (NAS-NS), NAS-NS 3021, Washington, DC.
6552 Bate, L.C. and Leddicotte, G. W. 1961. The Radiochemistry of Cobalt, National Academy of
6553 Sciences-National Research Council, (NAS-NS), NAS-NS 3041, Washington, DC.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 14-223 DO NOT CITE OR QUOTE
-------
Separation Techniques
6554 Booman, G.L. and Rein, I.E. 1962. "Uranium," in Treatise on Analytical Chemistry., Kolthoff,
6555 I.M. and Elving, P.J., Eds., Part II, Volume 9, John Wiley and Sons, New York, pp. 1-188.
6556 Cleveland, J.M. 1970. The Chemistry of Plutonium, Gordon and Breach Science Publishers, New
6557 York.
6558 Cobble, J.W. 1964. "Technetium," in Treatise on Analytical Chemistry, Kolthoff, I.M. and
6559 Elving, P.J., Eds., Part II, Volume 6, John Wiley and Sons, New York, pp. 404-434.
6560 Coleman, G.H. 1965. The Radiochemistry of Plutonium, National Academy of Sciences-
6561 National Research Council (NAS-NS), NAS-NS 3058, Washington, DC.
6562 Finston, H.L. and Kinsley, M.T. 1961. The Radiochemistry of Cesium, National Academy of
6563 Sciences-National Research Council (NAS-NS), NAS-NS 3035, Washington, DC.
6564 Grimaldi, F.S. 1961. "Thorium," in Treatise on Analytical Chemistry, Kolthoff, I.M. and Elving,
6565 P.J., Eds., Part II, Volume 5, John Wiley and Sons, New York, pp. 142-216.
6566 Grindler, J.E. 1962. The Radiochemistry of Uranium, National Academy of Sciences-National
6567 Research Council (NAS-NS), NAS-NS 3050, Washington, DC.
6568 Hahn, R.B. 1961. "Zirconium and Hafnium," in Treatise on Analytical Chemistry, Kolthoff, I.M.
6569 and Elving, P.J., Eds., Part H, Volume 5, John Wiley and Sons, New York, pp. 61-138.
6570 Hyde, E.K. 1960. The Radiochemistry of Thorium, National Academy of Sciences-National
6571 Research Council (NAS-NS), NAS-NS 3004, Washington, DC.
6572 Kallmann, S. 1961. "The Alkali Metals," in Treatise on Analytical Chemistry, Kolthoff, I.M. and
6573 Elving, P.J., Eds., Part II, Volume 1, John Wiley and Sons, New York, pp. 301-446.
6574 Kallmann, S. 1964. "Niobium and Tantalum," Kallmann, S., in Treatise on Analytical Chemistry,
6575 Kolthoff, I.M. and Elving, P.J., Eds., Part II, Volume 6, John Wiley and Sons, New York pp.
6576 183-406.
6577 Kirby, H.W. and Salutsky, M.L. 1964. The Radiochemistry of Radium, National Academy of
6578 Sciences-National Research Council (NAS-NS), NAS-NS 3057, Washington, DC.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 14-224 DRAFT FOR PUBLIC COMMENT
-------
Separation Techniques
6579 Kleinberg, J. and Cowan, G.A. 1960. The Radiochemistry of Fluorine, Chlorine, Bromine, and
6580 Iodine, National Academy of Sciences-National Research Council (NAS-NRC), NAS-NRC
6581 3005, Washington, DC.
6582 Metz, C.F. and Waterbury, G.R. 1962. "The Transuranium Actinide Elements," in Treatise on
6583 Analytical Chemistry, Kolthoff, I.M. and Elving, P.J., Eds., Part II, Volume 9, John Wiley
6584 and Sons, New York, pp. 189-440.
6585 Schulz, W.W. and Penneman, R.A. 1986. "Americium," in The Chemistry of the Actinides, Katz,
6586 J.J., Seaborg, G.T., and Morss, L.R., Eds.,Vol. 2, Chapman and Hall, London, pp. 887-961.
6587 Seaborg, G. T. and Loveland, W.D. 1990. The Elements Beyond Uranium, John Wiley & Sons,
6588 New York.
6589 Sunderman, D.N. and Townley, C.W. 1960. "The Radiochemistry of Barium, Calcium, and
6590 Strontium," National Academy of Sciences-National Research Council (NAS-NS), NAS-NS
6591 3010, Washington, DC.
6592 Steinberg, E.O. 1960. The Radiochemistry of Zirconium and Hafnium, National Academy of
6593 Sciences-National Research Council (NAS-NRC), NAS-NRC 3011, Washington, DC.
6594 Sedlet, J. 1966. "Radon and Radium," in Treatise on Analytical Chemistry, Kolthoff, I.M. and
6595 Elving, P.J., Eds., Part II, Volume 4, John Wiley and Sons, New York, pp. 219-316.
6596 Turekian, K.K. and Bolter, E. 1966. "Strontium and Barium," in Treatise on Analytical
6597 Chemistry, Kolthoff, I.M. and Elving, P.J., Eds., Part II, Volume 4,John Wiley and Sons,
6598 New York, pp. 153-218.
6599 14.12.5 Separation Methods
6600 Berg, E.W. 1963. Physical and Chemical Methods of Separation, McGraw-Hill, New York.
6601 Hermann, J.A. and Suttle, J.F. 1961. "Precipitation and Crystallization," in Treatise on
6602 Analytical Chemistry, Kolthoff, I.M. and Elving, P.J., Eds., Part I, Volume 3, John Wiley and
6603 Sons, New York, pp. 1367-1410.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 14-225 DO NOT CITE OR QUOTE
-------
Separation Techniques
6604 Irving, H. and Williams, R.J.P. 1961. "Liquid-Liquid Extraction", in Treatise on Analytical
6605 Chemistry., Kolthoff, I.M. and Elving, P.J., Eds., Part I, Volume 3, John Wiley and Sons,
6606 New York, pp. 1309-1364.
6607 Leussing, D.L. 1959. "Solubility," in Treatise on Analytical Chemistry, Kolthoff, I.M. and
6608 Elving, P.J., Eds., Part I, Volume 1, John Wiley and Sons, New York, pp. 675-732.
6609 Perrin, D.D. 1979. "Masking and Demasking in Analytical Chemistry," in Treatise on Analytical
6610 Chemistry, 2nd Ed., Kolthoff, I.M. and Elving, P.J., Eds., Part I, Volume 2, John Wiley and
6611 Sons, New York, pp. 599-643.
6612 Rieman, W. and Walton, H. 1970. Ion Exchange in Analytical Chemistry, Pergamon Press, New
6613 York.
6614 Salutsky, M.L. 1959. "Precipitates: Their Formation, Properties, and Purity," in Treatise on
6615 Analytical Chemistry, Kolthoff, I.M. and Elving, P.J., Eds., Part I, Volume 1, John Wiley and
6616 Sons, New York, pp. 733-766.
6617 Willard, H.H. and Rulfs, C.L. 1961. "Decomposition and Dissolution of Samples: Inorganic," in
6618 Treatise on Analytical Chemistry, Kolthoff, I.M. and Elving, P.J., Eds., Part I, Volume 2,
6619 John Wiley and Sons, New York, pp. 1027-1050.
6620 Yu.A. and Kuz'man, N.M. 1990. Preconcentration of Trace Elements, Zolotov, Vol. XXV of
6621 Wilson and Wilson's Comprehensive Analytical Chemistry, G. Svehla, Ed., Elsevier Science
6622 Publishers, Amsterdam.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 14-226 DRAFT FOR PUBLIC COMMENT
-------
15 NUCLEAR COUNTING INSTRUMENTATION
Portions of this chapter have been extracted, with permission, from D 3648-95-Standard
Practice for the Measurement of Radioactivity, copyright American Society for Testing and
Materials, 100 Barr Harbor Drive, West Conshohocken, PA 19428. A copy of the complete
standard may be purchased from ASTM (tel: 610-832-9585, fax: 610-832-9555, e-mail:
service@astm.org, website: www.astm.org).
7 15.1 Introduction
8 This chapter presents descriptions of counting techniques to help the user to determine what
9 radioanalytical measurement method(s) best suit a given need. References cited in the text
10 provide additional details of how these measurements are made. The primary focus here is on the
11 variables that ultimately affect the bias and precision of the counting data. The type of informa-
12 tion that is desired—in relation to the type of radiation to detect—will determine the type of
13 instrument and associated technique one will use to generate data. For example, samples
14 containing a single radionuclide of high purity, sufficient energy, and ample activity may only
15 require a simple detector system. In this case, the associated investigation techniques may offer
16 no complications other than those related to calibration and reproducibility. At the other extreme,
17 a sample or set of samples may require quantitative identification of many radionuclides or the
18 laboratory may need to prepare unique calibration standards. In the latter case, specialized
19 instruments are available. Typically, a radiochemical laboratory will encounter samples routinely
20 that require a level of information between the two extremes described above.
21 A typical laboratory may be equipped with the following nuclear counting instrumentation:
22 • Proportional or Geiger-Mueller detectors for alpha and beta counting;
23 • Sodium iodide or high resolution germanium detectors for gamma detection and
24 spectrometry;
25 • Solid state detectors for alpha spectrometry;
26 • Scintillation counters suitable for both alpha- or beta-emitting radionuclides; and
27 • Multichannel analyzers for alpha and gamma-ray spectrometry.
28 A basic requirement for accurate measurements is the use of high quality standards, traceable to a
29 national standards organization (Section 15.9; ANSIN42.22, ANSIN42.23), to calibrate
JUNE 2001 DRAFT MARLAP
NOT FOR DISTRIBUTION 15-1 DO NOT CITE OR QUOTE
-------
Nuclear Counting Instrumentation
30 instrumentation. Generally, with the present availability of good standards, radiochemistry
31 laboratories rarely require instrumentation suitable for producing their own calibration standards.
32 However, it is always advisable to compare each new standard received against the previous
33 standard. The next three main sections of the chapter describe counting instrumentation for alpha
34 (Section 15.2), beta (Section 15.3), and gamma (Section 15.4) radiation. In a number of cases the
35 same instrumentation is used for radionuclides with one or more types of radiation. Note that a
36 review covering descriptions of radionuclides, types of radiation, associated principles, and
37 definitions for related terminology is given in Appendix A of this manual. The discussion next
38 turns to several specific areas to cover spectrometry (Section 15.5), special instrumentation
39 (Section 15.6), and spectrometers and energy-dependent detectors (Section 15.7). Shielding
40 (Section 15.8) to reduce detector background of nuclear counting instruments and instrument
41 calibration (Section 15.9) follows. This chapter closes with a discussion of other nuclear
42 counting instrumentation considerations (Section 15.10) including a discussion on non-nuclear
43 instrumentation (Section 15.10.4).
44 15.2 Alpha Counting
45 15.2.1 Introduction
46 Alpha particles are relatively massive, expend their energy over short distances, and typically
47 exhibit limited penetration into neighboring materials. Alpha particles are also characterized by
48 an intense loss of energy while passing through matter (see ICRU, 1992, for a discussion of dose
49 equivalents and linear energy transfer). This loss of energy is used to differentiate alpha
50 radioactivity from other types through the dense ionization or intense scintillation it produces.
51 This high rate of loss of energy in passing through matter, however, also makes sample
52 preparation conditions for alpha counting more stringent than is necessary for other types of
53 radiation. An example of direct alpha counting to determine total alpha activity is given in
54 ASTM C799.
55 Alpha radioactivity normally is measured by one of several types of detectors in combination
56 with suitable electronic components. The detector devices most used are ionization chambers,
57 proportional counters, silicon semiconductor detectors, and scintillation counters. The associated
58 electronic components in all cases include high-voltage power supplies, preamplifiers, amplifiers,
59 sealers, analog-to-digital converters, and recording devices.
60 The measured alpha-counting rate from a sample will depend on a number of variables. The most
61 important of these variables are:
DRAFT MARLAP JUNE 2001
DO NOT CITE OR QUOTE 15 -2 NOT FOR DISTRIBUTION
-------
Nuclear Counting Instrumentation
62
63
64
65
66
67
Geometry;
Source diameter;
Self-absorption;
Absorption in air and detector window;
Coincidence losses; and
Backscatter.
68 These are discussed in detail in the literature (Blanchard et al., 1960; Hallden and Fisenne, 1963),
69 and can be measured or corrected for in many cases by holding conditions constant during the
70 counting of samples and standards.
71 Alpha counters have low backgrounds and high efficiencies. Thus, outside sources of alpha
72 radiation will not impact the counting process and the instrument essentially focuses on the alpha
73 source presented by the sample. However, some counters are easily contaminated internally and
74 care should be taken to avoid contamination. Silicon detectors operated in a vacuum may become
75 contaminated due to recoil from sources (Merritt et al., 1956). Some alpha counters are sensitive
76 to beta radiation depending on the detector (Blanchard et al., 1960; Hallden and Fisenne, 1963).
77 In these cases, electronic discrimination is often used to eliminate the smaller pulses due to beta
78 particles. A discussion of alpha particle attenuation can be found in Section 15.10.1.1.
79 15.2.2 Detectors for Alpha Counting
80 15.2.2.1 lonization Chambers
81 As the incident particle enters the ionization chamber, ionization occurs through the interaction
82 of the particle with the fill gas. The secondary electrons produced through these interactions are
83 accelerated toward the anode as a result of the bias applied to the system. An ion current is
84 produced at the anode as a result of the collection of the free electrons (negative ions) generated
85 through ionization interactions. The charge collected at the anode is collected across an RC
86 circuit resulting in a change in potential across a capacitor. The change in potential is thus related
87 to the charge produced from the collection of electrons produced through the ionization
88 interactions of the incident particle.
89 15.2.2.2 Proportional Counters
90 As the incident particle enters the proportional counter, ionization occurs through the interaction
91 of the particle with the fill gas. The secondary electrons produced through these interactions are
92 accelerated toward the anode as a result of the bias applied to the system. In proportional
JUNE 2001
NOT FOR DISTRIBUTION
15-2
DRAFT MARLAP
DO NOT CITE OR QUOTE
-------
Nuclear Counting Instrumentation
93 counters, the free electrons gain sufficient kinetic energy during acceleration to produce
94 secondary ionization as they migrate toward the anode. This effect, known as "gas multiplica-
95 tion," is used to amplify the charge collected at the anode. Similar to ionization chambers, the
96 charge collected at the anode is collected across an RC circuit resulting in a change in potential
97 across a capacitor. As a result of gas multiplication, the voltage pulse produced is considerably
98 larger than the pulse produced in an ionization chamber. The magnitude of the voltage pulse is
99 thus proportional to the original number of ion pairs formed by the incident particle.
100 Proportional detectors are generally constructed of stainless steel, oxygen free/high conductivity
101 (OFHC) copper, or aluminum. No additional shielding is required for alpha proportional
102 counting. The counter should be capable of accepting mounts up to 51 mm in diameter.
103 Proportional counters are available in two types, either with or without a window between the
104 sample and the counting chamber. The manufacturer's specifications for either type should
105 include performance estimates of background count rate, length and slope of the voltage plateau,
106 and efficiency of counting a specified electrodeposited standard source, along with the type of
107 gas used in the tests. For a window flow counter, the window thickness—in milligrams per
108 square centimeter—also should be specified. With a windowless flow counter the sample and
109 sample mount should be made of an electrical conductor in order to avoid erratic behavior due to
110 static charge buildup.
111 Typical parameters for the alpha windowless flow counter are:
112 b ackground count rate = 10 counts/h or 2.8 x 10"3 cp s
113 length of voltage plateau = 300V
114 slope of voltage plateau = 1 %/l 00 V for an electrodeposited source
115 For a window flow counter, typical values are:
116 window thickness = 0.08 to 0.5 mg/cm2
117 background count rate = 10 counts/h or 2.8^10"3 cps
118 length of voltage plateau = 300V
119 slope of voltage plateau = 1 %/l 00 V for an electrodeposited source
120 efficiency = 35 to 40 percent for an electrodeposited source
121 15.2.2.3 Scintillation Counters
122 In a scintillation counter, the alpha particle transfers energy to a scintillator, such as zinc sulfide
123 (silver activated). The transfer of energy to the scintillator results in the production of light at a
DRAFT MARLAP JUNE 2001
DO NOT CITE OR QUOTE 15 -4 NOT FOR DISTRIBUTION
-------
Nuclear Counting Instrumentation
124 wavelength characteristic to the scintillator, and with an intensity proportional to the energy
125 transmitted from the alpha particle. The scintillator medium is placed in close proximity to the
126 cathode of a multiplier phototube; light photons from the scintillator strike the photo cathode,
127 and electrons are emitted. The photoelectrons are passed through a series of dynodes resulting in
128 the multiplication of electrons at each stage of the multiplier phototube. After amplification, a
129 typical scintillation vent will give rise to 107 to 1010 electrons, which is sufficient to serve as a
130 signal charge for the scintillation event. The electrons are collected across an RC circuit, which
131 results in a change in potential across a capacitor, thus giving rise to a pulse used as the
132 electronic signal of the initial scintillation event.
133 The counter size is limited by the multiplier phototube size, a diameter of 51 mm being the most
134 common. Two types of systems may be employed. In the first, the phosphor is optically coupled
135 to the multiplier phototube and either is covered with a thin (<1 mg/cm2) opaque window or
136 enclosed in a light-proof sample changer. With the sample placed as close as possible to the
137 scintillator, efficiencies approaching 40 percent may be obtained. The second system employs a
138 bare multiplier phototube housed in a light-proof assembly. The sample is mounted in contact
139 with a disposable zinc sulfide disk and placed on the phototube for counting. This system gives
140 efficiencies approaching 50 percent, is associated with a slightly lower background, and less
141 chance of counter contamination.
142 A major advantage of alpha scintillation counting is that the sample need not be conducting. For
143 a 51 mm multiplier phototube with the phosphor coupled to the tube, typical values obtained are
144 a background count rate of 0.006 cps and an efficiency for an electrodeposited standard source of
145 35 to 40 percent. With a disposable phosphor mounted on the sample, typical values are a
146 background count rate of 0.003 cps and an efficiency for an electrodeposited standard source of
147 45 to 50 percent. For both systems, voltage plateau length is 150 V with a slope of 5
148 percent/100 V.
149 15.2.2.4 Liquid Scintillation Counters
150 Liquid scintillation counting of alpha emitters with a commercially available instrument
151 overcomes many of the problems inherent in other techniques (Passo and Cook 1994; Horrocks,
152 1974; DeFilippis, 1990; Friedlander et al., 1964; Curtis et al., 1955; Matt and Ramsden, 1964;
153 Overman and Clark, 1960; Price, 1964; Flynn et al., 1971). Typical background counting rates
154 range from 0.1 to 0.2 cps. Sample preparation, after radiochemical separation is performed,
155 involves mixing the sample aliquant with a suitable liquid scintillator solution or gel phosphor
156 before counting. In this way, planchet preparation is eliminated, volatile components are retained,
157 and the completely enclosed sample cannot contaminate the counting chamber. Ideally, the
JUNE 2001 DRAFT MARLAP
NOT FOR DISTRIBUTION 15-5 DO NOT CITE OR QUOTE
-------
Nuclear Counting Instrumentation
158 sample is uniformly distributed in the scintillator so there is no self-absorption. This results in a
159 counting efficiency of almost 100 percent. Because of the high alpha energies, considerable
160 chemical quenching effects can be tolerated before counting efficiency is reduced. Coincidence
161 losses are small in liquid scintillation counting at count rates up to 2><104 cps. For samples that
162 contain both alpha and high-energy beta emitters, difficulties do arise in distinguishing between
163 the two. The problem is due primarily to the broad continuum of beta energy distribution up to
164 the maximum energy and the poor resolution of liquid scintillation spectrometers. This problem
165 is aggravated because the light yield per million electron volts of alpha particles in most liquid
166 scintillators is approximately tenfold lower than a beta particle of equivalent energy, putting the
167 pulses from alphas and high-energy betas in the same region. Correction for beta activity may be
168 made by certain mathematical, graphical or electronic techniques (see discussion of pulse shape
169 discrimination in Section 15.5.4). It is preferable to separate the alpha emitter from the bulk of
170 the beta activity by chemistry.
171 15.2.2.5 Semiconductor Detectors
172 Semiconductor detectors used for alpha counting are essentially solid-state ionization chambers.
173 The ionization of the gas in an ionization chamber by alpha particles produces electron-ion pairs,
174 while in a semiconductor detector electron-hole pairs are produced. The liberated charge is
175 collected by an electric field and amplified by a charge-sensitive amplifier. In general, ion-
176 implanted-silicon or silicon surface barrier detectors are used for alpha counting. These detectors
177 are n-type base material upon which gold is evaporated to make a contact. The semiconductor
178 material must have a high resistivity since the background is a function of the leakage current.
179 This leakage current is present in an electric field since the starting material is a semiconductor,
180 not an insulator. The leakage current of silicon diodes doubles for every 5.5 to 7.5 °C change
181 in ambient temperature. Since the preamp HV bias resistor is a noise contributor, it is necessarily
182 of high value, typically 100 megohm. With a surface barrier detector having leakage current of
183 0.5 jiA, the change in bias voltage at the detector for a 2 °C change in ambient temperature can
184 be as much as 13V. This is enough bias change to affect overall gain of the detector-preamplifier
185 by a substantial amount. The reversed bias that is applied reduces the leakage current and a
186 depletion layer of free-charge carriers is created. This layer is very thin and the leakage current is
187 extremely low; therefore, the interactions of photons with the detector will have negligible effect.
188 Since the detector shows a linear response with energy, any interactions of beta particles with the
189 detector can be eliminated by electronic discrimination. The semiconductor is of special interest
190 in alpha counting where spectrometric measurements may be made since the average energy
191 required to produce an electron-hole pair in silicon is 3.5±0.1 eV compared to the 25 to 30 eV
192 needed to produce an ion pair in a gridded ionization chamber. Consequently, silicon detectors
193 provide much improved resolution and also normally have lower background count rates.
DRAFT MARLAP JUNE 2001
DO NOT CITE OR QUOTE 15-6 NOT FOR DISTRIBUTION
-------
Nuclear Counting Instrumentation
194 The detector size is generally less than 25 mm in diameter since the resolution decreases and cost
195 increases with detector size. For best results, the sample should be electrodeposited to make a
196 lower mass source (Puphal and Olson, 1972). However, micro precipitation as fluorides has been
197 reported with only slight lose of resolution (Sill and Williams, 1981; Hindman, 1983). The
198 detector is operated in a vacuum chamber. Typical backgrounds range from 8* 10"5 to 2* 10"4 cps.
199 15.3 Beta Counting
200 15.3.1 Introduction
201 This section covers the general techniques used to measure the beta particle activity resulting
202 from radiochemical separations of specific nuclides or groups of nuclides. Beta radioactivity may
203 be measured by several types of instruments that provide a detector and a combined amplifier,
204 power supply, and sealer. The most widely used detectors are proportional or Geiger-Mueller
205 counters—however, scintillation systems offer certain advantages (see discussion in Section
206 15.3.3). An example of the measurement of fission product activity by beta counting is given in
207 ASTM C799, D1890, and D3648.
208 15.3.2 Proportional Counter
209 Among the gas ionization-type detectors, the proportional type counter is preferable because of
210 the shorter resolving time and greater stability of the instrument. For preparing solid sources for
211 beta activity measurement, the sample is reduced to the minimum weight of solid material having
212 measurable beta activity by dissolution, radiochemistry, precipitation, or ion exchange tech-
213 niques. For measuring solid sources resulting from individual radiochemical separation
214 procedures, the precipitate is appropriately mounted for counting.
215 Beta particles entering the sensitive region of the detector produce ionization that is converted
216 into an electrical pulse suitable for counting. The number of pulses per unit time is directly
217 related to the disintegration rate of the sample by an overall efficiency factor. This factor
218 combines the effects of sample-to-detector geometry, sample self-shielding, backscatter,
219 absorption in air and in the detector window (if any), and detector efficiency. Because most of
220 these individual components in the overall beta-particle detection efficiency factor vary with beta
221 energy, the situation can become complex when a mixture of beta emitters is present in the
222 sample. The overall detection efficiency factor may be empirically determined with prepared
223 standards of composition identical to those of the sample specimen, or an arbitrary efficiency
224 factor can be defined in terms of a single standard such as cesium-137 (137Cs) or other nuclide.
JUNE 2001 DRAFT MARLAP
NOT FOR DISTRIBUTION 15-7 DO NOT CITE OR QUOTE
-------
Nuclear Counting Instrumentation
225 Gross counts can provide only a very limited amount of information and therefore should be used
226 only for screening purposes or to indicate trends.
227 15.3.3 Liquid Scintillation
228 Liquid scintillation counting (LSC) avoids many sources of error associated with counting solid
229 beta sources, such as self-absorption, back scattering, loss of activity during evaporation due to
230 volatilization or spattering, and variable detection efficiency over a wide beta-energy range. In
231 addition to the greatly improved accuracy offered by liquid scintillation counting, sample
232 preparation time and counting times are significantly shorter. Sample preparation involves only
233 adding a sample aliquant to the scintillator or gel phosphor. Because every radioactive atom is
234 essentially surrounded by detector molecules, the probability of detection is quite high even for
235 low-energy beta particles. Radionuclides having maximum beta energies of 200 keV or more are
236 detected with essentially 100 percent efficiency. Liquid scintillation can, at times, be disadvan-
237 tageous due to chemiluminescence, phosphorescence, quenching, or the typically higher
238 backgrounds.
239 The observed count rate for a liquid scintillation sample is directly related to the beta (plus
240 conversion electron) and positron emission rate in most cases. The important exceptions are: beta
241 emitters whose maximum energy is below 200 keV, and counting systems wherein quenching
242 decreases the expected photon yield, thereby decreasing the overall detection efficiency
243 significantly below 100 percent. Low-energy beta emitters, such as tritium (3H) or carbon-14
244 (14C), can be measured accurately only when the appropriate detection efficiency factor has been
245 determined with a known amount of the same radionuclide counted under identical conditions.
246 Quenching losses are greatest at low beta energies. Quenching may be evaluated by comparison
247 to known quench standards of the same radionuclide, using the channel ratio technique, or with
248 other techniques as described in the manufacturer's instructions.
249 For measurements in which data are expressed relative to a defined standard, the individual
250 correction factors cancel whenever sample composition, sample weight, and counting
251 configuration and geometry remain constant during the standardization and tests.
252 Liquid scintillation counting systems use an organic phosphor as the primary detector. This
253 organic phosphor is combined with the sample in an appropriate solvent that achieves a uniform
254 dispersion. A second organic phosphor often is included in the liquid scintillation cocktail as a
255 wavelength shifter. The wave length shifter efficiently absorbs the photons of the primary
256 phosphor and re-emits them at a longer wavelength more compatible with the multiplier
257 phototube. Liquid scintillation counting systems use either a single multiplier phototube or two
DRAFT MARLAP JUNE 2001
DO NOT CITE OR QUOTE 15-8 NOT FOR DISTRIBUTION
-------
Nuclear Counting Instrumentation
258 multiplier phototubes in coincidence. The coincidence counting arrangement is less likely to
259 accept a spurious noise pulse that occurs in a single phototube, and thus provides lower
260 background. The requirement that both multiplier phototubes respond to each has a slight effect
261 on the overall detection efficiency of betas with E-max >200 keV; however, system response to
262 beta E-max <200 keV will be significant. The need to minimize detectable radioactivity in the
263 detector and its surroundings is likewise important in liquid scintillation counting. To achieve
264 this, scintillation-grade organic phosphors and solvents are prepared from low 14C materials such
265 as petroleum. The counting vials are of low potassium glass or plastic to minimize counts due to
266 potassium-40 (40K). Liquid scintillation provides a fixed geometry from a given size counting
267 vial and liquid volume. The calibration of liquid scintillation counting detectors is given in
268 ASTM El 81. The use of an organic phosphor for liquid scintillation counting creates a mixed
269 waste. Chapter 20 of this manual addresses the proper disposal of these materials.
270 Another approach to LSC without the use of organic phosphors is Cerenkov counting. When
271 charged particles pass through a dielectric medium, such as water, and there is an exchange of
272 energy to the molecules of that medium, Cerenkov radiation is produced. This happens if the
273 charged particles are moving faster than the speed of light and the exchange of energy produces
274 electronic polarization, then when the polarized molecules return to a normal state the excess
275 energy is released as electromagnetic radiation (Kessler, 1986). Wave shifters are usually
276 employed to convert the ultraviolet Cerenkov radiation to the visible range. Although Cerenkov
277 counting efficiencies are about 20 to 50 percent (Scarpitta and Fisenne, 1996) lower than when
278 organic phosphors are used, mixed waste disposal is eliminated.
279 15.3.4 Solid Organic Scintillators
280 Organic scintillators, such as p-terphenyl plus a wave shifter in a plastic monomer, are
281 polymerized to form sheet material of any desired thickness. The plastic phosphor counting
282 system (Campion et al., 1960) has its widest use as a beta particle detector for separated, solid
283 samples rather than for beta spectrometry applications.
284 The plastic beta scintillator phosphor is mounted directly on the sample and is discarded after
285 counting. The phosphor-sample sandwich is placed in direct contact with the multiplier
286 phototube yielding essentially a 2-n configuration. Since the output pulse of the detector system
287 is energy dependent, the counting efficiency for a given phosphor thickness of 0.25 mm yields
288 the highest counting efficiency with the lowest background.
JUNE 2001 DRAFT MARLAP
NOT FOR DISTRIBUTION 15-9 DO NOT CITE OR QUOTE
-------
Nuclear Counting Instrumentation
289 Solid samples (precipitates from radiochemical separations) containing 3 to 5 mg/cm2 of stable
290 carrier are measured in such a system. For yttrium-90 (90Y) a solid sample of this type would
291 have a counting efficiency of 45 to 50 percent.
292 A plastic scintillator/phosphor system with a 25 mm multiplier phototube shielded with 12.7 mm
293 of lead has background in the order of 4x 10"2 cps. For very low backgrounds, about 4x 10"3 cps,
294 the multiplier phototube and sample assembly are fitted into a well-type hollow anode Geiger
295 tube operated in anti-coincidence. The entire assembly is then placed in a heavy shield.
296 The system has many advantages but reduction of background is probably most important. The
297 reduction occurs since the scintillator does not see the surrounding mechanical components of
298 the counter. The additional advantage of keeping the counter itself free from contamination by
299 enclosing the phosphor-sample sandwich is also important.
300 A note of caution is advisable at this point. Any beta particle detection system, whether internal
301 gas counters or scintillation counters, will detect alpha particles. It is not possible to
302 electronically discriminate against all the alpha pulses.
303 If a sample is suspected of containing alpha activity, a separate alpha measurement should be
304 made to determine the alpha contribution to the beta measurement.
305 15.3.5 Beta Particle Counter
306 The end-window Geiger-Mueller tube and the internal proportional gas-flow chambers are the
307 two most prevalent types of detectors. Other types of detectors include scintillators and solid-
308 state detectors. The material used in the construction of the detector and its surroundings should
309 contain a minimal level of detectable radioactivity. If the detector is of the window-type, the
310 window thickness may be used in calculating beta-ray attenuation; however, direct calibration of
311 the entire counting system with standards is recommended. The manufacturer should provide all
312 settings and data required for reliable and accurate operation of the instrument. Detectors
313 requiring external positioning of the test sample should include a support of low-density material
314 (aluminum or plastic), which ensures a reproducible geometry between the sample and the
315 detector. Because different sample to detector geometries are convenient for differing sample
316 activity levels, the sample support may provide several fixed positions ranging from 5 to 100 mm
317 from the detector.
318 The detection capability for both Geiger-Mueller and proportional counters is a function of the
319 background counting rate. Massive shielding or anti-coincidence detectors and circuitry, or both,
DRAFT MARLAP JUNE 2001
DO NOT CITE OR QUOTE 15-10 NOT FOR DISTRIBUTION
-------
Nuclear Counting Instrumentation
320 are generally used to reduce the background counting rate to increase the lower limit of detection
321 (Friedlander et al., 1964). ASTM E181 covers the procedure for the calibration of beta particle
322 counting detectors. An application of beta particle counting is given in ASTM El 005.
323 15.3.6 Associated Electronic Equipment
324 The high voltage power supply amplifier, sealer, and mechanical register normally are contained
325 in a single chassis. The power supply and amplifier sections are matched with the type of detector
326 to produce satisfactory operating characteristics and to provide sufficient range in adjustments to
327 maintain stable conditions. The sealer should have a capacity for storing and visually displaying
328 at least 9xl05 counts. The instrument should have an adjustable input sensitivity matched to that
329 of the detector, and variable high voltage power supply—an adjustable power supply and meter
330 are unnecessary for liquid scintillation systems. Counting chambers of Geiger-Mueller and
331 proportional counters contain a suitable counting gas and an electrode. Counting rates that
332 exceed 200 cps should be corrected for dead time loss when using a Geiger-Mueller tube. As the
333 applied voltage to the electrode is increased, the counting chamber exhibits responses that are
334 characteristic of a particular voltage region. At low voltages of the order of 100 V, there is no
335 multiplication of the ionization caused by a charged particle. At voltages approaching 1,000 V,
336 there is appreciable amplification of any ionization within the counting chamber; however, the
337 size of the output pulse is proportional to the amount of initial ionization. When operated in this
338 voltage region, the device is known as a proportional counter. Usually, there is a region at least
339 100 V wide, known as a plateau, wherein the count rate of a standard is relatively unaffected. The
340 operating voltage for proportional counters is selected to approximate the middle of this plateau
341 in order to maintain stable responses during small voltage shifts. The plateau region is
342 determined by counting a given source at voltage settings that differ by 25 or 50 V. The number
343 of counts at each setting is recorded, and the resultant counts versus voltage are plotted. Voltage
344 plateau curves are to be re-measured periodically to ensure continued instrument stability, or
345 whenever an instrument malfunction is indicated. If the voltage is increased beyond the
346 proportional region into the 1,500 to 2,000 V region, the pulse size increases and the dependence
347 on the initial ionization intensity disappears. This is the beginning of the Geiger counting region,
348 where a single ion pair produces the same large pulse as an intense initial ionization.
349 In order to eliminate alpha particle interferences a thin absorber between the sample source and
350 the detector can be used. The absorber diameter should exceed that of the detector window. The
351 absorber should be placed against the window to minimize beta particle scatter. Any absorber
352 that stops alpha particles will also attenuate low energy beta particles somewhat. For example, an
353 aluminum absorber of 7 mg/cm2 will absorb 48 percent of beta particles of 350 keV maximum
354 energy. Chemical separation of the alpha and beta particle emitters produces a higher degree of
JUNE 2001 DRAFT MARLAP
NOT FOR DISTRIBUTION 15-11 DO NOT CITE OR QUOTE
-------
Nuclear Counting Instrumentation
355 accuracy for internal detector measurements. Published information on beta particle absorption
356 (Friedlander et al., 1964) should be used as a guide for use of an absorber. In liquid scintillation
357 spectra, the alpha component appears as a peak on the beta continuum and thus provides a basis
358 for resolving the two (Bogen and Welford, 1971).
359 15.4 Gamma Counting
360 15.4.1 Introduction
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
This section covers the non-destructive measurement of gamma-ray radioactivity. Since gamma
radiation is a penetrating form of radiation, it can be used for non destructive measurements of
samples of any form and geometry as long as standards of the same form are available and are
counted in the same geometry to calibrate the detector. Because of this penetrating nature,
attenuation, because of variations in sample density or sample thickness, although usually not
significant, can be mathematically corrected.
When a standard cannot be obtained in
the matrix and density of samples being
counted, a correction for the different
absorption in the matrices should be
made (Modupe et al., 1993). Photons
interact with matter in one of three
ways: photoelectric, where all energy is
transferred; Compton scattering, where
only part of the energy is transferred;
and pair production, where the energy
creates a positron-electron pair. When
the positron annihilates the electron, two
511 keV photons are emitted. Figure
15.1 shows the relative probability of
each of the three predominant photon
interactions with germanium.
o
if: 1E+04
d>
o
O »6*03
c
"CD
^ IE+01
CD
§ 1E+00
S 1M1
1E-02
IIPOc
"v
10 100 1000
Energy (keV)
10000
FIGURE 15.1 Gamma-ray Interactions with Germanium
Since different nuclides emit distinct and constant spectra of gamma radiation, the use of an
energy discriminating system provides identification and measurement of all the components
present in a mixture of radionuclides. General information on gamma-ray detectors and gamma
counting is covered in the literature (Friedlander et al., 1964, and ICRU, 1994). Recent applica-
tions of gamma counting are given in several ASTM Test Methods (ASTM C758, C759, D3649).
DRAFT MARLAP
DO NOT CITE OR QUOTE
15-12
JUNE 2001
NOT FOR DISTRIBUTION
-------
Nuclear Counting Instrumentation
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
Gamma counting is generally carried out using solid detectors since a gas-filled detector will not
provide adequate stopping power for energetic gammas. In solids such as Nal(Tl) or Csl, the
gammas interact by excitation of atoms and energy is transferred to orbital electrons and then
released as light photons when the orbits are refilled. These scintillations are easily detected and
amplified into useable electrical pulses by a multiplier phototube. The Nal(Tl) detector is the
recommended detector for gross gamma counting because of its high efficiency and room
temperature operation.
In semiconductor detectors such as Si(Li) and high-purity germanium semiconductors (HPGe),
the gamma photons produce electron-hole pairs and the electrons are collected by an applied
electrical field. A charge-sensitive preamplifier is used to detect the charge transferred and
produce a useable electrical pulse. The semiconductor detectors are widely used in gamma
spectrometry.
100000
10000
1000 -
100 -
10 -.
Compton Continuum
10
ENERGY (keV)
Hundreds
The output pulses from the multiplier
phototube or preamplifier are directly
proportional to the amount of energy
deposited, which could either be total and
included in the photopeak, or fractional and
included in the continuum or escape peaks,
in the detector by the incident photon. The
pulses may be counted using a sealer or
analyzed by pulse height to produce a
gamma-ray spectrum.
Gamma photons interact with the detector
by three distinct processes. The photo-
electric effect results in complete absorption
of the photon energy and produces the full
energy or photopeak shown. The Compton
effect results in a partial absorption of the
photo energy and a scattered photon of
lower energy results. The scattered photon
carries energy away and the Compton continuum results (Figure 15.2). The third interaction is
pair production, which occurs at energies above 1,022 keV and results in the conversion of the
photon to mass as an electron-positron pair. The electron and positron give up their kinetic
energy to the detector and the resulting electron joins the electron population of the detector; the
positron, however, is annihilated in combining with an electron and produces two gamma
20
FIGURE 15.2 Gamma-ray Spectra of 60Co
JUNE 2001
NOT FOR DISTRIBUTION
15-13
DRAFT MARLAP
DO NOT CITE OR QUOTE
-------
Nuclear Counting Instrumentation
423 photons of 511 keV each. One or both of the 511 keV photons may escape from the detector
424 without interacting and the single escape and double escape peaks result.
425 The Comptons, from a higher energy photon, always present an interference problem in the
426 counting of gamma photons and appropriate corrections should be made for this effect. Pair
427 production can also be considered as an interference since the escape peaks may have an energy
428 equal to the lower energy gamma of interest. The Compton and pair production effects can be
429 very significant interferences and should be corrected.
430 The change of the absorption coefficient with gamma energy results in a wide variation of
431 detection efficiency. The detection efficiency falls rapidly as gamma energy increases for a fixed
432 size of detector. Two other important effects are seen as a result of the variation of the absorption
433 coefficient; firstly, low energy photons may be absorbed in massive samples as sample thickness
434 increases, such as large bottles of water, and erroneous results may be obtained. A similar
435 absorption effect is seen in HPGe systems where the can around the detector acts as an absorber
436 for very low-energy gammas and the efficiency passes through a maximum usually around 100
437 keV. The second result is that for low energy gammas a thin detector may be as efficient as a
438 much thicker one since the low-energy gammas are easily stopped in the thin detector.
439 Additionally, thin detectors will have better low energy detection limits because of reduced
440 background interactions.
441 Because of this variation in efficiency and the possible interferences from other activities, gross
442 gamma counting is only reliable when used to compare standards and samples of the same
443 nuclide. The use of gross gamma monitoring systems should be avoided when possible and, in all
444 cases, proper allowance should be made for the lack of accuracy.
445 At high count rates, random sum peaking may occur. Two absorptions may occur within the
446 resolving time of the detector and electronics and are summed and seen as one pulse. For a
447 detector of resolving time, t, and a count rate of A counts per unit time, the time window
448 available for summing is 2At (since the count summed could occur as early as t before or as late
449 as t after the other count) and the probability of another count at any time is simply A. Therefore,
450 the sum count rate will be 2A2t in unit time. Random summing is strongly dependent on the
451 count rate A and, if summing occurs, it can be reduced by increasing the sample to detector
452 distance. Modern electronics, both conventional analog and digital (preamplifiers, amplifiers, and
453 analog-to-digital converters) are capable of processing 100,000 cps without any significant lose
454 of resolution. This is because of the very short time constants (resolving time) these systems are
455 capable of producing. Over all detector performance can be affected by count rate because
456 reduced time constants are required which will cause some loss of resolution. When a photon
DRAFT MARLAP JUNE 2001
DO NOT CITE OR QUOTE 15-14 NOT FOR DISTRIBUTION
-------
Nuclear Counting Instrumentation
457 interaction takes place (an event is detected), charge carriers in the form of holes and electrons
458 are produced. The electrical field produced by the detector's high voltage bias supply causes
459 these carriers to be swept toward the P and N electrodes of the detector. The time it takes the
460 carriers to travel to the electrodes is called the "charge collection time." At very high count rates
461 the detector continues collecting events but the data is not valid. If a second (or third) event takes
462 place while the first set of charge carriers are still in transit, the energy from the two events get
463 added together. Therefore, if a 2,000 keV event arrives while a 1,000 keV event is in transit, the
464 detector would "see" a single 3,000 keV event, producing a random sum peak on pulse pileup.
465 When the detector starts reporting more sum peaks than valid events, you have exceeded its
466 count rate capability. Random pulse summing or pileup can also cause peak shape and risetime
467 problems. But the real upper limit to a detector throughput is pulse summing. This problem can
468 be reduced or eliminated by either reducing the number of events the detector "sees" by moving
469 the sample further away, collimate the detector, or use a smaller, less efficient detector; the
470 smaller the detector the shorter the charge collection time, which means a higher count rate limit.
471 Peak shifts may also occur with high count rates and short time constants. Another factor that
472 will affect high count rate performance is improper setting of the amplifier pole zero. Improper
473 setting of the pole zero with either under or over shooting of input pulse will effect peak
474 resolution.
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
Well counters that have very high efficiencies are prone to summing since, for a given source
strength, the count rate is higher than for a
detector of lower efficiency. For moderate
and high-source strengths, the trade-off is a
poor one and the well counter is best suited
for low-level work where its high efficiency
is an important advantage.
Cascade summing may occur when nuclides
that decay by a gamma cascade are counted.
Cobalt-60 (60Co) is an example; 1,173.2 keV
and 1,332.5 keV from the same decay may
enter the detector and be absorbed, giving a
2,505.7 keV sum peak. Another example of
Cascade summing occurs when counting
sodium-22 (22Na) close to the detector (see
Figure 15.3). Cascade summing may be
reduced and eventually eliminated by
increasing the source-to-detector distance.
10000
1000 -
100 -
10 -
10 15
ENERGY (keV)
Hundreds
20
JUNE 2001
NOT FOR DISTRIBUTION
15-15
FIGURE 15.3 Energy Spectrum of 22Na
DRAFT MARLAP
DO NOT CITE OR QUOTE
-------
Nuclear Counting Instrumentation
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
The resolution of a gamma detector is the effective limit to its utility even when complex data
reduction methods are used. A typical 76x76 mm Nal(TI) detector will give full-width half-
maximum (FWHM) of approximately 60 keV at 661.6 keV gamma energy and approximately 90
keV at 1,332.5 keV gamma energy.
15.4.2 Energy Efficiency Relationship
100
HPGeWeU
10
n-iypeHPGe
p-T^peHPGe
0.1
35% Re
70% Re
122cc
320cc
10
100
10000
Because of the rapid falloff in gamma
absorption as gamma energy rises, the
detection efficiency shows a similar effect.
Figure 15.4 shows a typical efficiency vs.
energy plot of a 70 percent FIPGe p-type, a
35 percent HPGe n-type, and HPGe well
detectors of 122 cm3 with a vespel well and
320 cm3 with a Mg well. The portion of the
curve for n-type and well detectors at low
energies shows that as the absorption
coefficient increases geometry becomes the
limiting factor. The maximum efficiency for
both co-axial detectors is well below 50
percent due to the presence of a beta
absorber, the containment of the detector
and the geometry effect. The p-type detector
shows significant low energy efficiency
drop off because of the absorption of
gamma rays in the detector's inactive Ge dead layer. The well detector shows excellent efficiency
below 100 keV because of the geometry effect and absence of an attenuating germanium dead
layer. The 76x76 mm Nal(TI) detector is the most widely used size. A large amount of data are
available in the open literature on both the use and results obtained with detectors of this size.
Heath (1964) has written a comprehensive review and supplied many gamma-ray spectra in both
graphical and digital form.
Other sizes of detectors may be used. However, the following should be noted: smaller detectors,
such as 38x38 mm, will give efficiencies that are low and fall off more rapidly as gamma energy
increases. Small or thin detectors are useful for the measurement of low-energy gammas since
they are less responsive to high-energy gammas and the interference from Compton effects is
reduced. This will result in a lower background.
1000
ENERGY (keV)
FIGURE 15.4 Efficiency vs. Gamma-ray Energy
DRAFT MARLAP
DO NOT CITE OR QUOTE
15-16
JUNE 2001
NOT FOR DISTRIBUTION
-------
Nuclear Counting Instrumentation
527 Larger detectors will give higher efficiencies and less falloff as gamma energy increases. Larger
528 detectors are useful for situations where the highest attainable efficiency is desired and for the
529 assembly of complete absorption detectors. The increase in efficiency is accompanied by an
530 increased background count rate and an increase in the probability of summing in the detector.
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
Well detectors will give very high efficiencies, up to about 80 percent for low and moderate
energy gammas. The well detector is useful for low levels of activity and the background of a
well detector is essentially the same as that of a plain cylindrical detector of the same overall
dimensions. Summing becomes a definite problem at high activities since both random and
cascade summing result from the high efficiencies and the high geometry of the well detector.
Detector efficiency will also vary as a function of sample geometry. Table 15.1 gives counting
efficiencies obtained with various sample geometries for a 55 percent HPGe detector.
TABLE 15.1 Typical Percent Gamma-ray Efficiencies for a 55 Percent High-Purity
Germanium Detector* with Various Counting Geometries
ENERGY (keV)
60
88
122
166
279
392
514
662
835
898
1115
1173
1333
1836
FILTER
PAPER
15.6
15.2
15.1
12.0
9.3
7.2
5.4
4.7
3.9
3.1
3.0
2.6
2.3
1.7
50cm3
PLANCHET
14.6
14.2
12.6
9.6
7.4
5.5
4.2
3.6
2.9
2.4
2.3
2.0
1.8
1.3
90cm3
ALCAN
11.6
11.3
10.2
8.0
6.0
4.5
3.5
3.0
2.4
2.1
1.9
1.7
1.5
1.2
600 cm3
MARINELLI BEAKER
5.0
7.4
8.4
7.9
6.1
4.8
3.8
3.1
2.7
2.2
2.1
1.8
1.6
1.3
Although the counting efficiencies listed above were obtained with a 55 percent (relative to a 3x3 inch Nal
detector) HPGe detector, the calculation of counting efficiencies by extrapolation for detectors with different
relative efficiencies is not possible. This is because detectors with the same relative efficiency may be of
significantly different dimensions thus producing a detector/sample solid angle very different than what was used to
prepare this table.
JUNE 2001
NOT FOR DISTRIBUTION
15-17
DRAFT MARLAP
DO NOT CITE OR QUOTE
-------
Nuclear Counting Instrumentation
560 15.4.3 Sodium Iodide Detector Assembly
561 A cylindrical 76x76 mm Nal detector is activated with about 0.1 percent thallium iodide, with or
562 without an inner sample well, optically coupled to a multiplier phototube, and hermetically
563 sealed in a light-tight container. The Nal(Tl) crystal should contain less than 5 ppm of potassium
564 and be free of other radioactive materials. In order to establish freedom from radioactive
565 materials, the manufacturer should supply a gamma spectrum of the background of the detector
566 between 0.08 and 3,000 keV. The resolution of the detector for the 662 keV gamma from 137Cs
567 decay should be less than 50 keV FWHM or less than 7 percent when measured with the source
568 in contact with the end cap.
569 The following components are required for a complete Nal(Tl) gamma-ray spectrometry system:
570 High-Voltage Power Supply
571 Preamplifier
572 Analyzer with Sealer and Timer
500 to 2,000 V dc regulated to 0.1 percent with a ripple of
not more than 0.01 percent
Linear amplifier system to amplify the output from the
multiplier phototube to a maximum output of 10 V.
A single-channel discrimination system will accept all or
any part of the output from the amplifier and pass it to the
sealer. Any pulses lying outside the preset limits are
rejected. The lower limit is usually referred to as the
thresholdand the difference between the two limits is the
window.
Sample mounts and containers may consist of any
reproducible geometry container that is commercially
available. Other considerations are cost, ease of use,
disposal, and effective containment of radioactivity for the
protection of the workplace and personnel from
contamination.
573
Beta Absorber
A beta absorber of 3 to 6 mm of aluminum, beryllium, or
poly(methyl methacrylate) should completely cover the
upper face of the detector to prevent betas from reaching the
detector.
DRAFT MARLAP
DO NOT CITE OR QUOTE
15-18
JUNE 2001
NOT FOR DISTRIBUTION
-------
Nuclear Counting Instrumentation
574 15.4.4 High Resolution Germanium Detectors
575 High resolution germanium detectors are produced from very high purity material, the required
576 level of impurities in the detector crystal is usually less than 109 atom/cm3. Any type of
577 germanium—either planar, co-axial or well-configuration—cannot be operated at room
578 temperature because of the large thermally induced leakage current that results. These detectors
579 should be cooled in order to reduce the thermal generation of charge carriers (thus reverse
580 leakage current) to an acceptable level. Otherwise, leakage current induced noise reduces the
581 energy resolution of the detector. The detector is mounted in a vacuum chamber which is
582 attached to or inserted into an liquid nitrogen (LN2) dewar or an electrically powered cooler. The
583 sensitive detector surfaces are thus protected from moisture and condensation contaminants.
584 The boiling point of liquid nitrogen (77 °K) is usually taken advantage of to reduce the operating
585 temperature of the detector. Since germanium detectors can be operated at temperatures as high
586 as 130 °K, mechanical closed-cycle refrigerators can also be used. These systems can cool a
587 detector to as low as 50 °K. Therefore, with proper thermal control the detector can be cooled to
588 its optimum operating temperature. The required preamplifier is normally included as part of the
589 cryostat. In this configuration the preamplifier can also be cooled to reduce electronic noise.
590 HPGe detectors are preferred for the analysis of complex gamma-ray spectra involving many
591 nuclides and peaks. However, for samples with only a few nuclides, the complexity of an HPGe
592 system may not be cost effective. The calibration of germanium detectors is given in ASTM
593 E181.
594 15.4.5 Low Background High Resolution Germanium Detectors
595 Environmental samples requiring the lowest possible minimum detection analyses (MDAs)
596 should be counted with large high efficiency germanium detectors in low background cryostats.
597 Most of the background from naturally occurring radionuclides such as 40K from building
598 materials, radon decay products, and cosmic rays can be reduced by proper shielding. However,
599 naturally occurring 235U, 238U, 232Th, and anthropogenic 137Cs and 60Co may be present in cryostat
600 materials. With careful selection and substitution of materials, low background gamma-ray
601 systems can be fabricated. Germanium crystal mountings and detector end caps have been
602 fabricated with magnesium to eliminate aluminum contaminated with radioactive thorium
603 isotopes. Figures 15.5 and 15.6 show shielded background spectra obtained with 56 percent
604 germanium detectors in standard and extra low background cryostats.
JUNE 2001 DRAFT MARLAP
NOT FOR DISTRIBUTION 15-19 DO NOT CITE OR QUOTE
-------
Nuclear Counting Instrumentation
80
64 -
48 -
16
0246
10 12 14 16 18 20 22 24 26
ENERGY (keV)
Hundreds
64 -
48 -
32 -
16 -
0246
10 12 14 16 18 20 22 24 26
ENERGY (keV)
Hundreds
FIGURE 15.5 Standard Cryostat HPGe Background
Spectrum
FIGURE 15.6 Low Background Cryostat HPGe
Background Spectrum
605 15.4.6 High Resolution Detectors for Low Energy Spectrometry
606 High resolution low gamma-ray energy detectors are available in various configurations. The
607 commonly used ones are either high purity germanium or silicon. The various detector types
608 include: planar (Ge or Si), low-energy germanium (LEGe), reverse-electrode germanium (REGe)
609 and extended-range germanium (XtGe). These detectors are equipped with beryllium entrance
610 windows to reduce attenuation. These detectors are especially useful for measuring nuclides that
611 emit gamma or X-rays from a few keV to about 150 keV.
612 15.4.7 CsI(Tl) Detectors
613 CsI(Tl) crystals have the highest light output of all known scintillators. However, because light
614 output is not well matched to the sensitivity of the photocathode of a multiplier phototube, the
615 yield for gamma rays is only 45 percent of the efficiency of Nal(Tl). With the proper electronics,
616 CsI(Tl) detectors can be used for a-particle energy discrimination.
617 15.4.8 CdZnTe Detectors
618 These gamma-ray detectors, in addition to only being produced in very small volumes, do not
619 have energy resolutions as good as HPGe but are better than Nal(Tl). Their greatest advantage is
DRAFT MARLAP
DO NOT CITE OR QUOTE
15-20
JUNE 2001
NOT FOR DISTRIBUTION
-------
Nuclear Counting Instrumentation
620 their ability to operate at room temperature. Because of their small size and resulting low
621 gamma-ray detection efficiency, they are useful for the analysis of very high level sources.
622 15.4.9 BGO Detectors
623 Because bismuth germanate (Bi4Ge3O12)is a high Z, high density (7.13 gem3), scintillation
624 material, it is a very efficient gamma-ray absorber. Although BGO crystals a have very good
625 peak-to-Compton ratio, their effective efficiency is only 10 to 15 percent as good as a Nal(Tl)
626 crystal. However, BGO is a relatively hard, rugged, non-hygroscopic crystal which does not
627 cleave or absorb any significant amount of the scintillation light. The crystal housing does not
628 require hermetic air-tight sealing. These crystals are useful in applications where high
629 photofraction is required.
630 15.5 Spectrometry Systems
631 This section will present a number of different type of detector systems commonly use for
632 gamma-ray spectrometry.
633 15.5.1 Alpha/Gamma Coincidence Systems
634 Alpha/Gamma Coincidence Systems have been used for the direct measurement of 224Ra and
635 226Ra. The counting technique is based upon the coincidence measurement of the characteristic
636 particle-photon emissions of these isotopes. Silver activated zinc sulfide for alpha detection is
637 combined with a Nal well for gamma-ray detection (McCurdy, 1981).
638 15.5.2 Beta/Gamma Coincidence Systems
639 Many radionuclides remain in an excited state after what may be considered beta decay. This
640 results in the emission of a gamma ray as the decay process goes to the ground state. A
641 beta/gamma coincidence system will have significantly improved lower limit of detection over a
642 beta or a gamma counting system because of its very low background. Systems have been
643 designed with both 2-rc and 4-rc geometry (McCurdy et al., 1980).
644 15.5.3 Gamma/Gamma Coincidence Systems
645 These counting systems can provide extremely low backgrounds and are very useful for
646 analyzing those radionuclides that decay with cascading (coincident) gamma rays. The systems
647 usually consist of two large Nal(Tl) detectors with a surrounding active anti-coincidence shield
JUNE 2001 DRAFT MARLAP
NOT FOR DISTRIBUTION 15-21 DO NOT CITE OR QUOTE
-------
Nuclear Counting Instrumentation
648 of either Nal(Tl) or plastic phosphor. However, HPGe detectors have also been used in place of
649 the two large Nal(Tl) detectors. Only gamma-ray pulses that are detected in both of the primary
650 detectors at the same time (coincident) and not in the active shield are recorded. Even though
651 these systems can be large, because of the shielding requirements for two detectors and an active
652 annulus, and require complex electronics, the improvement in lower limit of detection for certain
653 radionuclides is worth the investment (Perkins, 1965; Sanderson, 1969).
654 15.5.4 Photon-Electron Rejecting Alpha Liquid Scintillation Systems
655 Another technique for the analysis of alpha emitting radionuclides combines liquid scintillation
656 counting with pulse shape discrimination to significantly reduce background counts from photo-
657 electrons produced by ambient background gamma rays and to eliminate interferences from beta
658 emitters in the sample/scintillation cocktail. Pulse shape discrimination electronically selects only
659 pulses produced by alpha particles because of their longer decay times in the scintillation
660 solution. Typical alpha peak resolutions are about 5 percent. Typical detectable activities for
661 alpha emitters such as 234U and 241Am are 0.0037 and 0.37 Bq (0.1 and 10 pCi).
662 15.6 Special Instruments
663 This section covers some radiation detection instruments and auxiliary equipment that may be
664 required for special application in the measurement of radioactivity.
665 15.6.1 4-7T Counter
666 The 4-Ti counter is a detector designed for the measurement of the absolute disintegration rate of
667 a radioactive source by counting the source under conditions that approach a geometry of 4-7t
668 steradian. Its most prevalent use is for the absolute measurement of beta emitters. For this
669 purpose, a gas-flow proportional counter is commonly used. 4-7t counting systems consist of two
670 hemispherical or cylindrical chambers whose walls form the cathode, and a looped wire anode in
671 each chamber. The source is mounted on a thin supporting film between the two halves, and the
672 counts recorded in each half are summed.
673 Gamma-ray and hard X-ray counters with geometries approaching 4-7t steradian can be
674 constructed from both Nal(Tl) or germanium crystals in either of two ways. A well crystal (that
675 is, a cylindrical crystal with a small axial hole covered with a second crystal) will provide nearly
676 4-Tt geometry for small sources, as will two solid crystals placed very close together with a small
677 source between them. The counts from both crystals are summed as in the gas-flow counter. The
678 deviation from 4-7t geometry can be calculated from the physical dimensions. For absolute
DRAFT MARLAP JUNE 2001
DO NOT CITE OR QUOTE 15 -22 NOT FOR DISTRIBUTION
-------
Nuclear Counting Instrumentation
679 gamma-ray counting, the efficiency of the crystal for the gamma energy being measured and the
680 absorption in the detector end cap should be taken into account. The liquid scintillation counter is
681 also essentially a 4-rc counter for alpha and beta particles, since nearly all the radiations are
682 emitted into and interact with the detecting medium.
683 15.6.2 Low-Geometry Counters
684 This type of instrument is particularly useful for the absolute counting of alpha particles. The
685 alpha emitter, in the form of a very thin solid source, is placed at a distance from the detector
686 such that only a small fraction (<1 percent) of the alpha particles are emitted in a direction to
687 enter the counter. This solid angle is obtained from the physical measurements of the instrument.
688 The space between the source and the detector is evacuated to eliminate the loss of alpha
689 particles by absorption in air. The detector can be any counter that is 100 percent efficient for all
690 alpha particles that enter the sensitive volume—a gas-flow proportional counter with a window
691 that is thin (approximately 1 mg/cm2) compared to the range of the alpha particles or the
692 semiconductor alpha detector with a 1 mg/cm2 covering. The advantages of this instrument for
693 absolute alpha counting are that the effect of absorption of alpha particles in the source itself is
694 kept to a minimum since only particles that travel the minimum distance in the source enter the
695 detector (particles that have longer paths in the source are emitted at the wrong angle), and back-
696 scattered alpha particles (those that are emitted into the source backing and are reflected back up
697 through the source) lose sufficient energy so that they cannot enter the detector. One such
698 instrument is described in Curtis et al. (1955).
699 15.6.3 Internal Gas Counters
700 The internal gas counter is so named because the radioactive material, in the gaseous state, is
701 placed inside a counting chamber and thus becomes part of the counting gas itself. It is useful for
702 high-efficiency counting of weak beta- and X-ray emitting radionuclides. The radiations do not
703 have to penetrate a counter window or solid source before entering the sensitive volume of a
704 detector. The counter may be an ionization chamber, or it may be operated in the Geiger or
705 proportional mode. Most present-day instruments are of the latter type, and they generally take
706 the form of a metal or metal-coated glass cylinder as a cathode with a thin anode wire running
707 coaxially through it and insulated from the cylinder ends. A wire through the wall makes
708 electrical contact to the cathode. The counter has a tube opening through which it may be
709 connected to a gas-handling system for filling. The purity of the gas is important for efficient and
710 reproducible counting, particularly in the proportional mode.
JUNE 2001 DRAFT MARLAP
NOT FOR DISTRIBUTION 15-23 DO NOT CITE OR QUOTE
-------
Nuclear Counting Instrumentation
111 In a modification of the internal gas counter, scintillation counting has been used. The inner walls
712 of the chamber are coated with a scintillation material and the radioactive gas is introduced. An
713 optical window is made a part of the chamber, and the counting is done by placing this window
714 on a multiplier phototube to detect the scintillations. This system is particularly useful for
715 counting radon gas with zinc sulfide as the scintillator. Additional details on internal gas
716 counting may be found in Watt and Ramsden (1964).
717 15.7 Spectrometers and Energy-Dependent Detectors
718 The availability of energy-dependent detectors (detectors whose output signal is proportional to
719 the energy of the radiation detected) that are easy to operate and maintain and have good
720 resolution makes it possible to measure not only the total activity of a radioactive sample but the
721 energy spectrum of the nuclear radiations emitted. Nuclear spectrometry is most useful for alpha
722 particles, electromagnetic radiation (gamma and X-rays), and conversion electrons, since these
723 radiations are emitted with discrete energies. Beta spectra have more limited use since beta
724 particles are emitted from a nucleus with a continuous energy distribution up to a characteristic
725 maximum (E- max), making a spectrum containing several different beta emitters difficult to
726 resolve into its components. The advantages of spectrometric over total activity measurements of
727 radioactive sources are increased selectivity, detection limit, and accuracy because nuclide
728 identification is more certain, interference from other radioactive nuclides in the sample is
729 diminished or eliminated, and counter backgrounds are reduced since only a small portion of the
730 total energy region is used for each radiation.
731 The detectors for alpha spectra are gridded ion-chambers and silicon semiconductor detectors.
732 Gridded ion-chambers are no longer available commercially and should be constructed by the
733 user. A variety of semiconductor detectors for alpha spectrometry are commercially available.
734 These detectors have essentially replaced ion-chambers, although the chambers have the
735 advantages of high efficiency (nearly 50 percent) for large-area sources.
736 Silicon alpha particle detectors have a depletion region which is formed by applying a high
737 voltage bias. The electric field produced collects the electron-hole pairs produced by incident
738 alpha particles. Either surface barrier or passivated ion-implanted silicon are commonly used for
739 spectrometry.
740 The principal detectors used for gamma-ray spectrometry are thallium-activated sodium iodide
741 scintillation crystals, Nal(Tl), and high purity germanium semiconductors, HPGe. HPGe
742 detectors are available in n-type and p-type germanium. P-type germanium detectors have dead
743 layers which produce entrance windows from 500 to 1,000 jim thick. On the other hand, n-type
DRAFT MARLAP JUNE 2001
DO NOT CITE OR QUOTE 15 -24 NOT FOR DISTRIBUTION
-------
Nuclear Counting Instrumentation
744 detectors have extremely thin entrance windows of about 0.3 jim. These n-type detectors when
745 housed in an end cap with a beryllium window are excellent for measuring both low energy and
746 high energy (3 to 10,000 keV) gamma rays. However, applications which require the best
747 possible energy resolution, peak shape, and efficiency for gamma-ray measurements above 80
748 keV, p-type HPGe is the detector material of choice.
749 For X-rays and very low-energy gamma rays, lithium-drifted silicon semiconductor Si(Li), planar
750 germanium, and gas-filled thin window (approximately 1 mg/cm2) proportional counters are
751 used.
752 The electronic version of Heath's (1964) Ge(Li) and Si(Li) Detector Gamma-ray Spectrum
753 Catalogue is available in two forms. The document is on the Web at http://id.inel.gov/gamma; it
754 is also available on a CD-ROM.
755 The portion of the crystal end cap through which gamma rays enter is normally thinner, or
756 constructed of a low-Z material, like beryllium or magnesium, than the rest of the package in
757 order to reduce low-energy attenuation. Sodium iodide crystals are available in a large range of
758 sizes and shapes, from thin crystals for X-ray analysis and small 25 by 25 mm cylinders to
759 hemispheres and cylinders over 300 mm in diameter. Information on the types of crystal
760 packages and mountings is available from the manufacturers.
761 A complete Nal(Tl) detector spectrometer requires a high-voltage power supply for the phototube
762 (usually operated at 600 to 1,000 V), a preamplifier, linear amplifier, pulse-height analyzer, and
763 output recorder. Because Nal(Tl) detectors cannot resolve gamma-ray energies that are only a
764 few keV apart, a least-squares computer program should be used to quantify a complex gamma-
765 ray spectrum.
766 Germanium and silicon detectors are junction-type semiconductor devices. With silicon
767 detectors, a sensitive region is produced by drifting lithium under the influence of an electric
768 field at an elevated temperature (100 to 400 °C) into the crystal. The crystal then functions as a
769 solid ion chamber when a high voltage is applied. Today, germanium detectors are made with
770 very high purity material that does not require lithium drifting. In order to obtain high resolution,
771 these detectors should be operated at low temperatures to reduce thermal noise. At room
772 temperature, sufficient free electrons will be present in the crystal to obscure the measurement of
773 gamma and X-rays. Consequently, the detectors are operated at liquid nitrogen temperatures by a
774 cryostat consisting of a metallic cold-finger immersed in a Dewar flask containing liquid nitrogen
775 or mechanically refrigerated.
JUNE 2001 DRAFT MARLAP
NOT FOR DISTRIBUTION 15-25 DO NOT CITE OR QUOTE
-------
Nuclear Counting Instrumentation
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
30
The electronic components required to obtain spectra are similar to those for Nal(Tl) detectors,
except that because smaller pulses should be measured, high-quality electronics should be used.
A complete HPGe system includes a high-voltage bias supply for the detector, a preamplifier,
amplifier (usually charge-sensitive), pulse height analyzer, and recording device. With the
exception of extremely complex spectra, most high resolution spectra can be quantified by
simple integration of full energy gamma-ray peaks.
The resolution of gamma-ray detectors
is usually specified in terms of its
FWHM. Detector resolution, expressed
in percent, improves with increasing
energy and for Nal(Tl) detectors and is
usually determined from the 662 keV
gamma ray emitted in the decay of 137Cs.
This is shown graphically in the
gamma-ray spectrum in Figure 15.7. For
HPGe detectors, 60Co is measured
25 cm above the detector end cap.
Quality sodium iodide crystals have
resolutions in the range of 6.5 to 7
percent for 137Cs. Detection efficiency
for the same geometry and window
thickness is a function of several
parameters and much published
information on efficiencies for various
25 -
20 -
15 -
10 -
5 -
ENERGY (keV)
Hundreds
FIGURE 15.7 Nal(Tl) Energy Spectrum of 137Cs
energies, detector sizes, source-to-detector distances, and other variables is available
(Crouthamel et al., 1970). The efficiency for gamma-ray detection is expressed as full energy
peak efficiency; the fraction of incident gamma rays that give a full-energy peak for a particular
source-detector configuration. For a 102 mm thick Nal(Tl) crystal, with the source on the surface
(zero distance), this fraction is approximately 0.24 for the 661.6 keV gamma ray of 137Cs and
approximately 0.14 for the 1,332.5 keV gamma ray of 60Co. The peak-to-valley or peak-to-
Compton ratio is the ratio of counts at the maximum height of the full-energy peak to the counts
at the minimum of the Compton continuum. A high ratio indicates narrow peaks, that is, good
resolution, for that particular energy.
The efficiency specification of a FIPGe detector is expressed by comparing its 60Co, 1,332.5 keV
efficiency at 25 cm with that of a 76x76 mm cylindrical Nal(Tl) detector at the same distance.
DRAFT MARLAP
DO NOT CITE OR QUOTE
15-26
JUNE 2001
NOT FOR DISTRIBUTION
-------
Nuclear Counting Instrumentation
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
Photopeaks are spread over a much
smaller energy range in germanium
than in sodium iodide, the background
under the peak is much less (Figure
15.8). This means that for small
sources of moderately energetic
gamma rays, germanium is more
sensitive than sodium iodide.
Typical specifications for a
germanium gamma-ray detector could
include but should not be limited to
the following:
DETECTOR: The gamma-ray
detector should consist of High-
Purity n-type germanium.
80
60 -
40 -
20 -
FWHM .
1.5 keV
4 6
ENERGY (keV)
Hundreds
FIGURE 15.8 HPGe Energy Spectrum of 137Cs
SIZE: The germanium crystal
should be at least 5.5 cm in diameter and at least 7.0 cm long.
EFFICIENCY: The relative counting efficiency compared to a 3"x3" Nal detector at 25 cm for
60Co (1,332 keV) should be equal to or better than 50 percent.
RESOLUTION: The resolution (FWHM) of the detector should be equal to or better than
2.2 keV at 1,333 keV (60Co). The resolution (FWHM) at 122 keV (57Co) of the detector
should be equal to or better than 1.0 keV. The detector resolution at FWTM should be equal
to or better than 2 times the FWHM.
834
835
PEAK-TO-COMPTON RATIO: The peak-to-Compton ratio for 1,333 keV (60Co) should be equal
to or better than 50:1.
836
837
838
839
BACKGROUND: Low radioactivity materials should be used so that any full energy gamma-ray
line (excluding 511 keV and 1,460 keV) present in a 1,000-minute background spectrum
(100-2,000 keV) obtained in a graded 10 cm lead shield should not exceed 0.20 counts per
minutes.
JUNE 2001
NOT FOR DISTRIBUTION
15-27
DRAFT MARLAP
DO NOT CITE OR QUOTE
-------
Nuclear Counting Instrumentation
840 CONTACTS: The internal detector contacts should be DC-coupled ion implants so that low
841 energy gamma-ray attenuation is avoided.
842 PREAMP: A low-noise, cooled field-effect transistor preamplifier should be used to provide
843 the detector output signals.
844 CRYOSTAT: The cryostat should be constructed of low radioactivity materials throughout and
845 should contain sufficient lead shielding in order to minimize radiation from the dewar or
846 lower portion of the cryostat.
847 END CAP: The end cap should consist of a 20 mil beryllium window with 0.5 mm aluminum
848 side walls and be no greater then 7.6 cm diameter (OD). This diameter should be maintained
849 for at least 8 cm from the end cap. Below this point the outside diameter of the end cap may
850 be increased. The top of the end cap should be between 95 and 102 cm above the outside base
851 of the dewar.
852 TEMPERATURE: The cryostat should contain a temperature sensing circuit to provide high
853 voltage shut down in order to prevent preamplifier damage in case of warm-up due to loss of
854 liquid nitrogen.
855 Spectra of beta particles and conversion electrons can be obtained with sodium iodide and n-type
856 HPGe detectors. A germanium detector with a volume of 120 cm3 has an efficiency approxi-
857 mately 20 percent that of a 76x76 mm Nal(Tl) crystal. Larger HPGe detectors are available with
858 relative efficiencies over 150 percent when compared with a 76x76 mm Nal(Tl) crystal.
859
860 Presently available germanium detectors have resolutions of 1.5 to 2.5 keV at 1,332.5 keV. The
861 method used to measure the energy resolution is described in ANSI/IEEE 325. This greater
862 resolution makes this detector the one of choice for gamma-ray spectrometry and cancels to some
863 extent the higher efficiency available from sodium iodide. Since the pulses from a single
864 semiconductor detectors sufficiently thick (a few centimeters) to absorb the particles completely.
865 One disadvantage of sodium iodide detectors is their relatively thick entrance windows. Other
866 semiconductor detectors have thin beryllium entrance windows and can be used for beta
867 spectrometry.
868 Good spectra of low-energy beta particles, conversion electrons, and X-rays can be obtained with
869 a gas-flow proportional Counter provided that a linear preamplifier is used. The resolution is
870 intermediate between Nal(Tl) and HPGe. Organic scintillators, such as anthracene and
871 polystyrene polymerized with scintillating compounds, are also useful for beta spectrometry.
DRAFT MARLAP JUNE 2001
DO NOT CITE OR QUOTE 15-28 NOT FOR DISTRIBUTION
-------
Nuclear Counting Instrumentation
872 They are packaged with a phototube in a
873 manner similar to sodium iodide
874 crystals. Liquid scintillation mixtures
875 also give beta spectra, and the output of
876 a commercial liquid scintillation counter
877 is usually fed into a multichannel pulse-
878 height analyzer to obtain a beta energy
879 spectrum (Blanchard et al., 1960). A
880 spectrum of 210Pb, 210Bi, and 210Po in
881 Figure 15.9 shows the resolution
882 obtainable by liquid scintillation
883 counting of aqueous samples in a
884 dioxane-based solution. The 210Bi curve
885 is from a beta particle, and the 210Po
886 peak is from an alpha particle. Organic
887 scintillators are preferable to sodium
888 iodide for beta spectrometry because
889 less back scattering occurs.
890 15.7.1 Anti-Coincidence Counters
o
o
210
210
Po
LOG ENERGY
FIGURE 15.9 Spectrum of 210Pb, 210Bi, and 210Po
891 Substantial background reduction can be achieved in beta and gamma counters by surrounding or
892 covering the sample detector with another detector also sensitive to beta or gamma radiation, and
893 connecting them electronically so that any pulse appearing in both detectors at the same time is
894 canceled and not recorded as a count. This is referred to as anti-coincidence shielding, and is
895 recommended for obtaining very low backgrounds. This type of counter was used for many years
896 in directional studies of cosmic rays, and was first applied to reducing the background of beta
897 counters by Libby in his study of natural 14C. The thick metal shielding (lead, iron, or mercury)
898 ordinarily used to reduce cosmic-ray and gamma-ray background should also be present, and is
899 placed outside the anti-coincidence shielding.
900 Anti-coincidence shielding of gamma-ray detectors operates in a similar way, and is particularly
901 useful in reducing the Compton continuum background of gamma rays (Nielson, 1972). Gamma
902 rays that undergo Compton scattering and produce a pulse in both the detector and the anti-
903 coincidence shield are canceled electronically. Ideally, only those gamma rays that are completely
904 absorbed in the sample detector produce a count that is recorded with the total energy of the
905 gamma ray (full-energy peak). There are second-order effects that prevent complete elimination
JUNE 2001
NOT FOR DISTRIBUTION
15-29
DRAFT MARLAP
DO NOT CITE OR QUOTE
-------
Nuclear Counting Instrumentation
906 of Compton scattering, but the improvement is substantial (Perkins, 1965, and Cooper et al.,
907 1968).
908 15.7.2 Coincidence Counters
909 In coincidence counting, two or more radiation detectors are used together to measure the same
910 sample, and only those nuclear events or counts that occur simultaneously in all detectors are
911 recorded. The coincidence counting technique finds considerable application in studying
912 radioactive decay schemes; but in the measurement of radioactivity, the principal uses are for the
913 standardization of radioactive sources and for counter background reduction.
914 Coincidence counting is a very powerful method for absolute disintegration rate measurement
915 (Friedlander et al., 1964; IAEA, 1959). Both alpha and beta emitters can be standardized if their
916 decay schemes are such that p-y, y-y, p-p, a-p, or a-X-ray coincidence occur in their decay.
917 Gamma-gamma coincidence counting with the source placed between two sodium iodide
918 crystals, is an excellent method of reducing the background from Compton scattered events. Its
919 use is limited, of course, to counting nuclides that emit two photons in cascade (which are
920 essentially simultaneous), either directly as in 60Co, by annihilation of positrons as in 65Zn, or by
921 immediate emission of a gamma ray following electron capture decay. Non-coincident pulses of
922 any energy in either one of the crystals will be canceled, including cosmic-ray photons in the
923 background and degraded or Compton scattered photons from higher energy gamma rays in the
924 sample. Thus, the method reduces interference from other gamma emitters in the sample. When
925 two multichannel analyzers are used to record the complete spectrum from each crystal, singly
926 and in coincidence, then the complete coincident gamma-ray spectrum can be obtained with one
927 measurement. The efficiency for coincidence counting is low since it is the product of the
928 individual efficiencies in each crystal, but the detection limit is generally improved because of
929 the large background reduction (Nielsen and Kornberg, 1965). This technique is often referred to
930 as two-parameter or multidimensional gamma-ray spectrometry.
931 Additional background improvement is obtained if the two crystals are surrounded by a large
932 annular sodium iodide or plastic scintillation crystal connected in anti-coincidence with the two
933 inner crystals. In this case a gamma ray that gives a pulse, but is not completely absorbed in one
934 of the two inner crystals, and also gives a pulse in the surrounding crystal, is canceled
935 electronically (Perkins, 1965, and Nielsen and Kornberg, 1965). This provides additional
936 reduction in the Compton scattering background. Germanium detectors may be used in place of
937 the inner sodium iodide crystals for improved resolution and sensitivities (Cooper et al., 1968).
938 An example of an assay for plutonium content using passive thermal-neutron coincidence
DRAFT MARLAP JUNE 2001
DO NOT CITE OR QUOTE 15-30 NOT FOR DISTRIBUTION
-------
Nuclear Counting Instrumentation
939 counting is given in ASTM (C1207). Another example of passive thermal-neutron coincidence
940 counting using a moveable californium source is given in ASTM (C1316).
941 15.8 Shielding
942 The purpose of shielding is to reduce the background count rate of a measurement system.
943 Shielding reduces background by absorbing some of the components of cosmic radiation and
944 some of the radiations emitted from material in the surroundings. Ideally, the material used for
945 shielding should itself be free of any radioactive material that might contribute to the
946 background. In practice, this is difficult to achieve as most construction materials contain at least
947 some naturally radioactive species (such as 40K, members of the uranium and thorium series,
948 etc.). The thickness of the shielding material should be such that it will absorb most of the soft
949 components of cosmic radiation. This will reduce cosmic-ray background by approximately 25
950 percent. Cosmic-ray interactions in lead shields will produce lead X-rays that are in turn shielded
951 by cadmium and copper liners. Such a shield is referred to as a "graded shield." Six millimeters
952 of oxygen-free high-conductivity (OFHC) copper can also be used to reduce the cosmic-ray
953 produced lead X-rays without the cadmium liner. Shielding of beta- or gamma-ray detectors with
954 anti-coincidence systems can further reduce the cosmic-ray or Compton scattering background
955 for very low-level counting.
956 Detectors have a certain background counting rate from naturally occurring radionuclides and
957 cosmic radiation from the surroundings; and from the radioactivity in the detector itself. The
958 background counting rate will depend on the amounts of these types of radiation and on the
959 sensitivity of the detector to the radiations.
960 In alpha counting, low backgrounds are readily achieved since the short range of alpha particles
961 in most materials makes effective shielding easy. Furthermore, alpha detectors are quite
962 insensitive to the electromagnetic components of cosmic and other environmental radiation.
963 The size and interior dimensions of shields constructed for gamma-ray spectrometry or gamma
964 counting in general should be considered so that sample back scatter radiation from the shield
965 wall to the detector is minimized. In general, shield wall should be at least 10 cm from the
966 detector. Back scatter radiation will fall off as the square of the detector to shield wall distance.
967 15.9 Instrument Calibration
968 Calibrations of instruments should be made using reference materials of known and documented
969 value and stated uncertainty. These reference materials should be supplied by:
JUNE 2001 DRAFT MARLAP
NOT FOR DISTRIBUTION 15-31 DO NOT CITE OR QUOTE
-------
Nuclear Counting Instrumentation
970 • National Institute of Science and Technology (NIST) directly;
971 • A standard source supplier whose measurement capabilities and/or manufacturing processes
972 are periodically tested by NIST; and
973 • A user who documents derived materials with stated uncertainty and whose value has been
974 verified with analytical and measurement systems that have been periodically tested through
975 an unbroken chain of comparisons to the national physical standards.
976 Periodic testing of source manufacturers, whether they be commercial or agency suppliers or end
977 users, is most cost effectively implemented through measurement assurance programs that are
978 ultimately linked to NIST traceability (Hoppes, 1990).
979 A comprehensive discussion of germanium detector set up and calibration can be found in ANSI
980 N42.14.
981 15.10 Other Considerations
982 15.10.1 Alpha
983 15.10.1.1 Troubleshooting
984 A number of factors can influence alpha counting results. These include attenuation or self
985 absorption, detector contamination, and other radionuclide interference. Attenuation or self
986 absorption corrections need not be made if constant conditions are maintained for sample and
987 calibration standard counting. If conditions can not be held constant, then corrections will have to
988 be made in order to produce accurate results. For example, the gamma rays from 137Cs in a water
989 matrix counted in a 90 cm3 aluminum can will require a 15 percent correction. Individual
990 electrical line conditioners or uninterruptible power supplies as well as supplemental air
991 conditioning can be provided in the counting rooms to maintain electrical and environmental
992 stability. Additionally, humidity control can also provided. Temperature and humidity may be
993 recorded with a chart recorder.
994 Detector contamination can also be a problem in some cases and, therefore, detector backgrounds
995 should be periodically checked. Contaminated detectors will have higher background counts and
996 even when sample spectra are corrected for the presence of contamination the higher background
997 results in higher MDAs. Finally, some alpha counters may be sensitive to beta radiation, and
998 corrections may have to be made for this interference. For a routine operating alpha counting
DRAFT MARLAP JUNE 2001
DO NOT CITE OR QUOTE 15-32 NOT FOR DISTRIBUTION
-------
Nuclear Counting Instrumentation
999 system periodic instrument QC checks should be performed at some specified frequency. This
1000 would include, as appropriate, counting efficiency, background, resolution, gain, and voltage
1001 plateau.
1002 Solid state detectors used for alpha spectrometry can become contaminated by recoil. This recoil
1003 contamination, which increases the detector background, takes place when fragments from
1004 sources travel to the detector and are implanted in the detector surface by the recoil energy
1005 imparted to the nucleus of an alpha-emitting atom. The energy of the fragments may be sufficient
1006 to implant them in the detector so that they cannot be removed non-destructively. Recoil
1007 contamination can invalidate a count after only a single sample count and cause a constant need
1008 to decontaminate equipment.
1009 The application of a negative bias to the sample, in conjunction with an absorbing layer of air, or
1010 a thin film absorber (12 jig/cm3) helps to keep recoil particles from imbedding themselves into
1011 the detector. For better resolution and where recoil contamination is of no concern, it is advisable
1012 to maintain a low pressure. Typically, systems can pump down to under 50 jim and, by
1013 continuously running the pump, maintain that level indefinitely.
1014 Detector contamination dominated by two processes, alpha recoil and "volatilization" of
1015 polonium. Alpha recoil contamination occurs when an alpha-emitting nuclide on the source plate
1016 decays to an alpha-emitting daughter or string of progeny. Since the specific activity is inversely
1017 proportional to the half-life for a fixed number of atoms, recoil will produce the most background
1018 activity when relatively short-lived progeny are produced. However, if the half-lives in question
1019 are very short (say up to a few hours), they will decay away quickly enough to be of little concern
1020 in alpha spectrometry. Particularly serious are those cases that involve transfer of recoil progeny
1021 with half-lives from days to weeks, short enough that a reasonable amount of parent activity will
1022 produce a significant amount of recoil contamination, and long enough that decay back to normal
1023 background levels will require an inappropriately long time. In addition, the effect is chronic:
1024 similar recoil-producing samples counted in the same chamber will produce a long-term build-up
1025 of detector background which could eventually become serious.
1026 Some common examples of decay-chains that produce recoil contamination include 228Th, 229Th,
1027 and 226Ra. It is important to realize that even p-emitting nuclides ejected by alpha recoil can
1028 contribute to alpha background if they subsequently decay to alpha emitters. For example, the
1029 direct daughter of 229Th is 225Ra which decays by p-emission to the a-producing daughter 225Ac.
1030 Contamination of detectors by polonium isotopes, such as 210Po (t,/2 = 138.4 days), should occur
1031 by some other process than alpha recoil. Note that 210Po, the last radioactive member of the 238U
JUNE 2001 DRAFT MARLAP
NOT FOR DISTRIBUTION 15-33 DO NOT CITE OR QUOTE
-------
Nuclear Counting Instrumentation
1032 decay series, is the daughter of 210Bi, a beta-emitter. The transfer of polonium from a source to a
1033 silicon detector has been attributed to "aggregate" recoil and inherent volatilization of polonium
1034 at low pressure. Whatever the actual cause, it is clear that polonium activity is indeed transferred
1035 to detectors, a very serious problem with long-lived 210Po and even worse when working with
1036 209Po (t,/2 = 102 years) as a yield tracer.
1037 Liquid Scintillation Quenching
1038 Quenching, which is probably the most prevalent interference in liquid scintillation counting, can
1039 be defined as anything which interferes with the conversion of radionuclide decay energy to
1040 photons emitted from the sample vial, resulting in a reduction of counting efficiency. Two types
1041 of quenching may be encountered in liquid scintillation counting: optical or color quenching and
1042 chemical quenching. Color quenching results in a reduction of the scintillation intensity (as seen
1043 by the multiplier phototubes) due to absorption of the scintillation light by materials present in
1044 the scintillation solution resulting if fewer photons per quanta of particle energy and a reduction
1045 in counting efficiency. Chemical quenching results in a reduction in the scintillation intensity due
1046 to the presence of materials in the scintillation solution that interfere with the process leading to
1047 the production of light resulting in fewer photons per quanta of particle energy and a reduction in
1048 counting efficiency. The quenching process may be illustrated as follows.
1049 Radionuclide Decay Beta ->• Solvent & Scintillator Light Photon ->• Multiplier Phototube
Chemical Quench Chemical Quench
1050 One can have both types of quenching present in a sample. Note that in chemical quenching all
1051 energy radiations are equally effected, but in color quenching not all energy radiations are equally
1052 effected. Therefore, the measured sample counts should be corrected for quenching effects so
1053 that the radioactivity in the sample can be quantified. Typical quench corrections include
1054 Channels Ratio, External Standard and Internal Standardization.
1055 Attenuation
1056 Attenuation or self absorption corrections may be necessary for alpha counting. Attenuation
1057 corrections should be made whenever the sample matrix differs from that of the calibration
1058 standard. For example, when a gross alpha analysis is performed on an evaporated water sample
1059 of some thickness and an electroplated standard was used for the calibration. Attenuation
1060 corrections will have to be made. Alpha particle attenuation corrections will generally be
1061 necessary with a sample density thickness greater than about one mg/cm2.
DRAFT MARLAP JUNE 2001
DO NOT CITE OR QUOTE 15-34 NOT FOR DISTRIBUTION
-------
Nuclear Counting Instrumentation
1062 Figure 15.10 shows how severe the attenuation of alpha particles is in air.
1063
1064
1065
1066
1067
1068
1069
1070
1071
1072
1073
1074
1075
1076
1077
1078
1079
1080
1081
1082
1083
1084
1085
1086
1087
15.10.1.2 Calibration
Alpha counting instrumentation should
be calibrated with the specific radionuc-
lide of interest or a radionuclide of
similar alpha energy under the same
configuration that the sample will be
counted. The standard should contain
the same solid material as the sample
and be of the same weight. If the
samples and calibration standard are not
counted under identical conditions, then
corrections will have to be made. Also,
if there is a variation in weight from
sample to sample corrections will have
to be made, typically a calibration curve
relating sample weight to counting
efficiency is used.
4 6
ENERGY (MeV)
FIGURE 15.10 Range vs. Energy for Alpha Particles in Air
Alpha calibration standards are available from NIST or NIST-traceable commercial vendors.
Among the radionuclides available are 230Th, 241Am, 235U, 239Pu, 228Th, 238U, and 226Ra. Other
radionuclides are also available, NIST or a commercial vendor should be contacted regarding
procurement. Sources should be prepared in the manner in which the sample will be counted.
The source may be procured as a solution and then prepared in the appropriate counting
geometry, or the source may be procured directly in the appropriate geometry, such as an
electroplated standard.
15.10.1.3 Costs
1088 There are three major types of detectors used for alpha counting. Their cost will depend on the
1089 type of information wanted and the number of detectors in the unit.
1090 Solid state silicon surface barrier detectors are used to count and distinguish alpha particles of
1091 different energies. An alpha spectrometer consists of a vacuum chamber, detector, electronics to
1092 amplify the signal, a multichannel analyzer, and some means of collecting data. A system with
JUNE 2001
NOT FOR DISTRIBUTION
15-35
DRAFT MARLAP
DO NOT CITE OR QUOTE
-------
Nuclear Counting Instrumentation
1093 eight detectors, vacuum pump, computer and the software necessary for data collection and data
1094 reduction costs approximately $50,000.
1095 A liquid scintillation counter can be used to count alpha particles and in some cases provide
1096 some information about the energy distribution, although with poorer resolution than silicon
1097 surface barrier detectors. The LSC unit is typically set up to count samples sequentially, using
1098 one detector and automatic sample changer. The price depends on the background required, and
1099 will range from $25,000 to as much as $45,000. This price includes a computer and the
1100 appropriate software.
1101 A gas-flow proportional counter is used to count samples for a gross alpha (or beta) activity. The
1102 price of a unit depends on the number of detectors, the size of each detector, and the accessories.
1103 One major accessory could be an automatic sample changer. A system with 8 to 10 small
1104 detectors (1 inch in diameter) will cost from $35,000 to more than $60,000.
1105 There are no maintenance costs associated with an alpha spectrometer. If properly used and
1106 monitored, the system will retain its specifications for a long time. The detectors may need
1107 replacing eventually, if its resolution deteriorates or it becomes contaminated, at a cost of $500-
1108 $1,000 each.
1109 A liquid scintillation counter requires the use of an organic scintillation cocktail, which cannot be
1110 reused. The total cost of this cocktail, combined with the cost of the sample vials, should not
1111 exceed $500 for an annual throughput of approximately 1,000 samples.
1112 The operation expense associated with the use of a gas-flow proportional counter is for the ultra-
1113 high purity P10 gas, which is necessary if stable efficiencies and low backgrounds are required.
1114 All proportional counters should have calibrated gas regulators for accurate and reproducible
1115 settings of flow rates. The flow rate should be placed with the QC information that is with the
1116 other instrument QC. For almost constant operation of a system with eight detectors, as many as
1117 24 tanks of P10 gas per year will be required, at a total cost of approximately $7,000.
1118 All of the above instruments should be in a fairly constant temperature and low humidity
1119 environment, so that air conditioning and/or heating costs need to be factored in, as needed.
DRAFT MARLAP JUNE 2001
DO NOT CITE OR QUOTE 15-36 NOT FOR DISTRIBUTION
-------
Nuclear Counting Instrumentation
1120 15.10.1.4 Quality Control
1121 Statistical quality control (SQC) is discussed here to familiarize the reader with its application to
1122 nuclear counting instrumentation. More detailed information about SQC is provided in
1123 Chapter 19.
1124 The primary tool for statistical quality control is the control chart. A control chart is a graphical
1125 tool for monitoring the distribution of values produced by a measurement process or system. The
1126 distribution of values observed during a period when the system is in statistical control is used to
1127 set up the control chart. Subsequent values are then plotted on the chart and inspected to ensure
1128 that the system remains in control.
1129 Typically one ore more control charts for counting efficiency and background are maintained for
1130 each counting instrument. The instrument should be fully operational before the control charts
1131 are implemented. However, control charts should be in use before calibration of the instrument
1132 for a particular analysis to ensure that the instrument parameters are in statistical control during
1133 the calibration.
1134 The selection of the check source for monitoring counting efficiency is critical and should be
1135 made after considering guidance in this document. The source geometry, half-life, and radiation
1136 energy are important factors.
1137 A control chart should be based on an initial data set obtained from at least 15 measurements.
1138 Ideally, at least 10,000 counts per measurement are recommended to provide a relative counting
1139 uncertainty of no more than 1 percent. For some instruments, achieving the recommended 10,000
1140 counts may be impractical, especially for a background control chart. It may also be undesirable
1141 to place a high-activity efficiency check source in a low-background detector because of the
1142 potential for contamination.
1143 The initial measurements should represent the measurement system as it is used over time.
1144 Making the measurements over several days ensures that variability due to temperature and
1145 humidity changes is included. The source should be repositioned before each measurement to
1146 ensure that variability due to positioning error is included.
1147 The mean and standard deviation of the counts or count rates are estimated from the initial data
1148 set. The mean is used as the central line (CL) of the control chart. Warning limits are placed at ±2
1149 standard deviations from the central line, and control limits are placed at ±3 standard deviations
1150 from the central line.
JUNE 2001 DRAFT MARLAP
NOT FOR DISTRIBUTION 15-37 DO NOT CITE OR QUOTE
-------
Nuclear Counting Instrumentation
1151 Statistical tests of the data distribution should be performed at the time the warning and control
1152 limits are established. Tests for normality are common. It is also common to test whether the
1153 counts follow the Poisson model (Chapter 19).
1154 The central line and warning and control limits for an efficiency control chart should be adjusted
1155 for radioactive decay of the check source unless the source is very long-lived. Either the limits or
1156 the data points may be decay-corrected. It may also be necessary to adjust the counting time for
1157 the check source measurements if the source decays considerably during the period when the
1158 chart is in use. It is important to note that the relative standard deviation of the measured values
1159 increases as the mean number of counts per measurement decreases.
1160 When a measured value falls within the warning limits, the measurement system is considered to
1161 be in control. If a value falls outside the control limits, the system is considered out of control.
1162 These two rules are commonly used to evaluate control charts, although stricter evaluation
1163 criteria are sometimes used. Common sense should be exercised if the data meet the objective
1164 evaluation criteria but nevertheless demonstrate patterns or trends that might indicate developing
1165 problems. For example if a long increasing or decreasing sequence of values is observed, an
1166 investigation is probably warranted even if all of the values are between the warning limits.
1167 Generally, if a value falls within the control limits but outside the warning limits, the system may
1168 require more attention but it is not yet considered definitely out of control. The Westgard Rules,
1169 which are recommended by ASTM El329, provide more elaborate criteria for evaluating such
1170 measurements.
The Westgard Rules*
1. Is the measurement more than 2 sigma from the mean? If not, go to Step 7.
2. Is the measurement more than 3 sigma from the mean? If so, go to Step 8.
3. Are the last two measurements more than 2 sigma from the mean? If so, go to Step 8.
4. Is the range of the last two measurements more than 4 sigma? If so, go to Step 8.
5. Are the last four measurements more than 1 sigma from the mean? If so, go to Step 8.
6. Are the last ten measurements more on the same side of the mean? If so, go to Step 8. Otherwise, go to Step 7.
7. Accept the measurements. Stop.
8. The measurements are out of control. Stop.
1171
1172
1173
1174
1175
1176
1177
1178
1179
1180 'Adapted from ASTM E1329.
1181 The following two sections on proportional counting and liquid scintillation counting are
1182 applicable to both alpha and beta measurements.
DRAFT MARLAP JUNE 2001
DO NOT CITE OR QUOTE 15-38 NOT FOR DISTRIBUTION
-------
Nuclear Counting Instrumentation
1183 Proportional Counters
1184 The following should be considered when QC checks are not within limits.
1185 1. Is the standard decay corrected, correctly?
1186 2. Check log book to see what changes were made to counter and if the repairman recently
1187 changed any switch settings.
1188 3. If gas cylinder was changed recently, was system allowed to purge? Was correct gas (10P)
1189 obtained? Verify the correct regulator pressure, and ensure the gas cylinder valve is open all
1190 the way.
1191 4. If backgrounds are high, check for dirt or dust on the background planchet. Check window
1192 for contamination and replace if necessary.
1193 5. Check alpha and beta voltages.
1194 6. Check discriminator settings.
1195 7. Check voltages on nim bin power supply (±12 V, ±24 V).
1196 8. Check alpha and beta plateau voltage for drift.
1197 Liquid Scintillation Counters
1198 The following should be considered when QC checks are not within limits.
1199 1. Is the standard decay corrected, correctly?
1200 2. Has the quench value for the unquenched standard for the instrument changed? The quench
1201 value for the unquenched standard indicates the overall gain of the system. Run the
1202 autocalibration and verify the result with the historical result.
1203 3. Check for dirt or fingerprints on outside of vial.
1204 4. Check for dirt inside instrument.
JUNE 2001 DRAFT MARLAP
NOT FOR DISTRIBUTION 15-39 DO NOT CITE OR QUOTE
-------
Nuclear Counting Instrumentation
1205 5. Is sample two phase?
1206 6. Has standard dark adapted and reached temperature equilibrium?
1207 7. Check log book to see what changes were made to machine and if repairman recently
1208 changed any switch settings.
1209 15.10.2 Beta
1210 15.10.2.1 Introduction
1211 Accurate beta particle measurements will depend upon the degree and extent to which the
1212 parameters that affect the measurement process under considerations are quantified. These
1213 parameters may include:
1214 • Radiation detector used;
1215 • Material and shape of the final sample mount;
1216 • Form and thickness of final sample for analysis;
1217 • Radionuclide purity of final sample;
1218 • Final sample-to-detector distance; and
1219 • Beta particle energy.
1220 Beta particle attenuation or self absorption corrections to the detector efficiency may be
1221 necessary depending on the beta particle energy detection system and final sample form. The
1222 potential of detector contamination from sample measurements is a function of the type of
1223 detector used and the stability of the final sample composition. The inherent beta particle
1224 background of the various detection systems should be evaluated and its contribution removed
1225 from the sample measurement result. The beta particle measurement system should be calibrated
1226 with NIST-traceable standards and its subsequent performance held to established measurement
1227 quality requirements through the use of daily or prior-to-use quality control checks. In addition,
1228 appropriate instrument quality control should be established for background, voltage plateau,
1229 quenching, resolution and alpha-beta cross talk. Guidance on beta particle counting can be found
1230 in industry standards (ASTM D1890; D3648; E1329) and publications (NCRP Report 58; Knoll,
1231 1989; Lapp and Andrews, 1954; Price, 1989; USPHS, 1967).
1232 "Gross" alpha and beta counting of evaporated samples, wherein a multitude of alpha and beta-
1233 emitting radionuclides may exist, is typically used for screening of water samples. The
1234 application of such methods may be targeted for a specific radionuclide or a category of
DRAFT MARLAP JUNE 2001
DO NOT CITE OR QUOTE 15-40 NOT FOR DISTRIBUTION
-------
Nuclear Counting Instrumentation
1235 radionuclides such as the naturally occurring nuclides or a specific radionuclide in a facility
1236 effluent. However, extreme caution should be applied to the interpretation and use of such results
1237 without a full specific radionuclide characterization of the water source under investigation. The
1238 type of analysis is to be considered "gross" and, in most cases and for a variety of sound
1239 technical reasons, the gross measurement result does not equal the sum of the radionuclides
1240 contained in the sample.
1241 When specific radiochemistry is performed the beta-emitting radionuclide of interest will be
1242 isolated, concentrated and converted to a desired final chemical and physical form. Under these
1243 circumstances, the beta detection system should be calibrated for the radionuclide, chemical
1244 composition of the final sample form and the range of final sample weights expected from
1245 chemical recovery.
1246 15.10.2.2 Alpha Particle Interference and Beta Energy Resolution
1247 When properly operated or under optimal counting conditions (thin final samples or low LS
1248 quenching and high beta energy), most beta particle counting systems can separate alpha and beta
1249 particle detection events. However, the degree of alpha particle detection by the beta detector
1250 under consideration should be evaluated for each radionuclide, mixture of radionuclides or
1251 specific final sample form. Beta detection systems that are considered to have beta energy
1252 spectral resolution capabilities may be less affected by samples containing alpha-particle emitting
1253 radionuclides. However, for window gas proportional counters, alpha particle energy degradation
1254 by air, detector window or sample self absorption may lead to false beta detection without proper
1255 evaluation. A typical example would be a thick final sample matrix containing a mixture of alpha
1256 and low-energy beta-emitting nuclides.
1257 Some commercial window gas proportional counting systems have a feature for simultaneous
1258 alpha and beta particle counting that uses a voltage pulse height discrimination for the separation
1259 of beta and alpha particle detection events. A common and more historical means of separating
1260 alpha and beta particle events is to count the sample on the alpha proportional counting voltage
1261 plateau followed by a count on the beta (plus alpha) proportional counting voltage plateau. An
1262 alpha-to-beta crosstalk factor should be determine for the final sample weight and for the alpha
1263 and beta energies under consideration. The net beta count is determined by multiplying the alpha
1264 counts (from the alpha window for simultaneous counting or on the alpha counting plateau) by
1265 the alpha-to-beta cross talk factor.
1266 Window gas proportional counters typically are not used for beta spectrometers but instead
1267 record beta particle detection events giving rise a voltage pulse large than a discriminator setting.
JUNE 2001 DRAFT MARLAP
NOT FOR DISTRIBUTION 15-41 DO NOT CITE OR QUOTE
-------
Nuclear Counting Instrumentation
1268 Under most circumstances, liquid scintillation counters have sufficient energy resolution
1269 capability and electronic discrimination to fully separate beta and alpha particle detection events.
1270 However, due to the nature of the beta energy continuum of an emission process and the inherent
1271 resolution of a liquid scintillation spectrometer, identification and quantification of multiple
1272 nuclides contained in the same sample is complicated unless their beta energies are widely
1273 separated. Computer software and beta interference factors should be applied in such cases.
1274 A liquid scintillation counter is typically used for Cerenkov counting. However, the final sample
1275 solution contains no scintillator as would a full liquid scintillation-sample cocktail. Cerenkov
1276 counting, due to the nature of measurement process, will not detect alpha particles of any energy
1277 or beta particles having an average beta energy less than 260 keV. Cerenkov counting is typically
1278 applied to single nuclide evaluations or for a mixture of two nuclides that have a differential
1279 maximum beta energies greater than 700 keV (e.g., 89Sr and 90Y). Beta interference factors should
1280 be applied in such cases.
1281 15.10.2.3 Liquid Scintillation Quenching
1282 The information on liquid scintillation quenching provided in Section 15.10.1.1 is applicable for
1283 beta particle detection. The degree of quenching should be determined for each radiochemical
1284 method, radionuclide or application. An appropriate correction factor/curve should be calculated
1285 and applied to the measurement results for the samples being evaluated. The magnitude of the
1286 quench correction may approach 50 percent in certain severe quenching situations.
1287 Cerenkov counting is less sensitive to "quenching" than liquid scintillation counters using
1288 scintillation cocktails. Typically, the final sample solution is a result of a control radiochemical
1289 process that eliminates most sources of contamination, chemical impurities and variability in the
1290 final sample solution.
1291 15.10.2.4 Beta Particle Attenuation
1292 Beta particle attenuation should be considered for window gas proportional, plastic scintillator
1293 and solid state detector counting applications. Beta particle attenuation can result from the
1294 interaction of a beta particle with the air, detector window or the matrix atoms of the final
1295 sample. Beta particle air attenuation is a function of the distance between the sample or source
1296 and the detector's particle entrance window. Under most application for beta particle counting,
1297 this factor is typically insignificant compared to the other sources of beta particle attenuation.
DRAFT MARLAP JUNE 2001
DO NOT CITE OR QUOTE 15 -42 NOT FOR DISTRIBUTION
-------
Nuclear Counting Instrumentation
1298 Figure 15.11 shows the attenuation of
1299 beta particles is in air and water.
1300 Consideration of the detector window
1301 thickness and its beta particle
1302 attenuation becomes important when
1303 evaluating low energy beta particles
1304 such as 14C. Normally, the air and
1305 detector window attenuation factors
1306 are determined as a combined beta
1307 attenuati on-effi ci ency factor that
1308 includes the sample self absorption for
1309 a given application. In most
1310 applications, a back scatter factor for
1311 the material composition (Z value) of
1312 the final sample mount is included
1313 into a combined attenuation-
1314 backscatter-efficiency factor or, more
1315 simply, the combined detector
1316 efficiency correction factor.
1317 For the lower to intermediate beta
1318 particle energies, the combined detector
1319 effi ci ency factor i s a functi on of b eta
1320 energy, final sample mass and mass
1321 composition. For beta particles having a
1322 maximum beta energies greater than
1323 1,500 keV, the combined detector
1324 efficiency factor is nearly constant over
1325 a final sample weight range of 0 to 5
1326 mg/cm2. A typical combined beta
1327 detector efficiency curve for 131I (606
1328 keV pmax) as Cul over a weight range of
1329 0 to 50 mg is shown for a plastic
1330 scintillator beta detector in Figure 15.12.
1331 A compl ete revi ew of the detect! on
1332 method can be found in reference
1333 (McCurdyetal., 1980).
IJiJ
1E2
1E1
1EO
1E-1
P
-2. -.
1E-3
[•-"
!^
W
^*>
,x*"
Air
atei
,--'
'^
,x-
**
X1
^
^
^
0.1 1
^
^
-
10
ENERGY (MeV)
FIGURE 15.11 Range vs. Energy for Beta Particles in Air and
Water
Window Gas Proportional Counter
Gross Beta Data - Cs-137 Standard
Ojion
AZ\J \
OAf\r\
.4UU
.380
^^
^ O.ooO
CD
Q O.o40
P n oon
jj U.O^U
Oonn
.oUU
OOQH
.ZOU
.260
0
V
i
\
^
\
\
"\
\
\.
^v
N
\
X
50 100 150 200 250 300 350
Weight (mg)
FIGURE 15.12 Beta Detector Efficiency Curve for 131I vs.
Weight
JUNE 2001
NOT FOR DISTRIBUTION
15-43
DRAFT MARLAP
DO NOT CITE OR QUOTE
-------
Nuclear Counting Instrumentation
1334 15.10.2.5 Calibration
1335 Beta particle counting systems should be calibrated with the specific radionuclide under
1336 investigation or a surrogate radionuclide of similar beta energy having a comparable final sample
1337 composition and configuration. However, it should be mentioned that moderate to severe
1338 calibration biases may occur depending on the severity of the departure from the chemical
1339 composition of the final sample matrix and the beta energy of a surrogate. For this reason, the use
1340 of a surrogate radionuclide is discouraged unless the availability of the radionuclide of interest is
1341 non-existent. Corrections between the surrogate and radionuclide of interest should be
1342 determined and applied to sample results. For electroplated plated samples, a correction factor
1343 needs to be determined if the plating material of the surrogate is not the same as that used for the
1344 samples.
1345 Cerenkov counting normally involves a single radionuclide calibration (single energy calibration)
1346 for the final sample solution. Typically, the final sample solution is a result of a control
1347 radiochemical process that eliminates most sources of variability for the calibration process.
1348 Aqueous beta-emitting radionuclide calibration standards and sources are available from NIST or
1349 from a NIST-traceable commercial radioactive source manufacturers. The long-lived pure beta-
1350 emitting radionuclides available from NIST include: 3H, 14C, 63Ni, 1291,89Sr, 90Sr, "Tc, 228Ra, and
1351 241Pu. The majority of the gamma-emitting radionuclides also emit beta particles in the nuclear
1352 transformation process. Check Section 15.4 for the availability of known beta- gamma emitting
1353 radionuclides. Contact a NIST-traceable radioactive source manufacturer for the availability of
1354 other pure beta or beta/gamma-emitting radionuclides (ANSI N42.15, American National
1355 Standard Check Sources for and Verification of Liquid-Scintillation Counting Systems).
1356 Aqueous radioactive standards can be prepared in the appropriate geometry for LS or Cerenkov
1357 counting or through chemical processing precipitated or electroplated as final sample form for
1358 counting by a gas proportional, plastic or solid state beta detection system.
1359 15.10.2.6 Costs
1360 There are four principal beta detection methodologies available. Window gas proportional
1361 counting and liquid scintillation counting systems (Cerenkov counting as well) can be purchased
1362 with the option of readily available automatic sequential sample counting systems. Sample
1363 capacity is typically 100. These automatic sequential counting systems are available in the
1364 $30,000 to $50,000 range depending on options. Multiple detector window gas proportional
1365 counters having a simultaneous counting capability are available from some commercial
DRAFT MARLAP JUNE 2001
DO NOT CITE OR QUOTE 15 -44 NOT FOR DISTRIBUTION
-------
Nuclear Counting Instrumentation
1366 manufacturers. The basic unit contains four detectors but several units can be combined to give
1367 eight or 16 detector systems. The basic price for such units is in the range of $20,000 to $50,000
1368 depending on the number of detectors and options.
1369 Solid state silicon surface barrier and ion implanted Ge or Si detectors are used to perform
1370 spectral analysis of beta emitting radionuclides. A solid state beta spectrometry system consists
1371 of a vacuum chamber, solid state detector, high voltage-preamp-amplifier instrumentation
1372 modules, a multichannel analyzer (MCA) or equivalent computerized MCA using an analog-to-
1373 digital converter and electronic data storage. Individual ion-implanted Ge detectors having an
1374 active area of 450-2,000 mm2 and a 500 |im thickness range in price between $1,300 and $3,200.
1375 Beta resolution of these detectors is typically approximately 12 keV.O
1376 A beta spectrometry system consisting of eight detectors with vacuum pump and computer would
1377 be approximately $30,000-$40,000, without background reducing shielding. Solid state
1378 spectrometry systems for beta particle applications, unlike that for alpha particles, would be
1379 sensitive to external background from cosmic radiation, terrestrial radiation and inherent beta
1380 radioactivity in the surrounding materials.
1381 Automatic sample counting, plastic scintillator beta particle detection systems have not been
1382 commercialized for the radioassay laboratory setting. Most of these systems have been fabricated
1383 by the user from readily available components, electronic modules, multichannel analyzers and
1384 lead shielding. The cost of a single detector system is estimated to be less than $15,000.
1385 Maintenance costs for the liquid scintillation counters, window gas proportional counters and
1386 alpha spectrometry systems have been discussed in Section 15.10.1.3.2 for alpha counting
1387 applications. If a laboratory already has existing units for the alpha particle measurement
1388 applications, there will be no additional maintenance cost relative to their use for beta particle
1389 measurements.
1390 There is no maintenance cost associated with the operation of a plastic scintillator beta
1391 spectrometry system.
1392 Costs associated with the maintenance of the room environment for the nuclear detection
1393 equipment should be considered. Service maintenance relative to the constant voltage supply or
1394 uninterruptable power sources as well as having a dust free constant temperature and humidity
1395 environment should be considered.
JUNE 2001 DRAFT MARLAP
NOT FOR DISTRIBUTION 15-45 DO NOT CITE OR QUOTE
-------
Nuclear Counting Instrumentation
1396 15.10.2.7 Quality Control
1397 See section 15.10.1.4.
1398 15.10.3 Gamma
1399 15.10.3.1 Troubleshooting
1400 Once a gamma-ray spectrometry system has been established in accordance with the manufac-
1401 turer's or supplier's instructions, a daily count of a calibration or reference source should be
1402 performed to assure the system continues to operate properly. The three parameters that should
1403 be checked and recorded are: energy calibration (keV/channel), counting efficiency (count
1404 rate/decay rate), and gamma-ray peak resolution (FWHM). With the exception of a complete
1405 detector or electronic component failure (no pulses are detected at the preamp or multiplier
1406 phototube output), degradation of gamma-ray peak resolution will be the first indication that
1407 detector is not performing properly or that electronic noise has been introduced into the counting
1408 system by the preamplifier, amplifier, or multichannel analyzer. Any indications that the detector
1409 efficiency is not within statistical limits of expected values should be reported, since this value
1410 will be used to convert the observed count rate to decay rate. The energy calibration should either
1411 be recorded with sample spectral data or adjusted daily to a previously established constant
1412 value. This energy calibration should be accurately known so that nuclide identifications can be
1413 made. See page 51 for a list items to be checked if the counting system is out of specifications.
1414 Gamma-ray spectrometry systems are extremely sensitive to both electronic and environmental
1415 conditions. Temperature changes can cause spectral shifts and improper nuclide identifications
1416 because of incorrect energy calibrations. Excessive humidity in the detector preamplifer can
1417 cause high voltage arcing which results in poor peak resolution or complete system failure.
1418 Improper pole zero settings, which effects the shape of the pulses being analyzed, can cause
1419 degradation of peak shapes and resolution. Poorly conditioned NEVI power can introduce
1420 electronic noise which will also result in degraded peak resolution. Routing of cables between the
1421 detector, electronics, multichannel analyzer, computers, and monitors is very important. The
1422 introduction of any spurious electronic noise into any of the components that make up the
1423 gamma-ray spectrometry system can degrade the resulting data.
1424 The need to make corrections for self-absorption in environmental samples during routine
1425 gamma-ray spectrometry cannot be overemphasized (Modupe et al., 1993). The correction to be
1426 made for the difference in self-absorption between calibration standards and sample matrices is
1427 usually small for intermediate and high energy photons, but it is not negligible at low energies
DRAFT MARLAP JUNE 2001
DO NOT CITE OR QUOTE 15-46 NOT FOR DISTRIBUTION
-------
Nuclear Counting Instrumentation
1428 where the photoelectric effect is the most important mode of attenuation. The photoelectric
1429 process varies approximately as Z4"5 (Z is the atomic number of the elements in the medium) so
1430 that a change in the elemental composition of a sample relative to a calibration standard can
1431 require a correction factor for detector efficiency as high as a factor of 2.
1432 The quantities |i and H are the linear attenuation coefficient and the thickness of sample,
1433 respectively. I0 and I are the intensities of the beam emerging from the sample container without
1434 and with an absorbing matrix in place. This is the traditional self-absorption equation. For
1435 complex counting geometries of homogeneous materials, an estimated average H (sample
1436 thickness) can be used.
1437 The method for self-absorption correction at various energies requires that the linear attenuation
1438 coefficient, \i, of the sample matrix be known. Knowledge of |i usually requires that the
1439 elemental composition of the matrix be determined. The tedium and time required in elemental
1440 analysis may make it impractical for routine gamma-ray analyses involving large numbers of
1441 samples. Computer programs are available to calculate |i/p for various compounds when the
1442 percent elemental composition of the compound is known. |i/p is computed as a linear
1443 combination of the mass attenuation coefficients of the composite elements.
^iPi ' (152)
1444 where P; is the percent by weight of the ith element in the compound.
1445 The gamma-ray path length, H, is equal to the thickness of sample. When performing gamma-ray
1446 transmission measurements to determine |i a path length of H is used. To determine the self-
1447 absorption correction for radioactive samples, the corrections are integrated for a path length of 0
1448 to H.
1449 When a photon beam passes through a homogenous sample of mass attenuation coefficient, |i/p,
1450 density, p, and thickness, H, the percentage beam attenuation, A, is given by
A =^100% = (l-e~^pp)100% (15.3)
0
JUNE 2001 DRAFT MARLAP
NOT FOR DISTRIBUTION 15-47 DO NOT CITE OR QUOTE
-------
Nuclear Counting Instrumentation
1451
15.10.3.2 Calibration
1452
1453
1454
1455
1456
1457
1458
1459
1460
1461
1462
1463
1464
1465
1466
1467
1468
1469
1470
1471
1472
1473
1474
1475
1476
1477
1478
1479
Most gamma-ray spectrometry systems are calibrated with either single or mixed standards in an
exact matrix and geometric form as the samples to be analyzed. However, there are computer
codes that can calculate detector efficiency from the physical dimensions of the detector and
sample counting geometry (Mitchell, 1986 and 1988, Hensley et al., 1997). Commercial
standards of single or mixed gamma-ray emitters in a matrix of known chemical composition and
density can be prepared in user supplied containers. Calibrations based upon these standards can
then be adjusted to correct for any differences in composition and density between the calibration
source and the sample (Modupe et al., 1993).
Table 15.2 lists some gamma-ray emitting nuclides that can be used for energy and efficiency
calibration (Sanderson et al., 1993; Browne et al., 1986).
'ABLE 15.2 Nuclides for Gamma-ray Spectrometer Calibratio
NUCLIDE
210pb
241 Am
109Cd
57Co
141Ce
139Ce
203Hg
51Cr
113Sn
85Sr
137Cs
54Mn
88y
65Zn
60Co
40K
ENERGY (KeV)
46.5
59.5
88.0
122.1
145.4
165.9
279.2
320.1
391.7
514.0
661.7
834.8
898.1, 1836.1
1115.5
1173.2, 1332.5
1460.8
HALF-LIFE
22.3 years
432.2 years
462.6 days
273 days
32.5 days
137.7 days
46.6 days
27.7 days
115.1 days
64.8 days
30.0 years
3 12.5 days
106.6 days
243. 8 days
5.27 years
1. 28 xlO9 years
DRAFT MARLAP
DO NOT CITE OR QUOTE
15-48
JUNE 2001
NOT FOR DISTRIBUTION
-------
Nuclear Counting Instrumentation
1480 15.10.3.3 Software
1481 Most laboratories are now using commercially available software for the analysis of gamma-ray
1482 spectra. These programs are easy to use and do not require the user to be an expert in gamma-ray
1483 spectrometry. An evaluation of some of these programs in 1987 indicated there were substantial
1484 differences in the abilities of the programs to resolve multiplets of unequal intensity and to
1485 analyze complex spectra (Sanderson 1988). Another evaluation was completed in 1992 (Decker
I486 and Sanderson, 1992) since many of the programs had undergone numerous revisions and there
1487 were a few new programs available. The second evaluation indicated a substantial improvement
1488 in the deconvolution of doublets and the results of the analysis of a Chernobyl air filter were
1489 much more consistent than when a similar filter was analyzed in the first evaluation. The six
1490 programs analyzed in 1991 include GAMMA-W from Germany, INTERGAMMA from France,
1491 OSQ/Plus from Canada, SAMP090 from Finland (supplied by Canberra Industries, USA) and
1492 OMNIGAM and GDR from the United States. Some of the features which contribute to a good
1493 program included the ability to display the spectrum as well as calculated calibration files, the
1494 ability to manually insert peaks during the fitting procedure, an extensive nuclide library and the
1495 ability to easily transfer nuclides to smaller, working libraries, an analysis report which includes
1496 the names of the calibration files used, a peak fit report including any problems with the shape of
1497 the peaks, and identification of the peaks used in the activity calculation as well as any problems
1498 with interfering lines.
1499 In 1996 the Environmental Measurements Laboratory of the U. S. Department of Energy began a
1500 Gamma Spectra Data Evaluation program (Decker et al., 1996) whose goal was to test the ability
1501 of the present day software to accurately identify and quantify the nuclides in a complex spectra
1502 and the ability of the user to properly utilize the software. In order to do this, synthetic spectra
1503 were generated using the computer code SYNTH developed by Walt Hensley at the Pacific
1504 Northwest National Laboratory. The spectra were then converted to a variety of formats on disk
1505 and Digital Equipment Corporation (DEC) TK 50 tape and sent to DOE laboratories and DOE
1506 contractors. A calibration spectrum, a background spectrum and three sample spectra were sent
1507 to each participant. These spectra simulated those that would be obtained when an air filter was
1508 counted 10 cm from a 22 percent coaxial detector with a 0.5 mm beryllium window. Two of the
1509 samples contained fallout and naturally occurring nuclides with half lives greater than thirty days.
1510 The third sample contained both short and long lived fission product nuclides. Thirty one
1511 laboratories participated using 16 different software packages. The software packages included
1512 Aptec, Vertechs GDR/P, Nuclear Data ASAP, various Ortec packages, and various Canberra
1513 packages for both the PC and the DEC MicroVax. Most of the laboratories did fairly well with
1514 the first two samples. A few laboratories reported nuclides that were not present in the third
1515 sample and did not accurately quantify those that were. The results did not seem to be software
JUNE 2001 DRAFT MARLAP
NOT FOR DISTRIBUTION 15-49 DO NOT CITE OR QUOTE
-------
Nuclear Counting Instrumentation
1516 dependent but were due to the user utilizing or not utilizing available software features properly.
1517 There was quite a wide range of numbers for both the uncertainty terms and for the minimum
1518 detectable activities which seems to indicate we need a consistent way of calculating these terms
1519 to make them more meaningful.
1520 15.10.3.4 Costs
1521 Gamma-ray spectrometry systems can cost from $14,000 to well over $60,000 depending upon
1522 the choice of detector. For a 75 x 75 mm Nal(Tl) system the costs would be approximately
1523 $1,000 for the detector, $5,000 for a 10 cm graded lead shield, and $8,000 to $10,000 for a
1524 multichannel analyzer. Very large HPGe detectors will cost more than $50,000. The actual
1525 detector cost will depending upon the size of the germanium crystal, its resolution, and method of
1526 cooling. Data reduction costs (software and computer) would be an additional expense for either
1527 type of system.
1528 Nal(Tl) detector systems do not require any additional maintenance beyond what any laboratory
1529 electronic system requires. Each HPGe detector will require approximately $1,200 per year for
1530 liquid nitrogen to maintain their operating temperature. An electrical/mechanical cooler can be
1531 used in place of a liquid nitrogen cryostat but it will require 0.5 to 1.0 kW of power around the
1532 clock to operate. Both systems should be operated at constant temperature for reliable
1533 performance. This may require substantial air conditioning.
1534 15.10.3.5 Quality Control
1535 Initial data to prepare solid state gamma detector QC charts may be obtained by counting a mixed
1536 gamma point source between 20 to 30 times (Ideally, these counts should be over a period of
1537 several weeks. However, if time does not permit, the counts may be accumulated over 1 to
1538 7 days.) Two or three QC charts (depending on age of mixed gamma point source) are initially
1539 established for the mixed gamma point source and control limits are established for background.
1540 The three source charts cover the low energy (88 keV, 109Cd), the medium energy (661.6 keV,
1541 137Cs), and the high energy (1,332.5 keV, 60Co). The source is counted until between 10,000 to
1542 40,000 counts are obtained in each photopeak.
1543 Background QC charts are established according to the procedure already listed for Proportional
1544 and liquid scintillation counters with the following exception: the background is counted and the
1545 total counts in the spectrum are obtained by summing the counts in the entire spectrum.
DRAFT MARLAP JUNE 2001
DO NOT CITE OR QUOTE 15-50 NOT FOR DISTRIBUTION
-------
Nuclear Counting Instrumentation
1546 The resolution of the detector (FWHM) is measured each month and recorded, but it is not
1547 plotted. A NIST 60Co source is positioned 25 centimeters from the end-cap face and counted for
1548 100 minutes. The FWHM is calculated by the peak search program for the 1,173.2 keV and
1549 1,332.5 keV peaks, and recorded in the logbook.
1550 When the energy of the source QC exceeds the specified energy tolerance (for example
1551 ±0.75 keV) from its initial calibrated value, the analyzer system should be recalibrated. First
1552 determine whether a gain or zero shift has occurred. A gain shift is a nonlinear shift in channels
1553 of low and high energy peaks (i.e., 109Cd peak shifts ±1 channel and 60Co peak shifts ±3 channels).
1554 A zero shift is a linear shift in channels for both low and high energy peaks (ie., 109Cd peak shifts
1555 ±1 channel and 60Co shifts ±1 channel also). Make the appropriate adjustments to the amplifier
1556 (gain) or the Analog to Digital Converter (zero). Recalibrate the analyzer and record the slope
1557 (keV/channel) and the zero intercept in the log book. If the best fit of the recalibration curve is a
1558 nonlinear fit (quadratic), record the "Q" coefficient, keV/channel2, in the log book. Also record
1559 the updated FWHM calibration factors, slope, offset, and FWHM at 1,332.5 keV in the log book.
1560 The following should be considered when QC checks are not within limits.
1561 • Is standard decay corrected to the proper date?
1562 • Check sample positioning.
1563 • Check for zero shift.
1564 • Check for gain shift.
1565 • Check full width at half-maximum.
1566 • Check nim bin power supply voltages (±6 V, ±12 V, ±24 V).
1567 • Check efficiency tables.
1568 • Check for moisture on the detector due to recently filling the dewar with liquid N2.
1569 15.10.4 Non-Nuclear Instrumentation
1570 15.10.4.1 ICP-Mass Spectrometry
1571 ICP-MS is one of the most versatile and sensitive atomic spectroscopy techniques available. It
1572 can be used to determine the concentrations of over 70 elements. The detection limit of the
1573 technique extends down to the parts-per-billion range in soils and to the parts-per-trillion range in
1574 waters. This sensitivity makes ICP-MS an attractive complement to decay-counting techniques in
1575 the radiochemical analysis laboratory. For very long-lived radioisotopes (those with half-lives
1576 over 10,000 years, e.g., 244Pu, "Tc, 129I), ICP-MS may be faster and more sensitive than decay
1577 counting. In addition, sample preparation for ICP-MS can avoid some of the analyte separation
JUNE 2001 DRAFT MARLAP
NOT FOR DISTRIBUTION 15-51 DO NOT CITE OR QUOTE
-------
Nuclear Counting Instrumentation
1578 and purification steps required for decay counting, providing an additional dimension of time
1579 savings. Another important feature of ICP-MS is its ability to provide isotopic distribution
1580 information (e.g., 238U vs. 235U). This information is frequently useful in determining the age
1581 and/or origin of materials. (ASTM C758, C759, C799)
1582 The isotopic discrimination capabilities of ICP-MS make possible the calibration technique
1583 known as isotope dilution. In this procedure, a sample is analyzed for one isotope after having
1584 been spiked with a different isotope of the same element (e.g., analysis of 235U might involve
1585 spiking with 233U). The spiked sample is carried through all preparation and analysis steps; in this
1586 way, any matrix or procedural effects that might influence the 235U signal will influence the 234U
1587 signal to precisely the same extent. Final quantization relies on measuring the ratio of unknown
1588 (here the 235U signal) to the known (234U) signal. Isotope dilution is a way of generating highly
1589 precise and accurate data from a mass spectrometer and has been used in the characterization of
1590 many certified reference materials.
1591 Although an ICP-MS instrument is extremely delicate, with proper care and preventive
1592 maintenance system up time should range between 80 to 95 percent. An initial investment of
1593 about $200,000 will be required to obtain a current commercial state-of-the-art system. Annual
1594 maintenance costs will run from $5,000 to $20,000 depending on the purchase of a service
1595 contract.
1596 For more sophisticated measurements, at substantially higher cost, an ICP-MS with magnetic
1597 sector, instead of quadrupole, detection can be applied. Sector instruments are capable of
1598 resolving species of very similar mass. For example, "Tc might be resolved from a
1599 contamination of "Ru with a high-resolution mass spectrometric detector. More typically, high
1600 resolution instruments are employed for their higher signal/noise ratio, and therefore superior
1601 detection limits. A single-collector high-resolution ICP-MS can be purchased for roughly twice
1602 the cost of a quadrupole ICP-MS, or about $300,000. For enhanced sample throughput a
1603 multiple-collector instrument might be purchased for about $500,000. These instruments, like
1604 most analytical equipment, can be expected to require about 2 to 10 percent of their purchase
1605 costs in annual maintenance costs.
1606 Thermal ionization mass spectrometers are available at a cost of $500,000. These instruments
1607 rely not on a plasma for ionization, but rather for thermal ionization from a heated filament. They
1608 provide more precise measurements than routine quadrupole ICP-MS but require substantially
1609 more delicate operator involvement, leading to markedly reduced sample throughput.
DRAFT MARLAP JUNE 2001
DO NOT CITE OR QUOTE 15-52 NOT FOR DISTRIBUTION
-------
Nuclear Counting Instrumentation
1610 Time-of-flight plasma mass spectrometers have just recently appeared on the market; they have
1611 not yet built up a historical record of performance that would permit reliable comparison with the
1612 ICP-MS equipment described above. Likewise, Fourier-transform mass spectrometers are still in
1613 the research phase and cannot yet be considered practical options for routine radiochemical
1614 analysis.
1615 15.10.4.2 Laser
1616 APPLICATION
1617 Lasers can be used to excite uranium (ASTM D5174) and lanthanide complexes in solution.
1618 During or following excitation, the complex relaxes to a lower energy state by emitting photons
1619 of light that can be detected. The amount of light produced is proportional to the uranium or
1620 lanthanide element concentration.
1621 The light emitted can be detected by fluorescence or phosphorescence. With fluorescence and
1622 phosphorescence, the detector is at right angles to Laser excitation. Fluorescence light is emitted
1623 simultaneous to the excitation.
1624 Phosphorescence detecting differs from fluorescence in that the light emitted is not simultaneous
1625 to the excitation. This enables the light source to be pulsed and the measurement to occur when
1626 the Laser source is off. This provides improved signal to noise over fluorescence. The light signal
1627 from organic material will decay promptly, since they have a short relatively lifetime, and not be
1628 available to the detector which is gated off at this initial time. A pulsed nitrogen dye Laser can be
1629 used as the source. Other Lasers can also be used. Chloride ion and other ions may cause
1630 interferences and may need to be removed before measurement.
1631 Kinetic phosphorimetry measures the rate of decay of the uranium or lanthanide element
1632 complex signal. Measurements are taken at fixed time intervals. In aqueous solution, the uranium
1633 or the lanthanide element is complexed to reduce quenching and increase the lifetime of the
1634 complex.
1635 UP/DOWN TIME
1636 Some reagents may have relatively short shelf life and need to be ordered accordingly. The life of
1637 a plasma cartridge is one to three years.
JUNE 2001 DRAFT MARLAP
NOT FOR DISTRIBUTION 15-53 DO NOT CITE OR QUOTE
-------
Nuclear Counting Instrumentation
1638 COST
1639 The initial cost is about $34,500 with a computer and $53,500 with an automatic sample changer.
1640 The cost of replacing a plasma cartridge is $1,400.00. The cartridge lifetime is 1 to 3 years
1641 depending on usage.
1642 15.10.4.3 Radionuclides Analyzed By Neutron Activation
1643 TECHNETIUM-99
1644 Neutron activation analysis methods have been employed since 1972 (Foti et al. 1972a; 1972b).
1645 The method was developed and applied for the analysis of "Tc in mixed fission products (Bate,
1646 1979).
1647 The method employs chemical separation of "Tc from most fission products by a cyclohexanone
1648 extraction from a basic carbonate solution. "Tc is stripped into water by addition of CC14 to the
1649 cyclohexanone phase and then adsorbed on an anion exchange column in a concentrated form.
1650 Neutron irradiation of the isolated "Tc could be made in the pneumatic facility at a high flux
1651 isotope reactor (e.g., at a flux of 5xl014 ng/cm2/sec for approximately 11 seconds. Thus, after
1652 irradiation "Tc is induced to 100Tc, which, because of its 15.8 second half-life, requires an
1653 automatic process to measure its 540 and 591 keV gamma lines.
1654 The lower limit of detection of the analysis under these conditions is approximately 5 ng and
1655 samples up to 100 mL volume can be processed. The method has been applied successfully to
1656 reactor fuel solutions and off-gas traps containing 6.5xlO"4 to 240 jig "Tc/mL.
1657 IODINE-129
1658 Iodine-129 can be determined by neutron activation and subsequent measurement of the
1659 12.4 hour 130I produced by the neutron capture reaction. The method (Bate and Stokely, 1982)
1660 utilizes conventional I valence adjustments and solvent extraction to isolate the I fraction.
1661 Chemically separated 129I is adsorbed on an anion exchange resin before being loaded for
1662 irradiation. With a neutron flux of 5><1014 ng/cm2/s for 100 seconds a lower limit of detection of
1663 0.03 ng can be achieved.
1664 129I also can be determined directly by mass spectrometry (Strebin et al., 1988). The measurement
1665 limit by this technique is approximately 2 femtograms.
DRAFT MARLAP JUNE 2001
DO NOT CITE OR QUOTE 15-54 NOT FOR DISTRIBUTION
-------
Nuclear Counting Instrumentation
1666
1667
1668
1669
1670
1671
1672
1673
1674
1675
1676
1677
1678
1679
1680
1681
1682
1683
1684
1685
1686
Special counting techniques have also
been applied to the analysis of 129I.
Figure 15.13 shows an efficiency plot
using beta-gamma coincidence
counting.
URANIUM, THORIUM, AND PLUTONIUM
Neutron Activation analysis method was
employed to determine uranium in the
hydrogeochemical samples from
Savannah River Plants within the scope
of the National Uranium Resource
Evaluation Program sponsored by DOE.
Uranium was determined by cyclic
activation and delayed neutron counting
of the 235U fission products. The method
relied on the absolute activation
Beta/Gamma (Nal) Coinc.
1-131 Standard
0.450
A JAA .
0.4OO
^^ n ocn _
Q O.oOO
c
2* 0.300
O
0.250
UJ
0.200
.150
04 A A
t
•
i
X.
"— — *_
^^
A — *
^
^*
^-— .
•
•
BetaEff(c/d)
Gamma Eff (c/d)
CoincidEff(c/d)
.100
10 20 30 40 50 60
Weight (mg)
FIGURE 15.13 Beta-gamma coincidence efficiency curve for 129I
techniques using the Savannah River Reactor Activation Facility.
Neutron Activation Analysis followed by delayed-neutron detection was commonly used for
determination of 235U, 239Pu, and 232Th (Echo and Turk, 1957; Hochel, 1979; Alfassi, 1990).
15.11 References
15.11.1 Cited References
1687 American National Standards Institute (ANSI) N42.14. "Calibration and Use of Germanium
1688 Spectrometers for Measurement of Gamma-Ray Emitting Radionuclides," 1991, New York.
1689 American National Standards Institute (ANSI) N42.15. "American National Standard
1690 Performance Verification of Liquid-Scintillation Systems," 1990, New York.
1691 American National Standards Institute (ANSI) N42.22. American National Standard.
1692 Traceability of Radioactive Sources to the National Institute of Standards and Technology
1693 (NIST) and Associated Instrument Quality Control.
JUNE 2001
NOT FOR DISTRIBUTION
15-55
DRAFT MARLAP
DO NOT CITE OR QUOTE
-------
Nuclear Counting Instrumentation
1694 American National Standard Institute (ANSI) N42.23. Measurement and Associated
1695 Instrumentation Quality Assurance for Radioassay Laboratories. 1996.
1696 American National Standard Institute/Institute of Electrical and Electronics Engineers,
1697 (ANSI/IEEE) 325. Standard Test Procedures for Germanium Gamma-Ray Detectors. 1996.
1698 Alfassi, Zeev B. 1990 Use of Delayed Neutrons In Activation Analysis., Activation Analysis, Vol.
1699 I, Edit. Z. Alfassi, CRC Press,, Inc, Boca Raton, Florida.
1700 American Society for Testing and Materials (ASTM) C758. Standard Test Methods for
1701 Chemical, Mass Spectrometric, Spectrochemical, Nuclear, and Radiochemical Analysis of
1702 Nuclear-Grade Plutonium Metal.
1703 American Society for Testing and Materials (ASTM) C759. Standard Test Methods for
1704 Chemical, Mass Spectrometric, Spectrochemical, Nuclear, and Radiochemical Analysis of
1705 Nuclear-Grade Plutonium Nitrate Solutions.
1706 American Society for Testing and Materials (ASTM) C799. Standard Test Methods for
1707 Chemical, Mass Spectrometric, Spectrochemical, Nuclear, and Radiochemical Analysis of
1708 Nuclear-Grade Uranyl Nitrate Solutions.
1709 American Society for Testing and Materials (ASTM) C1207. Standard Test Method for
1710 Nondestructive Assay of Plutonium in Scrap and Waste by Passive Neutron Coincidence
llll Counting.
1712 American Society for Testing and Materials (ASTM) C1316. Standard Test Method for
1713 Nondestructive Assay of Nuclear Material in Scrap and Waste by Passive-Active Neutron
1714 Counting Using a 252.
1715 American Society for Testing and Materials (ASTM) D1890. Standard Test Method for Beta
1716 Particle Radioactivity of Water.
llll American Society for Testing and Materials (ASTM) D3648, Standard Practices for the
1718 Measurement of Radioactivity.
1719 American Society for Testing and Materials (ASTM) D3649. Standard Test Method for High-
1720 Resolution Gamma-Ray Spectrometry of Water.
DRAFT MARLAP JUNE 2001
DO NOT CITE OR QUOTE 15-56 NOT FOR DISTRIBUTION
-------
Nuclear Counting Instrumentation
1721 American Society for Testing and Materials (ASTM) El 81. Standard Test Methods for Detector
1722 Calibration and Analysis ofRadionuclides.
1723 American Society for Testing and Materials (ASTM) E1005. Standard Test Method for
1724 Application and Analysis of Radiometric Monitors for Reactor Vessel Surveillance, E706
1725 (IIIA).
1726 Bate, L. C. 1979. Determination ofTc-99 In Mixed Fission Products By Neutron Activation
1121 Analysis, Radioelements Analysis — Progress and Problems, Ed. W.S. Lyon, Ann Arbor
1728 Science Publishers Inc., p. 175-189.
1729 Bate, L. C., and Stokely, J. R. 1982. "Iodine-129 Separation and Determination By Neutron
1730 Activation Analysis," J. Radioanal. Chem., Vol. 72:1-2, p. 557-570.
1731 Blanchard, R., Kahn, B., and Birkoff, R. D. 1960. "The Preparation of Thin, Uniform
1732 Radioactive Sources by Surface Absorption and Electrodeposition," Health Physics, Vol. 2,
1733 p. 246.
1734 Bogen, D.C., and Welford, G. A. 1971. Application of Liquid Scintillation Spectrometry for
1735 Total Beta and Alpha Assay, Proceedings of the International Symposium of Rapid Methods
1736 for Measuring Radioactivity in the Environment, LAEA-SM-14/3, p. 383.
1737 Browne,E., Firestone, R. B., Shirley, V. S. 1986. Table of Radioactive Isotopes, John Wiley and
1738 Sons, Inc., New York.
1739 Campion, P. J., Taylor, J. G. V., and Merritt, J. S. 1960. "The Efficiency Tracing Technique for
1740 Eliminating Self-Absorption Errors in 4-7i Beta Counting," Int. J. Appl. Radiation and
1741 Isotopes, Vol. 8, p. 8.
1742 Cooper, J. A., Ranticelit, L. A., Perkins, R. W., Hailer, W. A., and Jackson, A. L. 1968. An Anti-
1743 Coincidence Shielded Ge(Li) Gamma-Ray Spectrometer and Its Application to Neutron
1744 Activation Analyses, Report BNWL-SA-2009, Pacific Northwest Laboratory, Richland,
1745 Wash.
1746 Crouthamel, C. E., Adams, F., and Dams, R. 1970. Applied Gamma-Ray Spectrometry,
1747 Pergamon Press, New York, N. Y., 2nd ed.
JUNE 2001 DRAFT MARLAP
NOT FOR DISTRIBUTION 15-57 DO NOT CITE OR QUOTE
-------
Nuclear Counting Instrumentation
1748 Curtis, M. L., Heyd, J. W., Olt, R. G., and Eichelberger, J. F. 1955. Nucleonics, Vol. 13, May, p.
1749 38.
1750 Decker, K. M., Sanderson, C. G. 1992. "A Reevaluation of Commercial IBM PC Software for
1751 the Analysis of Low Level Environment Gamma-Ray Spectra," Appl. Radial Isot., Vol. 43,
1752 No. 1/2, pp.323.
1753 Decker, K. M., Sanderson, C. G., Greenlaw, P. D. 1996. Report of the Department of Energy
1754 Office of Environmental Management Gamma Spectrometry Data Validation Program,
1755 EML-586, Environmental Measurements Laboratory, New York, NY.
1756 DeFilippis, S. 1990. "Activity Analysis in Liquid Scintillation Counting," Radioactivity and
1757 Radiochemistry, Vol.1, No.4, p. 22.
1758 Echo, M. W., and Turk, E. H. 1957. Quantitative Determination of U-235 by delayed neutron
1759 counting, U. S. AEC Report PTR-143.
1760 Flynn, K. F., Glendenin, L. E., and Prodi, V. 1971. Absolute Counting of Low Energy Beta
1761 Emitters Using Liquid Scintillation Counting Techniques in Organic Scintillators and Liquid
1762 Scintillation Counting, D. L. Horrocks and Chim-Tzu Peng, eds., Academic Press, New
1763 York, p. 687.
1764 Foti, S., Delucchi, E., and V. Akamian, V. 1972a. Determination of Picogram Amounts of
1765 Technetium-99 by Neutron Activation Analysis. Anal. Chim. Acta, 60:2, p 261-268.
1766 Foti, S., Delucchi, E., and V. Akamian, V. 1972b. Determination of Picogram Amounts of
1767 Technetium in Environmental Samples by Neutron Activation Analysis. Anal. Chim. Acta,
1768 60:2, p 269-276.
1769 Friedlander, G., Kennedy, J. W., and Miller, J. M. 1964. Nuclear and Radiochemistry, John
1770 Wiley and Sons, New York, 2nd ed.
1771 Hallden, N. A., and Fisenne, I. M. 1963. "Minimizing Self-Absorption in 4-n Counting," Int. J.
1772 Appl. Radiation and Isotopes, Vol. 14, p. 529.
1773 Heath, R. L. 1964. Scintillation Spectrometry Gamma-Ray Spectrum Catalog, IDO 16880, and
1774 ANCR- 1000.
DRAFT MARLAP JUNE 2001
DO NOT CITE OR QUOTE 15-58 NOT FOR DISTRIBUTION
-------
Nuclear Counting Instrumentation
1775 Hensley, W. K., McKinnon, A. D., Miley, H. S., Panisko, M. E., and Savard, R. M. 1997.
1776 SYNTH for Windows., Pacific Northwest National Laboratory, Richland, WA.
1777 Hindman, F. D. 1983. "Neodymium Fluoride Mounting for Spectrometry Determination of
1778 Uranium, Plutonium, and Americium," Anal. Chem., Vol. 55, pp. 2460-2461.
1779 Hochel, R. C. 1979. A High-Capacity Neutron Activation Facility, Radioelements Analysis —
1780 Progress and Problems, Ed. W.S. Lyon, Ann Arbor Science Publishers Inc., p. 343-348.
1781 Hoppes, D. D., 1990. "Demonstrated Measurement Traceability for Nuclear Power
1782 Radiochemistry Departments," Radioact. Radiochem., Vol. , Nol, p9.
1783 Horrocks, D. L. 1964. "Alpha Particle Energy Resolution in a Liquid Scintillator," Review of
1784 Scientific Instruments, Vol. 55, p. 334.
1785 Horrocks, D. L. 1970. Applications of Liquid Scintillation Counting, Academic Press.
1786 International Atomic Energy Agency (IAEA). 1959. Metrology of Radionuclides, Proceedings of
1787 a Symposium, Oct. 14-16, Vienna.
1788 ICRU. 1992. Measurement of Dose Equivalents from External Photon and Electron Radiations.
1789 Report 47.
1790 Kessler, M. 1986. Cerenkov Counting, Packard Application Bulletin, No. 7. Packard Instrument
1791 Co., Downers Grove, IL.
1792 Knoll, Glenn F. 1989. Radiation Detection and Measurement, John Wiley & Sons, New York,
1793 2nd ed.
1794 Lapp, R. E., and Andrews, H. L. 1954. Nuclear Radiation Physics, Prentice Hall, New York,
1795 Englewood Cliffs, N.J., 2nd ed.
1796 McCurdy, D. E., Mellor, R. A., Lambdin, R. W., and McLain, Jr., M. E. 1980. "The Use of
1797 Cuprous Iodide as a Precipitation Matrix in the Radiochemical Determination of Iodine-131,"
1798 Health Physics, Vol. 38, pp. 203-213.
JUNE 2001 DRAFT MARLAP
NOT FOR DISTRIBUTION 15-59 DO NOT CITE OR QUOTE
-------
Nuclear Counting Instrumentation
1799 McCurdy, D. E. and Mellor, R. A. 1981. "The Application of Coincidence Counting Techniques
1800 to the Determination of Radium-226 and Radium-228," Analytical Chemistry, Vol. 53, pp.
1801 2212.
1802 Merritt, J. S., Taylor, J. G. V., and Campion, P. J. 1956. "Self-Absorption in Sources Prepared
1803 for 4-Ti Counting," Canadian Journal of Chemistry., Vol. 37, p. 1109.
1804 Mitchell, D. J. 1986. Sodium Iodide Detector Ananlysis Software (SIDAS), Sandia Report
1805 SAND86-1473, Sandia National Laboratories, Albuquerque, NM.
1806 Mitchell, D. J. 1988. Gamma Detector Response and analysis Software (GADRAS), Sandia
1807 Report SAND88-2519, Sandia National Laboratories, Albuquerque, NM.
1808 Modupe, O. O.,Decker, K. M., Sanderson, C. G. 1993. "Determination of Self-Absorption
1809 Corrections by Computation in Routine Gamma-Ray Spectrometry for Typical
1810 Environmental Samples," Radioactivity & Radiochemistry, Vol. 4, No. 1, pp.38.
1811 Moghissi, A. A. 1971. Low Level Counting by Liquid Scintillation, Organic Scintillators and
1812 Liquid Scintillation Counting, Academic Press, New York, NY.
1813 Nielson, J. M. 1972. Gamma Ray Spectrometry in Physical Methods of Chemistry, A.
1814 Weissberger and B. W., Rossiter, eds. Vol. I, Part 1II D, Chap. X, John Wiley and Sons, Inc.
1815 New York, N.Y.
1816 Nielsen, J. M., and Kornberg, H. A. 1965. Multidimensional Gamma-Ray Spectrometry and Its
1817 Use in Biology, Radioisotope Sample Measurement Techniques in Medicine and Biology.
1818 Proceedings of a Symposium, May 24-28, International Atomic Energy Agency, Vienna, p. 3.
1819 Overman, R. T., and Clark, H. M. 1960. Radioisotope Techniques, McGraw-Hill Book Co., Inc.,
1820 New York, NY.
1821 Passo, Jr., C. J. and Cook, G. T. 1994. Handbook of Environmental Liquid Scintillation
1822 Spectrometry, PACKARD (Canberra) Instrument Company, Meriden, CT.
1823 Perkins, R. W. 1965. "An Anti-coincidence Shielded Multidimensional Gamma-Ray
1824 Spectrometer," Nuclear Instruments and Methods, Vol. 33 p. 71.
1825 Price, W. J. 1964. Nuclear Radiation Detection, McGraw-Hill, New York, N.Y., 2nd ed.
DRAFT MARLAP JUNE 2001
DO NOT CITE OR QUOTE 15-60 NOT FOR DISTRIBUTION
-------
Nuclear Counting Instrumentation
1826 Puphal, K. W., and Olson, D. R. 1972. "Electrodeposition of Alpha Emitting Nuclides from a
1827 Mixed Oxalate-Chloride Electrolyte," Analytical Chemistry., Vol. 44.
1828 Scarpitta, S. C., and Fisenne, I. M. 1996. "Cerenkov Counting as a Complement to Liquid
1829 Scintillation Counting," Appl. Radial Isot. Vol. 47, No.8, pp. 795-800.
1830 Sanderson, C. G. 1969. "Determination of 226Ra and 228Th in Food, Soil, and Biological Ash by
1831 Multidimensional Coincidence Gamma-Ray Spectrometer," Health Physics Vol. 16, No. 6,
1832 pp. 747-753.
1833 Sanderson, C. G. 1988. "An Evaluation of Commercial IBM PC Software for the Analysis of
1834 Low Level Environment Gamma-Ray Spectra," Environment International, Vol. 14, pp.379.
1835 Sanderson, C. G., and Decker, K. M. 1993. "A Mixed Gamma-Ray Standard for Calibrating
1836 Germanium Well Detectors," Radioactivity & Radiochemistry, Vol. 4, No. 2, pp.36.
1837 Sill, C. W., and Williams, R. L. 1981. "Preparation of Actinides for Alpha Spectrometry without
1838 Electrodeposition," Anal. Chem., Vol. 53, pp. 412-415.
1839 Strebin, R. S. Jr., Brauer, F. P., Kaye, J. H., Rapids, M. S., and Stoffels, J. J. 1988. "Neutron
1840 Activation and Mass Spectrometric Measurement of 129I," J. Radioanal. Nucl. Chem.., Vol.
1841 127:1, p. 59-73.
1842 Watt, D. E., and Ramsden, O. 1964. High Sensitivity Counting Techniques, The MacMillan
1843 Company, New York, N. Y.
1844 15.11.2 Other Sources
1845 Bell, C. G., and Hayes, F. N. 1958. Liquid Scintillation Counting., Pergamon Press, New York.
1846 Birks, J. B. 1964. The Theory and Practice of Scintillation Counting, Pergamon Press, New
1847 York.
1848 Bransome, E., Jr., 1970. The Current Status of Liquid Scintillation Counting, (ed.), Greene and
1849 Straiton, New York.
1850 Currie, L. A. 1968. "Limits for Qualitative Detection and Quantitative Determination,"
1851 Analytical Chemistry, Vol. 40, No. 3, pp. 586-593.
JUNE 2001 DRAFT MARLAP
NOT FOR DISTRIBUTION 15-61 DO NOT CITE OR QUOTE
-------
Nuclear Counting Instrumentation
1852 Flynn, K. F., and Glendenin, L. E. 1959. "Half-Life and B-Spectrum of Rubidium-87," Physics
1853 Review, Vol. 116, p. 744.
1854 Flynn, K. F., Glendenin, E., Steinberg, E. P., and Wright, P.M. 1964. "Pulse Height-Energy
1855 Relations for Electrons and a-Particles in a Liquid Scintillator," Nuclear Instruments and
1856 Methods, Vol. 27, p. 13.
1857 Gunnick, R., Colby, L. I, and Cobble, J. W. 1959. Analytical Chemistry, Vol. 31, p. 796.
1858 Harley, J. H., Hallden, N. A., and Fisenne, I. M. 1962. "Beta Scintillation Counting with Thin
1859 Plastic Phosphors," Nucleonics, NUCLA, Vol. 20, p. 59.
1860 ICRP 1994. Gamma-Ray Spectrometryin the Environment, Report 53.
1861 Jarrett, Allen A. Statistical Methods Used in the Measurement of Radioactivity with Some Useful
1862 Graphs and Nomographs, AECU 262.
1863 Katz and Penfelt. 1952. Reviews, Modern Physics, Vol. 24, p. 28.
1864 Lawson and Cork. 1940. Physics Review, Vol. 57, p. 982.
1865 Mitchell, R. F. 1960. "Electrodeposition of Actinide Elements at Tracer Concentrations,"
1866 Analytical Chemistry, Vol. 32, pp. 326-328.
1867 Pate, B. D., and Yaffe, L. 1955. "Disintegration Rate Determination by 4-rc Counting," Canadian
1868 Journal of Chemistry, Vol. 33, pp. 15, 610, 929, and 1656.
1869 Sill, C. W., and Olson, D. G. 1956. "Sources and Prevention of Recoil Contamination of Solid-
1870 State Alpha Detector," Analytical Chemistry, Vol. 42, p. 1956.
1871 U.S. Environmental Protection Agency (EPA). 1972. Environmental Radioactivity Surveillance
1872 Guide, ORP/SID 72-2.
1873 U. S. Government Printing Office (GPO). 1952. Tables for the Analyses ofB Spectra, National
1874 Bureau of Standards Applied Mathematics Series Reports No. 13, Washington, DC.
DRAFT MARLAP JUNE 2001
DO NOT CITE OR QUOTE 15 -62 NOT FOR DISTRIBUTION
-------
Nuclear Counting Instrumentation
1875 ATTACHMENT 15A FIELD MEASUREMENTS
1876 15 A.I Introduction
1877 The scope of environmental remediation work taking place across the country in the public and
1878 private sector has led to a need to improve the speed and cost-effectiveness of measurements for
1879 characterizing contaminant levels at sites and assessing the results of cleanup efforts. In
1880 particular, the time for decisions that are required during soil excavation and waste segregation
1881 should be kept short to avoid delays that tend to increase labor costs. Thus, the time it takes to
1882 collect, prepare and analyze samples can be a limiting factor. To this end, one can use mobile
1883 laboratories at the field site to reduce sample handling and transit times. However, even with
1884 these, the sheer volume of samples can overwhelm processing and analytical capacity. Therefore,
1885 measurements performed directly in the field (in situ) that do not require the collection and
1886 processing of a sample are an attractive alternative. Fundamentally, a field measurement gives
1887 the concentration of a contaminant at the same place where one might otherwise have collected a
1888 sample. In effect, the instrument is brought to the sample rather than the sample to the
1889 instrument. Frequently, the field measurement can be performed within minutes with a result
1890 obtained in what is essentially "real time."
i89i 15A.2 Analytical Level of Measurements
1892 Over the years, field measurements have formed an important component of standard
1893 radiological surveys. Typically, these measurements have comprised scans for gross levels of
1894 alpha or beta/gamma radiation. These types of measurements, particularly where judgment is
1895 used to evaluate a change in an instrument or audible signal, are semi-quantitative in nature and
1896 therefore would be designated at analytical level 1 under the EPA classification system used in
1897 the past or Analytical Support Laboratory(ASL) level A of the American National Standards
1898 Institute (ANSI). These levels reflect the fact that the measurement is intended for screening
1899 purposes.
1900 However, field measurements can be performed at a higher analytical level. For example, an
1901 exposure rate measurement using a pressurized ionization chamber (PIC) is definitive for
1902 assessing the external dose rate from penetrating (gamma) radiation. In this situation, the PIC
1903 provides a direct reading of the desired measurement quantity at the actual point of interest.
JUNE 2001 DRAFT MARLAP
NOT FOR DISTRIBUTION 15-63 DO NOT CITE OR QUOTE
-------
Nuclear Counting Instrumentation
1904 Another example of a field measurement technique that has been used successfully since the
1905 1960s is in situ gamma-ray spectrometry (ICRU, 1994). This technique provides radionuclide-
1906 specific information. In its simplest application, a spectrometer could be used to identify
1907 characteristic peaks in the energy spectrum that would point to the presence of a particular
1908 radionuclide at the measurement location. On a semi-quantitative basis, in situ spectrometry
1909 could serve as screening technique where the relative count rates—in particular spectrum
1910 peaks—are compared among measurement locations. At a higher analytical level, an appropriate
1911 calibration can be performed so that a spectrometer could be used to determine the radionuclide
1912 concentration in the media under study. Since this represents a contaminant-specific
1913 measurement where particular QA/QC checks can be made, it would be classified traditionally at
1914 the data quality objectives (DQO) analytical level 2, or ASL B.
1915 Despite a number of successful applications of in situ spectrometry over the years, issues have
1916 arisen regarding the level of data quality that is obtained with this or any other field measurement
1917 technique for the purposes of demonstrating RCRA, CERCLA, and other regulatory compliance.
1918 In the past, field measurements by definition have not been considered to possess the quality
1919 control that needs to be established at a DQO analytical level of 4 (analogous to ASL D) in the
1920 laboratory. However, the distinction between screening level and higher level measurements is
1921 based on factors relating to data quality, which should be demonstrable. In principle, the rigorous
1922 QA/QC protocols and documentation required for analytical level 4, using EPA Contract
1923 Laboratory Program (CLP) procedures, or ANSI ASL D, could be applied to radionuclide-
1924 specific field measurements. Using field techniques at a higher analytical level is also in keeping
1925 with the latest EPA proposals for performance based measurement systems.
1926 Typically, a projection of cost or time savings using a novel field method leads to its substitution
1927 for a more standard sampling/laboratory analysis method. In doing so, the intended applications
1928 of the field measurement method need to be established clearly. Using the DQO process, the
1929 requisite analytical level can be determined for the data that are to be collected. This analytical
1930 level should then be demonstrated through an objective judgement process whereby the data
1931 quality indicators are critically examined. Included would be those that arise when applying the
1932 DQO process (the "PARCC" parameters: precision, accuracy, representativeness, completeness,
1933 and comparability). Other related indicators or elements which can be broken out separately and
1934 which need to be addressed include documentation, instrument operating conditions, site
1935 conditions, interferences, limitations, calibration procedures, minimum detectable concentrations,
1936 reference measurements, record keeping, quality improvement, and management assessment. The
1937 following sections will provide some discussion on each of these elements as they apply to field
1938 measurement data quality level. Although the discussion is based on experiences with in situ
1939 spectrometry, the elements would generally apply to other field measurement techniques as well.
DRAFT MARLAP JUNE 2001
DO NOT CITE OR QUOTE 15 -64 NOT FOR DISTRIBUTION
-------
Nuclear Counting Instrumentation
1940 It would be expected that a demonstration of the data quality level of a field technique be
1941 performed in concert with regulatory bodies and stakeholders to obtain acceptance.
1942 15A.3 Documentation of Methodology
1943 A field measurement technique, like its counterpart in the laboratory, requires thorough
1944 documentation including the description of apparatus and materials, specification of personnel
1945 training/qualification level, listing of quality control checks, review of safety considerations, and
1946 issuance of non-conformance reports when necessary.
1947 Training materials, equipment manuals, reference texts, articles from technical journals, and
1948 laboratory reports are all potential sources of background information for describing a method. It
1949 would be expected that information be extracted from these sources and a comprehensive report
1950 issued that provides the necessary background and specifics for a particular site and application.
1951 This would essentially take the form of a written procedure. For multiple applications across a
1952 site, further detail may have to be provided in project-specific plans, as the conditions under
1953 which a technique is used may vary among areas. The guidance and recommendations given by
1954 standards groups can also form a key part of documentation. Adherence to these standard
1955 procedures allows one to proceed with some confidence in the measurements process. It is
1956 expected that standards groups will increasingly devote their efforts to field measurements
1957 techniques in the future.
1958 Individuals who will be working with the instruments and data collected need to be qualified.
1959 Educational backgrounds and necessary experience should be determined and appropriate
I960 training given for each area of work. Training and procedure manuals need to track revisions that
1961 invariably result as measurement programs progress.
1962 Quality systems documents would include a general site-wide quality plan with specific factors
1963 like performance tests, pre- and post-operational checks, frequency of calibrations, and replicate
1964 measurements addressed in a separate method-specific quality systems section or document.
1965 Quality systems includes documenting procurement specifications for apparatus and control of
1966 materials and services such as calibration sources. Also, the turnaround time for field
1967 measurements may be important to specify not only for cost and schedule control but for limiting
1968 the time lag between measurements under changing environmental conditions.
1969 Unforseen measurement conditions and unusual equipment malfunctions will lead to situations
1970 where doubt is cast on the validity of a field measurement. Tracking these failures will help to
1971 elucidate the problems over time and a provide a basis for corrective actions and modifications to
JUNE 2001 DRAFT MARLAP
NOT FOR DISTRIBUTION 15-65 DO NOT CITE OR QUOTE
-------
Nuclear Counting Instrumentation
1972 the procedures for future measurements. Situations where obvious bad data is collected despite
1973 the fulfillment of QC elements will require the writing and issuance of a non-conformance report
1974 with subsequent root cause analysis.
1975 15A.4 Instrument Operating Conditions
1976 Specification of instrument operating conditions is fundamental in the field as in the laboratory.
1977 These would include power and cooling requirements as well as an acceptable range in
1978 temperature and humidity conditions. The physical set-up of the instrument, such as a
1979 reproducible sample-detector geometry, also should be specified. For laboratory radioactivity
1980 counting systems, it is generally a planchet, can, bottle or similar small volume where the
1981 distribution of activity within the sample volume is assumed to be homogeneous. For a field
1982 measurement, the sample is in a form such as an area of ground, a storage drum, or wall.
1983 Distances and orientation to the measured area or object need to be specified and held within
1984 control limits.
1985 For a field measurement, the distribution of activity within the volume of measurement should be
1986 considered, since one does not usually have the luxury of mechanical blending as in the case of a
1987 laboratory sample. The field of view of the detector with respect to lateral and depth
1988 displacement within the volume under measurement needs to be established. For large volume
1989 sources, this generally means determining the response of the instrument across all angles or
1990 radiation incidence, not just the front face. While one cannot necessarily control the distribution
1991 of a contaminant, the instrument response needs to be established so that the integrated signal
1992 that it measures can be converted into a meaningful average result over the volume measured and
1993 the sensitivity to non-homogeneous activity distributions determined.
1994 15A.5 Site Conditions/Limitations
1995 Preparing a site for a field measurement is analogous to preparing a sample for analysis.
1996 Procedures need to be followed that will assure that the measurement will yield a valid result.
1997 This might include removing obstructions and accounting for topography and ground cover such
1998 as vegetation or surface water. The radiation absorption properties of the type of soil where
1999 measurements are made may have to be determined beforehand depending upon the energy and
2000 type of radiation being measured.
2001 A significant element which should be addressed in performing field measurements is changes in
2002 the "sample," i.e., changes in the field conditions at the measurement point. For example,
2003 measurements at the same location several days apart may be not be comparable if soil moisture
DRAFT MARLAP JUNE 2001
DO NOT CITE OR QUOTE 15-66 NOT FOR DISTRIBUTION
-------
Nuclear Counting Instrumentation
2004 conditions have changed. Precipitation events would increase soil moisture, while hot, dry
2005 conditions would lead to a decrease in soil moisture.
2006 Depending upon the instrumentation and the physical basis of the measurement, the effects of
2007 such variables as air and soil temperature, humidity, air pressure or related meteorological
2008 parameters may have to be taken into account.
2009 Limitations should be specified for a field measurement technique. They could include site
2010 conditions such as the water content of soil, the degree and type of ground cover, the size of an
2011 area, and the estimated depth of contamination. The radionuclide mix and the concentration level
2012 could also be limiting factors.
2013 15 A.6 Interferences
2014 The effects of interferences need to be assessed for proper QC in a field measurement. As
2015 compared to a laboratory setting where there is generally a controlled environment, adverse
2016 instrument effects may result from extraneous signals or electronic noise that could be produced
2017 by power line or other electromagnetic interference. Interferences in a measurement could also
2018 result from personnel—whether instrument operators or other workers—who enter into a
2019 measurement area and attenuate the measured radiation.
2020 Whereas a laboratory counting system may employ a shield to block out background radiation, a
2021 field measurement system is exposed to ambient radiation. If significant direct or scattered
2022 (shine) radiation is present from extraneous sources, collimation or shadow shielding may be
2023 necessary. In high radiation fields, the effects of ionization in electronic components may present
2024 a problem. In this case the sensor assembly could be kept at the measurement point with the
2025 signal processing and other electronics kept at a distance.
2026 As in the case of laboratory analysis, attention should be given to the mix of radionuclides that
2027 may be present. Interferences can result from the inability to resolve the primary energies emitted
2028 by the nuclides or because there is a high amount of secondary (scattered) radiation present.
2029 15A.7 Calibration
2030 Calibration requires that the instrument response to a known level of measured substance be
2031 determined. This generally takes the form of measuring standard reference materials or samples
2032 spiked with known quantities.
JUNE 2001 DRAFT MARLAP
NOT FOR DISTRIBUTION 15-67 DO NOT CITE OR QUOTE
-------
Nuclear Counting Instrumentation
2033 Direct calibrations using standards or spikes are usually applied to laboratory-based counting
2034 systems since only small quantities are needed for the sample volumes used. For field
2035 measurements, direct calibrations using large volume sources with a known concentration can be
2036 performed as well, although this is generally impractical and potentially expensive. In place of
2037 this, a field calibration factor for a particular source geometry and matrix composition can be
2038 derived using a two-step process. This entails determining the response to incident radiation
2039 (fluence) as a function of energy and angle (by experimental and/or theoretical means) and then
2040 calculating the fluence at the point of measurement from a given source geometry and matrix.
2041 Two-step calibration methods sometimes are applied to laboratory sample counting geometries as
2042 well.
2043 Although a two-step process may be used for field calibrations, traceability still can exist insofar
2044 as certified point or other sources can be used in the calibration process. The calibration factor
2045 may actually represent an integrated response to a collection of sources or a single source at
2046 many different positions. In the case of a spectrometer, calibration points will need to be spaced
2047 out across the energy range of interest. Depending upon saturation effects, the calibration may
2048 also have to extend across a range in concentrations to assess the effects of signal processing
2049 dead time and pulse pile-up.
2050 15A.8 Minimum Detectable Concentrations
2051 Standard to any high quality measurement technique and integral to the DQO process is an a
2052 priori estimate of the detection limits of the measurement system. This needs to be done for a
2053 field measurement technique, although it may be necessary to first obtain preliminary readings in
2054 the area where measurements are to be performed. For example, the minimum detectable
2055 concentration (MDC) for a particular radionuclide will be affected by the continuum of scattered
2056 radiation present in a spectrum from other radionuclides in the soil or from sources of scattered
2057 radiation outside the area under investigation.
2058 In some situations the sensitivity for a given count time can actually be higher for a field
2059 measurement as compared to a laboratory-based sample measurement, thus producing a lower
2060 MDC. This will result when the field detector gives a higher count rate per unit concentration
2061 because there is a far larger sample being analyzed.
DRAFT MARLAP JUNE 2001
DO NOT CITE OR QUOTE 15-68 NOT FOR DISTRIBUTION
-------
Nuclear Counting Instrumentation
2062 15A.9 Precision
2063 The precision of field measurements is determined with replicate measurements as in the case of
2064 laboratory measurements. To avoid potential changes in field conditions, replicate measurements
2065 can be performed sequentially with minimum time lag.
2066 In many cases, a field measurement is a non-destructive technique. Thus, replicate measurements
2067 are easily performed. Using the results from a successive set of measurements (5 to 10), a
2068 standard deviation about the mean can be calculated. This can then be compared to the counting
2069 error for a single measurement that is based on Poisson statistics to assess precision.
2070 Rather than perform many replicate measurements at one point, it can be more instructive to
2071 perform two or more measurements at several different points. The reproducibility can thus be
2072 judged for a variety of site conditions.
2073 15A. 10 Accuracy
2074 Estimates of accuracy for a field measurement can be obtained through uncertainty propagation
2075 just as in the laboratory. Factors to consider include potential bias due to uncertainties in the
2076 calibration source, variations in the assumed sample/detector geometry, uncertainties in the
2077 sample matrix composition, environmental conditions, as well as the statistical counting error.
2078 Overall system accuracy can be checked with comparisons to other techniques, or to results from
2079 an independent organization using the same technique.
2080 15 A. 11 Representativeness
2081 Representativeness refers to the degree to which a measurement reflects the condition at a
2082 location or whether a group of measurements reflects the conditions in a particular area.
2083 Generally, one desires that measurements (or samples) provide a value of a radionuclide
2084 concentration that in turn yields the best dose estimate (and thus risk) to a member of a critical
2085 group for a particular scenario. In order to achieve representativeness, a number of samples or
2086 measurements in a given area would be required in order to achieve a given confidence level or
2087 power using a statistical test.
2088 Representativeness is affected by the heterogeneity of the contaminants in the media under
2089 investigation. Perhaps more than any other factor, field and laboratory measurements may differ
2090 at any particular measurement location due to the effects of heterogeneity. Heterogeneity can
JUNE 2001 DRAFT MARLAP
NOT FOR DISTRIBUTION 15-69 DO NOT CITE OR QUOTE
-------
Nuclear Counting Instrumentation
2091 exist in both the lateral and depth distribution of a contaminant and can take the form of changes
2092 in concentration across various distances: a centimeter or less, as would result from hot particles;
2093 meters, as might occur from dumping and localized spills; and tens or hundreds of meters, as
2094 from up-wind airborne sources. Survey designs incorporate techniques and sample/measurement
2095 densities to accommodate these variations. The number of measurements and the standard
2096 deviation about the mean are fundamental parameters to judge whether the mean concentration
2097 that is measured is within a certain confidence limit. These parameters can be used to compute
2098 the t statistic or applied to other statistical tests.
2099 Where variations in concentration occur on a scale of tens of meters or more, it can be expected
2100 that either field measurements or soil sampling will give similar results. It is where the variations
2101 on the scale of a few meters or less occur that agreement between any particular pair of field
2102 measurement and soil sample results might suffer. However, if the mean concentration in an area
2103 should be determined, a sufficient number of measurements or samples can ultimately yield the
2104 same average result, regardless of where the measurements or samples are taken within the area
2105 under investigation.
2106 Depending upon the objectives of a measurement program, a field method could inherently have
2107 an advantage over discrete sampling. If the viewing area of a field instrument is significantly
2108 larger than the area of a soil sample, a set of field measurement results would tend to show a
2109 smaller standard deviation as compared to a set of soil sample data in a heterogeneous area. The
2110 mean obtained for a given number of measurements would then be more representative of the
2111 true mean. A wide measurement area represented by a field method could also be consistent with
2112 the assumptions of a dose model which averages over a large area.
2ii3 15A.12 Completeness
2114 Measurement losses can occur in the field just as sample losses can occur in the lab. They result
2115 from equipment failure, improper measurement procedures, or environmental factors beyond the
2116 control of operators. Survey designs should incorporate allowances for sample losses by
2117 specifying the collection of more than just the minimum number of samples needed to support a
2118 decision.
2119 There is somewhat of an advantage for a field technique in that QC checks can be performed at
2120 the time of the measurement. Problems can then be immediately identified and the data rejected
2121 on the spot. Another measurement can then be performed in place of the lost measurement.
DRAFT MARLAP JUNE 2001
DO NOT CITE OR QUOTE 15-70 NOT FOR DISTRIBUTION
-------
Nuclear Counting Instrumentation
2122 15A. 13 Comparability
2123 Comparability is a critical factor that readily establishes the validity of a field technique. It can be
2124 established by performing a study in which field measurement results are compared to those
2125 given by an independent technique, such as sampling and laboratory analysis. In some situations,
2126 it may be possible to compare two different field techniques.
2127 In performing a direct comparison study, it is important to establish that the two techniques are
2128 measuring the same thing. For instance, a technique that measures a contaminant concentration in
2129 the surface soil may compare poorly to one that is integrating down to greater depths. This
2130 situation would result where there is a non-uniform concentration depth profile of the
2131 contaminant. Where comparisons are made to soil samples, core depths can be adjusted to better
2132 match the effective viewing depth of the field measurement. The lateral distribution of the
2133 contaminant concentration across the ground could also be a factor. In this situation, compositing
2134 samples may be required to yield a better average with which to compare a field technique.
2135 Other factors to consider for data comparability include the soil moisture and stone content of
2136 soil. Where contaminant concentrations are determined with a field technique, the value is based
2137 on the wet weight of the soil in contrast to laboratory analysis which is performed on a dry
2138 weight basis. Corrections to one of the data sets therefore need to be applied. Similarly, one
2139 should consider the effects of soil sample preparation where large stones are screened out. The
2140 concentration which is then determined is based on the activity associated with the finer particle
2141 content of the soil. If there is little or no activity in the coarser fraction, a concentration for a soil
2142 sample would be higher than that given by a field technique which has averaged in the stone
2143 content.
2144 In place of comparing single field measurement points to single or composited samples, one can
2145 instead compare the averages of sets of field measurements to sets of soil samples over a
2146 particular size area. This would be useful to establish comparability where there is a known
2147 heterogenous distribution of the contaminant and the techniques under comparison are measuring
2148 very different areas of soil.
2149 15A. 14 Reference Measurements
2150 An important QC practice in the laboratory involves the regular analysis of reference materials to
2151 confirm system calibration and performance. In practice, an analogous check can be performed
2152 for measurements in the field. A reference measurement location at a site can be designated as a
2153 field quality control station where routine, perhaps daily, measurements of the contaminants of
JUNE 2001 DRAFT MARLAP
NOT FOR DISTRIBUTION 15-71 DO NOT CITE OR QUOTE
-------
Nuclear Counting Instrumentation
2154 concern are performed. QC charts can be kept which show the results of these measurements and
2155 control limits can be specified accordingly. Unusual trends can then be identified early and
2156 corrective actions taken before unusable data is generated. Measurements at a station such as this
2157 also serve to demonstrate the effects of environmental variables such as temperature and
2158 humidity.
2159 To further qualify a field station, intensive sampling can be performed with laboratory analyses
2160 to determine contaminant concentrations. In this situation, relatively homogeneous conditions
2161 (soil type, contaminant concentration) would make the comparison more favorable and help to
2162 trace any bias between measurement methods that might be observed.
2163 In addition to reference materials, the analysis of blanks is a regular feature of laboratory-based
2164 counting systems. This establishes that contamination of equipment and materials has not
2165 occurred. Similar contamination can occur to field instrumentation such as wind blown soil
2166 particles in crevices, encrusted mud on the underside of equipment, or soil plugs in tripod legs.
2167 For a field measurement technique, it may be possible to check self-contamination by performing
2168 measurements in a background area where the contaminant in the soil is essentially zero. If the
2169 contaminant is present in background, such as 137Cs from nuclear weapons fallout, an offsite area
2170 at least can serve to establish a regional baseline measurement. As a standard measure of
2171 precaution, routine scanning of equipment can be performed with friskers, especially after work
2172 in highly contaminated areas.
2173 15A.15 Record Keeping
2174 Field personnel need to use log sheets or books to record necessary information about the site
2175 conditions, measurement parameters, and data storage. In place of chain of custody forms for
2176 samples, analogous records may be required for data printouts or electronic files of results
2177 (spectral data) obtained in the field as they pass through different levels in the organization (data
2178 entry, data analysis, validation, etc.).
2179 Maintenance logs or files on specific pieces of equipment need to be kept. Factory repairs or in-
2180 house replacement of components should be noted as any changes to an instrument are likely to
2181 require recalibration. Equipment and component failures should also be tracked.
2182 15A.16 Quality Improvement
2183 Operating experience generally leads to fuller knowledge of instrument performance and
2184 characteristics as well as better recognition of precursors to problems. Based on control chart
DRAFT MARLAP JUNE 2001
DO NOT CITE OR QUOTE 15 -72 NOT FOR DISTRIBUTION
-------
Nuclear Counting Instrumentation
2185 records, observations and correlations with other factors associated with a measurement,
2186 breakdown and repair logs, and the information contained in any non-conformance report,
2187 procedures can be modified to improve data recovery and usability. In time, the net effect of
2188 changes incorporated in standard operating procedures will lead to improvements in performance
2189 tests. Along with the identification of limiting factors and the development of solutions, it may
2190 be possible to justify raising the analytical level of the measurement based on the quality control
2191 indicators.
2192 15A.17 Management Assessment
2193 In addition to the quality control elements in place when a field technique is demonstrated
2194 initially, systems need to be in place to insure that data quality is maintained in subsequent
2195 measurements once the technique is used routinely. Deployment of a field methodology on a
2196 broader scope generally entails use by non-experts, i.e., individuals not associated with the
2197 development or implementation of an instrument. For this reason, internal assessments may be
2198 needed in the form of independent oversight (audits). The data verification and validation process
2199 can be used to insure the fulfillment of QC checks. Ultimately, data may have to be reviewed and
2200 approved by individuals who have expertise with the measurement system.
2201 15A. 18 Combined Laboratory and Field Measurements
2202 Laboratory and field measurement techniques are not mutually exclusive. They can frequently be
2203 used in concert to achieve better and more cost-effective radiological surveys. A likely
2204 combination would be reliance on field methods which are faster with the laboratory method
2205 serving as a QC check. Appropriate ratios in the number of field to lab measurements would have
2206 to be established based on expert judgement and by reviewing the data quality objectives. The
2207 ratio could vary from area to area within a site depending upon the situation and the presence of
2208 complicating factors.
2209 15A. 19 References
2210 International Commission on Radiation Units and Measurements (ICRU), 1994. Gamma-Ray
2211 Spectrometry in the Environment, Report 53, Bethesda, MD.
JUNE 2001 DRAFT MARLAP
NOT FOR DISTRIBUTION 15-73 DO NOT CITE OR QUOTE
-------
i 16 INSTRUMENT CALIBRATION AND TEST SOURCE
2 PREPARATION
3 16.1 Introduction
4 This chapter provides guidance on the important functions of radiation detection instrument
5 calibration and test source preparation. In this chapter, the term "test source" will be used to
6 describe the radioactive material prepared to be introduced into a measurement instrument, and
7 "laboratory sample" will be used to identify the material collected for analysis. Thus, a test
8 source is prepared from laboratory sample material for the purpose of determining its radioactive
9 constituents. "Calibration source" is used to indicate that the prepared source is for the purpose
10 of calibrating instruments.
11 The continuing validity of calibrations should be checked on a periodic basis (Chapter 18,
12 Laboratory Quality Control) as specified in a laboratory's quality assurance manual. This is
13 usually done by counting a check source or some secondary standard in an instrument and
14 comparing the results to those previously obtained when the instrument was known to be in
15 calibration. The frequency and other aspects of calibrations and verifications may be specified in
16 project planning documents (Chapter 4, Project Plan Documents) and in analytical statements of
17 work (Chapter 5, Obtaining Laboratory Services).
18 Test sources may be prepared by destructive or nondestructive techniques. A destructive analysis
19 is performed when the original laboratory sample material is altered by ashing or dissolution,
20 which often is followed by chemical separations. Chemical separation usually is necessary when
21 analyzing for specific alpha- or beta-particle emitters. Nondestructive analyses can be used when
22 the laboratory sample is to be analyzed by gamma spectrometry or for gross analyses where the
23 laboratory sample is only dried and counted directly.
24 The requirements placed upon test source preparation are dictated primarily by the type and
25 energy of the radioactivity to be measured (alpha, beta, or gamma), the radiation detector
26 employed, and—to some degree—whether the measurement is simply a gross radioactivity
27 measurement or if specific radionuclide identification is required. The nature of the laboratory
28 sample material also will have an effect on the test source preparation. These are referred to as
29 "matrix effects" and can be caused by both the chemical and physical characteristics of the
30 laboratory sample. When matrix effects are encountered, one is faced with the choice of altering
31 the analysis methodology for that laboratory sample or possibly flagging the result to indicate a
32 high degree of uncertainty.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 16-1 DO NOT CITE OR QUOTE
-------
Instrument Calibration and Test Source Preparation
33 The significant characteristics affecting the bias and precision of radioactivity measurements will
34 be discussed in relation to each type of radioactivity. This includes counting efficiency, which
35 can be affected by the characteristics of the test source as well as those of the radiation detector
36 and the geometry of the source relative to the detector. Also, methods used to prepare radioactive
37 test sources for measurement from chemically separated (isolated) radionuclides will be
38 described.
39 A number of methods and techniques employed to separate and purify radionuclides contained in
40 laboratory samples, particularly in environmental samples, are described in Chapter 14 (Separa-
41 tion Techniques), and sample dissolution is discussed in Chapter 13 (Sample Dissolution).
42 Instruments that will be used to analyze the test sources prepared as outlined in this chapter are
43 described in Chapter 15 (Nuclear Counting Instrumentation). In the case of gross (non-nuclide
44 specific) and nondestructive measurements, chemical separation and purification procedures
45 often are not required. However, to accomplish these measurements, the test source still must be
46 prepared (mounted) in such a manner that the associated radioactivity can be quantified in a
47 reproducible and unbiased manner.
48 16.2 Instrument Calibration
49 Instrument calibrations generally are performed for the purpose of establishing the counting
50 efficiency of an instrument. The counting efficiency establishes the number of disintegrations
51 registered in the detector and electronics of a counting instrument compared to the number
52 emitted by the source. Counting efficiencies are specific to the radionuclide (or energy), the
53 geometrical relationship between the source and detector, and a number of characteristics of the
54 source material, especially those that affect absorption and scattering of the radiation. It is
55 common practice to have several different calibrations on a given detector in order to accommo-
56 date a number of radionuclides, source-to-detector distances, and counting containers that a
57 laboratory will be required to employ in order to meet project requirements for detection
58 sensitivity, specificity, and the variety of media encountered.
59 In cases where the efficiency of the detector varies with energy, it is necessary to perform the
60 calibration at a number of energies and establish an efficiency curve that covers the range of
61 energies to be encountered. Some radiation detection instruments require other types of
62 calibrations. These will be discussed under specific instrument calibrations. Generic issues which
63 govern the conduct of calibrations will be discussed below and instrument and test source
64 specific considerations will be provided in the appropriate sections in this chapter.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 16-2 DRAFT FOR PUBLIC COMMENT
-------
Instrument Calibration and Test Source Preparation
65 16.2.1 Standards
66 Instrument calibration should be performed as needed with only National Institute of Science and
67 Technology (NIST) traceable standards (ANSI N42.23). Calibrations of instruments shall be
68 made using certified reference materials of known and documented value and stated uncertainty.
69 These certified reference materials shall be supplied by:
70 • NIST directly;
71 • A standard source supplier whose measurement capabilities or manufacturing processes are
72 periodically tested by NIST; or
73 • A standard source supplier who documents derived materials with stated uncertainty, and
74 whose value has been verified with analytical and measurement systems that have been tested
75 periodically through an unbroken chain of comparisons to the national standards.
76 16.2.2 Correspondence
77 To assure that the instrument calibration is unbiased, calibration sources must be prepared and
78 counted in a manner that assures that they are virtually identical to the test sources in all respects
79 that could affect the counting efficiency determination (ANSI N42.23). The geometry, including
80 the size and shape of the calibration source and counting container (beaker, planchet, vial, etc.)
81 and source-to-detector distance and alignment, must be controlled. Backscatter, scattering, and
82 self-absorption present during test source counting must be duplicated in the calibration process.
83 The density of the calibration source material should be consistent with that of the test sources.
84 When possible, counting efficiency calibrations should be performed using the radionuclide,
85 whose activity is to be determined in test sources. This may not be possible when the radionuc-
86 lide is not available as a standard reference material or when gross analyses are performed. When
87 the actual radionuclide is not available, a surrogate radionuclide may be selected that has the
88 same type of particle or photon emission (a, p, or y) and a proximate energy. When calibrating an
89 instrument in this manner, corrections must be made for any differences between the decay
90 schemes of the two nuclides.
91 If any factor can vary throughout the test sources, calibrations must be performed which simulate
92 this variability over the range expected to be encountered during test source counting. An
93 example is the necessity to develop a self-absorption curve for alpha or beta counting to account
94 for the changing overall counting efficiency due to absorption in the variable source thickness.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 16-3 DO NOT CITE OR QUOTE
-------
Instrument Calibration and Test Source Preparation
95 16.2.3 Homogeneity
96 The calibration source must be prepared in a manner that assures that the material is uniformly
97 distributed throughout its volume. Any deviation from this requirement can result in a calibration
98 that is biased and contributes to the overall uncertainty of the laboratory results.
99 Liquid calibration sources are more likely to be homogeneous than are solids, particularly those
100 where reference material has been added to a solid material—soil, for example. In order to
101 minimize the overall uncertainty associated with calibration, care should be taken to assure the
102 reference material is thoroughly mixed into the calibration source and distributed uniformly
103 throughout its volume.
104 16.2.4 Uncertainty
105 The total uncertainty of calibration is affected directly by the uncertainty associated with the
106 activity of the reference material used in the calibration source. Furthermore, the uncertainties
107 related to the reproducibility of the counting geometry and the non-homogeneity of the
108 calibration source must be considered. Since the uncertainty associated with these factors is
109 difficult to quantify, it should be minimized.
110 The uncertainty associated with calibration can be reduced by the accumulation of as many
ill counts as practical during the calibration process. The two controllable factors for achieving this
112 are the amount of activity in the calibration source and the counting time allocated for the
113 calibration. As a general rule, at least 10,000 counts should be accumulated during the counting
114 of the calibration source. This may not always be practical when the activity of the calibration
115 source must be limited for reasons listed below.
116 The activity of calibration sources should be limited to an amount that will not lead to significant
117 dead-time losses and random summing in the instrument being calibrated. Unaccounted for,
118 dead-time losses and random summing could lead to an efficiency determination that is biased
119 and artificially low. In addition, one must be aware of the potential for detector contamination,
120 this is particularly true for semiconductor detectors used for alpha spectrometry.
121 16.3 General Test Source Characteristics
122 The goal of test source preparation is to achieve maximum detection capability while introducing
123 minimum bias and uncertainty into the measurement. To realize this goal, test sources must be
MARLAP JULY 2001
DO NOT CITE OR QUOTE 16-4 DRAFT FOR PUBLIC COMMENT
-------
Instrument Calibration and Test Source Preparation
124 prepared in a consistent manner relative to the geometry, disposition of test source material, and
125 the source container.
126 16.3.1 Geometrical Arrangement
127 The geometry of a test source must be suitable for the counting instrument and—particularly—it
128 must be reproducible. The radioactivity associated with test sources is measured in geometries
129 that have been standardized by measuring the instrument response to a known quantity of
130 radioactivity in the identical geometry as the calibration source, to the extent possible. Thus, for
131 this standardization to be accurate over time, the test source geometry must remain constant from
132 source to source and with respect to that of the calibration source. This requirement is necessary
133 for performing quantitative and unbiased measurements of all types of radioactivity and for all
134 types of measurement instruments.
135 16.3.2 Uniformity of Test Source Material
136 Test source uniformity is related to the physical nature of the source material. Uniformity of test
137 source material relative to its thickness, density (which can be influenced by water content), and
138 homogeneity is important. Nonuniformity can result from a variation in the thickness of the test
139 source material over its cross sectional area. If test sources are deposited in a nonuniform
140 manner, absorption characteristics will vary from source to source and acceptable reproducibility
141 may not be achieved.
142 Variation in test source thickness or density can have a particularly large effect in the
143 measurement of alpha-particle activity and, because of their smaller mass and charge, a lesser
144 effect in the measurement of beta-particle activity. Alpha and beta test sources, once prepared,
145 often are stored in a desiccator to maintain a constant moisture content. Test source uniformity is
146 relevant to gamma-ray measurements, not because of the absorption of gamma-rays, but because
147 nonuniformity (non-homogeneity) in the distribution of activity throughout a large source
148 changes the effective detection efficiency. For example, if the gamma-ray emitting radionuclides
149 are concentrated in the portion of the test source container nearest the detector, the counting
150 efficiency will be greater than if the radionuclides were uniformly distributed throughout the test
151 source. Thus, test source uniformity can have a large influence on the counting efficiency by
152 which the activity is detected and measured. Measurements of nonuniform sources are not
153 reproducible; thus, radioactive sources of all types must be homogeneous.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 16-5 DO NOT CITE OR QUOTE
-------
Instrument Calibration and Test Source Preparation
154 16.3.3 Self-Absorption and Scattering
155 Absorption and scattering within the source material are less important when measuring gamma
156 rays than when analyzing for charged particles. Particulate activity emitted in a source can be
157 scattered by elastic and inelastic collisions with nuclei of the source material, degrading the
158 energy of the particle (self-scatter) or—if sufficiently thick—the particle may be absorbed totally
159 by the source (self-absorption). A scattering/self-absorption factor can be used, however, to
160 correct the measured activity to that of an infinitely thin source. For beta counting, this factor is
161 proportional to (1 - e'^/jix, where |i is the linear absorption coefficient for beta particles in the
162 test source material and x is the source thickness (Friedlander and Kennedy, 1955, p. 278).
163 Because of the much smaller mass of beta particles, scattering is more pronounced in sources
164 emitting beta particles than in those emitting alpha particles. Depending on counter geometry,
165 measured beta activity can first increase as the source thickness increases, because of the
166 scattering of electrons out of the source plane and into the detector (Friedlander and Kennedy,
167 1955, pp. 276-278). At greater thicknesses, self-absorption begins to predominate, and the
168 activity eventually approaches a constant value. When this occurs, the source is said to be
169 "infinitely thick." Counting a source at infinite thickness refers to a measurement made with a
170 source thickness such that further increasing the amount of material added would have no effect
171 on the count rate. The minimum source thickness required for this type of measurement clearly is
172 not more than the maximum range R of the particle in the source material, and is often estimated
173 to be 0.75R (Friedlander and Kennedy, 1955, p. 278).
174 To assure that scattering does not lead to bias in test source results, it is important that standard
175 sources prepared for determination of counting efficiency and self-absorption corrections are
176 prepared identically in all aspects that affect absorption to test sources whose activities are to be
177 assayed.
178 Self-absorption increases with the density of the source material and with the size and charge of
179 the emitted particle. Thus, source thickness is of greater concern for measuring alpha particles
180 than for beta-particle emissions and has even less importance in measuring gamma rays, except
181 for low energy x- or gamma rays. Thus, test sources prepared for alpha-particle measurements
182 must be very thin and uniform for maximum detection capability and reproducibility.
183 The moisture content of the source material will affect the density of the source and the
184 absorption characteristics of the source. A change in source moisture content will alter the
185 density and affect the reproducibility of the measurement. Thus, the amount of moisture within
MARLAP JULY 2001
DO NOT CITE OR QUOTE 16-6 DRAFT FOR PUBLIC COMMENT
-------
Instrument Calibration and Test Source Preparation
186 the test source should be controlled. The following procedures often are followed in order to
187 maintain a low and constant moisture content of test sources to be counted.
188 • Test sources prepared by coprecipitation are dried by washing the precipitate first with ethyl
189 alcohol and then with acetone while in the filtering apparatus. Suction to the filter apparatus
190 is continued until the test source is dry. The filter with test source is removed from the
191 filtering apparatus, mounted on a planchet, and stored in a desiccator prior to counting.
192 • Electroplated test sources are dried by heating on a hot plate, in an oven, or under a heat
193 lamp, and then stored in a desiccator until cool and ready to count.
194 • Laboratory samples analyzed nondestructively are usually dried prior to measurement in
195 order to control moisture content and help ensure that test source characteristics are
196 reproducible. Laboratory samples, such as soil, biota, vegetation, etc., are usually dried in an
197 oven. Test sources not counted immediately, including those for gross alpha and beta
198 measurements, as well as for gamma-ray spectroscopy, should be desiccated to maintain a
199 constant moisture content.
200 • Evaporated test sources also are stored in a desiccator, after flaming, to maintain a constant
201 moisture content.
202 Another concern in measuring both alpha and beta particles from deposited test sources is back-
203 scattering: the scattering of particles from the source-mount back through the test source material
204 and into the sensitive part of the detector. Back-scattered beta particles have degraded energies,
205 but can have the apparent effect of increasing the counting efficiency. This may seem to have the
206 desired effect of improving the overall counting efficiency; however, the percent of back-
207 scattered beta particles from the test source must remain constant and be identical to that of the
208 standard source. The magnitude of backscatter is dependent on the beta-particle energy and the
209 thickness, density, and atomic number of the backing material (Faires and Boswell, 1981, p. 220-
210 222). Thus, to reduce the effect of backscatter on beta-particle measurements, the test source
211 often is mounted on a thin, low Z (atomic number), low density material, as for example
212 aluminum foil or thin organic films (Blanchard et al., 1960). For very precise measurements, a
213 conducting metal film is vaporized onto the organic film so that any electrical charge build up
214 due to the emission of charged particles can be eliminated.
215 As with absorption, backscatter increases with the thickness of the scattering material up to a
216 saturation level, beyond which it remains constant. The saturation level is reached at a thickness
217 that is about one-third the maximum range of the scattered particle (Faires and Boswell, 1981, p.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 16-7 DO NOT CITE OR QUOTE
-------
Instrument Calibration and Test Source Preparation
218 221). Therefore, due to the dependency of backscatter on atomic number and thickness, the
219 backing used for the standard source must be identical to that used for the test source mount. For
220 example, if the presence of HC1 in the test source requires changing from an aluminum planchet
221 to platinum, a platinum backing must also be used in counting the standard source.
222 16.3.4 Counting Planchets
223 A wide variety of planchets made of platinum, nickel, aluminum, and stainless steel can be
224 obtained in various sizes. It is normally not of great importance which type is used as long as
225 several factors are considered (PHS, 1967, p. 20). Some factors that should be considered in
226 selecting a planchet are:
227 • Chemical reactivity. The metal planchet must be inert to the chemicals in the test source, as
228 corrosion of the planchet surface radically alters test source absorption and geometry
229 characteristics.
230 • Radioactivity. The metal comprising the planchet should contain minimal radioactivity and,
231 although this is generally not a serious problem, the planchet background shall be measured.
232 • Size. Two-inch planchets (assuming the detector is at least that large) are often preferred for
233 gross alpha^eta counting to expedite and simplify the evaporation of liquid samples and
234 provide a greater surface area for solid samples, while 1-inch planchets are generally used for
235 alpha spectrometry test samples.
236 • Cost. Platinum planchets should not be used if stainless-steel ones are adequate for the
237 purpose.
238 It is usually impractical to reuse planchets, and it is generally not recommended. Except for those
239 made of platinum, planchets are inexpensive, and it is not cost effective to clean the planchets
240 and insure they are not contaminated from the prior test source. Platinum planchets are quite
241 expensive and usually can be cleaned effectively in acid and recounted prior to reuse to insure
242 that they are not contaminated.
243 16.4 Test Source Preparation and Calibration for Alpha Measurements
244 Several types of instruments are used for counting alpha particles (Chapter 15, Nuclear Counting
245 Instrumentation). Each type of instrument has characteristics that affect preparation and
246 mounting of sources. Similarly, these characteristics also affect the calibration of the instrument.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 16-8 DRAFT FOR PUBLIC COMMENT
-------
Instrument Calibration and Test Source Preparation
247 This section discusses the attributes of commonly used instruments and their effects on test
248 source and standard source preparation.
249 16.4.1 Proportional Counters
250 Proportional counters (Section 15.2.2.1) often are used to measure alpha particles, particularly
251 when gross analyses are desired. Proportional counters may be "internal," where the test source is
252 placed into the detector or "windowed," where a thin window covers a part of the detector and
253 separates the source from the detector.
254 16.4.1.1 Alpha Test Source Preparation
255 Test sources for proportional counters are usually prepared by electrodeposition, coprecipitation,
256 or evaporation, as described below in Section 16.7.6. For internal counters, since the source is
257 placed within the detector, care must be exercised in test source preparation to avoid the
258 inclusion of chemicals which may react with the detector materials. Likewise, any spillage of test
259 source material can result in contamination of the detector.
260 The absorption of alpha particles in the source material (self-absorption) is quite important when
261 using proportional counters, or other ionization counters, and must be addressed when preparing
262 a test source for counting. Self-absorption is primarily a function of source thickness (ts) and the
263 range (Rs) of the alpha particles in the source material. For a uniformly thick source, the fraction
264 of alpha particles absorbed by the source increases proportionately to 1/2^, when ts < R, (NCRP,
265 1978, pp. 104-105). Thus, to approach absolute counting in either 2n or 4n counting geometries,
266 test sources should be prepared as thinly and uniformly as possible.
267 Another method sometimes used for alpha-emitting test sources in ionization counters is to
268 perform the count at infinite thickness (Section 16.3.3). The count rate of a test source at infinite
269 thickness usually is related to the count rate of a standard source prepared and measured in the
270 exactly the same manner.
271 Backscatter from alpha sources increases with the atomic number of the backing or source
272 material and with decreasing alpha energy (NAS/NRC, 1962, p. 115). Scattering of alpha
273 particles from the source material itself is not a significant problem, and scattering from the
274 source backing has only a small affect for very thin sources (NCRP, 1978, p. 107). When
275 stainless-steel planchets are used, the increase in a count rate because of alpha backscatter is only
276 about 2 percent (PHS, 1967, p. 19).
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 16-9 DO NOT CITE OR QUOTE
-------
Instrument Calibration and Test Source Preparation
277 16.4.1.2 Proportional Counter Calibration — Alpha
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
Calibration sources prepared for calibrating counters for a specific nuclide measurement shall
contain a radionuclide of similar alpha energy and be measured under identical conditions as the
test sources to be measured (ASTM D3648). A variety of radionuclides have been recommended
for calibrating for gross alpha analyses (Table 16.1).
TABLE 16.1— Nuclides for alpha calibration
Purpose
Specific Nuclide and Gross Alpha
Gross Alpha
Gross Alpha
Gross Alpha
Nuclide
239Pu, 241Am, 210Po, 228Th, 226Ra, 233U,
235U,andUmt
241 Am
241 Am, 237Np, and Umt
241Am,239Pu,230Th,andUmt
Reference
ASTM D3648
EPA, 1980
ASTM D 1943
APHA (1995), Method 7110
To the extent possible, standard sources should be prepared in a manner identical to the method
used for test source mounting. The counting efficiency (e) is then determined by counting the
standard source for a sufficient time to accumulate approximately 10,000 counts and dividing the
derived counts per second (cps) by the a emission rate of source in disintegrations per second
(dps).
dps
In cases where finite test source thicknesses are unavoidable, alpha-source counts can be adjusted
to account for self-absorption (PHS, 1967, p. 19). This requires that a self-absorption curve be
prepared in order to determine the change in counting efficiency as a function of source thickness
or mass. Standard sources containing a known amount of the radionuclide of interest are prepared
in varying thicknesses (mass) and counted. Absorption curves for gross alpha-particle measure-
ments most often are constructed using reference material containing one of the nuclides listed
above. The absorption curve is constructed by counting planchets containing varying mass of
material but with constant added radioactivity. A curve is generated by plotting the efficiency at a
given source thickness divided by the efficiency at "zero" thickness versus source mass (mg) or
density thickness in jig/cm2 or mg/cm2 (NCRP, 1978, p. 105). Thus, the efficiency relative to the
"zero thickness" efficiency can be read directly from this curve for any measured test source
thickness. Test sources prepared for gross measurement are counted in the exact geometry as
those used to prepare the absorption curve. The material forming the matrix for the self-
absorption standard source should, when possible, be identical to that expected in the test sources
to be analyzed. Based on the test source mass or density thickness in units of |ig/cm2 or mg/cm2,
MARLAP
DO NOT CITE OR QUOTE
16-10
JULY 2001
DRAFT FOR PUBLIC COMMENT
-------
Instrument Calibration and Test Source Preparation
309 the correction factor determined from the absorption curve is applied to the test source count,
310 yielding the count rate equivalent to an infinitely thin source.
311 Most modern proportional counters are capable of simultaneous alpha and beta counting. This is
312 accomplished by identifying the two types of particles based on their pulse height. Those pulses
313 whose heights exceed an experimentally established discriminator level are registered as alpha
314 counts and those falling below this level are recorded as beta counts. Some fraction (usually less
315 than 10 percent for a weightless source) of the alpha particles is recorded as betas, even for
316 nearly weightless test sources. This fraction increases as the thickness (mass) of the source
317 increases. A much smaller (often insignificant) fraction of the beta interactions are registered as
318 alphas. This misclassification of alpha and beta counts is referred to as "crosstalk."
319 For simultaneous alpha and beta counting, corrections must be made to the beta count rate to
320 remove the portion contributed by alpha particles. Since the fraction of alpha counts occurring in
321 the beta channel is a function of the source mass, a crosstalk curve relating the fraction of alpha
322 particles counted as beta to source mass must be developed. This can be accomplished
323 concurrently with the self-absorption calibration if the radionuclide selected is an alpha emitter
324 only—no beta particles. This is done by recording the beta counts from the alpha self-absorption
325 determination at all source weights and plotting the fraction (beta counts/alpha + beta counts) as
326 a function of source mass (Section 17.4.1). Beta count rates then can be corrected for the
327 influence of the alpha particles at all source thicknesses.
328 16.4.2 ZnS(Ag) Scintillation Counter
329 This type of counter is discussed in Section 15.2.2.3. Because the alpha particle must be emitted
330 from the source and interact with the screen, as it does with the ionization chamber of an internal
331 proportional counter, the previous description concerning self-absorption and scatter of alpha
332 particles during analysis in an internal proportional counter may be applied to counting alpha
333 particles with a ZnS(Ag) scintillation counter. Additional advantages of this counting
334 arrangement are the very low backgrounds that are achievable and the small potential for
335 permanently contaminating the counter, because the zinc sulfide screens can be replaced.
336 A source mount shaped like a washer, with one side enclosed with a transparent ZnS(Ag) screen,
337 is an arrangement often used. The test source to be counted is placed in the hole of the "washer,"
338 in contact with the ZnS(Ag) screen. The other side of the test source mount is sealed, generally
339 with wide transparent tape, securing the test source within the source mount. The test source is
340 then placed on an appropriately sized photomultiplier tube and counted. Because of the
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 16-11 DO NOT CITE OR QUOTE
-------
Instrument Calibration and Test Source Preparation
341 availability of large photomultiplier tubes, sources up to 5 inches in diameter can be prepared for
342 measurement (PHS, 1967, p. 26).
343 The considerations related to alpha calibrations, discussed above under proportional counters,
344 apply equally to scintillation counter calibration.
345 16.4.3 Alpha Spectrometry With Semiconductor Detectors
346 Semiconductor detectors for alpha particle counting are discussed in Section 15.2.2.5. Alpha-
347 energy spectra of very high resolution are attainable with semiconductor detectors if the prepared
348 test source is essentially weightless, < 1 |ig/mm2 (Helpers, 1986, pp. 143-145). As the thickness
349 of the test source increases, the spectral energy is degraded due to self-absorption, which
350 broadens the peak and forms a "tail" on the lower-energy side (Chapter 17). The alpha-energy
351 spectral degradation will increase, as the source thickness increases, raising the possibility of
352 overlapping peaks with a loss of spectrum integrity. Thus, it is of utmost importance to prepare
353 very thin and uniform alpha test sources for spectrometry. This may be accomplished by
354 electrodeposition or coprecipitation (ASTM, D3084), if reagents are controlled so that only small
355 (milligram) quantities of precipitate are recovered (Sections 16.6.1 and 16.7.2). For example, in
356 the coprecipitation of actinide test sources for spectral analysis, source thicknesses of 0.4 to 1
357 |ig/mm2 (0.04-0.1 mg/cm2) are routinely achieved, which is quite adequate for producing well-
358 defined alpha spectral peaks (EPA, 1984a).
359 Semiconductor detectors used for alpha spectrometry require both efficiency and energy
360 calibrations. Calibration sources, traceable to NIST, often are prepared with multiple
361 radionuclides so they may be used for both types of calibration (ASTM D3084). Sources
362 containing 234U, 238U, 239Pu, and 241Am have been used for this purpose. When mixed-nuclide
363 calibration sources are used, the average counting efficiency is often calculated using the
364 efficiencies of the individual radionuclides. Some alpha spectrometry analysis programs calculate
365 an average efficiency where the individual radionuclide efficiency is weighted by the uncertainty
366 in its determination. Other radionuclide combinations may be used, but in addition to the
367 requirement for traceability for the disintegration value, the energies of the radionuclides must be
368 known with a high degree of certainty.
369 Calibration sources may be prepared by either electrodeposition or coprecipitation. Due to their
370 durability and stability, electrodeposited calibration sources are often chosen. It is important that
371 the area of deposition be consistent with that of test sources to be counted and that there are no
372 significant impurities present (ASTM D3084). See the additional discussion on alpha
373 spectrometer calibration in Section 17.3.2.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 16-12 DRAFT FOR PUBLIC COMMENT
-------
Instrument Calibration and Test Source Preparation
374 16.4.4 Liquid-Scintillation Spectrometer
375 With proper scintillators, liquid scintillation can be used to measure alpha-particle emitters
376 (Passo and Cook, 1994) (Section 15.2.2.4). Although the relatively high background of liquid
377 scintillation counting restricts the sensitivity relative to other counting techniques, e.g., internal
378 proportional counting or the use of ZnS(Ag) screens, the ease of source preparation and the
379 nearly 100 percent counting efficiency are advantages often exploited (Hemingway, 1975, p.
380 146). The separation of alpha- and beta-particle counts attained in the spectrometer can be
381 enhanced by proper scintillator choice. Ultima Gold AB™ was designed specifically to maximize
382 alpha/beta separation in aqueous solutions and, in other studies, poor alpha/beta separation has
383 been overcome by making the standard cocktail 20 percent in naphthalene (Passo and Cook,
384 1994, pp. 3-11 to 3-12). It is believed that naphthalene improves the alpha/beta separation by
385 acting as an intermediate in the energy transfer process between the solvent and the fluor
386 (McDowell, 1986).
387 EPA's (1978) recommended procedure for measuring 222Rn in water uses liquid scintillation
388 counting. The protocol is based on the solubility of radon in a number of scintillators. To
389 measure radon in air, the radon is first adsorbed onto activated charcoal and then mixed with an
390 appropriate scintillator and counted (EPA, 1987; Passo and Cook, 1994, pp. 8-5 to 8-10).
391 Utilizing the high solubility of 222Rn in organic solvents, concentrations of 222Rn in air have been
392 determined by bubbling air through the scintillator in a scintillation vial (Amano et al., 1985).
393 Concentration of 222Rn, determined by liquid scintillation, also can be used in the measurement of
394 its parent, 226Ra.
395 Some actinides (U and Th) and transuranics (Np, Pu, Am, and Cm) have been measured by a
396 procedure that involves "Extraction Scintillation Techniques" (Passo and Cook, 1994, pp. 6-1 to
397 6-2 and 13-1 to 13-6). An extraction agent, e.g., bis(2-ethylhexyl) phosphoric acid (HDEHP), is
398 mixed either with a toluene or a di-isopropylnaphthalene (DIN) based cocktail. The alpha
399 emitter, in the aqueous laboratory sample, is extracted into the scintillation mixture and counted
400 by liquid scintillation. The discussion in Section 16.5.2.1 can be applied to both alpha and beta
401 particles.
402 16.5 Characteristics of Sources for Beta Measurements
403 16.5.1 Proportional Counters
404 Beta decay generally is accompanied by gamma-ray emission; the latter normally is much easier
405 to identify and quantify. Beta-particle counting typically is more difficult, due to the additional
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 16-13 DO NOT CITE OR QUOTE
-------
Instrument Calibration and Test Source Preparation
406 source preparation and associated complications resulting from the effects of backscatter,
407 scattering, and absorption in the source material (NAS/NRC, 1962, p. 118-119). Beta particles
408 are not emitted monoenergetically and may result in additional difficulty in quantitative
409 measurements.
410 Beta counting in ionization-type counters often is used after chemical separations are performed
411 to isolate the beta-emitting radionuclide of interest from other radionuclides. Beta measurements
412 are performed on chemically isolated pure beta emitters (beta decay not accompanied by a
413 gamma-ray) and also in cases when increased sensitivities are required to meet detection limits,
414 such as, 89Sr, 90Sr, "Tc, 131I, 134Cs, and 137Cs (EPA, 1980). The proportional counter often is used
415 for measuring these beta-particle emitters. Test sources measured in a proportional counter are
416 usually prepared by electrodeposition, coprecipitation, or evaporation, as described below in
417 Section 16.7 (Blanchard et al., 1960). The comments on chemical reactivity of source contained
418 materials and contamination given in Section 16.3.1, apply here.
419 16.5.1.1 Beta Test Source Preparation
420 Although it remains a consideration, self-absorption of beta particles is not as pronounced as
421 with alpha particles, because the charge and mass of beta particles are significantly smaller.
422 Scattering, and particularly backscatter from the source mount, is much more pronounced for
423 beta counting than for alpha counting (Blanchard et al., 1957). To reduce scatter, plastic
424 mountings are often used to mount sources for beta counting (EPA, 1980). The effects resulting
425 from self-absorption and scattering can be minimized by preparing test sources in a standardized
426 constant thickness, or using a correction factor based on an empirical calibration curve for
427 different thicknesses (Friedlander and Kennedy, 1955, pp. 276-277; Tsoulfanidis, 1983, pp. 133-
428 134). (Section 16.3.3.)
429 For sufficiently thick sources, the beta particles emitted from the source reach a limit, and the
430 count rate becomes independent of the source thickness.
431 16.5.1.2 Proportional Counter Calibration — Beta
432 As in other calibrations, proportional counters used for beta-particle analysis shall be calibrated
433 with NIST traceable standards in a manner that is totally consistent with the counting of test
434 sources. When possible, the radionuclide to be quantified should be used as the calibration
435 source. For gross beta analysis, the radionuclides presented in Table 16.2 have been
436 recommended for calibration sources.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 16-14 DRAFT FOR PUBLIC COMMENT
-------
Instrument Calibration and Test Source Preparation
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
TABLE 16.2 — Nuclides for beta calibration
Purpose
Gross Beta
Gross Beta
Gross Beta
Gross Beta
Nuclide
137Cs
137Cs
137Cs
137Cs and 90Sr-90Y
Reference
ASTM D3648
EPA, 1980
ASTM D 1890
APHA (1995), Method 71 10
If test sources of varying mass are to be counted for beta activity determination, a self-absorption
curve must be prepared. The method used is identical to that described under alpha calibration
for proportional counters, except that a beta-emitting reference material is used instead of alpha.
16.5.2 Liquid-Scintillation Spectrometers
When beta measurements are required, especially those involving pure beta emitters of low
energy, they are often performed in a liquid scintillation spectrometer, because self-absorption
and backscatter are eliminated and counting efficiencies are relatively high (Herpers, 1986, pp.
133-135). Although it is the preferred instrument to measure low-energy, pure beta-emitting
radionuclides, e.g., 3H, 14C, and 35S, it is a well-established procedure for measuring numerous
other beta-emitting radionuclides, including 45Ca, 65Zn, 141Ce, 60Co, 84Sr, 55Fe, 87Rb, 147Pm, and
36C1 (Hemingway, 1975, pp. 145-146). The liquid scintillation spectrometer, applied to beta-
particle measurements, is described in detail in Section 15.3.3.
Tritium is the radionuclide most often measured by liquid scintillation counting (DOE, 1997;
EPA 1979; Lieberman and Moghissi, 1970, p. 319). The primary step in preparing water samples
for counting is distillation in the presence of an oxidizing agent, such as KMnO4, to separate the
tritium labeled water from dissolved solids, including interfering radionuclides, and any organic
material that may be present. An aliquant of the distillate is then mixed with a liquid scintillator
and counted in a liquid scintillation spectrometer. To measure tritium in samples of other
matrices, the water in the sample can be removed and collected by distillation as an azeotrope
with, for example, /7-hexane or cyclohexane (Moghissi, 1981; EPA, 1979). An aliquant of the
water collected is then mixed with a liquid scintillator and counted, as described above for water
samples.
Tritium can be concentrated in a sample of water if lower detection limits are required. The
concentration process, electrolysis, uses the isotopic effect caused by the large mass difference
(three times) between JH and 3H (DOE, 1997; EPA, 1984a). Tritium becomes enriched as
electrolysis continues. Generally, 50 mL of the laboratory sample is placed in an electrolysis cell
and a current of about three amps applied. Electrolysis is continued until the volume reaches
JULY 2001
DRAFT FOR PUBLIC COMMENT
16-15
MARLAP
DO NOT CITE OR QUOTE
-------
Instrument Calibration and Test Source Preparation
470 about 5 mL. More sample can be added to the cell during the electrolysis, if greater sensitivity is
471 necessary for the measurement. The concentrated laboratory sample is then distilled in the
472 presence of an oxidizing agent, such as KMnO4, and treated like a water sample (see above).
473 16.5.2.1 Liquid Scintillation Test Source Preparation
474 The preparation of a laboratory sample for a liquid-scintillation spectrometer usually is relatively
475 simple and fast. The radionuclide to be measured is isolated in a solution, which is then
476 introduced into and thoroughly mixed with one of a variety of ready-to-use commercially
477 available liquid scintillators. This mixture is often referred to as a scintillation "cocktail." The
478 liquid scintillator is an emulsion system, usually consisting of an aromatic solvent containing the
479 appropriate scintillator mixed with a detergent (NCRP, 1978, pp. 168-169). If a sample is
480 insoluble in the scintillator, it can be ground to a fine powder, stirred into the scintillator until a
481 homogeneous mixture is formed, and solidified with a gelling agent (Friedlander et al., 1981, p.
482 303).
483 Because much or our ecosystem consists of materials composed of carbon and hydrogen, the
484 measurement of 3H and 14C levels in biological materials is important. Water, for 3H analysis, can
485 be recovered efficiently from all types of environmental and biological samples by azeotropic
486 distillation. The laboratory sample is distilled with a hydrocarbon, such as benzene or
487 cyclohexane, which is compatible with the liquid scintillation process (Moghissi et al., 1973;
488 Moghissi, 1981). The distillate is mixed with the proper scintillator and counted in a liquid
489 scintillation counter. Tritium has been successfully measured by this technique in such samples
490 as animal and human tissues, soil, hay, grass, urine, and milk.
491 Environmental and biological samples also can be analyzed for total 3H (that contained in both
492 the water and fibrous fractions) by quantitatively combusting the laboratory sample, collecting
493 the water formed, and analyzing it by liquid scintillation spectrometry (DOE, 1997). In another
494 case, both 3H and 14C can be measured simultaneously (EPA, 1984b). The laboratory sample first
495 is freeze-dried to remove and collect the water fraction. The tritium in the water is measured
496 directly by liquid scintillation spectrometry. The fibrous (freeze-dried) material is combusted and
497 the H2O and CO2 are collected. As before, the 3H in the water is measured directly by liquid
498 scintillation spectrometry, while the 14C is first converted to benzene or captured as CO2 and then
499 counted by liquid scintillation spectrometry.
500 A primary problem with measurements using a liquid-scintillation spectrometer is "quenching."
501 Quenching occurs when the production of light is inhibited or the light signal is partially
502 absorbed during the light transfer process by a substance in the liquid. The two basic types are
MARLAP JULY 2001
DO NOT CITE OR QUOTE 16-16 DRAFT FOR PUBLIC COMMENT
-------
Instrument Calibration and Test Source Preparation
503 chemical and color quenching. Some of the stronger chemical quenchers are alkyl bromides,
504 iodides, nitrates, mercaptains, and ketones (NCRP, 1978, p. 46). Color quenching involves the
505 reduction of light transmission through the solution to the cathode of the phototube by the
506 absorption of the light photons. The two techniques most often used to correct for quenching
507 involve the use of internal or external standards.
508 Chemiluminescence, the production of light by a chemical reaction, can be troublesome in liquid-
509 scintillation counting. However, the duration of chemiluminescence is generally short, and a wait
510 of a few minutes after mixing the reagents will allow the effect to dissipate before counting
511 starts. Similarly, phosphorescence, the emission of light from certain chemicals caused by
512 exposure to light, will cease a short time after being placed in the dark. This is referred to as
513 "dark adapted" (Faires and Boswell, 1981, p. 182).
514 16.5.2.1 Liquid-Scintillation Spectrometer Calibration
515 When the quenching of a group of test sources is predictable, e.g., distilled drinking water (EPA,
516 1980; ASTM D4107), a counting efficiency is determined for the group by placing a known
517 quantity of reference material in the source medium and scintillation solution under identical
518 conditions (vials and volumes) as the sample medium.
519 Except for test sources with very predictable amounts of quenching, it may be necessary to
520 determine a counting efficiency for each laboratory sample. Two methods of determining
521 counting efficiency are available: internal standardization and external standardization (NCRP,
522 1978).
523 Internal standardization for quench correction is by the method of standard additions. This
524 involves the counting of two aliquants of a sample, one being the sample and the other is an
525 identical aliquant that has been spiked with a known amount of the radionuclide being
526 determined. The degree of quench can then be determined from the spiked aliquant and applied
527 to the unspiked sample (DOE, 1995). This method does not require a curve for correction but
528 decreases throughput because two test source counts are required. For these reasons, the use of an
529 external standard is the more widely used technique to correct for quenching (Horrocks, 1973).
530 One external standard method is also called the "external-standard channels-ratio" (Baillie, 1960;
531 Higashimura et al., 1962). In this method, a series of vials is prepared containing a known
532 amount of reference material and varying amounts of the medium being evaluated. Windows in
533 the energy spectrum are set for a high- and low-energy region. The vials are counted and the
534 ratios of low-to-high count rates are recorded for each quenched source. A quench curve is then
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 16-17 DO NOT CITE OR QUOTE
-------
Instrument Calibration and Test Source Preparation
535 prepared by plotting the ratios of low-to-high energies as a function of counting efficiency. The
536 efficiency of an unknown test source can then be determined from its low-to-high energy ratio
537 during counting.
538 The second external-standard method employs an external gamma-ray source that generates
539 Compton electrons in the scintillation solution. Count rates from the external source are
540 determined for a set of sources whose efficiency is known from the internal-standard method. A
541 quench curve is then prepared by plotting the external count rate vs. counting efficiency.
542 The external-standard methods should not be generalized beyond use for the media conditions
543 under which they were prepared.
544 16.6 Characteristics of Sources for Gamma-Ray Measurements
545 Backscatter and self-absorption, which must be addressed when measuring alpha and beta
546 emissions, cause less uncertainty in the measurement of most gamma-ray emitters. This is
547 because the penetrating nature of gamma rays is totally different from that of particles. For thick
548 samples or high-Z matrices, a detection-efficiency correction is necessary for low-energy photons
549 (especially below 200 keV) due to the self-absorption of photons in the sample. There is,
550 however, some backscatter of gamma-rays from the shield surrounding the detector, which
551 produces a small peak at about 200 keV (NAS/NRC, 1962, p. 32).
552 16.6.1 Gamma Test Source Preparation
553 No significant precautions usually are required in preparing test sources for gamma-ray
554 spectrometry, as long as the test source is homogenous and positioned reproducibly relative to
555 the detector. Although source properties (e.g., density and moisture content) are not as important
556 in gamma-ray spectrometry as in alpha or beta measurements, test source preparation for gamma
557 measurements may still include drying and ashing to control moisture content and to reduce the
558 test source size. Homogeneity of the test source can be attained by thoroughly mixing laboratory
559 samples that have been ashed (many combustible matrices not containing volatile radionuclides
560 are ashed), by grinding and mixing solids (e.g., soils and sediments), or by finely chopping and
561 mixing fresh vegetation. Also, calibrations are generally conducted using standard sources with
562 identical counting geometries and the same or similar matrices as the test source for analysis.
563 Important considerations in preparing test sources for gamma-ray spectrometry are geometry
564 (shape), size, and homogeneity (uniformity) of the source. Test sources can be in any
565 reproducible shape or size, but the radionuclides must be uniformly distributed throughout. A
MARLAP JULY 2001
DO NOT CITE OR QUOTE 16-18 DRAFT FOR PUBLIC COMMENT
-------
Instrument Calibration and Test Source Preparation
566 counting container that allows the source to surround the detector, thus maximizing the
567 geometrical efficiency, is referred to as the "Marinelli" or "reentrant" beaker (Hill et al., 1950). It
568 consists of a cylindrical sample container with an inverted well in the bottom of the beaker that
569 fits over the detector.
570 Counting efficiencies are determined by measuring a known quantity of the radionuclide(s) of
571 interest in the same matrix and source-detector configuration as the sources requiring analysis
572 (NCRP, 1978, pp. 243-244; ASTM, D3649). This eliminates any effect that might be caused by
573 differences in test and calibration source characteristics, e.g., density, moisture content, shape,
574 and size. Efficiency curves may be prepared for a detector by measuring a variety of standardized
575 sources having different photopeak energies under identical conditions as the unknown
576 (Coomber, 1975, p. 18; ANSI, 1991).
577 Two important advantages of gamma-ray spectrometry are the ability to measure more than one
578 radionuclide simultaneously and the elimination or reduction of the necessity for chemical
579 dissolution and radionuclide separations (nondestructive analysis). Source configurations for
580 nondestructive analyses generally are selected to optimize counting efficiency. Examples are
581 (PHS, 1967, p. 78):
582 • Marinelli beakers of various volumes to measure liquid sources, as water, milk, and food
583 samples blended to a slurry;
584 • Cylindrical plastic containers of various volumes, such as the 400 mL "cottage-cheese
585 container" frequently used for containing solid sources;
586 • Planchets of various diameters to measure precipitates, air filters, etc.; and
587 • Aluminum cans of a standardized volume into which solid sources can be compressed, and
588 sealed, if desired, to retain radon.
589 If greater counting efficiency is required, the test source size can be reduced, allowing a greater
590 amount of the laboratory sample to be counted and in a more favorable geometry. Examples of
591 such processes are:
592 • Reducing the volume of water samples by evaporation;
593 • Reducing the volume of water samples by co-precipitating the desired radionuclides;
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 16-19 DO NOT CITE OR QUOTE
-------
Instrument Calibration and Test Source Preparation
594 • Reducing the size of vegetation samples by compression into a large pellet or by ashing, if
595 volatile radionuclides are not of interest; and
596 • Reducing the size of filter samples by compressing the sample into a reduced standard
597 volume or by ashing, if volatile radionuclides are not of interest.
598 16.6.2 Gamma Spectrometer Calibration
599 Most gamma-ray spectrometry systems are calibrated with either single or mixed standards in an
600 exact matrix and geometric form as the samples to be analyzed. However, there are computer
601 codes that can calculate detector efficiency from the physical dimensions of the detector and
602 sample counting geometry (Mitchell, 1986; Hensley et al., 1997).
603 Commercial standards of single or mixed gamma-ray emitters in a matrix of known chemical
604 composition and density can be prepared in user-supplied containers. Calibrations based upon
605 these standards can then be adjusted to correct for any differences in composition and density
606 between the calibration source and the test source (Modupe et al., 1993).
607 MARLAP recommends that calibration data for gamma spectroscopy calibration be obtained
608 from the National Nuclear Data Center at Brookhaven National Laboratory (http://www.nndc.
609 bnl.gov/nndc/nudat/). Calibration data are readily available for common radionuclides, including
610 210Pb, 241Am, 109Cd, "Co, 141Ce, 139Ce, 203Hg, 51Cr, 113Sn, 85Sr, 137Cs, 54Mn, 88Y, 65Zn, 60Co, and 40K.
611 For more information on gamma spectrometry calibration see ANSI 42.14. (Also see Section
612 17.3.1.6 on gamma calibration.)
613 16.7 Methods of Test Source Preparation
614 16.7.1 Electrodeposition
615 High-resolution spectroscopy requires a very thin, uniform, flat, and nearly weightless source
616 mount. Ideally, the source plate to determine alpha activity by a spectrometer would be a flat
617 plate coated with a monolayer of radioactive atoms and with no foreign material above the layer
618 to attenuate the alpha radiation (Kressin, 1977). The electrodeposition of radionuclides on a
619 suitable metallic surface from an aqueous solution often can produce thin and uniform test
620 sources that approach these ideal conditions. Thus, this technique is very appropriate for
621 preparing sources of alpha emitters, especially the actinides, which include uranium, plutonium,
622 thorium, americium, and neptunium (ASTM, D3865; DOE, 1997; EPA, 1979).
MARLAP JULY 2001
DO NOT CITE OR QUOTE 16-20 DRAFT FOR PUBLIC COMMENT
-------
Instrument Calibration and Test Source Preparation
623 There are a number of electrolytic cell designs used to electrodeposit radionuclides. The cathode,
624 on which the radionuclide deposits is often a thin metal foil or disc, such as platinum or stainless
625 steel, or a metal-coated plastic film (Blanchard et al., 1960). The stirring rod, often made of
626 platinum, can also serve as the anode of the cell. Deposition of actinides for alpha spectrometry
627 also has been performed on disposable cells constructed form 20 mL polyethylene scintillation
628 vials and highly polished stainless steel planchets (Talvite, 1972). Disposal prevents cross
629 contamination. The composition of the electrolyte and the parameters applied in the electro-
630 deposition process, such as applied voltage, amperage, current density, and deposition time, are
631 dependent upon the chemical properties of the element, especially its reduction potential, and
632 foreign material that might be present. Thus, "Each element requires optimization of its own
633 procedure" (Adloff and Guillaumont, 1993, p. 158). Deposition time varies from 10 minutes to
634 two hours.
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
Actinides and similar elements are extremely hydrolytic and can deposit on the glass cell wall or
anode or precipitate during deposition (Puphal et al., 1983). Electrodeposition typically is
performed, therefore, in electrolytic solutions at low pH (~2) to prevent hydrolysis or
precipitation. The solution may contain complexing agents (such as fluoride) and chelates (such
as EDTA) to minimize the effect of interfering ions, commonly encountered in biological and
environmental samples (Puphal and Olsen, 1972). The procedure of Kressin (1972), however,
illustrates the admonition of Adloff and Guillaumont cited above: citrate and fluoride, a chelate
and complexing agent, respectively, each interferes with the electrodeposition of plutonium and
americium in his process.
Electrodeposition is applicable to more than 30 radionuclides. The main advantage of
electrodeposited sources over those from other methods of preparation is their extremely thin,
uniform deposit of a radionuclide on a plate, which permits high resolution spectroscopy;
however, the yield is often not quantitative (Adloff and Guillaumont, 1993, p. 158). Thus, the
yield must be monitored with the inclusion of a known quantity of an isotope, which is deposited
simultaneously with the analyte. Radioactive sources of the following elements have been
prepared successfully by electrodeposition (DOE, 1997; Blanchard et al., 1960; Johnston et al.,
1991.)
Actinium
Americium
Antimony
Bismuth
Cadmium
Cobalt
Copper
Curium
Gold
Hafnium
Indium
Iron
Lead
Neptunium
Nickel
Plutonium
Polonium
Promethium
Protactinium
Radium
Rhenium
Ruthenium
Selenium
Silver
Strontium
Tellurium
Thallium
Thorium
Tin
Uranium
Yttrium
Zinc
JULY 2001
DRAFT FOR PUBLIC COMMENT
16-21
MARLAP
DO NOT CITE OR QUOTE
-------
Instrument Calibration and Test Source Preparation
660 Particularly important to environmental analysis is a procedure by which virtually all alpha-
661 emitting nuclides—radium through californium—can be determined in soil in any combination
662 on a single sample with few interferences using electrodeposition to prepare the source (Sill et
663 al., 1974).
664 Although sources of radioactive isotopes of these elements have been prepared by electro-
665 deposition, it might not be the preferred technique in some of the examples cited. For various
666 reasons, other methods of test source preparation may be superior: yields can be low, the
667 presence of other metals sometime interferes, the quality of deposition might be poor (flaking),
668 the recovery can be low, the spectral resolution might be poor, and some procedures require
669 rather elaborate equipment, are expensive, and are time consuming, thus labor intensive (Sill and
670 Williams, 1981; Hindman, 1986). Interference will be caused by several factors: (1) "Any
671 element present in the separated fraction that is able to be electrodeposited will be present on the
672 metal disc;" (2) "Incomplete separation of rare earth elements or incomplete wet ashing for the
673 removal of organic material will decrease the efficiency of the electrodeposition and may result
674 in a thick deposit unsuitable for a-spectrometry measurement;" and (3) "Samples containing
675 more than 20 jig of U are unsuitable for measurement by a spectrometry due to the thickness of
676 the deposit" (DOE, 1997, p. 4.5-270). When stainless-steel planchets cannot be used, because of
677 the corrosive nature of the electrolyte, and platinum is required, the method can be quite
678 expensive and time consuming, since recycling of the expensive electrode material requires
679 thorough cleaning to prevent cross contamination.
680 Test sources of actinides are often prepared by electrodeposition with yields of 90 percent and
681 higher (DOE, 1997; EPA, 1979; Sill et al., 1974; Puphal and Olsen, 1972; Kressin, 1977; Talvite,
682 1972; Mitchell, 1960; Shinohara and Kohno, 1989, pp. 41-45). In addition, 54Mn sources have
683 been successfully prepared by the electrodeposition from mixed-solvent electrolytes onto
684 stainless steel planchets (Sahoo and Kannan, 1997, pp. 185-190).
685 If the redox couple between the metal cathode and the radionuclide to be deposited is positive,
686 the radionuclide will deposit spontaneously. That is, it will deposit quantitatively without using
687 any applied potential. Generally, a metal planchet is simply suspended in the solution that is
688 stirred with a glass stirring rod for a few hours (Blanchard, 1966; DOE, 1997). An example of
689 such a spontaneous reaction between polonium and nickel is given below.
690 Po+4 + 2 Ni ^ Po + 2 Ni+2 E° = 0.98 Volt
691 Polonium also will deposit quantitatively on silver planchets. 210Po is an important naturally
692 occurring radionuclide that is often included in environmental studies. Spontaneous deposition
MARLAP JULY 2001
DO NOT CITE OR QUOTE 16-22 DRAFT FOR PUBLIC COMMENT
-------
Instrument Calibration and Test Source Preparation
693 onto either nickel or silver is the preferred technique for preparing 210Po sources for
694 measurement.
695 A similar technique, called internal electrolysis, is preformed by selecting electrodes that have a
696 large difference in potential. A conventional electrolytic cell containing an acid solution of the
697 radionuclide to be deposited may be used. A magnesium (E° = +2.37 volts) strip, for example, is
698 inserted into the electrolyte and connected by an external circuit to the inert metal cathode
699 (planchet), usually platinum. A spontaneous current flows and deposition on the cathode will
700 occur. The conditions at the inert cathode are exactly the same as if an external voltage were
701 applied; however, longer electrolysis times are necessary to achieve quantitative recoveries. Very
702 thin and uniform sources of 106Ru, 110Ag, 203Hg, 60Co, 114In, 51Cr, 198Au, and 59Fe were prepared by
703 this technique, with greater than 96 percent recovery in all cases (Blanchard et al., 1957, pp. 46-
704 54; Van der Eijk et al., 1973).
705 16.7.2 Coprecipitation
706 Coprecipitation (Section 13.8) has been employed to mount sources for alpha spectrometry.
707 Some radiochemists prefer the method to electrodeposition, maintaining that, "The procedure is
708 faster and more reliable than those involving electrodeposition and gives consistently higher
709 yields" (Sill and Williams, 1981). Hindman (1986) asserts that the method is "more rapid, more
710 economical, and more efficient" ... "and yields good decontamination factors, high recoveries,
711 and excellent resolution of the a spectra for uranium, plutonium, americium, and thorium."
712 Although sources prepared by Coprecipitation are thicker than those prepared by electrodepo-
713 sition, sufficiently thin sources, even for alpha spectrometry, can be prepared by controlling the
714 amount of precipitate formed. Sources thinner than 0.5 jig/mm2 can be prepared of the actinides
715 by Coprecipitation (EPA, 1984a). Thicker sources lead to poor resolution of the spectra
716 (Hindman, 1983) and sources produced by any technique that are greater than 10 jig/mm2 lead to
717 attenuation of alpha particles (Adolff and Guiallaumont, 1993, p. 161).
718 After separations are completed, a slurried precipitate is poured quantitatively through a filtering
719 apparatus collecting the precipitate on a small (e.g., 25 mm dia.) filter. Vacuum filtration often is
720 used to speed the operation. With suction applied, the precipitate typically is washed with water,
721 then ethyl alcohol, and finally with acetone to dry the precipitate. The filter is removed from the
722 filtering apparatus and mounted on a metal planchet, commonly with double-stick tape, and
723 stored in a desiccator to await counting. Any 222Rn progeny that collects on the filter during the
724 filtration process will decay in a short period of time and not affect the measurement. Samples of
725 the following radionuclides have been prepared for quantitative analysis by Coprecipitation:
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 16-23 DO NOT CITE OR QUOTE
-------
Instrument Calibration and Test Source Preparation
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
Radionuclide
32p
51Cr
89/90Sr
90y
131j
137Cs
147Pm
2ioBi
226Ra
Th
Th
U
Np
Pu
Am
Cm
Th
Np
Pu
Am
Cm
U
a EPA (1984)
b EPA (1980)
Carrier
MgNH4PO4
BaCrO4
SrC03
Y2(C204)3
PdI2
Cs2PtCl6
Nd2(C204)3
BiOCl
BaSO4
Ce(I04)4
LaF3
LaF3 (NdF3)
LaF3
LaF3(NdF3)
LaF3(NdF3)
LaF3
Ce(OH)2
Ce(OH)2
Ce(OH)2
Ce(OH)2
Ce(OH)2
UF3
c DOE (1997)
d Hindman (1983)
References
a
a
a,b,c
a,b,c
a,b,c
b
a
a
b
d
a,b
a,b,(f)
b
a,b,d,(f)
a,b,d,(f)
b
e
e
e
e
e
e
e Sill (1981)
f Hindman (1986)
It should be emphasized that precipitated sources must be thoroughly dry before measurement,
otherwise, self-absorption and scattering will change with time as water evaporates. Also,
sources are often covered with a thin film, such as Mylar™ or Formvar™, to avoid sample loss and
contamination of counting equipment. Care must be taken to avoid excessive handling of the
source that can change the physical nature of the co-precipitate, producing an uneven thickness.
Another precipitation technique has been applied to preparing radioactive sources. Source
preparation by precipitation can be conducted in a desiccator fitted with a valve to allow first the
evacuation of the desiccator and then the admission of a precipitating gas, such as ammonia
(NH3) or hydrogen sulfide (H2S) (Blanchard et al., 1957, pp. 26-31; Van der Eijk et al., 1973). A
carrier is added to the sample and a know quantity is pipetted onto a planchet. The planchet
containing the test source solution is placed in the desiccator and exposed to a precipitating gas
for one to two hours. This period of time allows settling to occur. The test source is removed
from the desiccator and evaporated beneath a heat lamp. Using an A1C13 carrier in an ammonia
atmosphere, Yoshida et al. (1977) prepared uniformly deposited radioactive sources of 59Fe, 60Co,
95Nb, 103Ru, and 198Au by this technique.
MARLAP
DO NOT CITE OR QUOTE
16-24
JULY 2001
DRAFT FOR PUBLIC COMMENT
-------
Instrument Calibration and Test Source Preparation
766 16.7.3 Evaporation
767 When a high degree uniformity of the deposit is not a requirement for the measurement, sources
768 can be prepared by simple evaporation under a heat lamp (Bleuler and Goldsmith, 1952). This
769 procedure is easy, fast, and adequate for many type measurements. Water samples for gross alpha
770 and beta screening measurements are often prepared by this method (EPA, 1984a; EPA, 1980).
771 An aliquant of the water laboratory sample is evaporated on a hot plate until only a few milliliters
772 remain. The concentrated solution that remains is then transferred quantitatively with a pipette to
773 a tared stainless-steel planchet, usually 2-inch diameter, and evaporated to dryness under a heat
774 lamp. The planchet, with the evaporated test source, is then flamed over a burner until dull red to
775 reduce the amount of solids present and to convert the matrix to an oxide. (Insoluble hydroxides,
776 which are often bulky and gelatinous, are prime candidates for ashing, as the oxide formed is
777 much firmer, more uniform, and better defined.) The test source is cooled, weighed, and counted
778 for alpha and beta particles in a proportional counter. Planchets containing evaporated solids
779 cannot be flamed if volatile radionuclides are to be measured.
780 Most of the solids in an evaporated source deposit in a ring around the edge. Techniques to
781 improve uniformity include the addition of a wetting agent, such as tetraethylene glycol or a 5
782 percent insulin solution (Shinohara and Kohno, 1989), freeze drying the sample, or precipitation
783 and settling of the active material prior to evaporation (Friedlander et al., 1981, p. 305; Van der
784 Eijk and Zehner, 1977). The wetting agent is pipetted onto the spot to be covered by the test
785 source, then removed with the pipette. That remaining can be dried under a heat lamp. A known
786 quantity of the laboratory sample is then pipetted onto the spot and dried under a heat lamp.
787 Additional portions of the sample may be added and evaporated.
788 Sample spreading on the planchet, as it is heated, can result in depositing test source material on
789 the planchet walls or in the flow of the liquid over the edge of a flat, lipless planchet. Such
790 spreading can be controlled or restricted by outlining the desired source area with a wax pencil.
791 Metal planchets often are constructed with a small lip around their circumference that retains the
792 test source on the planchet. All sources prepared by evaporation should be flamed to a dull-red
793 color, cooled, and stored in a desiccator until counted, unless they contain volatile radionuclides,
794 in which case simply store the evaporated test source in a desiccator.
795 Source spreading during evaporation has been restricted by electrospraying a silica gel
796 suspension onto a thin film to produce a circular pad. The radioactive source solution is dropped
797 onto the circle and evaporated to dryness (Chen et al., 1989).
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 16-25 DO NOT CITE OR QUOTE
-------
Instrument Calibration and Test Source Preparation
798 EPA's (1980) prescribed Method 900.0 for measuring gross alpha and beta radioactivity in
799 drinking water suggests that the sample aliquant be limited to what will produce 5 mg/cm2 of
800 solids on the planchet. Thus, for a 2-inch planchet (20 cm2), an aliquant containing 100 mg of
801 non-volatile dissolved solids is the recommended maximum test source mass.
802 After a radionuclide in solution has been purified by chemical techniques, i.e., impurities
803 removed, the solution can be transferred to a planchet and evaporated to dryness, as described
804 above. Evaporation of a laboratory sample after purification is used by the EPA to measure 228Ac
805 in the analysis for 228Ra (EPA, 1984a), and sources of thorium, isolated from marine carbonates,
806 have been prepared by evaporation for measurement by alpha spectrometry (Blanchard et al.,
807 1967). Measured count rates of identified radionuclides, for which absorption curves have been
808 prepared, can be adjusted for self absorption in evaporated test sources.
809 In the case of all dry sources, steps should be taken to prevent solids from exiting the planchet,
810 which will affect the measurement and, in time, contaminate the detector. Sources consisting of
811 loose, dry material, or with a tendency to flake, should be covered with thin plastic or
812 immobilized by evaporating a few drops of a lucite-acetone solution on the solid deposit (PHS,
813 1967, p. 21).
814 16.7.4 Thermal Volatilization/Sublimation
815 Vacuum thermal volatilization or sublimation are often used when very thin and uniform sources
816 are required (Blanchard et al., 1957, p. 7-9 and Friedlander and Kennedy, 1955, p. 122). The
817 disadvantages of this technique are that it is time consuming and the recoveries are often less
818 than 50 percent (NAS/NRC 1962, pp. 126-127).
819 The apparatus used to perform this procedure consists of a demountable vacuum chamber that
820 contains either a ribbon filament, often with a shallow trough, or a crucible. The collector plate is
821 usually mounted less than an inch away. The source solution is first evaporated onto the filament.
822 As the required temperature of the filament is reached, the trough in the filament tends to
823 collimate the sublimed material onto the collecting plate, increasing the recovery of the sample.
824 Pate and Yaffe (1956) designed a system for volatilizing radionuclides from a crucible heated
825 with electrical resistance wire. Their design resulted in nearly 100 percent yields on thin
826 collecting films, and made it possible to prepare thin and uniform sources containing a known
827 aliquant of a stock solution (NAS/NRC 1962, p. 127).
MARLAP JULY 2001
DO NOT CITE OR QUOTE 16-26 DRAFT FOR PUBLIC COMMENT
-------
Instrument Calibration and Test Source Preparation
828 For very thin sources, it is necessary either to swing the collector plate away or have it covered
829 during initial heating in order to burn off impurities at low temperatures without volatilizing
830 them onto the source mount. Separation from contaminants can be accomplished at the time of
831 source preparation by considering differences in vapor pressure and carefully controlling the
832 temperature (Coomber 1975, p. 306). The temperature at which a radionuclide will volatilize
833 depends on the compound in which it exists, e.g., as a hydride, oxide, or halide. Sources have
834 been prepared by thermal volatilization/sublimation for radioisotopes of manganese, chromium,
835 cobalt, rhodium, arsenic, silver, ruthenium, technetium, and many others (Blanchard et al., 1957,
836 p. 9; Coomber 1975, pp. 306-308). See Section 13.5, Volatilization and Distillation, for further
837 discussion of this topic with examples.
838 A technique called vacuum evaporation has been used to prepare thin, uniform radioactive
839 sources (Van der Eijk, 1973). Radioactive substances are volatilized by heating a solution in an
840 oven under reduced pressure. Yields, usually rather low, can be improved by using a collimating
841 oven.
842 16.7.5 Preparing Sources to Measure Radioactive Gases
843 Gaseous radionuclides most often measured include tritium, both as a vapor (3HOH) and in the
844 elemental form (3H-H), 14C, as CO2, and the noble gases, 37Ar, 41Ar, 85Kr, 131mXe, and 133Xe.
845 Tritiated water vapor is often collected by condensation from a known volume of air (EPA
846 1984b). The air is drawn first through a filter to remove all particulates and then through a cold
847 trap submerged in a dry ice/alcohol bath. A measured aliquant of the water collected is analyzed
848 by liquid scintillation spectrometry (EPA, 1984b). Tritiated water vapor is sometimes collected
849 by pulling air through a trap containing silica gel (SC&A, 1994). After collection, the water is
850 distilled from the silica gel, collected, and counted in a liquid scintillation spectrometer.
851 Gaseous products of oxidation or combustion can be trapped in a suitable media, such as water
852 for 3H, ethanolamine for 14C, peroxide for 35S, and then analyzed by liquid scintillation
853 spectrometry (NCRP, 1978, p. 211). For this method, it is very important to de-aerate the liquid
854 prior to introducing the gas, and the temperature must be carefully controlled since gas
855 solubilities are temperature dependent (NCRP, 1978, p. 210), generally inversely proportional to
856 the temperature.
857 Although not as common nor convenient as liquid scintillation spectrometry, a gaseous
858 radionuclide can be measured in an internal proportional counter as a component of the counter-
859 filling gaseous mixture, usually argon, methane, or an argon-methane mixture (Friedlander and
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 16-27 DO NOT CITE OR QUOTE
-------
Instrument Calibration and Test Source Preparation
860 Kennedy 1955, p. 274; NAS/NRC 1962, p. 128; Bleuler and Goldsmith 1952). For example,
861 tritiated water can be reduced to hydrogen gas (3H2) by passing water vapor over a bed of hot
862 zinc, and sodium carbonate can be converted to carbon dioxide (14CO2) by the action of an acid
863 (NCRP, 1978, p. 211). These gases then can be mixed with a counting gas and introduced into
864 the proportional-counter chamber. The major disadvantage of this technique is that it requires a
865 gas handling system.
866 Concentrations of radioactive noble gases in the effluents of some nuclear facilities are
867 sufficiently high that source preparation simply involves filling an evacuated vessel with the
868 gaseous sample or flushing the vessel sufficiently to insure a 100 percent exchange (EPA, 1984b,
869 pp. 19-20). The counting geometries (efficiencies) of the collection vessels can be determined,
870 allowing the collected test sources to be measured directly in the vessels by gamma-ray
871 spectrometry.
872 For environmental samples collected downwind of a nuclear facility, concentrating the nuclides
873 in the gaseous sample is nearly always required prior to measurement. One example is a system,
874 called the "Penn State Noble Gas Monitor," which was designed to measure low concentrations
875 of radioactive noble gases (Jabs and Jester, 1976; Jester and Hepburn, 1977). Samples of
876 environmental air are compressed in SCUBA (high pressure) bottles to 3,000 psig, providing a
877 sample volume of 2.3 m3. The inlet air to the compressor passes through a scrubbing train that
878 contains particulate filters and activated charcoal to remove radioiodine. The noble-gas
879 measurement system consists of a spherical 14.69 L, high-pressure, stainless steel vessel with a
880 reentrant well in its base to permit insertion of a Ge detector connected to a spectrometry system.
881 The vessel is surrounded with 2 inches of lead shielding.
882 There may be occasions when radioiodine is discharged into the atmosphere in several chemical
883 forms. A molecular species filtering system, described by EPA (1990), collects four primary
884 species of iodine on separate cartridges so that they can be measured individually. Air is pulled
885 first through a particulate filter and then through the cartridges placed in series. The normal order
886 of the four cartridges in the filtering system is as follows: (1) cadmium iodide media (CdI2) for I2
887 retention, (2) 4-iodophenol (I • C6H4- OH) on alumina for HOI retention, (3) silver-salt (AgX)
888 loaded zeolite or impregnated charcoal for organic iodine retention, and (4) charcoal for a
889 breakthrough monitor. Air, at a calibrated flow, is passed through the system at a
890 rate of one to two cubic feet per minute (cfm). When the sample-collection period is complete,
891 the cartridges are separated, and the activities of each are measured separately by direct counting
892 of the individual cartridges using gamma-ray spectrometry.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 16-28 DRAFT FOR PUBLIC COMMENT
-------
Instrument Calibration and Test Source Preparation
893 16.7.6 Preparing Air Filters for Counting
894 Air filters containing particulates may be counted directly by a proportional or scintillation
895 detector. Minimal source preparation is normally required for directly counted filters. Some
896 project plans may require that the mass of the particulates on filters be determined. If so required,
897 the filters are weighed on receipt and the net particulate mass calculated by subtracting the mass
898 of an average filter mass or, if pre-weighed, the beginning filter mass.
899 Actual preparation may be limited to a reduction of the size of the filter and placing it in the
900 appropriate counting container, e.g., a planchet. If the filter is of the correct size and shape to fit
901 directly in a counting container, no preparation may be required. Since particulate matter is
902 deposited on the surface of the filter medium, care must be exercised in handling, particularly
903 during size reduction, so that particulate material is not removed.
904 Because potentially contaminated material is relatively easily removed from a filter surface,
905 caution is necessary to avoid contamination of detectors. If a filter is to be gamma counted it can
906 remain in the envelope or plastic bag in which it is received for counting. The filter may be
907 placed in such an enclosure if not received in that manner. The size of the filter may be reduced
908 by simply folding the filter to a standard size for gamma counting.
909 When specific alpha- and beta-emitting nuclide analyses are required (e.g., Pu, U, Th, Am, Sr),
910 the filter media along with the particulate material are usually ashed or dissolved and processed
911 as any digestate by the procedure used in the laboratory.
912 16.7.7 Preparing Swipes/Smears for Counting
913 Swipes are collected to determine the level of removable surface contamination. They are
914 normally taken on a filter paper or fabric pad by rubbing it over a predetermined surface area,
915 nominally 100 cm2. Swipes are routinely counted directly in a proportional counter for alpha and
916 beta activity determination. The size of the swipe is selected to allow it to be placed in a
917 standard-size planchet for counting. If elevated beta radioactivity is identified, a swipe may be
918 gamma counted to determine the contributing radionuclide. Elevated alpha activity may require
919 isotopic analyses for identification.
920 The precaution relative to detector contamination given above for air filters applies to swipes. All
921 swipes should be treated as if they are contaminated until proven otherwise. In some cases swipes
922 may be wetted with water or alcohol prior to collection of the sample. Wet swipes shall be
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 16-29 DO NOT CITE OR QUOTE
-------
Instrument Calibration and Test Source Preparation
923 allowed to air dry prior to counting in order to avoid the reduction of particles reaching the
924 detector due to absorption in the liquid remaining on the swipe.
925 16.8 References
926 Adolf, J.P. and Guillaumont, R. 1993. Fundamentals of Radiochemistry, CRC Press, Boca
927 Raton, Florida.
928 Amano, H., Kasal, A., and Matsunaga, T. 1985. "Simultaneous Measurement of Radon and its
929 Progeny in Cave Air by Liquid Scintillation Techniques and Alpha Spectrometry," Health
930 Phys. 49:3, pp. 509-511.
931 American National Standards Institute (ANSI) N42.14. "Calibration and Use of Germanium
932 Spectrometers for Measurement of Gamma-Ray Emitting Radionuclides," 1991, New York.
933 American National Standard Institute (ANSI) N42.23. "Measurement and Associated
934 Instrumentation Quality Assurance for Radioassay Laboratories." 1996.
935 American Society for Testing and Materials (ASTM). 1998. 1998 Annual Book ofASTM
936 Standards, Volume 11.02, D1890, D1943, D3084, D3648, D3649, D3865, D4107, and
937 D4962.
938 American Public Health Association (APHA). 1995. Standard Methods for the Examination of
939 Water and Wastewater, 19th Edition, pp. 7-1 to 7-40.
940 Baillie, L. A. 1960. "Determination of Liquid Scintillation Counting Efficiency by Pulse Height
941 Shift," Int. J. Ppl. Radiat. Isot, .8:1.
942 Blanchard, R.L., Kahn, B., and Birkhoff, R.D. 1957. "The Preparation of Thin, Uniform Sources
943 for a Beta-Ray Spectrometer," Oak Ridge National Laboratory Report, ORNL-2419, Oak
944 Ridge, TN.
945 Blanchard, R.L., Kahn, B., and Birkhoff, R.D. 1960. "The Preparation of Thin, Uniform,
946 Radioactive Sources by Surface Adsorption and Electrodeposition," Health Phys., Vol. 2,
947 pp. 246-255.
948 Blanchard, R.L. 1966. "Rapid Determination of Lead-210 and Polonium-210 in Environmental
949 Samples by Deposition on Nickel," Analytical Chemistry, Vol. 38, p. 189.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 16-30 DRAFT FOR PUBLIC COMMENT
-------
Instrument Calibration and Test Source Preparation
950 Blanchard, R.L., Cheng, M.H., and Potratz, H.A. 1967. "Uranium and Thorium Series
951 Disequilibria in Recent and Fossil Marine Molluscan Shells," J. Geophys, Vol. 72, pp. 4745-
952 4757.
953 Bleuler, E. and Goldsmith, GJ. 1952. Experimental Nucleonics, Rinehart & Company, New
954 York, pp 38-39.
955 Chen, Q.J., Nielson, S.P., and Aarkrog, A. 1989. "Preparation of Thin Alpha Sources by
956 Electrospraying for Efficiency Calibration Purposes," Radioanal. Nucl. Chem. 135:2, 117-
957 123.
958 Coomber, D.I. 1975. RadiochemicalMethods in Analysis, Plenum Press, New York.
959 U.S. Department of Energy (DOE). 1997. EML Procedures Manual, Chieco, Nancy A., Ed.,
960 HASL-300, 28th Edition, DOE Environmental Measurements Laboratory, New York.
961 U.S. Environmental Protection Agency (EPA). 1978. Radon In Water Sampling Program,
962 Eastern Environmental Radiation Facility., EPA/EERF-Manual-78-1.
963 U. S. Department of Energy (DOE). 1995. DOE Methods for Evaluating Environmental and
964 Waste Management Samples, Goheen, S.C. et al. (ed.), DOE/EM- 0089T.
965 U.S. Environmental Protection Agency (EPA). 1979. Radiochemical Analytical Procedures for
966 Analysis of Environmental Samples, Johns, F.B., Hahn, P.B., Thome, D.J., and Bretthauer,
967 E.W., EMSL-LV-0539-17, Environmental Monitoring and Support Laboratory, Las Vegas,
968 Nevada.
969 U.S. Environmental Protection Agency (EPA). 1980. Prescribed Procedures for Measurement of
970 Radioactivity in Drinking Water, Krieger, H.L. and Whittaker, E.L., Eds., EPA 600-4-80-032,
971 EPA Environmental Monitoring and Support Laboratory, Cincinnati, Ohio.
972 U.S. Environmental Protection Agency (EPA). 1984a. Radiochemistry Procedures Manual,
973 Lieberman, R.E., Ed., EPA 520-5-84-006, EPA Eastern Environmental Radiation Facility,
974 Office of Radiation Programs, Montgomery, Ala.
975 U.S. Environmental Protection Agency (EPA). 1984b. An Airborne Radioactive Effluent Study at
976 the Savannah River Plant, EPA 520-5-84-012, Office of Radiation Programs, Eastern
977 Environmental Protection Agency, Montgomery, Alabama.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 16-31 DO NOT CITE OR QUOTE
-------
Instrument Calibration and Test Source Preparation
978 U.S. Environmental Protection Agency (EPA). 1987. Interim Protocols for Screening and
979 Follow Up Radon and Radon Decay Product Measurements, Office of Radiation Programs,
980 Washington, DC., EPA 520-1-86-014-1.
981 U.S. Environmental Protection Agency (EPA). 1990. Standard Operating Procedures - Field
982 Operations and Emergency Activities, Office of Radiation and Indoor Air, National Air and
983 Radiation Environmental Laboratory, Montgomery, Alabama.
984 Faires, R.A. and Boswell, GJ. 1981. Radioisotope Laboratory Techniques, Butterworth & Co.
985 Publishers, Ltd., London.
986 Friedlander, G. and Kennedy, J.W. 1955. Nuclear and Radiochemistry, 1st Ed., John Wiley &
987 Sons, New York, pp. 274-282.
988 Friedlander, G., Kennedy, J.W., Macias, E.S., and Miller, J.M. 1981. Nuclear and
989 Radiochemistry, 3rd Ed., John Wiley and Sons, New York.
990 Hemingway, J.D. 1975. "Measurement Techniques and Instrumentation," International Review
991 of Chemistry - Radiochemistry, Inorganic Chemistry, Series Two, Volume 8, ed. A.G.
992 Maddock, University Park Press, Baltimore.
993 Hensley, W.K., McKinnon, A.D., Miley, H.S., Panisko, M.E., and Savard, R.M. 1997. SYNTH
994 for Windows, Pacific Northwest National Laboratory, Richland, WA.
995 Herpers, U. 1986. "Radiation Detection and Measurement," in Treatise on Analytical Chemistry,
996 Elving, P.J., Krivan, V., and Kolthoff, I.M., Eds., Part I, 2nd Edition, Volume 14, John Wiley
997 and Sons, New York, pp. 123-192.
998 Higashimura, T., Yamada, O., Nohara, N., and Shicei, T. 1962. "External standard method for
999 the determination of the efficiency in liquid scintillation counting," Int. J. Appl. Radiat. Isot.
1000 13, p. 308.
1001 Hill, R. F., Hine, G. F., and Marinelli, L. D. 1950. "The Quantitative Determination of Gamma-
1002 Ray Radiation in Biological Research," Am. Jour. Roentg, 63, pp. 160-169.
1003 Hindman, F.D. 1983. "Neodymium Fluoride Mounting for Alpha Spectrometric Determination
1004 of Uranium, Plutonium, and Americium," Anal. Chem. Vol. 55, pp. 2460-2461.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 16-32 DRAFT FOR PUBLIC COMMENT
-------
Instrument Calibration and Test Source Preparation
1005 Hindman, F.D. 1986. "Actinide Separations for a Spectrometry Using Neodymium Fluoride
1006 Coprecipatation," Analytical Chemistry, Vol. 58, pp. 1238-1241.
1007 Horrocks, D.L. 1973. "Measuring Tritium With Liquid Scintillator Systems," in Tritium,
1008 Moghissi, A.A. and Carter, M.W., Eds., Messenger Graphics, Publishers, Las Vegas, NV,
1009 pp.150-151.
1010 Jabs, R.H and Jester, W.A. 1976. "Development of Environmental Monitoring System for
1011 Detection of Radioactive Gases," Nuclear Tech. Vol. 30, 24-32.
1012 Jester, W.A. and Hepburn, F.J. 1977. "A Ge(Li) System for the Monitoring of Low Level
1013 Radioactive Gases," Trans. ANS, Vol. 26, 121.
1014 Johnston, P.N., Moroney, J.R. and Burns, P.A. 1991. "Preparation of Radionuclide "Sources" for
1015 Coincident High-Resolution Spectrometry with Low-Energy Photons and Electrons or Alpha
1016 Particles," Appl. Radial Isot. No. 3, 245-249.
1017 Kressin, IK. 1977. "Electrodeposition of Plutonium and Americium for High Resolution a
1018 Spectrometry," Analytical Chemistry, 49:6, pp.842-845.
1019 Lieberman, R. and Moghissi, A.A. 1970. "Low-Level counting by Liquid Scintillation, JJ.
1020 Application of Emulsions in Tritium Counting," Inter. J. Appl. Rad. Isotopes, Vol. 21, p. 319.
1021 McDowell, W.J. 1986. "Alpha Counting and Spectrometry Using Liquid Scintillation Methods,"
1022 Nuclear Science Series on Radiochemical Techniques.
1023 Mitchell, D. J. 1988. Gamma Detector Response and Analysis Software (GADRAS), Sandia
1024 Report SAND88-2519, Sandia National Laboratories, Albuquerque, NM.
1025 Mitchell, R.F. 1960. "Electrodeposition of Actinide Elements at Tracer Concentration,"
1026 Analytical Chemistry, 32:3, pp. 326-328.
1027 Modupe, O.O., Decker, K.M., and Sanderson, C.G. 1993. Determination of Self-Absorption
1028 Corrections by Computation in Routine Gamma-Ray Spectrometry for Typical
1029 Environmental samples, Radioactivity & Radochemistry, 4:1, pp. 38.
1030 Moghissi, A.A. 1981. "Application of Cyclohexane in Separation of Water From Biological and
1031 Environmental Samples," Health Phys., Vol. 41, p. 413.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 16-33 DO NOT CITE OR QUOTE
-------
Instrument Calibration and Test Source Preparation
1032 Moghissi, A.A., Bretthauer, E.W., and Compton, E.H. 1973. "Separation of Water From
1033 Biological and Environmental Samples for Tritium Analysis," Anal. Chem., Vol. 45, 1565-
1034 1566.
1035 National Academy of Sciences-National Research Council (NAS/NRC). 1962. Detection and
1036 Measurement of Nuclear Radiation, O'Kelley, G.D., NAS-NS 3105, Washington, DC.
1037 National Council on Radiation Protection and Measurement (NCRP). 1978. A Handbook of
1038 Radioactivity Measurements Procedures, Report No. 58, Washington, DC.
1039 Passo, CJ. and Cook, G.T. 1994. Handbook of Environmental Liquid Scintillation Spectrometry
1040 — A Compilation of Theory and Methods, Packard, 800 Research Parkway, Meriden,
1041 Connecticut.
1042 Pate, B.D. and Yaffe, L. 1956. "Disintegration-Rate determination by 4n Counting, Pt. IV. Self-
1043 Absorption Correction: General Method and Application to Ni63 p- Radiation," Can. J. Chem.
1044 Vol. 34, p. 265.
1045 U.S. Public Health Service (PHS). 1967. Common Laboratory Instruments for Measurement of
1046 Radioactivity, Environmental Health Series-Radiological Health, Interlaboratory Technical
1047 Advisory Committee, Report No. 2, Rockville, Maryland, pp. 19-27, 78.
1048 Puphal, K.W. and Olsen, D.R. 1972. "Electrodeposition of Alpha-Emitting Nuclides form a
1049 Mixed Oxalate-Chloride Electrolyte," Analytical Chemistry, Vol. 44, no. 2, pp. 284-289.
1050 Puphal, K.W., Filer, T.D., and McNabb, GJ. 1983. "Electrodeposition of Actinides in a Mixed
1051 Oxalate-Chloride Electrolyte," Analytical Chemistry, Vol. 56, No. 1, pp. 114-116.
1052 Sahoo, P. And Kannan, S.E. 1997. "Preparation of an Electrodeposited Source of 54Mn,"
1053 RadiochimicaActa76:4,pp. 185-190.
1054 S. Cohen & Associates (SC&A). 1994. Routine Environmental Sampling Procedures Manual
1055 For Radionuclides, Prepared for the U. S. Environmental Protection Agency, Office of
1056 Radiation and Indoor Air, under Contract No. 68D20155, Work Assignment No. 2-25.
1057 Shinohara, N. And Kohno, N. 1989. "Rapid Preparation of High-Resolution Sources for Alpha-
1058 Ray Spectrometry of Actinides in Spend Fuel," Appl. Radiat. Isot. 40:1, pp. 41-45.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 16-34 DRAFT FOR PUBLIC COMMENT
-------
Instrument Calibration and Test Source Preparation
1059 Sill, C. W., Puphal, K.W., and Hindman, F.D. 1974. "Simultaneous Determination of Alpha-
1060 Emitting Nuclides of Radium through Californium in Soil," Analytical Chemistry, 46:12, pp.
1061 1725-1737.
1062 Sill, C.W. and Williams, R.L. 1981. "Preparation of Actinides for a Spectrometry without
1063 Electrodeposition," Analytical Chemistry., 53:3, pp. 412-415.
1064 Talvite, N.A. 1972. "Electrodeposition of Actinides for Alpha Spectrometric Determination,"
1065 Analytical Chemistry, Vol. 44, No. 2, pp. 280-283.
1066 Tsoulfanidis, N. 1983. Measurement and Detection of Radiation, McGraw-Hill Book Company,
1067 New York.
1068 Van der Eijk, W., Oldenhof, W., and Zehner, W. 1973. "Preparation of Thin Sources, A
1069 Review," Nucl. Instr. AndMeth. Vol. 112, 343-351.
1070 Van der Eijk, W. and Zehner, W. 1977. "Preparation of Thin Sources for Absolute Beta
1071 Counting," Radiochimica Acta Vol. 24, 205-210.
1072 Yoshida, M., Miyahara, H. and Tamaki, W. 1977. "A Source Preparation for 4?ip-Counting With
1073 an Aluminum Compound," Intern. J. Appl. Rad. Isotopes Vol. 28, 633-640.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 16-3 5 DO NOT CITE OR QUOTE
-------
17 DATA ACQUISITION, REDUCTION, AND REPORTING
2 17.1 Introduction
3 This chapter provides information and guidance, primarily for laboratory personnel, on data
4 acquisition, reduction, and reporting. Its intent is to provide an understanding of the many
5 operational parameters which should be addressed in order that the data developed and reported
6 are compliant with project planning documents (Chapter 4), considered valid (Chapter 8), and
7 usable for their intended purposes (Chapter 9). These processes are all linked and each is
8 dependent upon the results of its predecessor. The material presented is intended to provide an
9 overview of the processes which are required in all radiochemistry laboratories, but are by no
10 means performed in the same way in all laboratories.
11 In this chapter, data acquisition refers to the results produced by the radiation detection process,
12 often referred to as counting. This chapter will provide guidance for laboratory personnel on
13 selecting and applying the operational parameters related to instrumentation and the determina-
14 tion of the radioactivity contained in the test source.1 Parameters that are applicable to counting
15 for essentially all radiation detection instrumentation are discussed in Section 17.2 and those that
16 are specific to a given type of instrumentation are covered in the appropriate section describing
17 that instrument. A detailed description of the instrumentation discussed in this chapter was
18 provided in Chapter 15.
19 Once test sources have been prepared (Chapter 16) and counted using laboratory measurement
20 instruments (Chapter 15), the basic information generated by the instrument should be reduced
21 (processed) to produce data which can be reviewed, verified, validated, and interpreted in light of
22 and in accordance with project planning documents and analytical statements of work (SOWs)
23 (Chapter 7). Data reduction is primarily mathematical in nature while data reporting involves the
24 presentation of the results of the data acquisition and reduction processes and nonmathematical
25 information necessary to interpret the data (e.g., sample identification and method of analysis).
26 Data reduction may be as simple as a division of the counts by the counting time, the sample
27 aliquant weight or volume, and the counter efficiency, thereby producing the radionuclide
28 concentration. On the other hand, it may also require more complicated processing such as the
29 fitting of an analytical function, or the unfolding of a differential spectrum (Tsoulfanidis, 1983,
1 The term "test source" will be used to describe the radioactive material prepared to be introduced into a measure-
ment instrument and "laboratory sample" will be used to identify the material collected for analysis. Thus, a test
source is prepared from laboratory sample material for the purpose of determining its radioactive constituents.
"Calibration source" is used to indicate that the prepared source is for the purpose of calibrating instruments.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 17-1 DO NOT CITE OR QUOTE
-------
Data Acquisition, Reduction, and Reporting
30 p. 327). In any case, the reduction process should continue by calculating the combined standard
31 uncertainty (Chapter 19).
32 The output of some laboratory instruments is highly simplistic and consists only of the number of
33 nuclear decay events recorded by the detector in the time interval allocated for the measurement.
34 An example of this might be a gas-proportional counter whose only output is an electronic sealer
35 and the available data consists of total counts or counts per minute. On the other extreme, some
36 laboratory counting instruments with computer components produce outputs consisting of
37 radionuclide concentration, uncertainty, and other information (see Chapter 19). Examples of
38 these types of data reducing instruments are alpha- and gamma-spectrometry and liquid-
39 scintillation systems.
40 ANSI N42.23 contains an outline of a minimal data report. Most project-specific planning
41 documents (Chapter 4) and/or analytical SOWs (Chapter 5) require that the radiochemical data
42 produced by laboratories be submitted in a specific format and form (i.e. electronic or hard copy,
43 or both). In some cases, the requirements are minimal and may consist of a data report which
44 gives only the sample identifier information, accompanied by the radionuclide concentration and
45 its associated uncertainty. Many projects require much more supporting information, primarily to
46 assist in the data validation (Chapter 8) process. Support material can include information on
47 calibration, background determination, sample processing, sample receipt, quality control sample
48 performance, raw-counting data, and chain-of-custody records.
49 This chapter gives an overview of data acquisition, reduction, and reporting in radiochemical
50 laboratories. The material presented is intended to be descriptive rather than prescriptive, since
51 these processes vary greatly between laboratories; depending upon the equipment, personnel,
52 project requirements, and the methods and analyses being performed.
53 17.2 Data Acquisition
54 Data acquisition refers to the process of collecting the basic information produced by nuclear
55 counting instruments. These data may be produced in hard copy or electronic format, or visually
56 displayed for the operator to record. As previously stated, this can be simply the number of
57 counts detected by the instrument within the allotted counting time or as conclusive as the
58 identification of the radionuclides contained in the sample along with their concentrations and
59 associated uncertainties.
60 Following generation, data requiring further processing may be electronically or manually
61 transferred to the next to the next data-reduction step. Electronic transfer should be employed as
MARLAP JULY 2001
DO NOT CITE OR QUOTE 17-2 DRAFT FOR PUBLIC COMMENT
-------
Data Acquisition, Reduction, and Reporting
62 often as possible to avoid the inherent errors associated with manual transfer. On the other hand,
63 the next step in the data reduction process may be performed manually, i.e., with a calculator.
64 The reliability of the data generated also depends upon the proper operation of the instrumenta-
65 tion and the associated data reduction programs. Data quality further depends upon the correct
66 input of associated information by laboratory personnel.
67 17.2.1 Generic Counting Parameter Selection
68 Instrument operators have choices, provided by instrument manufacturers, in the setup and
69 operation of nuclear counting instruments. These selections can affect the quality and
70 applicability of the data. Some selections can be made on a one-time basis and left unadjusted for
71 the processing of all samples and others require the operator to reevaluate the settings, possibly
72 for each test source counted. In some cases adjustments can be made following counting during
73 the processing of the derived information. Some adjustments can only be made before counting
74 or by extending the counting time. In making the proper selection, there are some overall
75 considerations relative to the project requirements, as specified in project planning documents
76 (Chapter 4) or in the analytical SOW (Chapter 5). Other operator decisions depend on the nature
77 of the test source itself. Caution should be exercised when changing operational parameters so
78 that the calibrations (counting efficiency, energy, self absorption, etc.) performed on the
79 instrument remain valid. For example, changing the source container or holder may affect the
80 counting efficiency and/or background. Determining the appropriate operating conditions
81 requires that the operator have a thorough understanding of the counting process and the
82 instruments and their operation for the production of valid and useable data. In addition, the
83 operator should be cognizant of the measurement quality objectives (MQOs) that have been
84 established.
85 Some of the factors that affect operational parameter selection are related to project requirements.
86 Planning documents and the analytical SOW may specify the limits on measurement uncertainty
87 and detection capability. In order to achieve compliance with the limits, instrument operating
88 parameter adjustment may be required for some or all the samples received. The number of
89 samples received during a time period may make it mandatory for adjustments to be made in
90 order to meet these requirements while complying with project defined turn-around-times.
91 Factors that may affect the selection of operational parameters include:
92 • Project and External
93 - project requirements for uncertainty, detection capability, and quantification capability
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 17-3 DO NOT CITE OR QUOTE
-------
Data Acquisition, Reduction, and Reporting
94 - laboratory backlog and contract turn-around times
95 • Sample Characteristics
96 - expected sample radionuclide concentration
97 - interfering radionuclides
98 - interfering stable constituents (e.g. liquid scintillation counting quenching)
99 - amount of sample available
100 - physical characteristics of the test source (e.g. density)
101 - half-life of the radionuclide of interest
102 • Analytical Process
103 - chemical separation process leading to counting source generation (Chapter 14)
104 • Instrumentation
105 - instrument adjustments available and their limits
106 - conditions and limits of an instrument's calibration
107 - time availability of instruments
108 - counting efficiency
109 - calibration geometries available
110 Taking into consideration the above, the operator has control over and should select certain
ill parameters for all radiation measurements. The selection of the basic parameters should be
112 carefully planned in advance to assure that the project requirements are met. The laboratory's
113 selection of parameters during the planning process may require alteration as the process of
114 sample analysis is actually taking place due to unavoidable changes in the samples and sample
115 characteristics throughout the duration of the study.
116 17.2.1.1 Counting Duration
117 For the Poisson counting model, the uncertainty associated with a given count determination is
118 proportional to the square root of the total number of counts accumulated (Chapter 19). The total
119 counts accumulated during counting are proportional to the activity of the source and the length
120 of the counting time. Counting duration is a controllable factor that allows one to achieve a given
121 level of counting uncertainty. The operator should then select a duration which is sufficient to
122 meet project objectives for detection capability and uncertainty. The length of time allotted for
123 determination of the instrument background will also affect the uncertainty associated with the
124 measurement (Chapter 19). Thus, when preparing an analytical protocol to meet the requirements
125 of a project, as expressed in the project planning documents, the laboratory will establish the
MARLAP JULY 2001
DO NOT CITE OR QUOTE 17-4 DRAFT FOR PUBLIC COMMENT
-------
Data Acquisition, Reduction, and Reporting
126 counting durations of both sample and background accordingly. An alternative to selecting a
127 counting duration, available on many instruments, is to count until a preset number of counts is
128 obtained.
129 17.2.1.2 Counting Geometry
130 The counting efficiency of a radiation detector depends on the geometry of the source and
131 detector arrangement, e.g., the solid angle subtended at the detector by the source. A given
132 radiation detector may have the counting efficiency established for several geometries. The
133 geometry selected among those available may depend upon the amount of sample available, the
134 detection capability required for the analysis, the radionuclide concentration in the sample, the
135 dictates of the radioanalytical method, the physical characteristics of the sample, the nature and
136 energy of the decay process, and the characteristics of the detector.
137
138 The choices to be made relative to geometry selection are usually the type of test source
139 container, the source mounting, and the detector to source distance. Choices are to be made
140 among those for which the detector has an established efficiency calibration.
141 17.2.1.3 Software
142 The use of properly developed and documented computer software programs for data acquisition
143 and reduction can lead to an enhancement in the quality of laboratory data. Guidance on software
144 documentation can be found in EPA (1995). Caution should be exercised in the selection and use
145 of undocumented programs and those which may not have been tested in laboratories performing
146 analyses similar to those for which MARLAP has been developed. For example, a spectral
147 analysis program may accurately identify and quantify the radionuclides in test sources
148 containing higher levels of radioactivity (which produce spectra with well defined peaks, easily
149 distinguishable from background) but may be inaccurate for samples with environmental levels.
150 When selecting software, a thorough review of the data reduction algorithms should be
151 performed. The user should not blindly accept the notion that all software performs the
152 calculations in an appropriate manner without this review. When evaluating software, it is often
153 helpful to review the software manual, particularly in regard to the algorithms used in the
154 calculations. While it may not be necessary that the user understand in detail all the calculations
155 performed by highly complex software programs, the user should understand the overall scheme
156 of analysis and reduction in order to assure data meet quality objectives and reporting
157 requirements. This understanding is also beneficial in assuring that user defined parameters are
158 properly selected.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 17-5 DO NOT CITE OR QUOTE
-------
Data Acquisition, Reduction, and Reporting
159 The output of some instruments is very basic, primarily counting data, i.e., counts or counts per
160 second. These data should be manipulated by external systems to convert them to the form
161 required by planning documents. The external system which performs the calculations may be a
162 calculator or a computer with the appropriate software to reduce the data to usable terms. In
163 either case, additional information relative to the processing of the sample should be input along
164 with the counting data (counting time, total counts, and background counts). This information
165 may include laboratory sample number, collection date, sample mass or volume, instrument
166 counting efficiency, and chemical yield.
167 For computer (processor) based systems, some of this information is generated and processed
168 internally and the remainder is manually entered or electronically transferred from the Laboratory
169 Information Management System (LDVIS) or some other adjunct system where it has previously
170 been stored. It is becoming increasingly common for much or all of this adjunct information to be
171 transferred to the counting instrument by reading a bar code affixed to the test source to be
172 counted. In this manner, the information which has previously been entered into a LEVIS is
173 electronically transferred to the counting instrument. For hand calculations, these data are simply
174 entered into the calculations.
175 17.2.2 Basic Data Reduction Calculations
176 The equations used for data reduction depend on the analytical methods used. The following
177 equations are provided as examples to illustrate the basic principles involved in data reduction.
178 Following counting, the radionuclide concentration may be calculated:
r
n _ UNet
c 77 (17.1)
e- V- Y-Kc-e l
179 where:
180 Rc = radionuclide concentration at time of collection (Bq/L or Bq/g)
181 Cnet = net count rate (cps)
182 E = counter efficiency for the radionuclide (cps/dps)
183 V = volume or mass of sample analyzed (L or g)
184 Y = chemical yield (when appropriate)
185 e = base of natural logarithm
186 k = the radioactive decay constant for the radionuclide (s~\ miir1, or d"1)
MARLAP JULY 2001
DO NOT CITE OR QUOTE 17-6 DRAFT FOR PUBLIC COMMENT
-------
Data Acquisition, Reduction, and Reporting
187 ^ = time lapse from sample collections to beginning of source count (units consistent
188 with A,)
189 Kc = the correction for decay during counting and is:
190 where:
191 tc = actual clock time (real time) of counting (units consistent with A)
192 This calculates the radionuclide concentration at the time of sample collection2. It compensates
193 for the fact that short-lived radionuclides may experience significant reduction in activity during
194 counting, when the counting duration is a significant fraction of the half-life. For long-lived
195 radionuclides, the term Kc approaches unity and can be ignored. The efficiency used in this
196 equation may be obtained from the specific radionuclide whose concentration, Rc, is to be
197 determined or it may be obtained from an efficiency curve which plots counter efficiency vs.
198 energy. In the latter case, the abundance, Ee, of the particle or photon being counted should be
199 considered. This is required because the energy dependent efficiency, e^ is developed in terms of
200 the fraction of particles or photons detected divided by the number emitted at that energy. Thus,
201 if the radionuclide emission being determined during the counting of a test source has an
202 abundance less than 100 percent, an adjustment should be made to Equation 17. 1, as shown in
203 Equation 17.3:
r
n _ _ UNet _
c -- 77 (17.3)
£
e
204 Most modern instrument systems contain preprogrammed software to perform data manipula-
205 tions that convert basic counting information to a form which can be compared to the project data
206 quality objectives, or at least to begin or promote this process. Certain sample-specific
207 information should be manually entered or transferred to the system electronically in order to
208 perform the necessary calculations.
2 For radionuclides with short half-lives detected at or near detection limits, it may be more appropriate to calculate
the concentration at the time of counting.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 17-7 DO NOT CITE OR QUOTE
-------
Data Acquisition, Reduction, and Reporting
209 17.3 Data Reduction on Spectrometry Systems
210 Software is available for resolving alpha, gamma, and liquid scintillation spectra and for
211 performing the attendant functions such as calibration, energy alignment, background acquisition
212 and subtraction, and quality control functions.
213 Spectroscopic analysis for alpha particles and gamma-rays is performed to identify and quantify
214 radionuclides in samples. Since these emissions occur at discrete energies, spectrometry is useful
215 for these purposes and can be applied to the analysis of a wide range of s radionuclides. Energy
216 spectra are produced when a detector absorbs a particle or photon and produces a signal that is
217 proportional to the energy absorbed. The resulting signal is digitized by an analog-to-digital
218 converter and processed by a multichannel analyzer. A differential spectrum is produced, where
219 the number of events within an incremental energy, AE, is recorded on the jaxis and the energy
220 is represented on the * axis (Tsoulfanidis, 1983, p. 327). In this way, radionuclides can be
221 identified by the characteristic energies of their emissions and quantified because the area under
222 the full energy peak is proportional to the emission rate (activity) of the source being analyzed.
223 The spectra for alpha and gamma emitters are quite different, due to the differences in the way
224 these two types of radiation interact with matter in transferring their energy to the detector
225 material. The process of resolving the spectra into its contributing components is referred to as
226 spectral analysis (NCRP 1978, p. 159) and unfolding (Tsoulfanidis, 1983, p. 342). Computer
227 programs for analyzing alpha and gamma spectra are available from several sources (Decker and
228 Sandderson, 1992). A method of performance testing of gamma analysis software is given in
229 ANSIN42.14.
230 17.3.1 Gamma Spectrometry
231 Gamma spectrometry on environmental samples requires the use of gamma spectral analysis
232 software for any reasonable degree of accuracy and detection capability. This is due to the
233 potentially large number of photopeaks to resolve, the low level of radioactivity in most
234 environmental samples, and the relatively low detection limits and stringent quality control
235 requirements of most project-specific planning documents. Spectral analysis by manual
236 techniques is only practical when the number of radionuclides is limited and the contributing
237 isotopes are predictable. An example is the analysis of milk samples for gamma-emitting
238 radionuclides, where the milk production process in the cow restricts the number of radionuclides
239 in the milk product (Hagee et al., 1960, p. 36; USPHS, 1967, pp. 1-51).
MARLAP JULY 2001
DO NOT CITE OR QUOTE 17-8 DRAFT FOR PUBLIC COMMENT
-------
Data Acquisition, Reduction, and Reporting
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
Gamma-rays interact with matter in
three ways, namely, by photoelectric
effect, Compton scattering, and pair
production (Tsoulfanidis, 1983, pp.
141-148). These interactions within a
gamma detector, usually a high-purity
germanium or sodium iodide (see
Chapter 16), result in varying amounts
of the gamma-ray energy being
absorbed. Only one of these inter-
actions, the photoelectric effect, results
in the total energy being absorbed in a
single interaction. The photopeak,
shown in Figure 17.1, due to a
photoelectric interaction in the detector,
results from the processing of the
detector signal through the linear
circuitry and the multichannel analyzer.
Full-Energy_
Photopeaks "~
Photopeak
Baseline
Channel Number and Gamma-Ray Energy
FIGURE 17.1 — Gamma-ray spectrum
This photopeak has a basic Gaussian shape (Gilmore and Hemmingway, 1995, p. 163) and may
be described by (Quittner, 1972, p.20):
(17.4)
260
261
262
263
264
265
266
267
268
269
where:
A =
x =
p =
the peak amplitude
the channel number
the peak centroid
(The width of the peak is related to the full-width at half-maximum (FWHM) of the detector, F,
where F = 2.355 a. The area under the peak is N= 1.064 A F.)
As can be seen in Figure 17.1, the photopeak (PI) may be displaced upward by its position on the
Compton continuum from other, higher-energy gamma-rays (P2) and background radiation.
The photopeak is the key element in gamma-ray spectrometry in that its location on the energy
axis provides a means for radionuclide identification, and the area under the peak is proportional
JULY 2001
DRAFT FOR PUBLIC COMMENT
17-9
MARLAP
DO NOT CITE OR QUOTE
-------
Data Acquisition, Reduction, and Reporting
270 to the number of gamma-ray events comprising the photopeak. This becomes the basis for
271 radionuclide identification and quantification.
272 The fundamental purposes of gamma-ray computer-based spectral analysis programs are to
273 identify the photopeaks in a spectrum and to measure the true area under the photopeaks. It
274 should do this in the presence of natural background, a potentially large number of sometimes
275 overlapping photopeaks, and a great number of Compton-scattering events. Once these initial
276 tasks have been performed, the computer program uses this information to determine the
277 radionuclide mix that contributed the complex spectrum and the individual concentrations in the
278 sample being analyzed.
279 Most computer programs for gamma-spectral analysis are provided by equipment manufacturers,
280 although some are supplied by independent providers. There are significant differences in the
281 structure of the programs. However, they all perform similar functions which are given below
282 and illustrated in Figure 17.2.
283 17.3.1.1 Peak Search or Identification
284 There are two basic methods of gamma spectral analysis. The first method is to allow the
285 analysis software to determine the existence of the peaks and their energy. The second method is
286 often referred to as a "library directed" search, where the operator identifies the peak energy
287 locations, e.g., regions of interest, to be searched for discernable peaks. The latter method may be
288 more sensitive (Gilmore and Hemmingway, 1995, p. 165) but, taken alone, will fail to identify
289 and report unspecified radionuclides. If the confirmation of the existence of a particular
290 radionuclide is required, the second method should be employed. Most software programs allow
291 either approach to be activated and used for each analysis.
292 A most important function performed by an analysis program is the identification of true
293 photopeaks. In the programs available, this is achieved in one of the four ways discussed below.
294 Many spectral analysis programs allow the operator to select among two or more of the four
295 methods for peak identification. Selection of the most accurate and sensitive method depends on
296 the radionuclides present in the source, detection capability requirements for individual
297 radionuclides, the number of radionuclides present, the nature of the background spectrum, the
298 degree to which the radionuclide mix can be predicted, and the activities of the isotopes. The
299 selection of a particular peak search method can be determined by experience with similar
300 sample types and past performance, particularly on performance evaluation (known) samples.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 17-10 DRAFT FOR PUBLIC COMMENT
-------
Data Acquisition, Reduction, and Reporting
Acquire Sample
Spectrum
Peak Search and
Identify
Assign
Define
Energy
Width
Peak
Library
Energy
Calibration Curve
Width
Calibration Curve
Multiple!
Resolve
Radioisotope
Identification
A .
I
^
Operation Info.
Sample ID
Batch ID
Calculate
Uncertainty
I,.
h Pen
Corrections for:
Random Summing
Live Time
Decay during Count
Decay to Collection Tim
ort
Concentration
A
e
Operator Input.
Reporting Units
Sample Volume
FIGURE 17.2 — Gamma-ray analysis sequence
301 REGIONS OF INTEREST (ROI) METHOD
302 This is the simplest form of peak identification, but can only be used when the radionuclides
JULY 2001
DRAFT FOR PUBLIC COMMENT
17-11
MARLAP
DO NOT CITE OR QUOTE
-------
Data Acquisition, Reduction, and Reporting
303 present in the sample are known and when the analysis system has been compensated for gain
304 drift. ROI analysis involves the establishment of predetermined energy regions, at least one for
305 each radionuclide present. Once the spectrum has been acquired, the number of counts in each
306 region is summed after subtracting the photopeak baseline (Figure 17.1). This method of spectral
307 analysis is more applicable to alpha rather than gamma spectrometry.
308 GAUSSIAN FUNCTION DERIVATIVE METHOD
309 As previously stated, the photopeak has a basic Gaussian shape; in reality it is a histogram with a
310 Gaussian-like shape. The most widely used peak identification technique was proposed by
311 Mariscotti (Mariscotti 1967, p. 309) and uses the Gaussian function derivative to assess the
312 presence of a photopeak. For most low-level radioactivity, this peak search method may provide
313 the best peak detection capability with the fewest false peak identifications or omissions of true
314 peaks (Gilmore and Hemmingway, 1995, p. 20).
315 CHANNEL DIFFERENTIAL METHOD
316 This method searches for a number of channels where the counts are significantly greater than the
317 preceding channels, and then looks for the expected decrease in counts corresponding to the
318 backside of the prospective photopeak. This method works relatively well for large, well-defined
319 peaks, but is limited for poorly defined peaks with counts barely above the background baseline
320 of the peak (Gilmore and Hemmingway, 1995, p. 163).
321 CORRELATION METHOD
322 In this method, a search function is scanned across the spectrum. Each channel count, over the
323 width of the search function, is multiplied by the corresponding value of the search function. The
324 sum of these products is then made a point on a correlation spectrum. A correction for the
325 baseline contribution leaves only positive counts within a photopeak. Although the scan function
326 is normally Gaussian in form, other forms may be applied (Gilmore and Hemmingway, 1995,
327 p. 164).
328 Spectral analysis programs usually have some user selected peak acceptance criteria. The
329 acceptance criteria may be based on peak shape, width uncertainty, or the number of standard
330 deviations above the background to be subtracted. Care is required in selection of the values for
331 these acceptance criteria. If the values are too high, valid photopeaks remain undetected. If the
332 values selected are too low, radionuclides may be reported which are not present in the samples.
333 Knowledge of the sample origin and experience with using the analysis program on similar
MARLAP JULY 2001
DO NOT CITE OR QUOTE 17-12 DRAFT FOR PUBLIC COMMENT
-------
Data Acquisition, Reduction, and Reporting
334 samples to those being processed is useful in establishing values for these user-selected
335 parameters. Peak searches may be standard or directed (Canberra, 1994). In a standard search, all
336 peaks identified are assigned to a library contained radionuclide. In a directed search, the user
337 specifies the energies and radionuclides over which the search is performed. If reporting of a
338 specific radionuclide is required, the directed search is appropriate; however, some radionuclides
339 could go unreported if only a directed search is performed.
340 17.3.1.2 Singlet/Multiplet Peaks
341 A peak is referred to as a singlet or multiplet according to whether it is composed of a single
342 photopeak or multiple photopeaks, respectively. Deconvolution is the term given to the process
343 of resolving a multiplet into its components (Gilmore and Hemmingway, 1995, p. 172). The
344 ability of a spectral analysis program to perform this function may well be the deciding point for
345 its selection. It is particularly important if the laboratory has analyses in which one of the critical
346 radionuclides has only one gamma-ray whose energy is very near to that of another radionuclide
347 expected to be present in all or most samples.
348 There are three primary ways that programs deal with the problem of resolving multiplets. The
349 first method is a deconvolution algorithm which is based on the peak-shape being the composite
350 of multiplet Gaussian distributions. The second method uses the gamma-ray library to anticipate
351 where peaks occur within a multiplet. The disadvantage of the first is in dealing with small ill-
352 defined peaks and the second cannot, of course, resolve peaks not included in the library. The
353 third method, peak stripping, again depends on defining all radionuclides whose gamma-rays
354 contribute to the multiplet. In peak stripping, one of the interfering gamma-ray's contribution is
355 subtracted from the multiplet area by using another of its gamma-rays to estimate the peak shape
356 and size in the multiplet area. The remaining peak is, presumably, that of the interfered
357 radionuclide which can then be identified and quantified. This method requires that one of the
358 interfering radionuclides have a second gamma emission which identifies and tentatively, for the
359 purpose of removing its contribution, quantifies it.
360 In some cases, the uncertainty of multiplet deconvolution can be avoided by selecting photopeaks
361 from gamma-rays which are not interfered with, even though they may have lower abundances.
362 The increase in uncertainty due to the lower number of accumulated counts may well overcome
363 the uncertainty of deconvolution (Gilmore and Hemmingway, 1995, p. 174).
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 17-13 DO NOT CITE OR QUOTE
-------
Data Acquisition, Reduction, and Reporting
364 17.3.1.3 Definition of Peak Centroid and Energy
365 Once a peak has been detected, the centroid of the peak will be defined, since it will rarely be
366 located at exactly a whole channel number. The centroid will be used to represent the gamma-ray
367 energy and should be calculated to the fraction of a channel. An algorithm is used to calculate the
368 centroid value may be expressed as (Gilmore and Hemmingway, 1995, p. 167):
,.
Centroid = ' (17.5)
369 where:
370 Cjis the count in the /h channel.
371 In order to assign a gamma-ray energy value to the peak centroid channel position, the analysis
372 program refers to a previously established energy calibration file. The detector's response to the
373 full range of gamma energies should be established by counting a source(s) having a number of
374 well-defined gamma-rays over the range of energies emitted by the radionuclides in the
375 calibration source. This calibration source is most often a "mixed-nuclide source," which also
376 has certified emission rates so that it may also be used for an efficiency calibration. The mixed-
377 nuclide source is counted on the detector, being sure to accumulate sufficient counts in the peaks
378 to obtain good statistical precision, and an energy-versus-channel relationship is established. The
379 operator will be required to provide information on the peaks to be used and their exact energies.
380 With modern spectrometry systems, the relationship between energy and channel number is
381 nearly linear. Both linear and quadratic fits have been included in available spectral analysis
382 programs.
383 17.3.1.4 Peak Width Determination
384 In order to calculate the area under the peak, an estimate of the peak width is required, unless the
385 analysis program is operating in the region-of-interest mode. The width of a photopeak is
386 normally quoted in terms of its FWHM. For a discussion of peak width (resolution) and the
387 factors affecting it, see Chapter 15.
388 There are several ways to determine the peak boundary. These are:
389 (1) A Gaussian shape is assumed and some number of standard deviations (2 or 3) are
390 allowed on each side of the peak centroid.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 17-14 DRAFT FOR PUBLIC COMMENT
-------
Data Acquisition, Reduction, and Reporting
391 (2) A standard width for each peak, based on its energy, is used.
392 (3) A five-point moving average is used to determine a minimum on each side of the peak,
393 which is set as the peak limits.
394 Each method has strengths and weaknesses, but all struggle with ill-defined (small number of
395 counts) peaks. Once the peak limits are defined, determining the area under the peak is
396 accomplished by summing the counts per channel for the channels contained in the peak and
397 subtracting the baseline (see Figure 17.1).
398 The determination of FWHM requires an assumption of peak shape and, as has previously been
399 stated, the acceptance of a Gaussian function is the norm for gamma spectrometry. In addition,
400 the peak width increases with the energy of the gamma-ray, so some function should be defined
401 for the analysis program to determine the width based on the energy of the peak. This
402 relationship, in practice, is found to be nearly linear (Gilmore and Hemmingway, 1995, p. 133)
403 and described by:
w=a + bE (17.6)
404 where:
405 w = width of the peak
406 E = the energy
407 a, b = empirical constants
408 For spectra developed by high-purity germanium semiconductors (HPGe) and alpha solid state
409 detectors, it is more appropriate to assume a peak shape which is a modification of the Gaussian
410 function to allow for the low energy tailing observed in these spectra. This type of tailing is
411 illustrated in Figure 17.3. Some spectroscopy programs have algorithms to fit peaks with lower
412 energy tailing.
413 When the "tailing" peak fit option is selected, the software algorithm for peak fitting changes
414 from the pure Gaussian form to a dual fit. The channels in the peak not affected by the tailing are
415 included in the Gaussian fit (Equation 17.7), and those that are affected by tailing are modified
416 according to Equation 17.8, below:
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 17-15 DO NOT CITE OR QUOTE
-------
Data Acquisition, Reduction, and Reporting
FIGURE 17.3 — Low-energy tailing
417
418
419
420
421
422
423
where:
x
A
PC
AC
XX) =
Ae
Ae
x>Pc-AC
, x< Pc - AC
(17.7)
(17.8)
the channel number
the peak amplitude
the peak centroid
the tailing factor (the distance from the centriod to the point where the tailing
joins the Gaussian peak)
the width of the Gaussian peak (~ 2.355 x FWHM)
MARLAP
DO NOT CITE OR QUOTE
17-16
JULY 2001
DRAFT FOR PUBLIC COMMENT
-------
Data Acquisition, Reduction, and Reporting
424
425
426
427
428
429
430
431
432
433
17.3.1.5 Peak Area Determination
For single peaks sitting on a Compton continuum, two methods of peak area determination are
available. The simpler method is the addition (integration) of the number of counts per channel in
each of the channels considered to be within the peak limits, and subtracting the natural
background and Compton contribution to those same channels (Baedecker, 1971; Loska, 1988).
However, this is rarely simple since the photopeak is usually offset by a baseline continuum
whose contribution is not easily determined. While the background may be subtracted by the
spectrometry program, the Compton continuum will be estimated by the software and then
subtracted. This estimation is often based on the number of counts per channel in those channels
immediately above and below the photopeak region as shown in Figure 17.4.
CHANNEL NUMBER
434
FIGURE 17.4 — Photopeak baseline continuum
The baseline contribution is then estimated as:
(17.10)
435 where:
436 B = the number of counts attributed to the baseline
437 N = number of channels in the peak
JULY 2001
DRAFT FOR PUBLIC COMMENT
17-17
MARLAP
DO NOT CITE OR QUOTE
-------
Data Acquisition, Reduction, and Reporting
438
439
440
441
n = the number of baseline channels considered on each side of the peak for calculating
BL and BH
BL = the sum of the number of counts in the baseline region on the low-energy side
BH = the sum of the number of counts in the baseline region on the high-energy side
442 In practice, the baseline continuum appears to have a step beneath the peak (Gilmore and
443 Hemmingway, 1995, p. 114), as illustrated in Figure 17.5. This type of function is estimated by:
N
444
445
446
447
448
449
450
V BH ~ BT
L +_^ L
nG
(17.11)
where:
G
N
n
sum of counts in the baseline region on the low-energy side
sum of counts in the baseline region on the high-energy side
counts per channel in channel j
gross counts in the peak
number of channels in the peak
number of channels in each of the two baseline regions
CHANNEL NUMBER
FIGURE 17.5 — Photopeak baseline continuum-step
function
MARLAP
DO NOT CITE OR QUOTE
17-18
JULY 2001
DRAFT FOR PUBLIC COMMENT
-------
Data Acquisition, Reduction, and Reporting
451 The second peak area determination method is the least-squares method, which fits a theoretical
452 peak shape plus background shape to the channels surrounding the peak (Kruse and Spettel,
453 1982; Helmer et al., 1983). Background is often subtracted prior to the fitting process (Loska and
454 Ptasinski, 1994).
455 17.3.1.6 Calibration Reference File
456 Three types of calibrations are required for gamma spectral analysis, namely those for efficiency,
457 energy, and FWHM. Efficiency and energy calibrations require a source whose gamma-ray
458 emission rate is known and referenced to a national standard, and whose gamma-ray energy lines
459 are well known. "Mixed radionuclide" reference material, containing eight or more gamma lines,
460 is available for performing these spectral calibrations. The operator is required to enter the
461 pertinent information, usually listed in the calibration source certificate, into the file prior to
462 performing the calibrations. The information generally consists of:
463 • Radionuclide name;
464 • Certified activity and units;
465 • Uncertainty in activity;
466 • Reference date and time;
467 • Gamma energies and branching ratios; and
468 • Half-life.
469 Once calibration files are established, the calibrations are performed according to methods
470 specific to individual software and as described in manufacturers manuals (also see Chapter 16).
471 17.3.1.7 Activity and Concentration
472 In order to convert the counts under a photopeak to activity, an efficiency calibration should be
473 performed on the detector. Since the efficiency varies with energy, the detector should be
474 calibrated over the range of energies to be used and a calibration curve developed for the
475 detector. In constructing an efficiency calibration curve, only calibration sources with singlet
476 peaks and well-known abundances should be selected. The efficiency, at a specific energy, is
477 simply the number of counts determined in a photopeak of known energy divided by the number
478 of gamma-rays emitted by the source in the same time period, or:
* = - (17.12)
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 17-19 DO NOT CITE OR QUOTE
-------
Data Acquisition, Reduction, and Reporting
479 where:
480 s = efficiency in cps/yps
481 Cr = cps in the photopeak
482 D = gamma emission rate of source in dps
483 The efficiency versus energy curve developed in most gamma software packages is in the form of
484 a polynomial. One such form is:
n
.-[\nE]i (17.13)
485 where:
486 E = full peak efficiency
487 bj = coefficient as determined by calculation
488 E = the energy of the photopeak
489 The efficiency curve for high-purity germanium detectors shows two distinctive slopes. The
490 polynomial fit in some analysis programs allows for a dual fit, i.e., a separate fit is made to the
491 two portions of the curve.
492 This efficiency curve is maintained in the calibration file of the spectral analysis program to be
493 applied to each analysis. An efficiency curve should be maintained for each test-source geometry
494 to be used for the calibrated detector.
495 To obtain the activity in the test source, the net counts (background subtracted) in the photopeak,
496 as determined by the software through the process described above, is divided by the geometry-
497 specific efficiency. The activity units are converted to those selected by the operator and
498 corrected for decay to the time of collection. Based on sample-aliquant size/volume information
499 supplied by the operator, sample concentration is calculated and reported.
500 17.3.1.8 Summing Considerations
501 Summing refers to the summing of the energy of two or more gamma-rays when they interact
502 with the detector within the resolving time of the spectrometer's electronics. There are two types
503 of summing: (1) random summing, where two unrelated gamma-rays are detected at the same
504 time, and (2) true coincidence summing, is due to the simultaneous emission of gamma-rays by a
505 radionuclide and their subsequent detection by the gamma detector.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 1 7-20 DRAFT FOR PUBLIC COMMENT
-------
Data Acquisition, Reduction, and Reporting
506 Random summing, sometimes referred to as pile-up, is due to gamma-rays, from different
507 radionuclides, being detected almost simultaneously. If two gamma-rays arrive at the detector
508 within the resolving time of the amplifier and both have a photoelectric interaction, instead of
509 having a count in both full-energy peaks a count will occur somewhere else in the spectrum equal
510 to the sum of the two energies. Random summing can also occur with other than photoelectric
511 interactions, e.g., photoelectric with Compton and Compton with Compton. Since this occurs
512 randomly in nature, the probability of random summing increases with the square of the total
513 count rate. Random summing can be reduced by the use of pile-up rejection circuitry which
514 examines the pulse shape of detector signals and rejects those which are distorted by summing
515 (Gilmore and Hemmingway, 1995). However, even with pile-up rejection random summing will
516 still be present. A mathematical correction for random summing is given by:
AT = Ae2Rl (17.14)
517 where:
518 AT = the true peak area (counts)
519 A = the observed peak area (counts)
520 R = the mean (total) count rate (cps)
521 T = the resolving time of the electronics (|is)
522 If unknown, the resolving time can be estimated by a method similar to that described in Gilmore
523 (1995).
524 True coincidence summing is a source of error when a source contains nuclides which emit
525 gamma-rays nearly simultaneously. Coincidence summing is geometry dependent and increases
526 as the source is positioned closer to the detector. Thus, the use of multi-gamma-ray calibration
527 sources for close geometry efficiency calibrations must be done with caution. True coincidence
528 summing also increases with detector volume and is very prevalent in a well detector. The use of
529 a detector with a thin entry window opens the possibility of coincidence summing with X-rays.
530 Since coincidence summing is independent of count rate, it is a mistake to assume that the
531 measurement of environmental media is immune from errors caused by this phenomena.
532 As is the case with random summing, true coincidence summing results in the loss of counts
533 from photopeaks and a corresponding loss in efficiency. The use of single gamma-ray emitting
534 radionuclides is recommended, to the extent possible, for developing calibration curves for
535 detectors at close geometries. In practice, even when the efficiencies are determined in this
536 manner, errors in analyzing for nuclides emitting more than one gamma-ray still exist. When a
537 multi-emitting gamma-ray source is to be measured with minimum bias, it may be necessary to
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 17-21 DO NOT CITE OR QUOTE
-------
Data Acquisition, Reduction, and Reporting
538 perform an efficiency calibration with the specific radionuclide to be measured in the specific
539 geometry desired.
540 In theory it is possible to mathematically correct for true coincidence summing; however, for
541 complicated decay schemes, the task is daunting (Gilmore and Hemmingway, 1995). Some data
542 have been published which give correction factors for coincidence summing for a number of
543 radionuclides (Debertin and Helmer, 1988). Unfortunately they only apply to the particular
544 detector and geometries for which they were developed.
545 17.3.1.9 Uncertainty Calculation
546 The various components of uncertainty in the determination of the source activity should be
547 propagated to obtain the combined standard uncertainty. The sources of uncertainty in the gamma
548 spectral analysis include those associated with the determination of the net peak area, which
549 includes the standard uncertainties of the gross counts, the background counts, and any
550 interference from other gamma radionuclides present; the uncertainty associated with the
551 unfolding of multiplets; the detector efficiency, which includes uncertainties of the net peak area,
552 the calibration source emission rate, and decay correction factor; and uncertainty in the
553 determination of the sample volume or mass.
/ 2 2 2 2 /i -7 i r\
= Jt/p + uv+ uf + Ur, (17.15)
Vp T UF -r u^
554 where:
555 uc = the combined standard uncertainty
556 Up = the component of combined standard uncertainty due to the net peak area
557 determination
558 uv = the uncertainty component for the volume or mass determination
559 UK = the uncertainty component for the efficiency determination
560 % = the uncertainly component for the unfolding routine for multiplets
561 Each of these factors may have a number of components of uncertainties included, for example,
562 the net peak uncertainty:
up
= J UG + UB + UE + uI (17.16)
563 where:
564 UG = the uncertainty component for the gross counts in the peak
MARLAP JULY 2001
DO NOT CITE OR QUOTE 17-22 DRAFT FOR PUBLIC COMMENT
-------
Data Acquisition, Reduction, and Reporting
565 UB = the uncertainty component for the baseline subtraction
566 UE = the uncertainty component for the background peak subtraction
567 Uj = the uncertainty component for the coincidence summing correction
568 The calculations of combined standard uncertainty typically are performed by the spectrometry
569 software for an alpha-spectrometry analysis. It should be noted that not every available software
570 package will incorporate all the listed uncertainty contributions listed.
571 17.3.2 Alpha Spectrometry
572 This section deals with alpha spectrum reduction as applied to semiconductor detectors, since it
573 is likely that this is the type of detector that will be employed for environmental analyses.
574 Since the range of alpha particles is a few centimeters in air and their energy is significantly
575 degraded in passing through a few millimeters of air, alpha spectrometry is conducted in a partial
576 vacuum and on extremely thin sources prepared by electrodeposition or coprecipitation (see
577 Chapter 16).
578 The number of full energy peaks is usually not large, three to four, in an alpha spectra and they
579 are normally well separated in energy. This, coupled with the fact that the test source subjected to
580 counting has gone through a chemical separation (Chapter 14), makes the radionuclide identifica-
581 tion relatively simple when compared to gamma spectrometry. However, it is still of great benefit
582 to have alpha spectrometry software to identify s radionuclides, subtract background, perform
583 calibrations and energy alignments, determine radiochemical yields, and perform and track
584 quality control functions. In production laboratories where hundreds of alpha spectra may be
585 generated each week, it is almost imperative that alpha spectra are resolved by properly designed
586 computer software. An alpha spectrum produced by a semiconductor detector by the counting of
587 a thin source containing 234U, 238U, 239Pu, and 241Am is shown in Figure 17.6.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 17-23 DO NOT CITE OR QUOTE
-------
Data Acquisition, Reduction, and Reporting
588 The spectrum demonstrated contains
589 four peaks which are distorted from
590 their basic Gaussian shape because
591 each of the isotopes emits more than
592 one alpha particle whose energies
593 are within the resolving power of the
594 detector and electronics. The
595 FWHM of the peaks shown is
596 approximately 30 keV. Of particular
597 note is the fact that the peaks are
598 essentially sitting on the baseline.
599 Spectral analysis programs usually
600 have routines for the identification
601 of full-energy peaks. However, in
602 the case of alpha spectrometry,
603 because the locations of peaks in the
604 spectrum are known and the peaks
605 may contain a small number of
606 counts, an ROI-type of analysis is usually performed. However, peak fitting programs are
607 available and may be beneficial when overlapping of peaks is possible. The algorithms used for
608 peak fitting of alpha spectra should take into account the low energy tailing present in most alpha
609 sources (Equation 17.8). The algorithms which account for tailing are modified Gaussian
610 functions and require a peak shape calibration where a number of well-defined singlet peaks
611 covering the full energy range are acquired. The calibration program then calculates the tail
612 parameter values (see discussion entailing in Section 17.3.1.4, "Gamma Spectrometry").
613 Alpha peaks are normally sitting on the baseline (no background continuum) and display
614 minimal overlapping for well-prepared sources. For a given analysis (Pu, U, Am, Th, and etc.),
615 ROIs are established for all energies of the alpha emissions in the source being counted and the
616 count rate in a given ROI represents the emission rate of the alpha whose energy falls within that
617 ROI.
6000
Energy (keV)
FIGURE 17.6 — Alpha spectrum
618 Given these qualifications, the spectral analysis software performs essentially the same functions
619 as that for gamma analysis, described above. The programs may also perform system control
620 function, e.g., maintaining vacuum in the chambers. Databases related to procedures, chemical
621 tracers, and efficiency and energy calibration standards are normally maintained for calculational,
MARLAP
DO NOT CITE OR QUOTE
17-24
JULY 2001
DRAFT FOR PUBLIC COMMENT
-------
Data Acquisition, Reduction, and Reporting
622 documentation, and quality control purposes. The general analysis sequence for alpha
623 spectrometry will be briefly discussed below.
624 An efficiency calibration is not an absolute necessity if a standard/reference material is used for a
625 tracer in each sample and an accurate determination of the yield is not required. In some cases,
626 the laboratory may perform an energy and efficiency calibration for an alpha spectrometry
627 analysis. This requires the operator to establish a calibration certificate file for the program to
628 reference. It should refer to this file for both energy and efficiency calibrations. Calibration
629 sources are necessary for performing the required calibrations, and the appropriate certificate
630 information should be entered into the certificate files in order to perform the calibrations and to
631 analyze test sources. This information should be supplied with calibration sources. Calibration
632 sources , consisting of three to four radionuclides, are available in the form of plated discs from
633 several commercial suppliers.
634 Information typically required by the analysis program consists of the following:
635 • Radionuclide
636 • Activity
637 • Assay date
638 • Half-life
639 • Energy
640 • Energy uncertainty
641 • Emission probability per event
642 • Emission rate uncertainty
643 • Activity units
644 This information should be entered for each of the radionuclides included in the calibration
645 source. Once the library file has been established, an energy calibration can be performed as
646 directed by the software program.
647 The efficiency for alpha particles varies only slightly with energy, within the range of alpha
648 energies usually encountered. While the calibration source may contain several certified
649 radionuclides, during an efficiency calibration, the mean efficiency for the full-energy peaks may
650 be calculated and used as the alpha efficiency for a given detector (Chapter 16).
651 Once the alpha spectrometry system has been calibrated and a spectrum of a test source acquired,
652 either a peak search is performed to identify alpha peaks or, if operating in a ROI mode, the
653 counts in the ROI are determined. ROIs to be used for a given analysis are established prior to the
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 17-25 DO NOT CITE OR QUOTE
-------
Data Acquisition, Reduction, and Reporting
654 spectrum acquisition by selecting an analysis protocol where the radionuclides and their alpha
655 energies are preestablished.
656 In the ROI mode, the counts accumulated during the preset counting duration in each of the
657 designated regions are corrected for background contribution and, in some cases, for reagent
658 blank activity. If a tracer has been added to the test source, the counts in the tracer ROI are
659 summed, background corrected, and the effective efficiency (yield times counting efficiency)
660 determined using certificate information previously entered by the operator and/or from a
661 protocol file. The yield, if required, is then computed by the use of an efficiency which has been
662 previously determined during an efficiency calibration process. The radionuclide concentration is
663 then calculated by3:
CR
..
V- e
(17.17)
664 where:
665 RC = radionuclide concentration of the radionuclide at time of collection (Bq/L or Bq/g)
666 CR= net count rate in the designated ROI for the radionuclide (cps)
667 £e '= effective efficiency (s • Y) for the tracer (cps/dps)
668 V = volume or mass of sample analyzed (L or g)
669 e = base of natural logarithm
670 'kj = the radioactive decay constant for the radionuclide (s~\ miir1, or d"1)
671 fj = time lapse from sample collection to beginning of source count (units consistent
672 with ^)
673 Following the spectrum acquisition process, spectral analysis programs may either automatically
674 process the data and present the results, or they may store the spectral data and await interaction
675 from the operator for processing. In either case, post-acquisition review of the analysis results is
676 recommended. This review may include the following items:
677 • Assure that the alpha peaks fall within the ROIs;
678 • Confirm the absence of unexpected peaks (contamination);
679 • Verify that there are no interfering peaks;
680 • Confirm that peak centroids are within requirements (energy alignment);
681 • Verify that all requirements are met with regard to FWHM and chemical yield; and
For certain alpha-emitting radionuclides, Ra for example, a decay-correcting term is needed.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 17-26 DRAFT FOR PUBLIC COMMENT
-------
Data Acquisition, Reduction, and Reporting
682 • Check units and sample aliquant information.
683 The FWHM of a given peak may depend greatly on the source preparation. However, since an
684 ROI-type of peak search is normally used, and the limits of the peak determined by the setting of
685 the ROI rather than some algorithm, the peak width definition is not significantly affected by
686 reasonable peak broadening. As a precautionary measure, the above review of each test-source
687 spectrum assures that the peaks appear within the ROIs. Alpha spectrometry analysis software
688 allows for the adjustment of the ROIs to account for peak broadening and slight displacement. A
689 review of the FWHM of the alpha peaks, as calculated by the software, will also reveal peak
690 broadening due to matrix effects and poor test-source preparation.
691 17.3.2.1 Radiochemical Yield
692 Alpha spectrometry test sources are usually prepared by radiochemical separation and the
693 chemical recovery may be less than 100%. Therefore, a radiochemical tracer, which is an isotope
694 of the radioactive species for which the analysis is being performed, may be added to the sample
695 prior to preparation and radioanalysis. The tracer is normally a certified standard solution whose
696 recovered activity is determined during the alpha spectrometric analysis in the same manner as
697 the activities of the isotopes for which the analysis is being performed. The radiochemical yield
698 is then calculated by the spectral analysis program according to:
Y= (17.18)
699 where:
700 Y = radiochemical yield
701 AR = calculated activity recovered
702 As = certified activity added (decay corrected to time of counting)
703 The calculation of the chemical yield is normally performed by the alpha spectrometry analysis
704 software using operator input information relative to the alpha energy and abundance, activity,
705 uncertainty, and date of certification of the radiochemical tracer.
706 For some types of radionuclide analyses, no suitable alpha-emitting radionuclide may be
707 available for use as a chemical yield tracer. In this case, the chemical yield may be determined by
708 some other method, such as beta counting, and the resulting yield value provided to the alpha
709 analysis program so the source activity may be calculated from the alpha spectrometry data.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 17-27 DO NOT CITE OR QUOTE
-------
Data Acquisition, Reduction, and Reporting
710 When a reference material is used for the chemical tracer, the effective efficiency is measured for
711 each test source. If the chemical yield is to be reported, an independent measure of the counting
712 efficiency should be made.
713 17.3.2.2 Uncertainty Calculation
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
The calculation of the combined standard uncertainty for alpha spectrometry is similar to that for
gamma-ray spectrometry as reported in Section 17.3.1.8 above. One additional source of
uncertainty which should be taken into account for alpha spectrometry is that associated with the
determination of radiochemical yield. Since a tracer is added to the sample and the yield
determined by a counting process, the uncertainty involved in this analysis should be accounted
for in the total uncertainty. The uncertainty of the yield determination involves that associated
with the net count of the tracer, the counting efficiency, and that of the emission rate of the tracer
material. The combined standard uncertainty of the radionuclide concentration, RC , is given by
either
or
where:
s
Y
V
e
N
V2
(17.19)
\
u2(V)
V2
u2(Y) ^ 2u(i
Y2
(17.20)
= net count rate in the designated ROI for the radionuclide (cps)
= the alpha counting efficiency
= the chemical yield
= effective efficiency (s • Y) for the tracer (cps/dps)
= volume or mass of sample analyzed (L or g)
= base of natural logarithm
= the radioactive decay constant for the radionuclide (s"1, min"1, or d"1)
= time lapse from sample collection to beginning of source count (units consistent
with ^)
denotes the standard uncertainty of a quantity
denotes the covariance of two quantities
MARLAP
DO NOT CITE OR QUOTE
17-28
JULY 2001
DRAFT FOR PUBLIC COMMENT
-------
Data Acquisition, Reduction, and Reporting
736 The two uncertainty equations are equivalent. However, when the yield is determined using an
737 alpha-emitting tracer, Equation 17.19 generally is easier to implement.
738 17.3.3 Liquid Scintillation Spectrometry
739 17.3.3.1 Overview of Liquid Scintillation Counting
740 All modern counters are computer controlled for data acquisition, spectral unfolding, data
741 reduction, sample changer control, external quench correction, and performing the multifarious
742 other functions associated with liquid scintillation counting.
743 Liquid scintillation has traditionally found its primary use in the analysis of low-energy beta
744 emitters, such as 3H and 14C. In spite of the complicating factors of high background and
745 quenching (Chapter 15), procedures for other beta- and alpha-emitting isotopes have been
746 developed over the years (Holm, 1984; Harvey, 1970).
747 Liquid scintillation has also been applied to the simultaneous analysis of alpha and beta emitters
748 in environmental media (Leyba, 1992). Discrimination between alpha and beta radiation is based
749 on differences in the fluorescence decay pulses. Pulse height is proportional to particle energy,
750 and high counting efficiency results from 4n (4-pi) geometry and the absence of test-source self-
751 attenuation (McDowell and McDowell, 1993). Because of these characteristics, liquid
752 scintillation counting can be utilized as an alternative to gas proportional counting (Section 17.4)
753 and alpha semiconductor counting (Section 17.3.2).
754 17.3.3.2 Liquid Scintillation Spectra
755 The amount of light produced by alpha and beta particles in a liquid scintillation cocktail is
756 proportional to the particle energy. Beta spectra convey the energy continuum from zero to their
757 maximum energy. Alpha liquid scintillation spectra are similar in shape to those obtained by
758 semiconductor spectroscopy, but with greatly decreased resolution. Because alpha particles are
759 only about one-tenth as efficient as beta particles in producing scintillation light pulses, there is
760 an overlap of alpha and beta spectra (Passo and Kessler, 1992; McDowell and McDowell, 1993).
761 Gamma radiation interactions within the scintillation cocktail depend on energy and path length,
762 with lower energy gamma rays being more efficient in transferring their energy. Gamma events
763 are recorded in the same energy range as alpha and beta particles; therefore, discrimination
764 between alpha, beta, and gamma radiation based solely on scintillation spectra is not possible
765 (Passo and Kessler 1992; McDowell and McDowell, 1993).
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 17-29 DO NOT CITE OR QUOTE
-------
Data Acquisition, Reduction, and Reporting
766 17.3.3.3 Pulse Characteristics
767 Excited triplet and singlet energy states are formed by the fluor molecules when ionizing
768 radiation interacts with the scintillation cocktail. The excited singlet states dissipate their energy
769 very rapidly and produce short lifetime decay pulses, whereas triplet states lose their energy more
770 slowly, resulting in longer lifetime pulses. Because alpha particles have a higher linear energy
771 transfer than gamma or beta radiation, they produce a higher ratio of triplet to singlet excitation
772 states and therefore have a longer pulse duration. Differences in the decay time and shape of the
773 decay pulse are the basis for discriminating of alpha particles from beta and gamma radiation in
774 liquid scintillation counting (Passo and Kessler 1992; Passo and Cook 1994).
775 17.3.3.4 Coincidence Circuitry
776 Most modern liquid scintillation counters employ two photomultiplier tubes 180 degrees apart
777 for the detection of pulses. The light produced when ionizing radiation in the test source interacts
778 with the scintillation cocktail is emitted in all directions. A sample event should therefore
779 produce electronic pulses in both photomultiplier tubes simultaneously, or in coincidence.
780 Electronic noise pulses are produced randomly by the photomultiplier tubes, but the probability
781 that both tubes will produce noise pulses simultaneously is very low. An electronic gate can be
782 set to allow only pulses that are in coincidence to be registered. The rejection of random pulses
783 keeps background counts produced by electronic noise to a minimum.
784 17.3.3.5 Quenching
785 Chemical quenching reduces the amount of energy transferred to the fluor molecules. Halogens,
786 water, solvents, and oxygen are common agents that cause a decrease in the counting efficiency.
787 Color quenching is caused by impurities not removed during test-source preparation or by carrier
788 compounds such as iron chloride. Photons emitted from the fluor molecules are absorbed,
789 reducing the amount of light reaching the photomultiplier tubes.
790 Quenching causes a shift in the scintillation spectrum to lower energies and a reduction in the
791 number of counts. Quenching has a minimal impact on alpha counting, but significantly increases
792 as the energy of the beta particle decreases.
793 The most common method for monitoring quench is through the analysis of the Compton
794 spectrum. After the test source is loaded into the counter, it is irradiated by an external gamma
MARLAP JULY 2001
DO NOT CITE OR QUOTE 17-30 DRAFT FOR PUBLIC COMMENT
-------
Data Acquisition, Reduction, and Reporting
795 emitting source located in the instrument. The test-source spectrum is collected and compared
796 with factory or user-generated quench standards stored in the instrument library. Both color and
797 chemical quenching cause a shift to lower energies, but the color quench broadens the spectrum
798 as well. The efficiency of the test source is extrapolated and applied to normalize the test-source
799 count rate.
800 17.3.3.6 Luminescence
801 Photoluminescence is produced by ultraviolet light from the environment reacting with the
802 scintillation cocktail. The effect can be minimized by dark adapting the test sources prior to
803 counting.
804 Chemiluminescence is produced by reactions between the scintillation cocktail and chemicals
805 introduced from the test-source preparation. To minimize this effect, oxidizers and alkaline
806 conditions should be avoided.
807 Both photoluminescence and chemiluminescence cause random scintillation events. At low
808 levels, the coincidence gate should reject most of their contribution. However, at very high
809 levels, the probability increases that two events may pass through the gate. Manufacturers use a
810 method of spectral stripping to correct for the false counts, but it is best to avoid the conditions
811 that create the problem.
812 17.3.3.7 Test Source Vials
813 Glass test-source vials contain naturally occurring impurities such as potassium-40, thorium, and
814 uranium. Their contribution appears at the lower energy portion of the spectrum. Plastic vials
815 have a lower background, but they should be compatible with the liquid scintillation cocktail
816 being used. Teflon vials are also available from most manufacturers.
817 17.3.3.8 Data Reduction for Liquid Scintillation Counting
818 Liquid scintillation counters normally provide minimal data reduction in their output. Basic data
819 include the counting duration, count rate in one or more selected windows, and the date and time
820 of counting initiation. A blank source (background) is normally counted with each counting batch
821 and the output will provide the count rate of the blank to be subtracted from each test source.
822 The counting efficiency will also be provided by the output information. Its form of presentation
823 in the output will depend on the calibration/counting (quench correction) method for determining
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 17-31 DO NOT CITE OR QUOTE
-------
Data Acquisition, Reduction, and Reporting
824 counter efficiency4. If the internal (standards addition) method is used the data generated by the
825 counter must be further manipulated in order to develop the counting efficiencies for each test
826 source. When using the external-standards method (quench curve), the scintillation spectrometer
827 will apply the quench corrected efficiency and give the test sample disintegration rate by applying
828 the corrected efficiency.
829 The radionuclide or gross concentration is provided by the following equation:
830 where:
831 CG = the gross counting rate (source + background) (cps)
832 CB= the counting rate of the blank (cps)
833 zq = the radionuclide quench corrected counting efficiency (c/d)
834 Ac = radionuclide or gross concentration (Bq/L or Bq/kg)
835 V = the volume or mass analyzed (L or kg)
836 17.4 Data Reduction on Non-Spectrometry Systems
837 Proportional counters are primarily used for counting of test sources for alpha and beta emitters.
838 Proportional counters may have entry windows for allowance of the emitted radiation into the
839 active portion of the detector or they may be windowless. These instruments are described in
840 Chapter 15. They are used for the determination of specific radionuclides, following chemical
841 separation to isolate the radionuclide, and for nonspecific (gross) analyses (Chapter 16). Counters
842 are equipped to count alpha and beta simultaneously in a given source and report the activity of
843 both.
844 The basic information obtained from a determination in a proportional counter is the number of
845 counts recorded in the detector within the allotted counting duration. However, modern
846 proportional counters take the data reduction process to the point of finality, i.e., producing the
847 test-source concentration and associated counting uncertainty, providing automatic instrument
848 background subtraction, and correcting for source self-absorption and alpha/beta crosstalk.
4 For a discussion of liquid scintillation efficiency determination, see MARLAP Chapter 16, Section 16.5.2.1.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 17-32 DRAFT FOR PUBLIC COMMENT
-------
Data Acquisition, Reduction, and Reporting
849 The instruments may also have protocols for developing the correction factors for self-absorption
850 and for crosstalk. In addition, they should have the capacity to track and evaluate the periodic
851 quality control checks (check source and background) performed on the instrument.
852 The basic equation used to calculate test-source concentrations is:
C - C
A = — (17.22)
853 where:
854 A = the activity of the radionuclide or gross activity (Bq)
855 CG = the gross counting rate (source + background) (cps)
856 CB = the instrument background counting rate (cps)
857 E = the gross or radionuclide counting efficiency (c/d)
858 And the radionuclide or gross concentration is provided by the following equation:
Ac= G£VB (17.23)
859 where:
860 Ac = radionuclide or gross concentration (Bq/L or Bq/kg)
861 V = the volume or mass analyzed (L or kg)
862 The associated combined standard uncertainty is given by:
(17.24)
^
£
2
863 The above simple equations apply to counting either pure alpha or beta emitters and when no
864 correction for self-absorption is necessary (weightless sources). Modifications should be made in
865 the activity and concentration calculations when both alpha and beta particles are emitted by the
866 source, and when absorption and scattering within the source cause a reduction in the effective
867 efficiency.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 17-33 DO NOT CITE OR QUOTE
-------
Data Acquisition, Reduction, and Reporting
868 Self-absorption factors are applied for sources where the internal attenuation of the alpha or beta
869 particle is sufficient to affect the overall efficiency (Chapter 16). Commercially available
870 proportional counters have a protocol for developing the self-absorption correction factors. These
871 protocols process the data generated by counting a series of alpha calibration sources and a series
872 of beta calibration sources, which both have varying masses of material, from "zero" to the
873 maximum to be encountered in test sources (Chapter 16). The instrument is programmed to then
874 fit the data to a mathematical function so the counting efficiency correction factor can be applied
875 at any test-source mass within the range covered by the calibration source masses. A cubic
876 polynomial is one option used for both alpha and beta counting efficiencies. A cubic polynomial
877 has the form
(17.25)
878 where:
879 m = is the residual mass of the test source
880 £m = the counting efficiency at mass m
881 at = constants determined by the data fit
882 The combined standard uncertainty of £m is given by
t/2(a0) + £ m2lu\a) + 2^ £ m1+Ju(apa) + (al + 2a2w + 3a3w2)2 u\m) (17.26)
883 When the identities of the alpha or beta emitting radionuclides are unknown, an additional
884 component of uncertainty is needed to account for the dependence of the counting efficiency (and
885 self-absorption) on the unknown particle energy.
886 Another option that is often used for the beta counting efficiency is an exponential curve, which
887 has the form
£ =£ e am (1727}
m zero V1 ' -^')
888 where:
889 m = is the residual mass of the test source
890 £m = the counting efficiency at mass m
891 szero = the "zero" mass counting efficiency
892 a = constant determined by the data fit
893 Then the combined standard uncertainty of £m is:
MARLAP JULY 2001
DO NOT CITE OR QUOTE 17-34 DRAFT FOR PUBLIC COMMENT
-------
Data Acquisition, Reduction, and Reporting
t/c(8j = e -™a2u\ni) + t/2(8zero) + /n V(a) - 2/73 t/(8zero, a) (17.28)
894 Again, an additional uncertainty component may be needed when the identity of the beta-emitting
895 radionuclide is unknown.
896 Crosstalk, sometimes called "spill over," refers to the misclassification of alpha- and beta-
897 produced counts in a proportional counter which is designed to count both particles
898 simultaneously. It occurs when counts produced by alpha interactions in the detector are
899 registered as beta counts and vice versa. In order to accurately record the alpha and beta activities
900 of sources containing radionuclides emitting both particles, corrections must should be made for
901 crosstalk.
902 The number of alpha interactions registered as beta counts will increase as the source self-
903 absorption increases. The opposite is true for beta crosstalk, in that the number of beta
904 interactions falsely designated as alpha counts decreases with source self-absorption. Thus,
905 crosstalk correction factors vary with test-source mass and should be developed for the range of
906 test-source masses to be encountered. Commercially available proportional counters have
907 established programs to assist in the establishment of alpha and beta crosstalk factors. The
908 algorithms to correct for crosstalk are presented below.
909 The alpha in beta crosstalk, Xa, is defined as:
910 The respective counts in the alpha channel (a) and those in the beta channel (p) counts are
911 measured with a pure alpha-emitting source. Likewise, the beta in alpha crosstalk, Xp, is:
X' = (1730>
912 The respective alpha (a) and beta (p) count rates are measured with a pure beta-emitting source.
913 The relationship between Xa and Xp is given by:
a = a
-------
Data Acquisition, Reduction, and Reporting
P = P, - P,Xp + a,Xa (17.32)
914 Equation 17.3 1 states that the recorded alpha count rate, a, consists of the actual alpha count rate,
915 c^ (the total alpha count rate in both the alpha and beta channels due to only alpha interactions),
916 minus those alpha interactions recorded in the beta channel, plus those beta counts recorded in
917 the alpha channel. Equation 17.32 states the equivalent of Equation 17.3 1 for beta counts.
918 Solving the equations simultaneously for ad and pd gives:
a-X(a + p)
ad= , (17.33)
~ ~
R «
P)(! - Xp)2
~~
" Xo " Xp
(17.36)
V • /
920 Since crosstalk factors vary with radionuclide, additional uncertainty components may be needed
921 when the identities of the alpha and beta emitting radionuclides are unknown.
922 Processors execute many other functions for instruments which do not perform spectrometry.
923 These instruments include proportional counters, scintillation detectors, ionization chambers, and
924 special instruments (Chapter 15). The functions performed by processors may include instrument
925 control (sample change, gas flow control, etc.) and the calculations necessary to convert the basic
926 counting information to final form data or to some intermediate step.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 17-36 DRAFT FOR PUBLIC COMMENT
-------
Data Acquisition, Reduction, and Reporting
927 Data reduction functions which may be performed for scintillation detectors, ionization
928 chambers, and special instruments include the following:
929 • Background determination and subtraction;
930 • Conversion of total counts to counts per second;
931 • Calculate activity using calibration data;
932 • Calculate concentration using activity and operator input data;
933 • Perform efficiency calibrations;
934 • Calculate counting and total uncertainty;
935 • Cross talk determination and correction;
936 • Self-absorption corrections;
937 • Radioactive decay corrections; and
938 • Quality control (QC) functions (efficiency and background verification).
939 The output of manual systems usually requires further reduction to render it usable. The
940 information generated by processor-based systems may also need further processing.
941 These additional calculations may be performed using a calculator or by a computer using
942 general or custom software programs. The data may be electronically transferred to the
943 processing computer by a local area network (LAN) or on a computer disk. In some cases the
944 processing software may be part of the LEVIS.
945 17.5 Reporting Data
946 Quality assurance planning documents will give the level of data reporting required. This level
947 will vary from simply confirming the presence or absence of an analyte to a complete reporting
948 of all measurements, calibration data, documentation of the performance of laboratory processes,
949 provision of certain instrument counting reports, and QC sample results and analysis. Another
950 way of viewing this is as a tiered approach where preliminary studies or site surveys may only
951 require a minimum of data reporting, while a final site survey may require a detailed reporting of
952 the results. The necessary elements for data reporting are connected to the purpose for which the
953 data will be used (data quality objectives).
954 MARLAP recommendations for data reporting are that the reported value of a measurement
955 result: (1) be reported directly as obtained, with appropriate units, even if they are negative
956 values, (2) be expressed in an appropriate number of significant figures, and (3) include an
957 unambiguous statement of the uncertainty. The appropriate number of significant figures is
958 determined by the magnitude of uncertainty in the reported value. Each reported measurement
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 17-37 DO NOT CITE OR QUOTE
-------
Data Acquisition, Reduction, and Reporting
959 result should include the value and an estimate of the uncertainty (expanded uncertainty) (ANSI
960 42.23).
961 17.5.1 Sample and Analysis Method Identification
962 Sample data are normally reported by sample number, including both the field (project) and
963 laboratory assigned identifiers. In addition, the submitting laboratory should be identified as well
964 as the analysis method (ANSI 42.23, p. 38). Other information which can assist in the review and
965 interpretation of the data may be requested. This could include sample collection date (decay
966 correction reference date), analysis date, chain-of-custody (COC) number, and site or project
967 name.
968
17.5.2 Units and Radionuclide Identification
969 The individual radionuclides should be identified or, for gross analyses, the category, e.g., gross
970 alpha/beta, should be reported. Reporting units are likely specified by project planning
971 documents. If not specified, when possible, International System of Units (SI) units are preferred.
972 However, since regulatory compliance levels are usually quoted in traditional radiation units, it
973 may be appropriate to report in both SI and traditional units with one being placed within a
974 parenthesis. Both the SI and non-Si units are shown in Table 17.1 for common matrices.
TABLE — 17.1 Units For Data Reporting
Matrix
Airborne Particulates and Gas
Liquids
Solids
Surfaces
In Non-Si Units
pCi nT3
pCi L-1
pCikg-'orpCig-1
dpm/100cm2
In SI Units
BqnT3
BqL~'
Bqkg-1
Bq / 100 cm2
Conversion Factor From
Non-Si to SI Units
3.70 x 10~2
3.70 x 10~2
3.70x 10-2or37
1.67 x 1Q-2
975 17.5.3 Values, Uncertainty, and Significant Figures
976 The value, as measured, including zero and negative numbers, and the measurement uncertainty
977 (either expanded uncertainty or the combined standard uncertainty) should be reported in the
978 same units (Chapter 19). In general, environmental radiation measurements seldom warrant more
979 than two or three significant figures for the reported value, and one or two significant figures for
980 the uncertainty. As recommended in Chapter 19, Section 19.3.6, the measurement uncertainty
981 should be rounded to two significant figures, and both the value and uncertainty reported to the
MARLAP
DO NOT CITE OR QUOTE
17-38
JULY 2001
DRAFT FOR PUBLIC COMMENT
-------
Data Acquisition, Reduction, and Reporting
982 resulting number of decimal places. For example, a value of 0.8961 pCi/L with an associated
983 measurement uncertainty of 0.0234 should be reported as 0.896 ± 0.023 pCi/L. The MDC should
984 be reported to two significant figures (ANSI 42.23, p38). It should be noted that truncation
985 should only occur in reporting the final results (Section 18.3.6).
986 17.5.4 Other Information to be Provided on Request
987 Information which should be documented and retained for provision, if requested, includes
988 (ANSI 42.23, p38):
989 • Total weight or volume of the sample submitted and analyzed;
990 • Identification and documentation of specific analysis processes and analyst;
991 • Specific analytical parameters, i.e., chemical yields, counting times, decay factors, efficiency
992 of detectors used;
993 • Date, time, and place of sampling;
994 • Sample receipt information; and
995 • QC data demonstrating the quality of the measurement.
996 17.6 Data Packages
997 Project planning documents (Chapter 4) and analytical statements of work (Chapter 5) will
998 usually define the requirements of the final data submittal. Many projects will specify a data
999 package which contains not only the data reports described in the preceding section, but other
1000 supporting information to further describe, document, and define the analytical process. These
1001 additional requirements may be instituted to provide a basis for data verification/validation
1002 (Chapter 8), the purpose of which is to confirm that the data meet project quality objectives
1003 (Chapter 2). Material which may be required as part of a data package is discussed in Chapter 5.
1004 17.7 Electronic Data Deliverables
1005 Many project planning documents and SOWs require that laboratory data be delivered in
1006 electronic format, commonly called electronic data deliverables (EDD). This allows the data to
1007 be directly entered into a project database or, in some cases, into validation programs, and avoids
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 17-39 DO NOT CITE OR QUOTE
-------
Data Acquisition, Reduction, and Reporting
1008 transcription errors. There is no universal format for presenting EDDs, so the laboratory may be
1009 required to produce them in various formats. While the record structure of the EDD may vary in
1010 terms of the length and order of the fields, it is likely that the following are examples that may be
1011 requested:
1012 • Field Sample Number
1013 • Laboratory Sample Number
1014 • Sample Collection or Reference Date
1015 • Sample Receipt Date
1016 • Analysis Date
1017 • Result Identifier (sample or type of QC sample)
1018 • Radionuclide
1019 • Result
1020 • Results Units
1021 • Measurement Uncertainty
1022 • Sample Aliquant Size
1023 • Aliquant Size Units
1024 • Minimum Detectable Concentration
1025 • Minimum Quantifiable Concentration (MQC)
1026
1027 More information on EDDs may be found at the following websites:
1028 More information on EDDs may be found at the websites listed here. The U.S. Department of
1029 Energy EDD may be found at: (http://www.em.doe.gov/namp/pitimp.html) or (http://www.
1030 em.doe.gov/namp/deemmeet.html). Another EDD that is more general has been developed. It
1031 is called the General Electronic Data Deliverable (GEDD) and may be found at the website:
1032 (http://ersmo.inel.gov/edd/gedd.html#Entity Relationship Diagram). The EPA Environmental
1033 Data Registry may be found at: (http://www.epa.gov/edr/). U.S. Air Force Environmental
1034 Resources Program Management System (ERPREVIS) website: (http://www.afcee.brooks.af
1035 mil/ms/msc_irp.htm) also provides useful information on environmental databases and
1036 EDDs.
1037 EDDs may be transmitted by direct electronic transfer, e-mail, or by diskette.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 17-40 DRAFT FOR PUBLIC COMMENT
-------
Data Acquisition, Reduction, and Reporting
1038 17.8 References
1039 17.8.1 Cited References
1040 American National Standards Institute (ANSI) N42.14. "Calibration and Use of Germanium
1041 Spectrometers for Measurement of Gamma-Ray Emitting Radionuclides," 1991, New York.
1042 American National Standard Institute (ANSI) N42.23. Measurement and Associated
1043 Instrumentation Quality Assurance for Radioassay Laboratories. 1996.
1044 Baedecker, P. A. 1971. Digital Methods of Photopeak Integration in Activation Analysis,
1045 Analytical Chemistry, 43:405.
1046 Canberra Nuclear. 1994. VMS Spectroscopy Applications Package User's Manual, Canberra
1047 Industries, Inc.
1048 Decker, K. M., Sanderson, C. G. 1992. A Reevaluation of Commercial IBM PC Software for the
1049 Analysis of Low Level Environmental Gamma-Ray Spectra, Appl. Fadiat. Isot, Vol. 43, No.
1050 !/2, p. 323.
1051 U.S. Environmental Protection Agency (EPA). 1995. Good Automated Laboratory Practices.
1052 Directive 2185, Office of Information Resources Management, Research Triangle Park, NC.
1053 http://www.epa.gov/irmpoli8/irm_galp/index.html
1054 Gilmore, G. and Hemmingway, J. D. 1995. Practical Gamma-Ray Spectrometry, John Wiley,
1055 Chichester.
1056 Hagee, G. R., Karches, G. J., and Goldin, A. S. 1960. Determination of 1-131, Cs-137, and Ba-
1057 140 in Fluid Milk by Gamma Spectroscopy, Talanta, Vol 5,, p. 36.
1058 Helmer, R. G. and McCullough, C. M. 1983. Gauss VII, a Computer Program for the Analysis of
1059 Gamma-Ray Spectra From Ge Semiconductor Spectrometers, Nucl. Instr. andMeth., 206:
1060 Loska, L., 477.
1061 Kruse, H. and Spettel, B. 1982. A Combined Set of Automatic and Interactive Programs for
1062 INAA, J. Radioanalytical Chemistry, 70: 427.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 17-41 DO NOT CITE OR QUOTE
-------
Data Acquisition, Reduction, and Reporting
1063 Leyba, J. D. 1992. Gross alpha/beta determination by liquid scintillation counting. Westinghouse
1064 Savannah River Company, WSRC-TR-92-079.
1065 Loska, L. 1988. A Modification of the Total Peak Area Method for Gamma Ray Spectra, Appl.
1066 Radial Isot, 39: 475.
1067 Loska, L. and Ptasinski, J. 1994. A Simple Method for Peak-Area Determination of Multiplets,
1068 Radioactivity & Radiochemistry, Vol. 5, No. 4, p. 26.
1069 Mariscotti, M. A. 1967. A method for automatic identification of peaks in the presence of
1070 background and its application to spectrum analysis, Nuclear Instruments and Methods, 50:
1071 309-320.
1072 McDowell, J. And McDowell, B. L. 1993. "The Growth of a Radioanalytical Method: Alpha
1073 Liquid Scintillation Spectrometry." In Liquid Scintillation Specrometry 1992, ed. Noakes, J.
1074 E., Schoenhofer, F. And Polach, H. A. Tucson: Radiocarbon, 1993.
1075 National Council on Radiation Protection and Measurements (NCRP). 1978. A Handbook of
1076 Radioactivity Measurements Procedures, Report No. 58, p. 159.
1077 Passo, C. J. And Kessler, M. 1992. The Essentials of Alpha/Beta Discrimination.
1078 Passo, C. J. and Cook, G. T. 1994. Handbook of Environmental Liquid Spectrometry, Meriden,
1079 CT.
1080 Tsoulfanidis, N. 1983. Measurement and Detection of Radiation, McGraw-Hill, New York.
1081 U.S. Public Health Service (USPHS). 1967. Radioassay Procedures for Environmental Samples,
1082 Publication No. 999-RH-27.
1083 17.8.2 Other Sources
1084 A Handbook of Radioactivity Measurements Procedures, Report 58 (1984), National Council on
1085 Radiation Protection and Measurements, Washington, DC.
1086 American National Standard Calibration and Usage of Alpha/Beta Proportional Counters,
1087 ANSI/42.25-1997, IEEE, New York.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 17-42 DRAFT FOR PUBLIC COMMENT
-------
Data Acquisition, Reduction, and Reporting
1088 Browne, E. and Firestone, R. B. 1986. Table of Radioactive Isotopes, Wiley, New York.
1089 Calibration and Use of Germanium Spectrometers for the Measurement of Gamma-ray Emission
1090 Rates for Radionuclides, ANSI/42.14-1991, IEEE, New York.
1091 Debertin, K. and Helmer, R. G. 1988. Gamma- and X-ray Spectrometry with Semiconductor
1092 Detectors., North Holland, Amsterdam.
1093 Dewberry, A. 1997. Measurement of Uranium Total Alpha-Particle Activity by Selective
1094 Extraction and Photon/Electron-Rejecting Liquid Scintillation (PERALS) Spectrometry,
1095 Radioactivity and Radiochemistry, Vol. 8, No. 2.
1096 Escobar, G, Tome, F. V., and Lozano, J. C. 1999. Extractive Procedure for Radium-226
1097 Determination in Aqueous Samples By Liquid Scintillation Counting, Radioactivity and
1098 Radiochemistry, Vol 10, No. 1.
1099 Galloway, R. B. 1993. Correction for Sample Thickness in Activity Determination by Gamma-
1100 Ray Spectrometry, Radioactivity & Radiochemistry, Vol. 4, No. 3, p. 32.
1101 Harbottle, G. 1993. A Marinelli Beaker Modified for Easier Mathematical Modeling for Self-
1102 Absorption in Environmental Radioactivity Measurements, Radioactivity & Radiochemistry,
1103 Vol. 4, No.3, p. 20.
1104 Harvey, B. R. And Sutton, G. A. 1970. Liquid scintillation counting of nickel-63. Intl. J. Appl.
1105 Rad. Isot. 21: 519-523.
1106 Holm E., et. al. 1984. Determination of Tc-99 in environmental water samples. Nucl. Instrum.
1107 Phys. Res. 23: 204-207.
1108 IAEA. 1991. X-ray and Gamma-ray standards for Detector Calibration, IAEA-TECDOC-619,
1109 IAEA, Vienna.
1110 American National Standard Institute/Institute of Electrical and Electronics Engineers,
1111 (ANSI/IEEE) 325. Standard Test Procedures for Germanium Gamma-Ray Detectors. 1996.
1112 Killian, E. W., Koeppen, L. D., and Fermec, D. A. 1994. Quality-Assurance Techniques used
1113 with Automated Analysis of Gamma-Ray Spectra, Radioactivity & Radiochemistry, Vol. 5,
1114 No. 4, p. 34.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 17-43 DO NOT CITE OR QUOTE
-------
Data Acquisition, Reduction, and Reporting
1115 Kocher, D. C. 1981. Radioactive Decay Tables, U. S. Department of Energy Report DOC/TIC -
1116 11029.
1117 Knoll, G.F. 1989. Radiation Detection and Measurement, 2nd edn., John Wiley, New York.
1118 Koskelo, M. I, Burnett, W. C., and Cable, P. H. 1996. An Advanced Analysis Program for
1119 Alpha-Particle Spectrometry, Radioactivity & Radiochemistry, Vol. 7, No. 1, p. 18.
1120 Nuclear Data Sheets, Academic Press, Orlando.
1121 Oxford Instruments Inc. 1995. LB4100- W - Low Background System, Version 1.10.
1122 Quittner, P. 1972. Gamma-Ray Spectrometry, Adam Hilger LTD, London.
1123 Shirley, V. S. and Lederer, C. M. 1978. Table of Isotopes, 7th edition, Wiley Interscience, New
1124 York.
1125 Standard Practices for the Measurement of Radioactivity, ANSI/D 3648 - 1995, New York.
1126 Standard Test for High-Resolution Gamma-Ray Spectrometry of Water, ANSI/D 3 649 - 1991,
1127 New York.
1128 Standard Test Methods for Detector Calibration and Analysis of Radionuclides, ANSI/E 181-
1129 1993, New York.
1130 U.S. Environmental Protection Agency (EPA). 1980. Upgrading Environmental Radiation Data-
1131 Health Society Committee Report HPSR-1 (1980), Watson, I.E., Chairman, EPA 520-1-80-
1132 012, Office of Radiation Programs, Washington, DC.
1133 Yule, H. P. 1995. Could Your Gamma-Ray Spectrum Analysis Reports Survive an Audit?,
1134 Radioactivity & Radiochemistry, Vol. 6, No. 4, p. 4.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 17-44 DRAFT FOR PUBLIC COMMENT
-------
18 LABORATORY QUALITY CONTROL
2 18.1 Introduction
3 This chapter addresses internal laboratory quality control (QC), the purpose of which is to
4 monitor performance, identify problems, and initiate corrective action. If project requests are
5 more stringent than typical laboratory QC needs, the project manager and the laboratory should
6 confer to see whether the laboratory can accommodate the tightened QC requirements. Labora-
7 tory data should be produced in a quality system1 that incorporates planning, implementing, and
8 internal assessment of the work performed by the laboratory, including QC. While this chapter
9 focuses on laboratory QC, MARLAP fully endorses the need for a laboratory quality system and
10 a Quality Manual that delineates the quality assurance (QA) policies and QC practices of the
11 laboratory. General requirements for testing laboratories can be found in ISO/IEC 17025.
12 The chapter's purpose is to provide guidance to laboratory staff on those activities and profes-
13 sional practices a radioanalytical laboratory should undertake to produce data of known quality.
14 This chapter also shows how to use statistical techniques to monitor specific measures of the
15 analytical process to indicate the level of control of the analytical process within the laboratory.
16 These measures are called "performance indicators," and the statistical techniques involve the
17 use of control charts. Monitoring performance indicators through control charts enables the
18 identification of trends. The laboratory can then address analytical problems and help improve
19 the analytical process. Section 18.3.2 and Attachment ISA at the end of this chapter provide
20 examples of several types of charts. The use of statistical techniques is the preferred method for
21 implementing quality control in the laboratory (Attachment 18B). The chapter also identifies
22 specific performance indicators, the principles that govern their use, indications and underlying
23 causes of excursions, statistical means of evaluating performance indicators, and examples of
24 root-cause evaluations.
25 The control of the analytical process in the laboratory is distinct from meeting the typical
26 analytical needs of a specific project. This chapter addresses the former, to the extent that QC
27 provides quantitative estimates of analysis and measurement controls that can be used to
28 determine compliance with project objectives.
JA quality system is a structured and documented management framework that describes the policies, objectives,
principles, organizational authority, responsibilities, accountability, and implementation plan of an organization for
ensuring quality in its work processes, products (items), and services. The quality system provides for planning,
implementing, and assessing the work performed by the organization and for carrying out required quality assurance
and quality control (ANSI/ASQC E4, 1994).
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 18-1 DO NOT CITE OR QUOTE
-------
Laboratory Quality Control
29 18.1.1 Organization of Chapter
30 Chapter 18 has five major sections in addition to this introduction. Section 18.2 provides a
31 general overview of QC and its application in the laboratory setting. Section 18.3 discusses the
32 importance of evaluating performance indicators and provides statistical means for their evalua-
33 tion. Sections 18.4 and 18.5 identify primary radiochemistry and instrumentation performance
34 indicators, respectively, and discuss each in detail. Section 18.6 discusses other aspects of the
35 analytical process that require scrutiny but are not formally considered performance indicators.
36 18.1.2 Format
37 The chapter is presented in a different format than the preceding chapters in order to highlight the
38 performance indicators and to give examples. For each performance indicator, general guidance
39 is provided in the format shown below.
40 Issue: Defines and summarizes the performance indicator
41
42 Discussion: Identifies those matters important to the performance indicator, including:
43 • What is the performance indicator and how does it work?
44 • Why is the performance indicator important, and what is its impact on the quality of the
45 measurement?
46 • What is the relationship of the performance indicator and the combined standard uncertainty
47 derived for the analytical method?
48 • What are the acceptable limits of the performance indicator?
49 • What are the key assumptions underlying the performance indicator?
50 • What limits and cautions are associated with the assumptions made?
51 • How sensitive is the quality of the measurement to the assumptions made?
52 • What is the appropriate frequency for assessing this performance indicator?
53 Excursions: "Excursions" are departures from the expected condition. This section addresses the
MARLAP JULY 2001
DO NOT CITE OR QUOTE 18-2 DRAFT FOR PUBLIC COMMENT
-------
Laboratory Quality Control
54 likely types of excursions encountered during laboratory analysis and explains what each may
55 indicate. This section also discusses the potential reasons for these excursions and the
56 implications for the analytical results.
57 Examples: Where appropriate, this section provides typical examples of excursions, potential
58 reasons for excursions, and additional information.
59 18.2 Quality Control
60 Quality control includes all technical activities that measure the attributes and performance of a
61 process, item, or service against defined standards to verify that they meet the stated require-
62 ments established by the customer. It also includes operational techniques and activities that are
63 used to fulfill requirements for quality (ANSI/ASQC E4, 1994).
64 QC may not always detect blunders. Good laboratory practices, in addition to adherence to
65 standard operating procedures (SOPs), are part of the overall QA/QC aspects needed to check the
66 laboratory's performance. To monitor and control quality, laboratories use performance indica-
67 tors, which are instrument- or protocol-related parameters that are routinely monitored to assess
68 the laboratory's estimate of measurement uncertainty, precision, bias, etc. Initially, these
69 parameters are used to maintain or demonstrate control over the analytical process. The
70 performance indicators should be tracked by appropriate personnel. If the performance indicator
71 control limits are exceeded, management should be informed and corrective action should be
72 initiated.
73 Table 18.1 lists some of the potential causes for radioanalytical control excursions. By no means
74 is the list complete, and the reader should be aware of additional potential causes of excursions
75 that are presented in the rest of this chapter and the other chapters. Many problems are complex
76 and have multiple components that could complicate the search for causes of protocol or instru-
77 ment related excursions. A metrologist or radiochemist should be consulted to identify and
78 remedy any analytical problems.
79
80
81
82
83
84
r
Radiochemical
Processing
Laboratory blunder
Processing
difficulty
r ART F 18.1 —
Source
Preparation
Laboratory
blunder
Poor mounting
Poor Dialing
'rohlems leading to loss of analytical control
Instrument Related
Laboratory blunder
Electronic malfunction
• preamplifier
• power supply
• guard
Other
Laboratory
blunder
Data transcription
error
JULY 2001
DRAFT FOR PUBLIC COMMENT
18-2
MARLAP
DO NOT CITE OR QUOTE
-------
Laboratory Quality Control
Radiochemical
Processing
Questionable
reagent purity
Low tracer/carrier
recovery
Excessive
tracer/carrier
recovery
Inaccurate
aliquanting of
tracer/carrier
Sample aliquanting
inaccuracy
Cross-
contamination
Inadequate
dissolution of
sample
Complex matrix
Sample
heterogenity
Source
Preparation
Improper
geometry
Incorrect thin
plastic film
thickness
Improper
plating on the
planchet
Excessive
source mass
Uncorrected
self absorption
Quenching
Recoil
contamination
Instrument Related
analog to digital converter (ADC)
gain
high voltage
discriminator
pole zero
shape constant
Improper source or sample geometry
Poor counting statistics
Poor detector resolution
Detector contamination
Inappropriate/out-of-date efficiency, background or
calibration factor
Background shift
Incorrect nuclear transformation data or other constants
Variable memory effects
Peak/calibration shift
Counting gas
• pressure too high, too low, or variable
• gas impurity
Loss of vacuum/coolant
Temperature and humidity fluctuation
Measurement uroblem
Other
Incorrect units
Calculation error
Software
limitation
Computer
problem
Loss of electrical
power
Electrical power
fluctuations
Mislabeling
Loss of sample
Insufficient
sample
information
Data processing
problem
Interfering
radionuclides
85
86
87
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
18.3 Evaluation of Performance Indicators
106 18.3.1 Importance of Evaluating Performance Indicators
107 As stated previously, performance indicators are measures of the analytical process that the
108 laboratory monitors as part of its routine QC program. Performance indicators demonstrate
109 whether the analytical process is performing as planned, when it has exhibited a statistical
110 anomaly that requires investigation, and when a system has failed. Accordingly, monitoring
111 performance indicators using established statistical techniques provides the laboratory with an
112 effective tool for self assessment that allows the identification of trends or conditions that, while
113 still within the established bounds of acceptability, are drifting or trending out of control. These
MARLAP
DO NOT CITE OR QUOTE
18-4
JULY 2001
DRAFT FOR PUBLIC COMMENT
-------
Laboratory Quality Control
114 conditions can be addressed prospectively, allowing the laboratory to maintain analytical control.
115 Additionally, this process allows the development of a data base regarding a protocol's or
116 system's behavior over time or under a specified set of conditions.
117 18.3.2 Statistical Means of Evaluating Performance Indicators — Control Charts
118 The primary tool for statistical quality control is the control chart (see Attachment ISA). The
119 theory that underlies a control chart is statistical hypothesis testing (see Appendix C). The
120 implementation of a control chart makes the theory transparent to the average user and reduces
121 the process of statistical inference to answering simple questions, such as, "Is the measured
122 parameter greater than the upper control limit?" or "Is the measured parameter in the warning
123 region?"
124 In theory, to test whether a parameter 6 is above or below a certain value 60, a test statistic is
125 defined and its distribution is determined under the assumption that 6 = 60 (the null hypothesis).
126 The value of the statistic is calculated and compared to critical values to test the assumption. In
127 practice, a control chart is designed so that a non-statistician can perform these tests easily by
128 comparing the measured value of the parameter to control limits and warning limits.
129 Most control charts do not implement hypothesis tests in a rigorous manner that allows decision
130 error rates to be precisely determined. The charts are intended to be simple and practical tools for
131 use even in situations where the assumptions needed for a rigorous test are not verifiable.
132 Every control chart has control limits, which define the acceptable range of the monitored
133 variable. Many charts have both upper and lower limits. However, when changes in only one
134 direction are of concern, only one limit is necessary. Most control charts have a central line, or
135 reference line, which is an estimate of the expected value of the monitored variable. Many
136 control charts also have warning limits, which lie between the central line and the control limits.
137 By definition, control limits are action limits. A single measured value that falls outside these
138 limits requires that one stop the measurement process, investigate the problem, and if necessary
139 take corrective action. The warning limits are optional but recommended, since they help one to
140 identify and investigate possible problems before control limits are exceeded.
141 Types of Control Charts: Control charts based on grouped observations often are more power-
142 ful tools for detecting shifts of the monitored variable than charts based on individual observa-
143 tions. Average charts, or X charts, are used to monitor the arithmetic means of measured values
144 obtained in "rational subgroups," which are subgroups of equal size chosen to ensure that the
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 18-5 DO NOT CITE OR QUOTE
-------
Laboratory Quality Control
145 measurement variability within each subgroup is likely to represent only the inherent variability
146 of the measurement process produced by non-assignable causes (see Attachment ISA). When an
147 X chart is used, a range chart, or R chart, is generally used in tandem to monitor within-group
148 variability. (The range of a set of values is the difference between the largest value and the
149 smallest.)
150 A control chart for individual values (X chart or I chart) is used when it is impractical to obtain
151 measured values in the groups needed for an X chart. In this case, a moving range chart (MR
152 chart) is often used as well to monitor variability. The moving range chart is an R chart based on
153 the absolute differences between consecutive measured values.
154 A control chart may or may not be based on a particular type of data distribution. Most control
155 charts use limits derived from the normal distribution but are intended to be used for data with
156 almost any distribution (ISO 8258). However, when data obtained from radiation counters are
157 monitored, the Poisson distribution may often be assumed. The standard types of control charts
158 for Poisson data in industrial applications are called "c charts" (for total counts) and "u charts"
159 (for count rates). A third type of Poisson control chart, which is a variant of the u chart, is
160 frequently used to monitor radiation counter efficiency. When the data distribution is Poisson,
161 separate charts for monitoring the value of the parameter and its variability are generally
162 unnecessary because the mean and variance of a Poisson distribution are equal.
163 The following documents provide more guidance on the use of control charts:
164 • ASTMD6299. Standard Practice for Applying Statistical Quality Assurance Techniques to
165 Evaluate Analytical Measurement System Performance.
166 • ASTM E882. Standard Guide for Accountability and Quality Control in the Chemical
167 Analysis Laboratory. ANSI/ISO/ASQC A3 534-2. Statistics- Vocabulary and Symbols-
168 Statistical Quality Control.
169 • ISO 7870. Control Charts - General Guide and Introduction.
170 • ISO 7873. Control Charts for Arithmetic Average with Warning Limits.
Ill • ISO 7966. Acceptance Control Charts.
172 • ISO 8258. Shewhart Control Charts.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 18-6 DRAFT FOR PUBLIC COMMENT
-------
Laboratory Quality Control
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
• American Society for Testing and Materials (ASTM) MNL 7, Manual on Presentation of
Data and Control Chart Analysis ASTM Manual Series, 6th Edition, 1990.
Figure 18.1 illustrates a typical control chart using counting data of a standard reference material
(with limits corrected for decay) showing the statistical nature of the chart. The applicability of
control chart techniques is based on the assumption that laboratory data approximate a normal
distribution like that shown on the left of the vertical axis in the figure. The counting data plotted
graphically represent the test results on the vertical axis and the scale order or time sequence in
which the measurements were obtained on the horizontal axis. The mean of the measurements is
represented by the central line (CL), and the limits of dispersion in terms of standard deviation
are represented by the upper and lower warning and control limits (UWL, UCL, LWL, LCL). The
warning limits are usually 2 standard deviations from the mean and the control limits are 3
standard deviations from the mean.
UCL
x+30
x+20
(0
4M>
c
3
O
o
x-20
x-3o
Time
FIGURE 18.1 — Control chart for daily counting of a standard reference source, with limits corrected for
decay. Statistical nature of chart is illustrated on the left by the Gaussian curve.
18.3.3 Measurement Uncertainty
Issue: Since laboratory radioactivity measurements always involve uncertainty, every measured
result is uncertain to some degree. If the measurement uncertainties are large relative to the
tolerances needed for decision making, the data may not be useful for their intended purpose. A
discussion of measurement uncertainty is contained in Chapter 19, and the terms used in this
section are defined in that chapter and in the Glossary.
JULY 2001
DRAFT FOR PUBLIC COMMENT
18-7
MARLAP
DO NOT CITE OR QUOTE
-------
Laboratory Quality Control
191 Discussion: In order to determine the significance of a sample result, all reported values should
192 be accompanied by the laboratory's best estimate of the uncertainty associated with the result.
193 The "combined standard uncertainty" (one-sigma uncertainty) is obtained by propagating the
194 uncertainties of all the input quantities that contribute to the calculation of the derived value
195 (Chapter 19).
196 The combined standard uncertainty is used to indicate the statistical confidence in interpreting
197 the performance indicator's ability to assess analytical quality. The estimated statistical confi-
198 dence level that is usually associated with 1 combined standard uncertainty is about 68 percent,
199 the confidence level for 2 combined standard uncertainties is about 95 percent, and the confi-
200 dence level for 3 combined standard uncertainties is about 99 percent. It is important that the
201 combined standard uncertainty be a fair estimate because it will indicate when the analytical
202 process could be approaching the limits of statistical control and corrective actions should be
203 initiated. A performance indicator exceeding ±2 combined standard uncertainty limits from the
204 indicator's historical mean value may indicate that corrective action should be considered, and a
205 performance indicator exceeding ±3 combined standard uncertainty limits from the indicator's
206 historical mean value may indicate that an investigation must be conducted and corrective action
207 may be necessary. Because statistical confidence never reaches 100 percent, it probably would be
208 prudent to confirm the measurement for the performance indicator when it exceeds ±2 combined
209 standard uncertainty limits. If the performance indicator value for repeat measurements do not
210 exceed ±2 combined standard uncertainty limits, one may conclude that the first measurement
211 was a statistically allowable event. However, if the excursion is repeated, appropriate investiga-
212 tive actions should be considered.
213 Most of the significant sources of uncertainty in radiochemical data are known to a laboratory
214 and can be estimated. These include uncertainties associated with sample and background count-
215 ing, radiochemical yield determination, efficiency calibration, and blank assessment. Other less
216 easily defined but significant sources of uncertainty include those associated with self-absorption
217 and quench correction, sample density correction, sample geometry variation, gamma photopeak
218 area determination, determination of sample volume or weight, and dead time correction.
219 The uncertainty of a measured value is controllable, within certain limits, by decreasing the
220 uncertainty associated with some input parameters. For samples containing low levels of radio-
221 activity, a large component of the combined standard uncertainty may be associated with the
222 instrumental assessment (counting) of the sample aliquant, i.e., the standard uncertainty of the net
223 count (gross sample count minus background count). Increasing the total net count accumulated,
224 or decreasing the uncertainty of the instrument background, or both, will decrease the counting
225 uncertainty. Changes that may be made to decrease the counting uncertainty include increasing
MARLAP JULY 2001
DO NOT CITE OR QUOTE 18-8 DRAFT FOR PUBLIC COMMENT
-------
Laboratory Quality Control
226 the counting time for the sample or background, increasing the sample aliquant size (unless the
227 sample geometry, quench, or self-absorption factors offset the gain in total radioactivity counted),
228 using a more efficient geometry or detector, using an instrument with a lower background, and
229 reanalyzing the sample to obtain a greater radiochemical yield. It also may be possible to
230 concentrate the sample, which has the equivalent effect of increasing the sample aliquant size.
231 18.4 Radiochemistry Performance Indicators
232 Section 18.3 discussed how to evaluate radiochemistry performance indicators using statistically
233 based control chart techniques. Any of the indicators below (blanks, replicates, laboratory control
234 samples, matrix spikes, certified reference material, or tracer yield) can be evaluated using the
235 control chart techniques. Analysts can observe individual Z score values to identify loss of
236 control. Control charts will assist laboratory personnel in identifying the quality trends and
237 excursions of any performance indicator.
238
239 18.4.1 Method and Reagent Blank
240 Issue: A method blank is a sample of a matrix as similar as practical to the associated samples
241 that is free from the analytes (radionuclides) of interest to the extent possible. The method blank
242 is processed simultaneously with, and under the same conditions as, samples through all steps of
243 the analytical procedures. A reagent blank consists of the analytical reagent(s) in the procedure
244 without the target analyte or sample matrix, introduced into the analytical procedure at the
245 appropriate point and carried through all subsequent steps to determine the contribution of the
246 reagents and of the involved analytical steps.
247 Blank samples are used to determine whether any radionuclide contamination is introduced by
248 the measurement process. They assist in the control of any contamination introduced by the
249 laboratory. Ideally, no target analytes should be present in the blank at detectable concentrations.
250 If that is not possible (e.g., for naturally occurring radionuclides), those radionuclides should be
251 extremely well-characterized and tracked. Control charts can be used to track these radionuclide
252 levels in blanks. Using A'charts, the laboratory can establish a program that evaluates the levels
253 and trends of radionuclides in the different laboratory blanks. The techniques for establishing
254 such a control chart program are described in Attachment ISA.
255 Discussion: The method blank is assumed to be representative of all samples in the batch with
256 respect to the matrix and contamination assessment. When practical, it consists of the same or
257 equivalent medium as the analytical samples, such as a deionized water blank for aqueous
258 samples. Soil blanks are often prepared using "clean sand," commercially available fine-grained
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 18-9 DO NOT CITE OR QUOTE
-------
Laboratory Quality Control
259 or beach sand whose inherent concentrations of target radionuclides are small and have been
260 characterized sufficiently by the laboratory to allow its use as a blank. This approach may not be
261 appropriate for very low-level analyses. Powdered, natural-matrix Standard Reference Materials
262 (SRMs) are commercially available from National Institute of Standards and Technology (NIST)
263 and also may be suitable (Section 18.4.5). However, due to the natural variability of soils, each
264 choice of method blank medium must be evaluated by the laboratory prior to use. The results of
265 method blanks are not used to correct sample activities but only to monitor for contamination.
266 Reagent blanks are matrix-independent and assess any contamination only from the reagents and
267 lab-ware. They are used to correct sample activities for the contribution of naturally occurring
268 radionuclides in the reagents, and used like method blanks, to check for unexpected contamina-
269 tion. When reagent blank results are used to correct sample activities, it is important that the
270 blank results be carefully monitored using control charts.
271 It is common practice for some laboratories to add the reagents into a volume of deionized water
272 equal to the sample volume, while other laboratories simply add the required reagents to an
273 empty container and process it as an analytical sample. In either case, it should be noted that the
274 reagent blank is not monitoring the entire analytical process. The fundamental issue for each
275 laboratory is to decide on the appropriate reagent blank necessary to obtain the needed informa-
276 tion on the measurement system. Considerable variability exists among laboratories in the use
277 and preparation of reagent blanks.
278 In general, the reagent blank's concentration of analyte is expected to be small compared to that
279 of the sample. However, for some low-activity environmental samples this may not be the case,
280 and the correction becomes increasingly important as the concentration of the analyte in the
281 sample approaches background concentrations. In these cases, care should be taken to accurately
282 quantify the levels of radionuclides in the reagent blanks.
283 It is important to minimize radionuclide concentrations in the blanks and bring these levels under
284 control. This is usually achieved through careful selection of reagents, maintaining laboratory
285 and counting areas free from contamination, and by segregating high and low activity samples.
286 Thorough documentation of all blank values is essential to allow for the application of statistical
287 tests to evaluate potentially anomalous values and delineate their extent.
288 Ideally, the analyte concentration in a method or reagent blank should be as close to zero as
289 possible, and replicate measurement of the blanks should be consistent within counting statistics.
290 Acceptance criteria for blank results should be established and applied to all data, and should
291 include warning and control limits (Section 18.3.2, "Statistical Means of Evaluating Performance
MARLAP JULY 2001
DO NOT CITE OR QUOTE 18-10 DRAFT FOR PUBLIC COMMENT
-------
Laboratory Quality Control
292 Indicators — Control Charts"). Blank values require scrutiny as part of the data evaluation and
293 validation process for each analytical batch. Should restocking of reagents or other wholesale
294 laboratory changes occur during a project, the method and reagent blanks prepared under the new
295 conditions should be re-evaluated to ensure that they continue to be within established criteria.
296 An example of a numerical performance indicator for a method blank or a reagent blank used to
297 monitor for unexpected contamination is
Zelank =
298 where x denotes the measured blank activity and uc(x) denotes its combined standard uncertainty.
299 Warning limits for ^lank are ±2 and control limits are ±3. As mentioned earlier, if a reagent blank
300 is used to blank-correct sample results, the blank results should be evaluated using control charts.
301 Typically, one method blank and/or reagent blank is analyzed with each batch or grouping of
302 analytical samples regardless of batch size. Situations may occur where more frequent blanks are
303 required to ensure that analytical conditions are stable, particularly when analyzing high and low
304 concentration samples in the same analytical batch, or when instruments, reagents, or analytical
305 method are suspect.
306 In general, corrective actions include procurement control of reagents, good laboratory cleaning
307 practices, sample segregation according to anticipated concentrations, and instrument-related
308 concerns, as discussed in this section. Good laboratory cleaning protocols should incorporate the
309 evaluation of method and reagent blank performance to indicate if current practices are adequate.
310 Instrument background data indicate a system's stability, and can be used to pinpoint the source
311 of contamination, as can routine contamination (removable and fixed) surveys of laboratory and
312 counting areas that are performed by the organization's health physics or radiation safely
313 personnel.
314 Excursion: Blank changes can be grouped into three general categories: rapid changes, gradual
315 increase or decrease, and highly variable changes. These are represented in Figure 18.2 and
316 described below.
317 Rapid Changes: A sudden change in a blank value indicates the existence of a condition
318 requiring immediate attention. Sudden changes often are caused by the introduction of a
319 contaminant from high concentration samples, impure reagents, or contaminated sample
320 preparation areas. Laboratory cleaning practices and new or recently restocked reagents
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 18-11 DO NOT CITE OR QUOTE
-------
Laboratory Quality Control
321
322
323
324
325
326
327
328
329
330
331
332
333
should be checked. When a sudden, significant increase in the blank occurs in conjunction
with the introduction of new reagents through restocking or other changes, the causes should
be investigated and if the reagent is contaminated, the reagent contributing the activity should
be discarded and replaced. Particular attention should be paid to the samples counted directly
prior to the contaminated blank, since small amounts of residues from these samples can
contaminate the detector and have large effects on subsequent results when analyzing
samples at or near environmental background. It may be necessary to take swipe or smear
samples of questionable areas to identify the contaminant's source followed by a thorough
cleaning or decontamination of all affected areas. Additionally, method or reagent blank
values that are suddenly depressed should be investigated and may indicate other problems,
including instrument malfunction like a loss of counting gas, incomplete chemical separation
during the chemical preparation, or the failure to add necessary reagents. These other prob-
lems may be reflected in other areas, such as instrument performance checks or tracer yields.
BLANK CONTAMINATION
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
RAPID CHANGES
CROSS
CONTAMINATION
- Residual contamination
from high concentration
samples
PROCEDURE FAILURE -
INCOMPLETE
SEPARATION
INSTRUMENT
INSTABILITY
INTRODUCTION OF
CONTAMINATED
REAGENT
GRADUAL CHANGES
BUILDUP OF
CONTAMINATION
- Glassware/Laboratory
areas require thorough
cleaning
SUSPECTED REAGENTS
INAPPROPRIATE
PROCEDURES
INSTABILITY OF
CHEMICAL YIELD
MONITOR
INSTRUMENT DRIFT &
DETERIORATION
HIGH VARIABILITY
PROCEDURE FAILURE
INSTRUMENT
INSTABILITY
IMPROPER
SEGREGATION OF HIGH
& LOW ACTIVITY
SAMPLES
FIGURE 18.2 — Three general categories of blank changes
MARLAP
DO NOT CITE OR QUOTE
18-12
JULY 2001
DRAFT FOR PUBLIC COMMENT
-------
Laboratory Quality Control
349 Gradual Changes: Gradually increasing blank values indicate the need to inspect all sample
350 preparation and counting areas for sources of residual contamination. Often housekeeping or
351 routine contamination control details such as cleaning glassware or instrument counting
352 chambers are sufficient to bring blank values under control. Alternatively, gradually decreas-
353 ing blank values warrant scrutiny with respect to proper instrument settings and procedural
354 related problems like a lack of tracer/sample exchange, failure of chemical separation reac-
355 tions, or the addition of all necessary reagents. The importance of documenting method and
356 reagent blank values in this regard cannot be overemphasized, since data evaluation and
357 trending analyses are impossible without complete records.
358 High Variability: Because method blank values are expected to be near zero, the degree of
359 variability they exhibit should reflect the statistical variation inherent in radiometric
360 determinations near these levels. Large variations in blank values typically indicate problems
361 related to instruments or sample processing, as discussed in the two previous sections.
362 18.4.2 Laboratory Replicates
363 Issue: A laboratory replicate is two or more aliquants taken at the first subsampling event,
364 normally after homogenization. In the event that there is no subsampling (when the method calls
365 for using the entire sample) replicate analysis typically involves counting the prepared sample
366 twice. The results of laboratory replicates are used to evaluate the precision of the measurement
367 process. Note that counting a sample twice only assesses the instrument portion of the measure-
368 ment process.
369 Precision is a measure of agreement among replicate measurements of the same property under
370 prescribed similar conditions. Precision is a fundamental aspect of the analytical process and
371 should be evaluated routinely as part of the laboratory's quality system. Evaluation typically is
372 performed using multiple analysis of the same sample (blanks, spikes, blinds, reference
373 materials, performance evaluation samples, etc.), in whole or part, and evaluating the analyses
374 relative to a statistically based criterion. The range of sample types requires that the sample
375 matrix's effects on the precision be captured and evaluated by the laboratory's routine quality
376 control practices. The reproducibility of analytical results should be evaluated by replicates to
377 establish this uncertainty component.
378 Discussion: The purpose for measuring precision is to determine whether the laboratory can
379 execute an analytical method consistently and obtain results of acceptable variability. Analytical
380 samples cover a range of physical forms or matrices, from homogeneous samples like finished
381 drinking water to complex soils or heterogeneous wastes, and each matrix has the potential to
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 18-13 DO NOT CITE OR QUOTE
-------
Laboratory Quality Control
382 affect a protocol' s precision.
383 In general, precision for aqueous samples tends to be less affected by sample heterogeneity than
384 other media because if the sample's constituents are dissolved the sample is essentially homo-
385 geneous. This facilitates dividing the samples into equivalents fractions or aliquants. Multi-phase
386 and high-solid-content samples that are heterogeneous are more problematic.
387 The acceptance criterion for precision should be related to the combined standard uncertainties of
388 the measured results. The uncertainty of a result may depend on many factors (e.g., dissolved
389 solids in water or particle sizes of soil), but such factors should affect the acceptance criterion
390 only through their effect on the standard uncertainty.
391 As an alternative to sample duplicates, a matrix spike duplicate is sometimes used as an indicator
392 of the analytical precision, as discussed in Section 18.4.3. A matrix spike duplicate is treated in
393 the same manner as an unspiked replicate: both samples (original and duplicate) are processed
394 identically to the other samples in the batch, and each aliquant is treated as an individual sample.
395 If the sample has multiple phases, the phases should be separated for individual analysis. For
396 heterogenous materials, multiple analyses should be used, or the combined standard uncertainty
397 of the results should be increased, to account for subsampling error (Appendix F). A typical
398 frequency for replicate analyses is a minimum of one per analytical batch, regardless of batch
399 size. Batch is defined as samples of similar matrix type with associated QC samples analyzed
400 under the sample conditions at approximately the same time.
401 All analytical batches should be evaluated with respect to precision, whether by using replicates
402 or matrix spike duplicates. This is done typically by the use of an acceptance criterion that
403 derives a statistic that quantifies the difference between two values obtained by analyzing the
404 same sample. Limits are then placed on the criterion, and data for any batch in excess of the
405 criterion require investigation and corrective action as appropriate. An example of a numerical
406 performance indicator for laboratory replicates is
xi ~ xi
407 where xl and x2 denote the two measured activity concentrations and tfc(X) and Ue(x2) denote their
408 respective combined standard uncertainties. Warning limits for ZRep are ±2 and control limits
409 are ±3 .
MARLAP JULY 2001
DO NOT CITE OR QUOTE 18-14 DRAFT FOR PUBLIC COMMENT
-------
Laboratory Quality Control
410 Excursions: A regularly scheduled evaluation of precision with respect to the acceptance
411 criterion should be an integral part of the laboratory quality system. Careful attention should be
412 paid to the nature and anticipated analyte concentrations of all samples processed by the
413 laboratory. Prospective identification of samples where precision is expected to be problematic
414 often can address difficulties in this area. The choice of appropriate analytical method and analyst
415 training are also important. An analyst needs to be familiar with specific steps in the procedure
416 that provide an indication of incomplete processing.
417 Precision exhibits a range of values and depends in part on sample matrix and activity, assuming
418 correct execution of the analytical method. Small changes, positive and negative, are expected
419 and should be captured in the acceptance criterion's range. It is also sensitive to sample hetero-
420 geneity or errors in processing, such as incomplete chemical separation or sample dissolution,
421 and lack of tracer or carrier equilibration. When performance indicators for precision are outside
422 acceptance criteria, the laboratory should determine the reasons why and implement corrective
423 actions.
424 Certain samples will exhibit higher variability because of their matrix, or the proximity of their
425 analyte concentration to ambient background, as discussed previously. Consideration should be
426 given to cases where a matrix requires the development and implementation of a specific accep-
427 tance criterion. The main causes for lack of precision (Figure 18.3) can be grouped as follows:
428
429
430
431
432
433
434
435
436
437
438
439
440
Laboratory subsampling — subsampling techniques produced two dissimilar aliquants from
one sample, and the original and duplicate are not the same. An analyst should be careful to
ensure that the sample is thoroughly homogenized before subsampling.
DECREASE IN PRECISION
I
PROCEDURE PROBLEM
• Incomplete separation
• Improper processing
• Inappropriate tracer/carrier
• Inadequate analyst training
• Wrong reagent concentration
• Wrong ambient laboratory
conditions
• Reagent/labware change
INSTRUMENT FAILURE
• Counting instability
LABORATORY
SUBSAMPLING
Replicates not equivalent
1
MATRIX PROBLEM
• Matrix incompatible
• Excessive heterogeneity
PROCEDURE PROBLEM
• Analyst training required
• Procedure requires
modification
FIGURE 18.3 — Failed performance indicator: replicates.
JULY 2001
DRAFT FOR PUBLIC COMMENT
18-15
MARLAP
DO NOT CITE OR QUOTE
-------
Laboratory Quality Control
441 • Matrix - Sample constituents interfere with preparation chemistry, e.g., coprecipitation of
442 interfering non-analyte radionuclides from sample or excessive dissolved solids.
443 • Counting statistics - Sample activity is so low that small statistical variations in background
444 cause disproportionate responses.
445 • Contamination - Intermittent contamination from measurements system, glassware, etc.,
446 produces anomalous data for the original sample, but not the duplicate/replicate.
447 • Other - Failed chemical process, failed instrumentation, training, failed lab environment,
448 failed procurement control.
449 18.4.3 Laboratory Control Samples, Matrix Spikes, and Matrix Spike Duplicates
450 Issue: A laboratory control sample (LCS) is a QC sample of known composition (reference
451 material) or an artificial sample, created by fortifying a clean material similar in nature to the
452 environmental sample. The LCS is prepared and analyzed in the same manner as the environ-
453 mental sample. A matrix spike (MS) is an aliquant of a sample prepared by adding a known
454 quantity of target analytes to a specified amount of sample and subjected to the entire analytical
455 procedure to establish if the method or procedure is appropriate for the analysis of the particular
456 matrix. A matrix spike duplicate (MSD) is a second replicate matrix spike prepared in the lab-
457 oratory and analyzed to evaluate the precision of the measurement process.
458 An important performance indicator is the ability to ensure that the analytical methods employed
459 obtain data that are representative of the true activity in a sample, i.e., produce data that are
460 accurate. The routine analysis of spiked samples provide data for an evaluation of the labora-
461 tory' s reported measurement uncertainty and allow for the determination of bias, if one exists.
462 Evaluation is typically performed using prepared samples consisting of media equivalent to a
463 routine analytical sample with a known, measurable amount of the analyte of interest. Upon
464 completion of the analysis, the results are compared to the known or accepted value, and the
465 agreement is evaluated using a predetermined criterion. The range of sample types assayed in a
466 laboratory may require that spikes are prepared using several sample media. Use of matrix spiked
467 samples will reflect the analytical method's ability to make accurate quantitative determinations
468 in the presence of the matrix.
469 Discussion: As stated previously, analytical samples cover a range of physical forms or matrices,
470 and each matrix can change a method's expected bias. Tracking sets of LCS and matrix spike
471 results can give laboratory personnel an indication of the magnitude of bias. Care must be taken
MARLAP JULY 2001
DO NOT CITE OR QUOTE 18-16 DRAFT FOR PUBLIC COMMENT
-------
Laboratory Quality Control
472 when analyzing site specific matrix spike results because these matrices may be very complex
473 and subject to large variability. In general, aqueous samples tends to be less affected than other
474 media like soils or heterogeneous materials. However, multi-phase fluids, high solid content, and
475 brackish or saline waters may be more problematic.
476 The analyst should carefully consider the spiking levels for laboratory control samples and matrix
477 spikes. Spikes and LCSs may be prepared near the lower limits of detection to test the methods
478 performance on clean or slightly contaminated samples. Conversely, matrix spikes and LCSs
479 may be spiked at high levels for groups of highly contaminated samples. The laboratory should
480 try to spike at or near the action level or level of interest for the project.
481 Possible numerical performance indicators for laboratory control samples and matrix spikes are
r-, X ~ d
{'
(3)
x - XQ - d
ZMS = 2 2 =5= (4)
(x) + uc(x0) + uc(d)
482 where xis the measured value of the spiked sample, dis the spike concentration added, x0 is the
483 measured concentration of the unspiked sample, and uc2(x), u2(d), and U^(XQ) are the squares of
484 the respective standard uncertainties. The warning limits for either of these indicators are ±2 and
485 the control limits are ±3 .
486 Excursions: Excursions in the LCSs and MSs can be used to identify various out of control
487 situations. The advantage to the LCS is that the sample matrix is always the same so matrix
488 effects should not be a factor in evaluating excursions. A rapid and one-time excursion in the
489 LCS usually indicates that a mistake was made in the procedure. A rapid change with continued
490 occurrences suggest that something occurred that is out of the ordinary, such as a new analyst
491 performing the procedure or a new standard solution or new reagents being used. If an LCS
492 shows elevated concentrations, analysts should check for contamination sources or poorly
493 prepared spiking solutions. Slow changes showing a trend usually indicate degradation or
494 contamination of equipment or reagents and may be indicative of bias and should be investigated.
495 Excursions of MSs can be difficult to interpret if the matrix changes from batch to batch.
496 However, an excursion may indicate that the method is not appropriate for a particular matrix. If
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 18-17 DO NOT CITE OR QUOTE
-------
Laboratory Quality Control
497 the MS shows lower than expected concentrations, the analyst should check for poor techniques
498 or expired or poorly prepared reagents and spiking solutions.
499 Elevated or depressed results for site-specific MSs need to be interpreted with the results from
500 LCSs. If both the LCS and site-specific MS results are elevated or depressed then the cause is
501 usually internal to the laboratory. If only the site-specific MS is depressed or elevated, the cause
502 usually is due to the matrix.
503 18.4.4 Certified Reference Materials
504 Issue: Certified reference materials (CRMs) are well-characterized, stable, homogeneous
505 materials with physical or chemical properties determined within specified uncertainty limits.
506 Laboratories that analyze CRMs can compare their performance to the certified concentration
507 and uncertainty levels. CRMs are used for the calibration of an apparatus or the assessment of a
508 measurement method.
509 Discussion: Metrology organizations issue CRMs in various matrices with critically evaluated
510 concentration values for the radionuclide constituents. A CRM issued by NIST or under license
511 from NIST is called a "standard reference material" (SRM). The usefulness of a reference
512 material depends on the characterization of the radionuclide source, activity levels, and their
513 estimated uncertainties.
514 CRMs can be used as internal laboratory QC samples to evaluate the ability of analytical methods
515 to handle the matrix. CRMs need not be known to the analyst but can be introduced into the
516 analytical stream as a blind. Comparison of analytical results of CRMs to their certified values
517 provides linkage to the national scale of measurements and a measure of method accuracy.
518 The planning that goes into the preparation of a CRM involves the selection of analytical
519 techniques that have adequate sensitivity and precision for specific analyses. It has become
520 increasingly important to have available well-characterized CRMs of a natural "matrix" type,
521 which may be used in laboratory tests of measurements of environmental radioactivity. Such
522 materials may be used in the evaluation of competing analytical methods, and also in the
523 cross-comparison of interlaboratory data—both at the national level and the international level.
524 The Ionizing Radiation Division of NIST has constructed several SRMs for radiation
525 measurements. These are included in the 4350 series and can be ordered through NIST. One
526 widely used SRM is the natural matrix ocean sediment (4357). The radionuclides in the NIST
527 natural matrix SRMs are not spiked into the matrix but are incorporated through natural
MARLAP JULY 2001
DO NOT CITE OR QUOTE 18-18 DRAFT FOR PUBLIC COMMENT
-------
Laboratory Quality Control
528 processes to present the analyst with the combination of species that may be faced on a routine
529 basis. The SRM 4357 has two sediment sources: the Chesapeake Bay (benign) and the Irish Sea
530 ("hot").
531 The NIST natural matrix SRM project has certified actinides, fission and activation radionuclides
532 in soils, freshwater lake and river sediments, human tissues, and ocean sediment, and is working
533 on additional unique matrices: ashed bone, ocean shellfish, and Rocky Flats Soil-H
534 A numerical performance indicator for the analysis of a CRM is essentially the same as that for a
535 laboratory control sample. An example is
7 _ x - d
CRM , = (s.\
r^r\ 27T i?)
\juc(x) + uc(d)
536 where xis the measured value, dis the certified value, and uc\x) and uc\d) are the squares of the
537 respective combined standard uncertainties. Warning limits for ZCRM are ±2 and control limits
538 are ±3.
539 Excursions: Excursions in the CRM results can be used to identify various out-of-control
540 situations. The advantage of the CRM is that the sample matrix is always the same, and the levels
541 of analytes are known to a high degree, so uncertainties in matrix effects and radionuclide
542 content should not be a factor in evaluating excursions. A rapid and one-time excursion in the
543 SRM usually indicates that a mistake was made in the procedure. A rapid change with continued
544 occurrences suggest that something occurred that is out of the ordinary, such as a new analyst
545 performing the procedure or the use of a new batch of calibration solutions or reagents. Slow
546 changes showing a trend usually indicate degradation or contamination of equipment or reagents.
547 If a CRM result shows elevated concentrations, analysts should check for contamination sources
548 or poor instrument calibration. If the results show decreased concentrations, the analyst should
549 check for poor techniques or expired or poorly prepared reagents and solutions.
550 CRM results may indicate a bias in the measurement process. Tracking the performance of
551 several consecutive CRM measurements will show if the method or the laboratory consistently
552 obtains high or low results. If the results are consistently higher or lower than the certified values,
553 they should be evaluated for a statistical difference, e.g., ^-tested. When the test indicates a
554 statistical difference, a bias is indicated and the laboratory should investigate the cause of the bias
555 and correct or characterize it.
556 Example: The NIST ocean sediment SRM 4357 offers a good example of a material for
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 18-19 DO NOT CITE OR QUOTE
-------
Laboratory Quality Control
557 evaluating a laboratory performance using a specific analytical method. The blended sediment
558 sample has been analyzed by a number of laboratories, and 10 radionuclides have certified
559 activity values (Lin et al., 2001). The six "natural" radionuclides concentrations tended to have
560 normal distributions (Table 18.2a), while the four "man-made" radionuclides tended to have
561 Weibull distributions (Table 18.2b). There are also 11 other radionuclides where the activity
562 concentrations are not certified at this time but may be at some future time (Table 18.2c).
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
TABLE 18.2a — Certified Massic activities for natural radionuclides
with a normal distribution of measurement results
Radionuclide
40K
226Ra
228Ra
228Th
230Th
232Th
Mean ± 2sm
(mBqg1)
225 ±5
12.7 ±0.4
13.3 ±0.8
12.1 ±0.3
12.0 ±0.5
13.0 ±0.3
Tolerance Limit
(2.5 to 97.5%)
(mBqg1)
190-259
10.3-15.0
9.2-17.4
9.7-14.6
9.6-14.4
11.6-14.3
Number of
Assays
31
21
20
40
18
18
Half-Life ± Is
(In years)
(1.277 ±0.008) x 109
1600 ± 7
5.75 ±0.03
1.9131 ±0.0009
75380 ± 300
d.405± 0.006) x 1010
Table 18.2b — Certified Massic activities for anthropogenic radionuclides
with a Weibull distribution of measurement results
Radionuclide
90Sr
137Cs
238Pu
239pu
+ 240Pu
Mean ± 2sm
(mBqg1)
4.4 ±0.3
12.7 ±0.2
2.29 ±0.05
10.4 ±0.2
Tolerance Limit
(2.5 to 97.5%)
(mBqg1)
2.1-8.4
10.8-15.9
1.96-2.98
9.3-13.2
Number of
Assays
49
76
65
84
Half-Life ± Is
(In years)
28.87 ±0.04
30.07 ±0.03
87.7 ±0.3
24110 ±30
6564 ±11
Table 18.2c — Uncertified Massic activities. Radionuclides for which there are insufficient data
or for which discrepant data sets were obtained. Uncertainties are not provided because
no meaningful estimates could be made.
Radionuclide
129T
155Eu
210p0
210pb
212pb
214Bi
Mean
(mBq g-1)
0.009
1.4
14
24
14
15
Range of Reported
Results (mBq g1)
0.006
1.2
12
14
13
9
-0.012
-1.5
-15
-35
-14
-20
Number of
Assays
6
2
5
19
5
5
Half-Life ± Is
(In years unless listed
as minutes, hours, or
days)
(1.57 ±0.04) x 107
4.68 ±0.05
138.376 ± 0.002 d
22.3 ±0.2
10.64 ± 0.01 h
19.9 ± 0.4m
MARLAP
DO NOT CITE OR QUOTE
18-20
JULY 2001
DRAFT FOR PUBLIC COMMENT
-------
Laboratory Quality Control
590
591
592
593
594
595
596
597
Radionuclide
234U
235U
237Np
238U
241Am
Mean
(mBq g-1)
12
0.6
0.007
12
10
Range of Reported
Results (mBq g1)
9-15
0.1-1.4
0.004-0.009
7-16
7-18
Number of
Assays
68
63
9
76
97
Half-Life ± Is
(In years unless listed
as minutes, hours, or
days)
(2.45 ± 0.02) x 105
(7.038 ± 0.006) x 108
(2.14 ±0.01) x 106
(4.468 ± 0.003) x 109
432.7 ±0.6
SRM 4357. Data for these radionuclides are provided for information only. The Massic activities are not
certified at this time, but may be certified in the future if additional data become available.
18.4.5 Chemical/Tracer Yield
598 Issue: Some methods require that radionuclides should be separated chemically from their
599 sample matrix and purified before measurement. During chemical processing, some of the
600 analyte radionuclide will be lost due to sample spillage, evaporation, incomplete chemical
601 reactions (i.e., precipitation or extraction), etc., as discussed in Chapter 12. While these losses
602 may correlate with a group of samples of similar chemical composition or from the same
603 sampling area, they can be sample specific. For quantitative analysis, it is necessary to correct
604 observed instrument responses for these losses for each analytical sample. Corrections are made
605 using compounds that are stable (carriers) or radioactive (tracers). An inappropriate method for
606 determining chemical yield may result in an analytical bias.
607 Discussion: Most alpha- and beta-emitting radionuclides require chemical separation prior to
608 measurement, in part because of the short effective range of the radiation.
609 CARRIERS. Since it is impossible to determine exactly how much of the analyte is lost during
610 processing, and because the physical mass of the radionuclide is too small to measure gravi-
611 metrically, a compound is added to the sample at the start of the chemical processing, and is
612 carried through the analytical process and assayed. The added compound typically is stable and
613 exhibits the same chemical properties as the analyte and therefore "carries" the analyte
614 radionuclide—for example, stable barium that carries radium isotopes, or stable yttrium that
615 carries 90Y. These added compounds are called "carriers" and are added in sufficient quantity to
616 allow gravimetric assay upon completion of the analysis. The ratio of the carrier recovered to the
617 amount added is the chemical recovery, or yield. Because the carrier and analyte exhibit similar
618 chemical behavior, the chemical yield of both should be equal, i.e., if 85 percent of the stable
619 barium is recovered, then it follows that the observed instrument response represents 85 percent
620 of the radium present in the sample.
JULY 2001
DRAFT FOR PUBLIC COMMENT
18-21
MARLAP
DO NOT CITE OR QUOTE
-------
Laboratory Quality Control
621 TRACERS. For radionuclides above atomic number 83, stable isotopes do not exist, and a different
622 approach is taken to determine the analyte's yield. For these radionuclides, an isotope other that
623 those being measured is added to the sample in the same manner as described above, e.g., 232U
624 used as a tracer for isotopic uranium (234U, 235U, and 238U), 236U, or 242Pu used as a tracer for
625 isotopic plutonium (238Pu, 239Pu, and 240Pu).
626 This approach to chemical yield determination is based on the following assumptions regarding
627 the carrier/tracer:
628 • It exhibits similar chemical behavior as the analyte under the protocol's conditions.
629 • The energy emission of the tracer and progeny should not interfere with the resolution of the
630 analytes of interest.
631 • It is chemically and physically equilibrated with the sample before losses of either occur.
632 • Indigenous concentrations of carrier or tracer are insignificant, or are well known and can be
633 quantified and corrected for during subsequent data analysis.
634 • The chemical form of carrier or tracer precipitates are consistent with what was used during
635 the material's preparation and standardization.
636 Care should be taken during the analytical procedure to ensure that these assumptions are valid.
637 Different conditions, such as a lack of equilibrium between the tracer and sample analyte, can
638 result in inaccurate data. If there is indigenous tracer or carrier in the sample, this quantity should
639 be known so that the appropriate correction can be made for its contribution to the chemical
640 yield. In some cases, this will prevent the procedure's use, as described below. As stated
641 previously, the quantity of tracer or carrier added to the sample should overwhelm its indigenous
642 concentration, which cannot be determined for samples with unknown tracer or carrier content. A
643 separate analysis for trace elements or interfering radionuclides could provide information to
644 estimate the uncertainty contributed by the sample's indigenous tracer or carrier.
645 It should be noted that some analytical methods exclude direct assessment of the procedure's
646 chemical recovery for each sample analysis, e.g., Procedure 908.1 for Total Uranium in Drinking
647 Water (EPA, 1980b). In such cases, chemical recovery is typically addressed by analyzing a
648 group of prepared standards by the same protocol and the results are analyzed statistically to
649 derive a chemical recovery factor. The recovery factor is applied to routine samples based on the
650 assumption that the standards used for its derivation are representative of routine samples. This
MARLAP JULY 2001
DO NOT CITE OR QUOTE 18-22 DRAFT FOR PUBLIC COMMENT
-------
Laboratory Quality Control
651 approach precludes the empirical assessment of a sample specific chemical recovery, and would
652 probably require scrutiny and periodic verification.
653 Acceptance limits for chemical/tracer yields should be specified in the laboratory's Quality
654 Manual. While it is customary to establish lower limits for chemical yield, upper limits may also
655 be necessary since excessive yields indicate a loss of analytical control. All limits developed by
656 the laboratory should be either statistically based or based on historical data, and should include
657 warning and control limits. The inherent differences among sample matrices generally require the
658 use of matrix specific criteria, i.e., finished drinking water limits may differ from limits for high
659 solid content waters, sandy soils or heterogeneous media. Irrespective of medium, where
660 practical, the chemical yield and its uncertainty should be determined, recorded and tracked for
661 each radiochemical measurement.
662 Excursions: There are several possible reasons for the yield to be outside of the acceptance
663 limits. These are summarized in Figure 18.4 and discussed below.
664
665
666
667
668
669
670
671
672
673
674
CHEMICAL YIELD
EXCURSIONS
1
1
EXCESSIVE YIELDS
I
LOW YIELDS
INTERFERENCE
- Containment
Radionuclide
- Indigenous carrier in
sample
CHANGED
CALIBRATION
- Source thickness
- Source diameter
- Source-detector distance
I
PROCEDURE FAILURE
- Reagent problem
- Not following procedure
- Incompatible matrix/
interference
- Instrument failure
- Incomplete separation
HIGHLY VARIABLE YIELDS
I
• NEW MATRIX/
INTERFERENCE
- Reagent concentration
• NOT FOLLOWING
PROCEDURE
• CONTROL OF VARIABLE
- Temperature
- Concentration
- Time
- Technique
675
FIGURE 18.4 — Failed performance indicator: chemical yield
JULY 2001
DRAFT FOR PUBLIC COMMENT
18-23
MARLAP
DO NOT CITE OR QUOTE
-------
Laboratory Quality Control
676 EXCESSIVE YIELDS: A chemical yield significantly greater than 100 percent indicates a
677 problem. Typical causes of excessive chemical yields are provided below:
678 • Interference. The sample may contain an interfering radionuclide that cannot be
679 distinguished from the tracer and therefore biases the tracer response; the sample may
680 contain an indigenous concentration of the tracer or carrier used; or large amounts of
681 another stable element are present.
682 • Counting. Changes in instrument calibration factor or other factors that affect counting,
683 e.g., source thickness, diameter, source-detector distance or change in chemical form of
684 final sample precipitate.
685 • Instrument failure.
686 Low YIELDS: A very low yield usually indicates a procedural failure caused by incomplete or
687 unsuccessful chemical separation, matrix interference, missing reagents, or the exclusion of a
688 key element in the sample processing. A significantly lower yield will increase the overall
689 measurement uncertainty and degrade the procedure's effective detection capability unless
690 the counting time is appropriately extended, which may be impractical or even ineffective in
691 many cases. Furthermore, measurement of the recovered carrier or tracer becomes
692 increasingly more adversely affected by background, stable element, water absorption, and
693 other corrections as the yield decreases. Fixed lower limits for yields often are established
694 and should be specific to analytical procedures and sample matrices. Setting an upper limit is
695 recommended for the acceptable relative uncertainty in a yield measurement.
696 HIGHLY VARIABLE YIELDS: High variability in procedural temperature, concentration, time,
697 reagent concentration, or laboratory technique can have dramatic effects on yield. Highly
698 variable yields indicate a lack of procedural control and should be investigated and corrected.
699 A simple step such as heating samples on a hotplate can lead to variability in yield because
700 the hotplate surface is thermally uneven. Samples can be dried and reconstituted several
701 times during the course of the preparation protocol, and samples may require different
702 amounts of heat or water, which introduces additional variability. When highly variable
703 chemical yields are observed, a careful examination of the analytical procedure's application
704 is recommended to determine critical variables and the controls needed to re-establish
705 adequate management over yields.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 18-24 DRAFT FOR PUBLIC COMMENT
-------
Laboratory Quality Control
706 18.5 Instrumentation Performance Indicators
707 Radiometric and non-radiometric instruments are used currently to quantify radionuclides in a
708 variety of environmental matrices, and quality control measures are necessary to ensure proper
709 instrument performance. This section presents radiometric instrument performance measures that
710 indicate a measurement system is in control. For detailed information on instrument concepts and
711 specific techniques, see Chapters 15 and 16 as well as ASTM standard practices (e.g., D3648, for
712 the Measurement of Radioactivity). The specific quality control procedures to be followed
713 depend on the measurement equipment. Sufficient checks are needed to demonstrate that the
714 measurement equipment is properly calibrated, the appropriate background has been recorded,
715 and that all system components are functioning properly. QC measures for instrumentation
716 should include at a minimum: (1) instrument background measurements, (2) instrument
717 calibration with reference standards, and (3) periodic instrument performance checks subsequent
718 to the calibration. Acceptable control limits should be specified in the laboratory Quality Manual.
719 18.5.1 Instrument Background Measurements
720 Issue: In general, radionuclide detection covers more than 17 orders of magnitude of sample
721 activity, from irradiated material that produces high radiation fields to environmental samples.
722 All radiation detection instruments have a background response even in the absence of a sample
723 or radionuclide source. To determine the instrument's response to the radioactivity contributed
724 by the sample alone (net), the instrument background response is subtracted from the sample-
725 plus-background response (gross). For discussions on possible contamination, refer to Section
726 18.4.1. Background corrections become more critical when the instrument net response is small
727 relative to the background. Careful control of contamination and routine monitoring of
728 instrument background are therefore integral parts of a control program. Inappropriate
729 background correction results in analytical error and will increase the uncertainty of data
730 interpretation.
731 Discussion: Every radionuclide detector produces a signal response in the absence of a sample or
732 radionuclide source. These signals are produced by electronic dark current, cosmic radiation,
733 impurities in the instrument construction materials, crosstalk between the detector's alpha and
734 beta channels, sources in the general vicinity of the detector, and residual contamination from
735 previous counting episodes. The majority of these contributors to instrument background produce
736 a fairly constant count rate, given sufficient measurement time (i.e., dark current, cosmic
737 radiation, construction material impurities). For other sources, instrument backgrounds vary as a
738 function of time (i.e., from decay or ingrowth of residual contamination or as radon levels
739 fluctuate throughout the day and season). For low-level measurements, it is imperative that the
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 18-25 DO NOT CITE OR QUOTE
-------
Laboratory Quality Control
740 background be maintained as low as feasible. Active or passive detector shielding, removing or
741 adequately shielding radioactive sources in the vicinity of the detector, and good laboratory
742 practices to prevent residual contamination are necessary to maintain low instrument background.
743 The instrument's background should be determined in the absence of a radionuclide source.The
744 instrument background should be well characterized. The instrument background is an important
745 factor in determining the ability to achieve a specific minimum detectable concentration (MDC).
746 Control limits for the background should be specified in the laboratory's Quality Manual, as
747 appropriate. The background population considered in the statistical calculations should cover a
748 sufficient period of time to detect gradual shifts in the measurement system's background
749 contamination or detector instability. Additionally, backgrounds should be determined in such a
750 way that they mimic actual sample measurement conditions as closely as possible, i.e., using
751 appropriate sample containers, geometries, and counting times.
752 Background measurements should be made on a regular basis and monitored using control
753 charts. For instruments with well established background performance records and a low
754 probability of detector contamination, this frequency may be modified by the laboratory. For
755 mass spectrometry and kinetic phosphorimetry analysis, background measurements should be
756 performed on a real time basis. See ASTM E181, ANSI N42.12, and NELAC (2000) Quality
757 Systems Appendix D for more information on the suggested frequency of background
758 measurement.
759 Excursions: Variations in instrument backgrounds may indicate instrument malfunction. Variations
760 may take the form of rapid increase or decrease in background, slow increase or decrease in back-
761 grounds, and highly variable or erratic backgrounds. These variations can result in the measurement
762 system's reduced precision and decreased detection capability. Rapid or significant increases in
763 background measurements may be due to instrument or blank contamination, insufficient shielding with
764 relocation of nearby radionuclide sources, or large scale equipment malfunction (e.g., a broken window
765 on a gas proportional system).
766 Instrument background data should be evaluated for trends, which is facilitated by regular
767 observation of control charts. A slowly changing background could alert laboratory personnel to
768 a potentially serious instrument failure. A sufficient number of data points (Chapter 15) taken
769 over time should be included in any trend analysis. Slowly changing instrument backgrounds
770 could be caused by low counting-gas flow rates, small incremental instrument contamination, or
771 electronic drift or noise.
772 When the instrument background is more variable than expected, the reliability of measurements
773 becomes questionable, resulting in loss of confidence and increased uncertainty. This indicates a
MARLAP JULY 2001
DO NOT CITE OR QUOTE 18-26 DRAFT FOR PUBLIC COMMENT
-------
Laboratory Quality Control
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
loss of control over the measurement environment, or limitations of the data handling software.
The root cause of the variability should be identified and corrected to re-establish statistical
control over the instrument background. Table 18.3 presents reasons for changing backgrounds.
TABLE 18.3 — Instrument background evaluation
Instrument Background Failed Performance Indicator
Rapid Change in Background
Electronic failure
Detector failure
Loss of coolant/vacuum
Instrument contamination
Counting gas changes
Temperature/humidity fluctuation
Laboratory contamination
External sources
Insufficient shielding
Personnel with nuclear medicine dose
Slow Change in Background
Instrument contamination
Electronic drift
Low counting gas flow rate
Excessively Variable Background
Sources being moved
Radon fluctuation
Insufficient shielding
Insufficient counting statistics
Interfering radionuclides
Poor peak deconvolution
Intermittent electrical short
Failing electronics
18.5.2 Efficiency Calibrations
Issue: This section discusses selected aspects of instrument calibration that are pertinent to
laboratory quality control. A more in-depth, technical discussion is provided in Chapter 16. The
number of events (counts) recorded by a detector is converted to activity (actual radionuclide
transformations) by empirically determining this relationship with NIST-traceable radionuclide
sources when available. This relationship is expressed in the system's efficiency calibration. A
separate efficiency is determined for each detector-source combination and is typically energy or
radionuclide specific.
Detector efficiency is critical for converting the detector's response to activity. As discussed
above, routine performance checks can evaluate several aspects simultaneously (sample
geometry, matrix, etc.) and provide a means to demonstrate that the system's operational
parameters are within acceptable limits. These are typically included in the assessment of the
analytical method's bias and are specified in terms of percent recovery based on the source's
known disintegration rate. Performance checks for measurement efficiency are usually
determined statistically based on repeated measurements with a specific check source. Detection
of a shift in measurement efficiency should be investigated.
The frequency of performance checks for efficiency calibrations is instrument specific. The
frequency of these checks is often based on a standardized time scale or a percentage of the total
JULY 2001
DRAFT FOR PUBLIC COMMENT
18-27
MARLAP
DO NOT CITE OR QUOTE
-------
Laboratory Quality Control
808 number of analyses performed using that method.
809 Performance checks for instrument efficiency typically are performed on a day-of-use basis. The
810 level of activity in the check source should be sufficient to allow the accumulation of enough
811 counts in a short time so that daily performance checks do not impose an unnecessary burden on
812 the laboratory. However, the source strength for spectrometry systems should be such that
813 instrument dead time is not significant and gain shifts do not occur (ANSI 42.23). For detectors
814 that are used infrequently, it may be necessary to perform a check before and after each set of
815 measurements.
816 Control charts provide a useful tool for documenting and evaluating performance checks for
817 efficiency calibrations, and should be established and maintained for the intrinsic efficiency of
818 each detector. There are several methods available for evaluating performance using control
819 charts (see Attachment ISA).
820 Discussion: Most radiation detectors do not record all of the nuclear transformations that occur
821 in samples undergoing measurement, i.e., they are not one hundred percent efficient. This occurs
822 for several reasons, and the prominent reasons are discussed briefly below.
823 • Intrinsic or absolute efficiency2 - In the absence of all other factors, a detector will only
824 record a fraction of the emissions to which it is exposed due to its composition and other
825 material-related aspects. Intrinsic efficiency is a measure of the probability that a count will
826 be recorded when a particle or photon of ionizing radiation is incident on a detector (ANSI
827 Nl.l).
828 • Geometry - The spatial arrangement of sample, shielding, and detection equipment, including
829 the solid angle subtended by the detector and sample configuration, largely determines what
830 fraction of the emissions from the source actually reach the detector (ANSI N15.37).
831 Geometry includes the source's distance from the detector and its spatial distribution within
832 the counting container relative to the detector and shielding components.
833 • Absorption - Radiation emitted by the sample can be absorbed by the sample itself (self
Efficiency measures the fraction of emitted photons or particles that are actually detected. It is affected by the
shape, size, and composition of the detector as well as by the sample-to-detector geometry. There are two ways that
efficiency can be expressed: "Absolute efficiency" is the fraction of all the photons or particles emitted by the
source that are actually detected, and "intrinsic efficiency" is the ratio of photons or particles detected to the number
that actually fall on the detector.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 18-28 DRAFT FOR PUBLIC COMMENT
-------
Laboratory Quality Control
834 absorption), as well as other materials placed between the source and the detector, i.e.,
835 sample container, detector housing and shielding (NCRP 58).
836 • Backscatter - Radiation emitted by the sample can hit the sample container and scatter into
837 the detector.
838 The detector response is a composite of these factors.
839 Each radiation detector should be calibrated to determine the relationship between the observed
840 count rate of the detector and the disintegration rate of the source being assayed. This
841 relationship is called the efficiency calibration—typically expressed in counts per second/
842 disintegration per second, or cps/dps—and is an integral part of the measurement protocol. For
843 alpha spectrometry systems, the efficiency of detection is energy-independent. Efficiencies for
844 gamma spectrometry are energy dependent, and an efficiency calibration typically covers a range
845 for a specific counting geometry, e.g., 50 to 1,800 kilo electron volts (keV).
846 Once this relationship is established, it should be checked at regular intervals using what is called
847 a performance or calibration check. The performance check does not seek to reestablish the
848 detector's efficiency but simply demonstrates that the relationship is within acceptance limits.
849 When designed properly, an efficiency performance check evaluates the intrinsic efficiency,
850 geometry and absorption in a single measurement. Accordingly, it takes the form of a single
851 value that incorporates all effects for a target radionuclide and a specific detector-sample
852 configuration. Detectors that are energy dependent and measure radionuclides with multiple
853 energies, such as photon or alpha spectrometers, should have performance checks at several
854 energies throughout the measurement range. For these detectors, the performance check can
855 simultaneously address the system's efficiency, energy calibration and resolution using a single
856 source. An internal pulser can be used to check the electronics.
857 Because the performance check's purpose is to demonstrate that the system's efficiency remains
858 constant, the source's absolute disintegration rate need not be known, provided its purity can be
859 established, its half-life is known, and its activity is sufficient to provide adequate precision.
860 Accordingly, it is not necessary to use a NIST-traceable check source for this purpose. Check
861 sources that are non-NIST-traceable can meet the precision objectives of the performance check
862 and they are less expensive.
863 Excursions: Changes in the efficiency of a detector can only be corrected by determining the
864 root cause of the problem and repeating the efficiency calibration. Gradual changes in geometry
865 usually indicate a problem with the technique of sample mounting or preparation. A visual
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 18-29 DO NOT CITE OR QUOTE
-------
Laboratory Quality Control
866 inspection of the prepared sample is often helpful in eliminating sample geometry as a source of
867 the problem. For example, a precipitated sample counted on a gas proportional counter has an
868 expected appearance, i.e., a circle of precipitate centered on the planchet and often covered with
869 thin plastic film. If the prepared sample does not have the correct appearance, there could be a
870 problem with the geometry, self-absorption, and backscatter. This can sometimes be corrected by
871 preparing the sample a second time, inspecting it and presenting it for counting a second time.
872 Re-training personnel responsible for the error may also be indicated. Because samples that have
873 been improperly prepared for counting can result in contamination of or physical damage to the
874 detector, it is strongly recommended that every sample be visually inspected prior to counting.
875 Significant changes in geometry caused by modifications to the source preparation method can
876 only be corrected by recalibrating the detector. Examples of modifications to source preparation
877 methods are (1) using a new filter so that the geometry of the test source is different than the
878 geometry used for calibration, and (2) replacing the containers used for gamma spectrometry with
879 containers that have a different wall thickness or are made from different materials.
880 Changes in intrinsic efficiency generally result from a physical change to the detector and often
881 result in rapid changes in efficiency. In many cases, changes that affect the intrinsic efficiency of
882 a detector render it inoperable. These are specific to a detector type and are listed below:
883 • HPGe, Ge(Li), and surface barrier detectors - Real or apparent changes in intrinsic efficiency
884 caused by vacuum leaks or failure of field effect transistor.
885 • Thin window detectors (gas proportional counters, low-energy photon) - Changes in
886 measurement efficiency are typically associated with damage to the detector window.
887 • Gas proportional systems - Problems with efficiency related to the quality or flow of
888 counting gas.
889 • Anti-coincidence systems with guard detectors - Electrical problems with the anti-
890 coincidence circuits that may produce apparent changes in efficiency.
891 • Scintillation detectors - Gradual changes in efficiency are associated with the scintillator or
892 the photomultiplier tube. For example, Nal(Tl) crystals may gradually turn yellow over time
893 resulting in a lower intrinsic efficiency, and liquid scintillation counters may have residue
894 gradually build up on the surface of the photomultiplier tube affecting the detection of
895 photons by the tube.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 18-30 DRAFT FOR PUBLIC COMMENT
-------
Laboratory Quality Control
896 18.5.3 Spectrometry Systems
897 18.5.3.1 Energy Calibrations
898 Issue: This section discusses selected aspects of instrument calibration that are pertinent to
899 laboratory quality control. A more in depth, technical discussion is provided in Chapter 16. All
900 radiation measurements are energy dependent to a certain extent. However, spectrometric
901 techniques such as gamma and alpha spectrometry identify radionuclides based on the energy of
902 the detected radiations. For these techniques a correct energy calibration is critical to accurately
903 identify radionuclides. Problems with energy calibration may result in misidentification of peaks.
904 Discussion: Spectrometry systems should be calibrated so that each channel number is correlated
905 with a specific energy. To identify radionuclides correctly, this energy calibration needs to be
906 established initially and verified at regular intervals. The energy calibration is established by
907 determining the channel number of the centroid of several peaks of known energy over the
908 applicable energy range. Typically, a minimum of three peaks is used, and commercially
909 available sources contain nine or ten photopeaks. The relationship between energy and channel
910 number can be determined by a least squares fit. To account for non-linearity, a second or third
911 order fit may be used. However, these require more points to define the curve. For example, a
912 first order calibration requires at least two points, while a second order calibration requires a
913 minimum of three points. The end points of the curve define a range of applicability over which
914 the calibration is valid, and peaks identified outside the curve's range should be used carefully.
915 The uncertainty associated with the curve should be available at any point along the calibration
916 curve.
917 Quality control checks for energy calibration may be combined with checks for efficiency cali-
918 bration and resolution. Radiations emitted over the range of energy of interest are measured, and
919 two or more peaks are used to demonstrate that the energy calibration falls within acceptable
920 limits. Check sources may consist of a single radionuclide (e.g., 137Cs or 60Co) or a mixture of
921 radionuclides (e.g., mixed gamma). Because only the location of the peak is of concern, there is
922 no requirement that the check source be calibrated or certified, except for ensuring that it does
923 contain the radionuclide(s) of interest at a specified level of purity.
924 The energy calibration is determined when the system is initially set up by adjusting the gain of
925 the amplifier, analog-to-digital conversion (ADC) gain, and zero. Criteria that indicate when
926 readjustment is required because of gradual and abrupt changes in the energy versus channel
927 calibration should be established as an integral part of the system's operating procedure. These
928 changes usually are monitored by the measurement system's software, and the user specifies the
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 18-31 DO NOT CITE OR QUOTE
-------
Laboratory Quality Control
929 allowable difference between that the system's response and the radionuclide's known energy.
930 The tolerable difference often relates to the instrument's resolution. For example, a high resolu-
931 tion instrument such as an intrinsic germanium detector typically will have acceptable limits on
932 the order of a few keV, while a low resolution instrument such as a Nal(Tl) detector typically
933 will have acceptable limits on the order of several tens of keV.
934 Spectra also can be analyzed by identifying each peak manually. With manual identification, the
935 acceptable limits for the energy calibration are determined for each spectrum based on the pro-
936 fessional judgment of the person analyzing the spectrum.
937 The frequency of QC checks for energy calibrations can be related to the expected resolution of
938 the instrument, the electronic stability of the equipment, or the frequency needs of QC
939 measurements for efficiency calibration or resolution. These are specified typically in the
940 laboratory's Quality Manual or other typical project-related documentation. Examples for three
941 detector types are provided below and in Table 18.5.
942 • HPGe and Ge(Li) Photon Detectors. Energy calibrations are typically verified using a check
943 source on a day of use basis. Every sample spectrum should include verification of the energy
944 calibration as part of the data review process, when possible. Under extreme conditions (e.g.,
945 in situ measurements in bad weather), it may be necessary to perform checks at the beginning
946 and end of each measurement period or day the instrument is used.
947 • Surface Barrier Alpha Spectrometry Detectors. The energy calibration is often performed
948 using an alpha source when the instrument is setup initially and when a detector has been
949 serviced or replaced. Electronic pulsers can be used for daily checks on energy calibration.
950 Most alpha spectra include a chemical yield tracer with a peak of known energy that can be
951 used to verify the energy calibration during data review. Alpha spectrometers have a lower
952 resolution than germanium detectors, and newer spectrometers are sufficiently stable to allow
953 weekly or monthly performance checks. The frequency of performance checks should be
954 based on the number and frequency of measurements and historical information on the
955 stability of the instrument.
956 • Low-Resolution Nal(Tl) Detectors. These typically are less stable than HPGe detectors and
957 may require more frequent quality control checks, depending on the conditions under which
958 they are used.
959 For all detectors where energy calibrations are performed daily, plotting the channel numbers of
960 peak centroids can be useful for identifying trends and determining the need for adjusting the
MARLAP JULY 2001
DO NOT CITE OR QUOTE 18-32 DRAFT FOR PUBLIC COMMENT
-------
Laboratory Quality Control
961 system. Changes in peak location may result in mis-identification of radionuclides. When this is
962 observed, all spectra obtained since the last acceptable energy calibration check should be
963 reviewed. If there is sufficient information within the spectrum to determine the acceptability of
964 the energy calibration, no further action may be required for that spectrum. If the spectrum con-
965 tains too few peaks of known energy, reanalysis should be initiated.
966 Gradual changes in peak location are not unexpected and the rate of these gradual changes can be
967 used to establish the appropriate frequency of energy calibration checks. The acceptable limits on
968 peak location established during the initial system setup may be used to indicate when the energy
969 calibration needs to be readjusted.
970 Excursions: Changes in the energy calibration can be the result of many factors including power
971 surges, power spikes, changes in the quality of the electrical supply, variations in ambient condi-
972 tions (e.g., temperature, humidity), physical shock to the detector or associated electronics, and
973 electronic malfunction.
974 Rapid changes in energy calibration are usually caused by power surges, power spikes, or physi-
975 cal shocks to the system. Corrective actions typically involve recalibrating the system and repeat-
976 ing the analysis. If changes result due to loss of cryostat vacuum, the instrument may need to be
977 returned to the manufacturer to be refurbished or replaced.
978 Gradual changes in the energy calibration are usually the result of a variable or poorly condi-
979 tioned power source, changes in the ambient conditions, or electronic malfunction. Corrective
980 actions generally begin with identifying the root cause of the problem. Gradual changes that
981 begin following relocation of the instrument are more likely to be caused by the power source or
982 the ambient conditions. Installing a line conditioner, surge protector, and uninterrupted power
983 supply is recommended to address problems related to the system's electrical power source.
984 Problems with low humidity can be corrected through the use of a humidifier in dry climates or
985 cold weather; conversely, high or variable humidity may require the use of a dehumidifier. Prob-
986 lems associated with fluctuations in temperature may require significant changes to the heating
987 and cooling system for the room or building containing the instrument in order to stabilize the
988 temperature. Gradual changes that occur following physical shocks to the system or following a
989 rapid change in peak location with an unidentified cause are more likely to be the result of prob-
990 lems with the electronic equipment. In most cases the amplifier is the source of these problems,
991 but the analog-to-digital converter, pre-amplifier, power supply voltages, and multi-channel (or
992 single-channel) analyzer may also cause this type of problem. However, they could also be the
993 result of crystal or detector failure. Systematic switching out of components and discussions with
994 the instrument manufacturer will often help to identify which component may be the source of
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 18-33 DO NOT CITE OR QUOTE
-------
Laboratory Quality Control
995 the trouble. It may be especially difficult to identify the source of problems with new instruments
996 in a new facility.
997 18.5.3.2 Peak Resolution and Tailing
998 Issue: The shape of the full energy peak is important for identifying radionuclides and quantify-
999 ing their activity with spectrometry or spectrometry systems. Poor peak resolution and peak
1000 tailing may result in larger measurement uncertainty. If consistent problems with peak resolution
1001 are persistent, then an analytical bias most likely exists. Many factors will affect peak resolution
1002 and these are discussed below.
1003 Discussion: Detectors with good resolution permit the identification of peaks which are close in
1004 energy. When a monoenergetic source of radiation is measured with a semiconductor, scintilla-
1005 tion, or proportional spectrometer, the observed pulse heights have a Gaussian distribution
1006 around the most probable value (Friedlander et al., 1981). The energy resolution is usually
1007 expressed in terms of the full width at half maximum (FWHM) or the full width at tenth maxi-
1008 mum (FWTM).
1009 In a semiconductor detector, fluctuations in output pulse height result from the sharing of energy
1010 between ionization processes and lattice excitation (Friedlander, et al., 1981). The number of
1011 charge pairs created by radiation of a given energy will fluctuate statistically. This fluctuation
1012 occurs because the energy causes lattice vibrations in the semiconductor as well as the formation
1013 of charge pairs. This sharing of energy causes a variation in the number of charge pairs created
1014 and gives rise to the width of a measured peak. The magnitude of the statistical fluctuation is pro-
1015 portional to the energy of the radiation. There is also a variation in the number of charge pairs
1016 collected by a detector. This variation is accounted for by the Fano factor. Because several poorly
1017 understood factors degrade resolution in a semiconductor detector, an empirical value of the
1018 Fano factor should be used.
1019 In a scintillation detector, the statistical fluctuations in output pulse heights arise from several
1020 sources. The conversion of energy of ionizing radiation into photons in the scintillator, the elec-
1021 tronic emission at the photocathode, and the electron multiplication at each dynode are all subject
1022 to statistical variations. Note that the distance of the sample to the detector also impacts the
1023 resolution.
1024 In a proportional counter, the spread in pulse heights for monoenergetic rays absorbed in the
1025 counter volume arises from statistical fluctuations in the number of ion pairs formed and the gas
1026 amplification factor (Friedlander, et al., 1981). If the gas gain is made sufficiently large, the
MARLAP JULY 2001
DO NOT CITE OR QUOTE 18-34 DRAFT FOR PUBLIC COMMENT
-------
Laboratory Quality Control
1027 fluctuations in the number of ion pairs determine the resolution.
1028 The FWHM is typically used as a measure of resolution, while the FWTM is used as a measure
1029 of tailing for the full energy peak. For Gaussian peaks with standard deviation a, the FWHM is
1030 equal to 2.35a. The resolution of a detector is the ratio of the FWHM to the most probable peak
1031 height. The sources of fluctuations that contribute to the standard deviation are dependent on the
1032 type of detector.
1033 Resolution affects the ability to identify individual peaks in two ways (Gilmore and Heming-
1034 way, 1995). First, it determines how close together two peaks may occur in energy and still be
1035 resolved into the two components. Second, for gamma spectrometry, when a peak of small mag-
1036 nitude sits on the Compton continuum of other peaks, its ability to be detected can depend on its
1037 signal-to-noise ratio. With good resolution, the available counts are distributed in fewer channels,
1038 thus those counts will be more easily identified as a peak by the spectrometry analysis software.
1039 If resolution degrades significantly the efficiency may be in error. This is especially true when the
1040 spectrum analysis involves the region of interest (ROI) concept. When the calibration is per-
1041 formed, the full energy peak may fit within the defined ROI limits, whereas the resolution
1042 degraded peak may have counts which fall outside them. Thus, the detector efficiency will be
1043 effectively decreased and inconsistent with the previously determined efficiency.
1044 Tailing is another observable feature of the peak shape. Tailing is an increased number of counts
1045 in the channels on either side of the full energy peak. Tailing affects the FWTM more than the
1046 FWHM, so the ratio of FWTM to FWHM can be used as a measure of tailing. For a Gaussian
1047 distribution the ratio of FWTM to FWHM is 1.823. For most germanium detectors this ratio
1048 should not exceed 2.0. Tailing may be caused by imperfect or incomplete charge collection in
1049 some regions of the detector, escape of secondary electrons from the active region of the detector,
1050 electronic noise in the amplification and processing circuitry, loss of vacuum and escape of
1051 bremsstrahlung from the active region of the detector. Tailing may also result from the source's
1052 self-absorption for alpha emitting radionuclides.
1053 The resolution (FWHM) is routinely calculated for gamma and alpha spectrometry peaks by the
1054 spectrum analysis software and can be monitored by observing the FWHM calculated for the
1055 check sources routinely counted. Resolution monitoring and charting is normally an integral part
1056 of a measurement quality system. Acceptance parameters may be established for resolution and
1057 incorporated in the analysis software. For alpha spectrometry, where radionuclide tracers are used
1058 for chemical yield determination, the FWHM can be monitored for each analysis, if desired.
1059 Some projects may specify FWHM limits for internal tracer peaks on each sample run.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 18-35 DO NOT CITE OR QUOTE
-------
Laboratory Quality Control
1060 The shape of the peak is important for quantifying the activity, and resolution is important for
1061 identifying peaks in a spectrum. The shape of the peak is also important for monitoring the per-
1062 formance of a detector. Germanium detectors have very good resolution on the order of 1 per-
1063 cent. The FWHM at specific energies is provided by the manufacturer. The FWHM should be
1064 established at several energies throughout the range being measured because the FWHM is
1065 directly proportional to the energy. These energies are usually the same as those used for check-
1066 ing the energy calibration and the efficiency calibration. Control limits for FWHM and the ratio
1067 of FWTM to FWHM may be developed based on statistics using multiple measurements
1068 collected over time.
1069 The resolution of an alpha spectrum is dominated typically by self-absorption in the source. This
1070 is indicated by low energy tailing and elevated FWTM and FWHM. Most surface barrier detec-
1071 tors are capable of resolutions on the order of 30-40 keV for monoenergetic nuclides and 80-100
1072 keV for unresolved multiplets. Acceptance of sample resolution is usually monitored by visual
1073 inspection of individual spectra. For well-prepared samples, the FWHM of the alpha peaks may
1074 be expected to be from 30 to 80 keV.
1075 The resolution of scintillation detectors is not as good as the resolution of semiconductor detec-
1076 tors, but peak shape and tailing are just as important for analyzing samples. The FWHM should
1077 be established at several energies throughout the range being measured because the FWHM is
1078 inversely proportional to the energy. These energies are usually the same as those used for check-
1079 ing the energy calibration and the efficiency calibration. Control limits for FWHM and the ratio
1080 of FWTM to FWHM may be developed based on statistics using multiple measurements
1081 collected over time.
1082 Proportional counters are not used as spectrometers in many laboratories, so it is not necessary to
1083 perform checks for resolution and peak shape.
1084 Performance checks for resolution and tailing should be performed for all instruments used as
1085 spectrometers. These measurements are usually combined with the performance checks for
1086 energy calibration and efficiency calibration. Quality control activities should include visual
1087 inspection of all spectra to evaluate peak shape and tailing.
1088 Control charts for FWHM and the ratio of FWTM to FWHM can be developed and used to mon-
1089 itor the performance of any detector used as a spectrometer. Because the concern is when the
1090 resolution degrades (i.e., the FWHM increases) or tailing becomes a problem (i.e., the ratio of
1091 FWTM to FWHM increases), control limits are necessary. Limits can be developed based on
1092 historical performance for a specific type of detector. Control charts offer a convenient method
MARLAP JULY 2001
DO NOT CITE OR QUOTE 18-36 DRAFT FOR PUBLIC COMMENT
-------
Laboratory Quality Control
1093
1094
1095
1096
1097
1098
1099
1100
1101
1102
1103
1104
1105
1106
1107
1108
1109
1110
1111
1112
1113
1114
1115
for monitoring the results of the performance checks. As mentioned previously, the concern is
associated with an increase in the FWHM or the ratio of FWTM to FWHM. This means that only
an upper control limit or tolerance limit is required for the chart.
Excursions: Changes to the FWHM are associated with malfunctioning or misadjusted elec-
tronics, excessive noise or interference, or detector or source problems. Electronics problems
include changes in the high voltage applied to the detector, noise (including cable noise and high
voltage breakdown), and electronic drift. Electronics problems may be caused by changes in the
high voltage, improper adjustment of the pole zero or baseline restorer, or drift of the amplifier
gain or zero during acquisition. Source problems are usually only associated with alpha spectra
and result in excessive self-absorption resulting in low-energy tailing. This can result in counts
being identified with an incorrect peak. Problems that are not electronic or source related imply
that the detector is malfunctioning.
Changes to the ratio of FWTM to FWFDVI indicate problems associated with tailing. Tailing can
occur on the high- or low-energy side of the peak. High-energy tailing indicates electronics prob-
lems that may be caused by excessive activity in the sample, incorrect adjustment of the pole zero
or pile-up rejector, or drift of the amplifier gain or zero while acquiring the spectrum. Low-
energy tailing indicates an electronic or a source problem—a possible corrective action is to
check to see if the vacuum is set properly. Table 18.4 lists common problems, the implied root
cause of the problem, and possible corrective actions.
TABLE 18.4 — Root cause analysis of performance check results
1116
1117
1118
Observed Problem
Efficiency changed
Peak centroid moved
FWHM changed
FWTM: FWHM
changed
Implied Root Cause
Unknown
Electronics degradation
Geometry changed
Poor source
Software application
Gain changed
Offset shifted
Electronics problem
Electronics problem
Possible Corrective
Actions
Ensure the correct check source was used
Check to ensure the efficiency was evaluated using the
correct geometry
Ensure high voltage is set properly
Pulser check of electronics
Check amplifier gain
Check conversion gain
Check stability of amplifier for gain
Check zero offset
Check digital offset
Check stability of amplifier for gain
shifts or drifting
shifts or drifting
Ensure high voltage is set properly
Detector problem
Ensure high voltage is set properly
Detector problem
JULY 2001
DRAFT FOR PUBLIC COMMENT
18-37
MARLAP
DO NOT CITE OR QUOTE
-------
Laboratory Quality Control
1119
1120
1121
1122
1123
1124
1125
1126
1127
1128
1129
1130
1131
1132
1133
1134
1135
1136
Observed Problem
No peak or broad
peaks
Low-energy tailing
High-energy tailing
Spectra shifted
uniformly
Spectra stretched or
compressed
Implied Root Cause
Source problem
Electronics problem
Electronics problem
Source problem
Electronics problem
Source problem
(too much activity)
Offset shifted
Gain changed
Possible Corrective Actions
Repeat sample preparation and recount
Reanalyze sample
Check with weightless (plated) source
Ensure that high voltage is correct
Detector problem
Ensure that high voltage is correct
Check pole zero adjustment
Check baseline restorer
Check stability of amplifier for gain shifts or drifting
Check for loss of vacuum
Repeat sample preparation and recount
Reanalyze the sample
Check pole zero adjustment
Check pile-up rejector
Check stability of amplifier for gain shifts or drifting
Reduce volume of sample analyzed
Increase distance between the source and detector
Check zero offset
Check digital offset
Check amplifier for zero drift
Check amplifier gain
Check conversion gain
Check amplifier for sain shifts
18.5.4 Gas Proportional Systems
18.5.4.1 Voltage Plateaus
Issue: The accuracy of the results produced by a gas proportional system can be affected if the
system is not operated with its detector high voltage adjusted, such that it is on a stable portion of
the operating plateau.
Discussion: The operating portion of a detector plateau is determined by counting an appropriate
source at increasing increments (e.g., 50 volts) of detector high voltage. For detectors which will
be used to conduct analyses for both alpha- and beta-emitting radionuclides, this should be done
with both an alpha and beta source. The sources used should be similar in both geometry and
energy to that of the samples to be counted in the detector.
1137 A plot of the source count rate (ordinate) versus high voltage (abscissa) rises from the baseline to
MARLAP
DO NOT CITE OR QUOTE
18-38
JULY 2001
DRAFT FOR PUBLIC COMMENT
-------
Laboratory Quality Control
1138 a relatively flat plateau region, and then rises rapidly into the discharge region for both the alpha
1139 and beta determinations. From the plateau, the operating voltage is selected or verified. The oper-
1140 ating potential is usually selected in the middle of the plateau. It remains advisable to assure that
1141 the operating point is as far as practical above the plateau knees, and in any case not less than 50
1142 to 100 volts. Operation of the counter at the upper end of the plateau is not recommended and
1143 can result in the generation of spurious discharge counts. Modern high-voltage supplies, oper-
1144 ating properly, experience little actual potential variance. The detector response should be
1145 checked after repairs and after a change of gas. The detector plateau should again be determined
1146 and plotted (voltage vs. count rate) after repairs, particularly to the detector unit.
1147 The historical tracking of the establishment and maintenance of this operating parameter is
1148 recommended; it aids in determining the probable cause of quality control failure and the identi-
1149 fication of long-term instrument deterioration. Items to be recorded include date/time, instrument
1150 detector designation, source number, check source response at the operating point, and pertinent
1151 instrument parameters, such as lower level discriminator setting, alpha discriminator setting,
1152 length of the plateau, operating high voltage setting, etc.
1153 Excursions: Voltage changes of short- or long-term duration will affect reliability of a propor-
1154 tional counter. If the potential is lowered sufficiently, there is a danger of operating below the
1155 plateau knee which, in effect, reduces the efficiency and would bias the results of any sample
1156 count low. Should the voltage applied to the proportional detector be driven up to a point where
1157 the slope of the plateau is sufficiently great enough to increase the efficiency of the detector,
1158 sample counts may be biased high. A transient voltage increase of great enough magnitude could
1159 introduce spurious counts.
1160 Shifts in the operating voltage along the plateau or length of the plateau could also result from
1161 long-term detector deterioration or electronic drift or failure.
1162 18.5.4.2 Self-Absorption, Backscatter, and Crosstalk
1163 Issue: The accuracy of alpha and beta activity determinations in samples with discernable solids
1164 in a gas proportional system depends in large part on the determination and maintenance of self-
1165 absorption and crosstalk curves.
1166 Discussion: Samples counted for alpha and beta activity in a gas proportional system are typi-
1167 cally prepared as inorganic salts, e.g., nitrates, carbonates, oxides, sulfates, or oxalates, and
1168 contain on the order of tens to hundreds of milligrams of solids when counted, which result in
1169 absorption and scattering of the particles in the sample material and mounting planchet (Chapter
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 18-39 DO NOT CITE OR QUOTE
-------
Laboratory Quality Control
1170 16). Thus, for gas proportional systems, the detection efficiency for a given sample depends on
1171 the self-absorption occurring within each sample volume/mass. To establish the correction factor,
1172 a calibration curve is generated using a series of standards consisting of an increasing amount of
1173 solids and known amounts of radionuclide. The relative efficiency for each calibration source is
1174 plotted against the amount of solids, and these data are used to determine a sample's efficiency as
1175 a function of sample weight. The diameter and the composition of the sample planchette, not just
1176 the weight, should be identical with what was used for routine samples. This allows calculation
1177 of the corrected amount of activity regardless of the sample mass (mass/efficiency curves).
1178 The counting of alpha and beta particles simultaneously in a proportional counter requires that an
1179 electronic discriminator be adjusted, such that pulses of heights below that represented by the
1180 discriminator are registered as betas, and those of greater heights are counted as alphas. Crosstalk
1181 occurs when alpha particles are counted in the beta channel or betas are registered as alphas. For
1182 electroplated sources, crosstalk may be as low 1 percent for betas in the alpha channel and 3
1183 percent for alphas in the beta channel. However, this relationship is energy dependent, and care
1184 should be taken to identify samples that differ significantly from the sources used to establish the
1185 crosstalk ratio. For example, 90Sr/90Y (Emax 2.28 meV) is typically used as a beta source for
1186 instrument calibration. However, samples containing natural uranium in equilibrium with its
1187 progeny produce beta emissions that are considerably more energetic from the 3.28 MeV Emax
1188 betas of 214Bi. The crosstalk ratio established with 90Sr will be inadequate for such samples.
1189 As the amount of solids in the sample increases, the alpha into beta crosstalk increases, due to the
1190 degradation of the alpha particle energy by interaction with sample material. Similarly, the beta
1191 into alpha crosstalk decreases. Thus, crosstalk should be evaluated as a function of sample
1192 weight to correct the observed relative alpha and beta counts. This is normally determined in
1193 conjunction with the self-absorption curve. To check these parameters, test samples should be
1194 prepared at the low and high ends of the calibration curve, and the limit of their acceptability
1195 should be better than 1 percent (one sigma). These checks should be performed annually at a
1196 minimum, following detector replacement or significant repair. The historical tracking of the
1197 establishment and maintenance of these operating parameters is recommended. This aids in
1198 determining the probable cause of quality control failure and the identification of long-term
1199 instrument deterioration. In addition, items to be recorded include date/time, instrument detector
1200 designation, source number, operating point, and pertinent instrument parameters, such as lower
1201 level discriminator setting, alpha discriminator setting, etc.
1202 Excursions: Any change in the detector-source geometry or adsorption characteristics between
1203 the source and detector, can affect the self-absorption and crosstalk correction factors. For
1204 example, the replacement of a detector window with one whose density thickness is different
MARLAP JULY 2001
DO NOT CITE OR QUOTE 18-40 DRAFT FOR PUBLIC COMMENT
-------
Laboratory Quality Control
1205
1206
1207
1208
1209
1210
1211
1212
1213
1214
1215
1216
1217
1218
1219
1220
1221
1222
1223
from the original window can necessitate the reestablishment of these parameters. Electronic drift
of the alpha discriminator can also affect the crosstalk ratios.
18.5.5 Liquid Scintillation
Issue: A liquid scintillation counter is essentially a spectrometer that utilizes a multi channel
analyzer to differentiate alpha or beta emission energies. These samples are subject to interferen-
ces from a variety of sources for which corrections should be made to produce useful data. A
detailed discussion of liquid scintillation counting is provided in Chapter 15.
18.5.6 Summary
Table 18.5 provides some example calibration needs, performance frequency, and performance
criteria, listed by detector type. Individual laboratories may be more or less stringent. These items
are just presented as examples for consideration in this section. The table is presented mainly for
the reader to establish their own criteria and is not intended to be a set of minimum requirements.
For additional sources of information, see the calibration frequencies for several detector systems
given in ASTME181 and ANSI N42.12.
TABLE 18.5 — Instrument calibration: example frequency and performance criteria
1224
1225
Example
Calibration Needs
Measurement Parameters
Performance Frequency
Performance Criteria
Gas Proportional System
Initial calibration
Background
counting
Plateau checks as applicable
Crosstalk or sensitivity as
applicable
Counting efficiency to
calculate activity in sample
Weight of solids, when mass
loading is applicable, to
calculate sample activity
Count detector background
using contamination-free clean
planchet
After repairs or major
maintenance on control of
system is re-established
After repairs or major
maintenance on control of
system is re-established
Upon incorporation of new or
changes protocols
One per week or batch when
the system is in use
Plot voltage versus counting
activity to estimate proper
operating voltages for both
alpha and beta
Crosstalk of alpha in beta:
less than 10%; Crosstalk or
sensitivity of beta in alphas:
less than 1%
Counting uncertainty <1%;
<3% uncertainty (2s) over
calibration range
Establish a curve for
efficiency versus mass
loading; <3% uncertainty
(2s) over calibration range
Establish a background
count rate value for total
alpha and beta, with
N>1000
JULY 2001
DRAFT FOR PUBLIC COMMENT
18-41
MARLAP
DO NOT CITE OR QUOTE
-------
Laboratory Quality Control
Example
Calibration Needs
Counter control or
control standard
Measurement Parameters
Use a source of appropriate
energies
Performance Frequency
One per day when the system
is in use
Performance Criteria
Control limits: three sigma
or ± 3%, whichever is
greater
Gamma Spectrometry
Initial calibration
Background
Counter control or
control standard
Detector energy calibration
Counting efficiency matrix-
and geometry-specific
Counter detector background
to establish background level
Multi energy source covering
the general energy calibration
range
After repairs or major
maintenance if control of
system cannot be re-
established
Minimum of every week or
after analytical run, whichever
is longer
One per week or after
analytical run, whichever is
longer
Covers energy range of
desired nuclides; resolution
should be sufficient to
separate gamma-ray lines of
interest from background
peaks and other interfering
lines
Span energy range of
nuclide of interest
Control limits: three sigma
or ± 3%, whichever is
greater
Alpha Spectrometry
Initial calibration
Background
Counter control or
control standard
Energy calibration
Counting efficiency matrix-
and geometry-specific
Counter detector background
to establish background level
At least two isotopes
Monitor peak location,
resolution and efficiency
(where counting efficiency is
an analytical requirement).
After repairs or major
maintenance if control of
system cannot be re-
established
Minimum of every other week
or after analytical run,
whichever is longer
One per week or after
analytical run, whichever is
longer
No specific criteria, pending
on total channel and range
of energy spectrum of
desired nuclides
Span energy range of
nuclide of interest
Control limits: three sigma
or ± 3%, whichever is
greater
Liquid Scintillation
Initial Calibration
Calibration
Dark blank to check
photomultiplier tube
External (instrumental)
calibration
After mechanical or electronic
repairs
After repairs or major
maintenance if control of
system cannot be re-
established
Check against
manufacturer's
specifications
Check against
manufacturer's
specifications
MARLAP
DO NOT CITE OR QUOTE
18-42
JULY 2001
DRAFT FOR PUBLIC COMMENT
-------
Laboratory Quality Control
Example
Calibration Needs
Method Calibration
(Determining
quenching)
Background
Counter control or
control standard
Batch-approach
calibration
(Alternative
aroroach)
Measurement Parameters
Quench curve (at least five
points)
Internal standard
Counter detector background
Minimum two matrix-matched
standards and blanks
Performance Frequency
If matrix or cocktail changes
Add to each sample type
One per day or analytical
batch when the system is in
use
One per day or batch when
system is in use
One per batch
Performance Criteria
Control limits: three sigma
or ± 3%, whichever is
greater
Counting efficiency control
limits: three sigma or ± 5%,
whichever is greater
1241
1242
1243
1244
1245
1246
1247
1248
1249
1250
1251
Sources: ASTME181; ANSIN42.12.
1252
18.5.7 Non-Nuclear Instrumentation
1253 Radioactivity and radionuclide measurement techniques also employ the use of non-nuclear
1254 instrumentation such as mass spectrometry, fluorimetry, phosphorimetry, and fission tract.
1255 Although these instruments are not covered in MARLAP, analysts can apply many of the
1256 laboratory QC techniques discussed in Sections 18.3, 18.4, and 18.6 because they are basic to any
1257 laboratory method. A quality program using statistically based control charts of the performance
1258 indicators will identify out of control situations, assist in improving laboratory performance and
1259 aid in identifying the causes of trends and biases for any laboratory method. Analysts also need to
1260 consider detection capabilities, radionuclide secular equilibrium, half-life, interferences, and
1261 blind samples when using non-nuclear instrumentation.
1262 18.6 Related Concerns
1263 18.6.1 Detection Capability
1264 Issue: The detection capability of an analytical procedure is its ability to distinguish small
1265 amounts of analyte from zero (Chapter 19). The detection capability of a procedure can be
1266 estimated nominally and will depend on many factors.
1267 Discussion: In radioanalysis, the most commonly used measure of detection capability is the
1268 minimum detectable concentration (Chapter 19). The MDC is defined as the smallest concentra-
1269 tion of an analyte that has a specified probability of detection, typically 95 percent. The MDC is
1270 usually estimated as a nominal scoping performance measure of an analytical procedure, but a
JULY 2001
DRAFT FOR PUBLIC COMMENT
18-43
MARLAP
DO NOT CITE OR QUOTE
-------
Laboratory Quality Control
1271 sample-specific version is reported routinely by many laboratories.
1272 Detection capability is affected by many factors, including counting times, instrument back-
1273 ground levels, aliquant volume, yield, decay times, and interferences. The nominal MDC is
1274 presumably based on conservative assumptions about these factors, but measurement conditions
1275 vary. The sample-specific MDC is calculated using the actual measured values of all these
1276 factors. A high MDC by itself does not indicate that a sample result is invalid or that it cannot be
1277 used for its intended purpose. However, if an analysis fails to detect the analyte of interest and
1278 the sample-specific MDC is greater than a detection limit required by contract or other
1279 agreement, it may be necessary to reanalyze the sample in a way that reduces the MDC. Such
1280 decisions should be made case-by-case, since it is not always cost-effective or even possible to
1281 reanalyze a sample, or it may not be feasible to achieve the desired MDC.
1282 Excursions: A high sample-specific MDC can be caused by many factors, including:
1283 • Small sample aliquant;
1284 • Low chemical/tracer yield;
1285 • Short counting times;
1286 • Long decay/short ingrowth time;
1287 • High background or blank value; and
1288 • Low counting efficiency or sample self-attenuation.
1289 18.6.2 Secular Equilibrium
1290 Issue: It is sometimes necessary to ensure that target radionuclides are in secular equilibrium
1291 with their progeny, or to establish and correct for disequilibrium conditions. This is particularly
1292 applicable for protocols that involve the chemical separation of long-lived radionuclides from
1293 their progeny. This is also applicable for nondestructive assays like gamma spectrometry where
1294 photon emission from progeny is used to determine the concentration of the non-gamma ray
1295 emitting parent.
1296 Discussion: Some radionuclides that have long physical half-lives decay to species whose half-
1297 lives are shorter by several orders of magnitude. Following chemical separation of the parent, the
1298 progeny can "grow in" within a time frame relevant to analysis and provide measurable radio-
1299 active disintegration which should be considered in the analytical method. The condition where
1300 the parent and progeny radionuclide are equal in activity is called "secular equilibrium." An
1301 example is 226Ra, a common, naturally occurring radionuclide in the uranium series with a half-
1302 life of about 1,600 years. 226Ra is found in water and soil, typically in secular equilibrium with a
MARLAP JULY 2001
DO NOT CITE OR QUOTE 18-44 DRAFT FOR PUBLIC COMMENT
-------
Laboratory Quality Control
1303 series of shorter-lived radionuclides that begins with the 3.8-day-half-life 222Ra and ends with
1304 stable lead. As soon as 226Ra is chemically separated from its progeny in an analytical procedure
1305 via coprecipitation with barium sulfate, its progeny begin to reaccumulate. The progeny exhibit a
1306 variety of alpha, beta and gamma emissions, some of which will be detected when the precipitate
1307 is counted. The activity due to the ingrowth of radon progeny should be considered when evalua-
1308 ting the counting data (Kirby, 1954). If counting is performed soon after chemical separation,
1309 secular equilibrium will be substantially incomplete and a sample-specific correction factor
1310 should be calculated and applied. In some cases, it may be necessary to derive correction factors
1311 for radioactive ingrowth and decay during the time the sample is counting. These factors are
1312 radionuclide specific, and should be evaluated for each analytical method.
1313 Secular equilibrium concerns also apply to non destructive assays, particularly for uranium and
1314 thorium series radionuclides. Important radionuclides in these series (e.g., 238U and 232Th) have
1315 photon emissions that are weak or otherwise difficult to measure, while their shorter-lived
1316 primary, secondary or tertiary progeny are easily measured. This allows for the parents to be
1317 quantified indirectly, i.e., their concentration is determined by measuring their progeny and
1318 accounting for the amount of parent-progeny equilibrium. The amount of parent-progeny secular
1319 equilibrium is fundamental to these analyses, and data should be scrutinized to insure that the
1320 amount is valid.
1321 When several radionuclides from one decay chain are measured in a sample, observed activity
1322 ratios can be compared to those predicted by decay and ingrowth calculations, the history of the
1323 sample and other information. For example, undisturbed soil typically contains natural uranium
1324 with approximately equal activities of 238U and 234U, while water samples often have very
1325 different 238U/234U ratio. Data from ores or materials involved in processing that could disrupt
1326 naturally occurring relationships require close attention in this regard.
1327 All calculational protocols (electronic and manual) should be evaluated to determine if there is
1328 bias with respect to correction factors related to equilibrium concerns. This includes a check of
1329 all constants used to derive such correction factors, as well as the use of input data that unam-
1330 biguously state the time of all pertinent events (chemical separation and sample counting). The
1331 analyst should ensure that samples requiring progeny ingrowth are held for sufficient time before
1332 counting to establish secular equilibrium. Limits for minimum ingrowth and maximum decay
1333 times should be established for all analytical methods where they are pertinent. For ingrowth, the
1334 limits should reflect the minimum time required to ensure that the radionuclide(s) of interest has
1335 accumulated sufficiently to not adversely affect the detection limit or uncertainty. Conversely, the
1336 time for radioactive decay of the radionuclides of interest should be limited such that the decay
1337 factor does not elevate the MDC or adversely affect the measurement uncertainty. These will
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 18-45 DO NOT CITE OR QUOTE
-------
Laboratory Quality Control
1338 vary depending on the radionuclide(s) and analytical method.
1339 Excursions: Samples where equilibrium is incorrectly assumed or calculated will produce data
1340 that do not represent the true sample concentrations. It is difficult to detect errors in equilibrium
1341 assumptions or calculations. Frequently, it takes anomalous or unanticipated results to identify
1342 these errors. In these cases, analysts need to know the sample history or characteristics before
1343 equilibrium errors can be identified and corrected. Some samples may not be amenable to
1344 nondestructive assays because their equilibrium status cannot be determined; in such cases, other
1345 analytical methods are indicated.
1346 Examples:
1347 Isotopic Distribution - Natural, Enriched and Depleted Uranium: Isotopic distribution is
1348 particularly important with respect to uranium, an element that is ubiquitous in nature in soils
1349 and also a contaminant in many site cleanups. The three predominant uranium isotopes of
1350 interest are 238U, 234U, and 235U, which constitute 99.2745, 0.0055, and 0.72 atom percent,
1351 respectively, of "natural" uranium3, i.e., uranium as found in nature (General Electric, 1984).
1352 However, human activities related to uranium typically involve changing the ratio of natural
1353 uranium by separating the more readily fissionable 235U from natural uranium to produce
1354 material "enriched" in 235U, for use in fuel cycle and nuclear weapons related activities.
1355 Typical 235U enrichments range from 2 percent for reactor fuels to greater than 90 percent 235U
1356 for weapons. The enrichment process also produces material that is "depleted" in 235U, i.e.,
1357 the uranium from which the 235U was taken.4 While the 235U concentrations of depleted
1358 uranium are reduced relative to natural ores, they still can be measured by several assay
1359 techniques. This gives rise to uranium with three distinct distributions of 238U, 235U, and 234U,
1360 referred to as "natural," "enriched," and "depleted" uranium. Because 238U, 235U, and 234U are
1361 alpha emitters with considerably different physical half-lives and specific activity, a measure-
1362 ment of a sample's total uranium alpha activity cannot be used to quantify the sample's
1363 isotopic composition or uranium mass without knowing if the uranium is natural or has been
1364 enriched or depleted in 235U. However, if this information is known, measurement and
1365 distribution of the sample's uranium alpha activity can be used to infer values for a sample's
1366 uranium mass and for the activities of the isotopes 238U, 235U, and 234U. This ratio can be
1367 determined directly or empirically using mass or alpha spectrometry, techniques which are
3 The "natural abundance" of 235U of 0.72 atom percent is a commonly accepted average. Actual values from
specific ore samples vary.
4 Enriched and depleted refer primarily to 235U.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 18-46 DRAFT FOR PUBLIC COMMENT
-------
Laboratory Quality Control
1368 time and cost intensive, but which provide the material's definitive isotopic distribution. It is
1369 often practical to perform mass or alpha spectrometry on representative samples from a site to
1370 establish the material's isotopic distribution, assuming all samples from a given area are
1371 comparable in this respect. Once established, this ratio can be applied to measurements of
1372 uranium alpha activity to derive activity concentrations for 238U, 234U, and 235U data.
1373 18.6.3 Half-Life
1374 Issue: Radionuclides with short half-lives relative to the time frame of the analysis may decay
1375 significantly from the time of sample collection or chemical separation to counting. In some
1376 cases, this decay will cause the ingrowth of other short-lived radionuclides. In both instances,
1377 sample-specific factors should be applied to correct the sample's observed counting/disintegra-
1378 tion rate. Also, determination of half-life could indicate sample purity. If radioactive impurities
1379 are not appropriately corrected, analytical errors will occur. Consecutive counting of the sample
1380 may confirm the radionuclide impurity by analyzing the decay rate between counting events.
1381 Discussion: When assaying for short-lived radionuclides, data should be corrected for decay over
1382 the time period between sample collection and counting. For example, operating power reactors
1383 routinely assay environmental samples for 131I, a fission product with about an eight-day half-life.
1384 Samples may be counted for several days up to two weeks, during which time their 131I concen-
1385 tration is decreasing via radioactive decay. Using the eight-day half-life, the counting data should
1386 be decay-corrected to the time of collection in the field. If desired, environmental samples can be
1387 decay-corrected to a time other than sample collection.
1388 Half-life considerations also apply to radionuclide ingrowth. Certain radionuclides are assayed by
1389 an initial chemical separation which begins a period over which their direct progeny are allowed
1390 to come to secular equilibrium; this is followed by chemical separation, purification and counting
1391 of the progeny. After counting, the degree of the progeny's ingrowth is calculated, based on the
1392 radionuclides' half-lives and the elapsed time between separation and counting. Allowance
1393 should also be made for the progeny's decay from separation to counting and for decay that
1394 occurred while counting, if applicable. Two examples are the beta emitting radionuclides 228Ra
1395 and 90Sr: they are quantified by measuring the direct progeny of each, 228Ac and 90Y, respectively.
1396 For airborne concentrations of 222Rn, sample collection and analytical methods should incorpor-
1397 ate concerns related to the short-lived progeny of other radon species, such as 220Rn. Other half-
1398 life related considerations apply to alpha spectrometry when assaying samples for uranium and
1399 thorium chain radionuclides. Samples that have been allowed to sit for several weeks may
1400 accumulate short-lived radionuclides that have alpha emissions whose energies are in close
1401 proximity to target radionuclides. These can interfere with quantitative analyses of the target
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 18-47 DO NOT CITE OR QUOTE
-------
Laboratory Quality Control
1402 radionuclides. Chemical yield tracers used in alpha spectrometry, such as 234Th and 232U, can
1403 cause this effect due to their short-lived progeny and all chemical yield tracers should be
1404 scrutinized for this potential prior to their use in analytical methods. Radionuclide specific limits
1405 for minimum ingrowth and maximum decay times should be established for all analytical
1406 methods where they are pertinent. These should be based on limiting the adverse effect of such
1407 calculations on the detection limit and measurement uncertainty. All analytical methods
1408 involving computational corrections for radioactive decay of the target species should be
1409 evaluated relative to half-life and secular equilibrium related concerns. This evaluation should be
1410 incorporated in the routine data review process that is performed on all analytical results.
1411 A good source for radionuclide half-lives and other nuclear data can be found at the Brookhaven
1412 National Laboratory's National Nuclear Data Center (http://www.nndc.bnl.gov/nndc/nudat/).
1413 Using this data source will ensure consistency within and among laboratories, and will provide
1414 analysts with the current values.
1415 Excursions: Samples that are assayed by "non destructive" techniques like gamma spectrometry
1416 may provide indications of potential complications due to half-life related considerations.
1417 Because the assay provides information on photon emitting radionuclides in the sample, the
1418 analyst can develop appropriate corrections for half-life related phenomena. However, non-
1419 spectrometric techniques like gas flow proportional counting are essentially gross counting
1420 procedures that record all events without any indication of their origin. Therefore, these data
1421 should be evaluated to ensure they are free from half-life related considerations.
1422 Samples with short-lived radionuclide concentrations at or near environmental background will
1423 experience elevated detection limits and increased measurement uncertainty if there is excessive
1424 elapsed time between sample collection and counting. Because there is an additional correction
1425 factor in the algorithms for these samples (decay factor), they are more susceptible to
1426 measurement uncertainty than longer-lived radionuclides.
1427 18.6.4 Interferences
1428 Issue: Chemical or radionuclide interferences can produce erroneous results or increased
1429 measurement uncertainty.
1430 Discussion: Analytical samples, particularly environmental samples, are often chemically
1431 complex. This complexity may include chemical constituents or other physical aspects that
1432 interfere with an analytical method to the point that they require modification of the method.
1433 Examples of modifications include limiting the size of the sample aliquant, quantifying
MARLAP JULY 2001
DO NOT CITE OR QUOTE 18-48 DRAFT FOR PUBLIC COMMENT
-------
Laboratory Quality Control
1434 interfering compounds through other analyses (radiometric and non-radiometric) and changing
1435 time periods to allow adequate ingrowth of target radionuclides or decay of interferences.
1436 A common example is groundwater or well water that contains high concentrations of salts or
1437 dissolved solids, so that screening for gross alpha activity produces erratic or anomalous results.
1438 For such samples, it may be necessary to limit the aliquant volume with the resulting increase in
1439 detection limit and measurement uncertainty. There is a concentration at which this procedure
1440 cannot overcome the interferences and should not be used.
1441 Samples that contain natural concentrations of stable or unstable compounds that an analytical
1442 procedure adds to the sample for a specific purpose (carrier or tracer) may also be problematic
1443 because the sample's concentration interferes with the analysis. Because barium is used as a
1444 carrier, water samples that contain high concentration of barium may provide inaccurate carrier
1445 yields when screened for alpha-emitting radium isotopes. Quantifying the sample's barium
1446 content prospectively via a non-radiometric technique (e.g., atomic absorption) would be
1447 required to correct for this interference. With respect to unstable compounds, two examples are
1448 provided. The first involves the radiochemical procedure for determining 228Ra in drinking water
1449 that separates radium via coprecipitation with barium sulfate. The precipitate is allowed to come
1450 to equilibrium with its direct progeny 228Ac, which is separated via co-precipitation with yttrium
1451 oxalate, purified, mounted and counted. The yttrium precipitate also carries 90Y, the direct
1452 progeny of 90Sr, a fission product often found in environmental samples as a result of
1453 atmospheric weapons testing and nuclear fuel cycle activities. Samples assayed for 228Ra may
1454 contain measurable amounts of 90Sr that require corrections based on differences in half-life
1455 (228Ac with a 6-hour half-life versus 90Y with a half-life of about 64 hours) or other parameters.
1456 The second example involves alpha spectrometry procedures that use tracers to determine
1457 chemical yield. For example, 234Th is used as a chemical yield tracer for isotopic thorium
1458 analyses. The approach assumes that the sample's inherent concentration of the tracer
1459 radionuclide is insignificant such that it will not interfere with the tracer's ability to accurately
1460 represent the sample's chemical recovery. Samples that contain measurable amounts of these
1461 radionuclides may produce excessive interference and may not be amenable to this procedure.
1462 Alpha spectra should be checked for radionuclide interferences, e.g. look for 238U peak in a Pu
1463 spectra. If the 238U peak is present, 234U might be an interference in the 239Pu and 240Pu
1464 determinations. Data can be corrected or the sample may require reanalysis.
1465 Each analytical method should be evaluated with respect to interferences, when its use is
1466 proposed or at least prior to their implementation in the laboratory. Such evaluations can be
1467 based on available information and, if properly documented, can serve as the basis for developing
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 18-49 DO NOT CITE OR QUOTE
-------
Laboratory Quality Control
1468 the range of applicability, which becomes an integral part of the protocol. Evaluating
1469 performance indicators aids in the identification of samples that have interferences. All
1470 performance criteria would be protocol specific, and have clearly established acceptance ranges
1471 that incorporate the potential interferences discussed above.
1472 Excursions: Interfering elements can affect measurement results in several ways. For example,
1473 large amounts of non-analyte elements may overload ion exchange resins, affecting the resin's
1474 ability to collect all of the analyte. In addition, spiking elements, already in the sample prior to
1475 preparation, may cause matrix spike results to exceed acceptance limits.
1476 Carrier/tracer yields exhibiting gradual changes that appear to be correlated with a batch or group
1477 of samples from the same sampling location may indicate potentially interfering conditions. A
1478 significant decrease in the carrier/tracer recovery may indicate that the analytical method is not
1479 functioning as planned. Yields that are significantly low or in excess of 100 percent may be
1480 caused by competing reactions within the sample matrix, or by the presence of inherent
1481 concentrations of carrier/tracer within the sample.
1482 For screening analyses, e.g., gross alpha or beta, large changes in counting efficiencies or erratic
1483 counting data can reflect the presence of salts. Samples of this type are hydroscopic, and continue
1484 to gain weight following preparation in planchettes as they absorb moisture from the air. These
1485 changes could be detected by reweighing the planchettes directly prior to counting. These
i486 samples can be converted to oxides by carefully holding them over the open flame of a laboratory
1487 burner; however, this will cause losses of volatile radionuclides, predominantly 210Po and 137Cs,
1488 which have alpha and beta emissions, respectively. An alternative approach is to thoroughly dry
1489 each planchette, record the weight and count it immediately, followed by a post-counting
1490 weighing to ensure that the weight did not change significantly over the measurement period.
1491 This approach may not be practical for all laboratories.
1492 18.6.5 Negative Results
1493 Issue: When an instrument background measurement is subtracted from a measurement of a low-
1494 activity sample, it is possible to obtain a net activity value less than zero.
1495 Discussion: Many factors influence the evaluation of negative results. The simplest case occurs
1496 when the background measurement is unbiased and both the gross counts and background counts
1497 are high enough that the distribution of the net count rate is approximately normal. In this case,
1498 normal statistics can be used to determine whether a negative result indicates a problem. For
1499 example, if a sample contains zero activity, there is a very small probability of obtaining a net
MARLAP JULY 2001
DO NOT CITE OR QUOTE 18-50 DRAFT FOR PUBLIC COMMENT
-------
Laboratory Quality Control
1500 count rate more than two-and-a-half or three standard deviations below zero Since the combined
1501 standard uncertainty is an estimate of the standard deviation, a result that is less than zero by
1502 more than three times its combined standard uncertainty should be investigated. In fact, if a blank
1503 sample is analyzed using an unbiased measurement process, negative results can be expected
1504 about 50 percent of the time. As long as the magnitudes of negative values are comparable to the
1505 estimated measurement uncertainties and there is no discernible negative bias in a set of
1506 measurements, negative results should be accepted as legitimate data and their uncertainty should
1507 be assessed. On the other hand, if a sample activity value is far below zero, there may be a reason
1508 to investigate the result. A large percentage of negative results may also indicate a problem, even
1509 if all of the results are near zero. When instrument backgrounds are extremely low, statistics
1510 based on a normal distribution may not be appropriate (Chapter 19).
1511 A preponderance of results that are negative, even if they are close to zero, indicates either a
1512 systematic error or correlations between the results. If the results are measured independently, a
1513 pattern of negative results indicates a bias, which requires investigation.
1514 Excursions: Negative results occur routinely when samples with low levels of activity are
1515 analyzed, but a result should seldom be more than a few standard deviations below zero. Possible
1516 causes for extremely negative results or for an excessive number of negative values include:
1517 • Instrument failure (low sample counts or high blank counts);
1518 • Positive bias in the background or reagent blank measurement;
1519 • Overestimation of interferences;
1520 • Data transcription error; or
1521 • Calculation error.
1522 18.6.6 Blind Samples
1523 Issue: The performance of the analytical method should be assessed independently on a regular
1524 basis. This assessment is achieved through the use of blind samples that provide an objective
1525 means of evaluating the laboratory's performance for specific analytes and matrices. Blind
1526 samples can be internal or external, and either single or double. External blind PE samples are
1527 used for QA purposes and also can provide information that is useful to laboratory QC.
1528 Discussion: A blind sample is a sample whose concentration is not known to the analyst, and
1529 whose purpose is to assess analytical performance. Regardless of their nature, blind samples are
1530 effective only when their contents are unknown to the analysts. The preparation of all blind and
1531 other performance assessment samples is usually designated as a QA function. The QA staff
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 18-51 DO NOT CITE OR QUOTE
-------
Laboratory Quality Control
1532 functions independently from personnel responsible for sample processing and analysis. Blind
1533 samples consist of a matrix routinely processed by the laboratory that contains a known amount
1534 of one or more analytes (radionuclides). A blind also may take the form of a replicate sample that
1535 is submitted for analysis such that its composition and origin are unknown to the analyst. These
1536 can be split samples (if run in the same batch) or spiked samples, and are prepared and submitted
1537 by an independent group either within the organization (internal), or from an independent
1538 organization (external). Performance on blind samples should be an integral part of the labora-
1539 tory's quality system, which includes routine evaluation of them against specific performance
1540 criteria. For example, analysis of blind samples should be evaluated for relevant performance
1541 indicators. Data that fall outside an acceptance criterion may indicate loss of control in sample
1542 chemical processing, radiometric determination (counting) or other aspects of the analytical
1543 process. The ability to prepare blind samples depends fundamentally on the ability to obtain the
1544 appropriate combination of matrix with a radionuclide of a well-known concentration, ideally
1545 traceable to NIST or other appropriate certifying body. Also important are the expertise and
1546 experience of the preparer of the blind samples, proven and verified methodologies used for the
1547 blind samples, and detailed documentation. The use of blind samples assumes that their physical,
1548 chemical and radiological nature are compatible with the analytical methods employed at the
1549 laboratory.
1550 When the analyst is aware that the sample is a blind sample but does not know the concentration,
1551 these samples are called single blinds. In the case of replicates , the analyst is not aware that two
1552 samples are the same; for spiked samples, the analyst may know what analytes the blind sample
1553 contains, but not the analyte's concentration. Single blinds and other internal samples of this type
1554 are generally prepared by an organization's QA personnel that are independent of the samples'
1555 analyses. External single blind samples are available and can be obtained from several sources.
1556 A double blind sample is the same as a single blind except that it is submitted for analysis as a
1557 routine sample. The sample should be identical in appearance to a routine sample, and the analyst
1558 is not forewarned of the analytes in the sample. In general, a double blind is thought to be a more
1559 rigorous indication of the laboratory's performance, since analysts and other laboratory personnel
1560 may take special precautions when analyzing known PT samples, in anticipation of the greater
1561 scrutiny associated with such samples. This should not happen with double blind samples, since
1562 there should be no way to distinguish them from routine samples. However, true double blind
1563 samples are difficult to prepare.
1564 INTERNAL BLIND SAMPLES. Internal blind samples are prepared by the laboratory's QA
1565 personnel. Internal blind samples assess several aspects of the analytical process. They allow
1566 the laboratory to demonstrate that it can successfully process routine samples for a specific
MARLAP JULY 2001
DO NOT CITE OR QUOTE 18-52 DRAFT FOR PUBLIC COMMENT
-------
Laboratory Quality Control
1567 analysis; in other words, they get a measured result within accepted limits. They provide an
1568 auditable, empirical record against specific quality performance criteria. They also demons-
1569 trate the efficacy of analytical methods and areas in need of adjustment. Double blind
1570 samples can pose logistical problems. It may be difficult to prepare internal double blind
1571 samples and submit them to the laboratory for analysis successfully disguised as routine
1572 samples. Evaluation criteria should be established to identify when conditions are out of
1573 acceptance limits.
1574 EXTERNAL BLIND SAMPLES. External blind samples are those prepared by an organization
1575 outside that laboratory. This may be helpful with respect to ensuring that the analyte
1576 concentrations are truly unknown to the analyst; external blinds may offer a greater variety of
1577 matrices and analytes than can easily be produced within the laboratory and augment the
1578 laboratory's internal quality control program. Alternatively, if external blinds are not
1579 appropriate to the laboratory's programs, they will be of limited utility.
1580 If differences between observed and known values typically arise, these should be
1581 investigated thoroughly, as they indicate areas where important details of the analytical
1582 process may have been overlooked. Often a laboratory's observed values agree with the
1583 known value within acceptable tolerances, but are biased high or low. Careful documentation
1584 of the laboratory's performance in this regard can assist in characterizing the fluctuations of a
1585 measurement system or analytical method. Like other performance indicators, large or sudden
1586 changes in bias require scrutiny.
1587 Blind samples should be an integral part of the laboratory's quality control program and they
1588 should be processed according to a predetermined schedule. Important sources of external blind
1589 samples include the NIST Radiochemistry Intercomparison Program (NRIP), National Voluntary
1590 Accreditation Program (NVLAP/EPA), Food and Drug Administration, DOE Lab Accreditation
1591 Program (DOELAP), Quality Assessment Program (DOE QAP), and Multi-Analyte Performance
1592 Evaluation Program (DOE MAPEP).
1593 Excursions: The excursions typically encountered with analytical methods for specific
1594 parameters (carrier/tracer recovery, lack of precision, elevated backgrounds, etc.) apply to blind
1595 samples as well. Additionally, instances where the analysis of external blinds produces values
1596 that do not agree with the known values, may indicate that instrument calibrations or other
1597 correction factors require reevaluation. Problems revealed by the analysis of blind blank samples
1598 can indicate a problem (e.g., bias, blunder) within the laboratory, or conditions where the current
1599 protocol is inadequate. Excursions discovered while analyzing samples from external PE
1600 programs should be addressed.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 18-53 DO NOT CITE OR QUOTE
-------
Laboratory Quality Control
1601 18.6.7 Calibration of Apparatus Used for Weight and Volume Measurements
1602 Issue: Fundamental to all quantitative analysis is the use of the proper weights and volumes.
1603 Analysts should perform careful gravimetric and volumetric measurements (especially in the
1604 preparation of calibration solutions, test sources, and reagents) in order to achieve the desired
1605 levels of precision and bias in each analytical method. Therefore, laboratory balances and
1606 volumetric glassware and equipment should be calibrated and checked periodically to maintain
1607 the desired method performance levels. This section discusses the calibrations of laboratory
1608 balances and volumetric glassware and equipment.
1609 Discussion: Laboratory balances should be periodically calibrated and checked. Most balances
1610 are typically calibrated and certified by the manufacturer once a year. These calibrations are
1611 performed to achieve the manufacturer's specified tolerances for each balance. A calibration
1612 certificate is supplied to the laboratory. In addition to this yearly calibration, daily calibration
1613 checks should be performed by the laboratory. Some laboratories check the balances once a day
1614 or at the time of each use. Any balance failing the daily calibration check should be taken out of
1615 service. Ordinarily, ASTM E617 Class 1 or 2 weights are used to perform the daily calibration
1616 check, depending on application. Over time, daily wear and tear on the weights can affect
1617 calibration, so it is a good idea to get them periodically re-certified or to purchase new weights.
1618 Volumetric glassware and equipment, especially those used in the preparation of instrument
1619 calibration solutions and laboratory control samples, should be calibrated to the desired level of
1620 accuracy. Calibration can either be performed by the manufacturer of the equipment or by
1621 laboratory personnel. Calibration certificates for volumetric pipets and flasks are provided by the
1622 manufacturer at the time of purchase. Borosilicate and pyrex volumetric glassware will hold its
1623 calibration indefinitely provided that it is not exposed to hydrofluoric acid, hot phosphoric acid
1624 or strong alkalis, and that it is not heated above 150 °C when drying. Any glass volumetric pipet
1625 with a damaged tip should be discarded or re-calibrated. The manufacturer of volumetric
1626 automatic pipetting equipment calibrates the equipment and provides a certificate at the time of
1627 purchase. The re-calibration of automatic equipment should be performed annually and can be
1628 performed by the manufacturer, calibration specialty companies, or in-house laboratory
1629 personnel. Outside calibration services should provide a calibration certificate.
1630 Laboratory personnel can calibrate and check volumetric apparatus using procedures like those
1631 specified in ASTM E542. Typically calibrations use volumes of water and are gravimetrically
1632 based. Volumes are corrected for temperature and atmospheric pressure and require thoroughly
1633 cleaned glassware, standard procedures for setting and reading the water meniscus, and accurate
1634 balances and thermometers.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 18-54 DRAFT FOR PUBLIC COMMENT
-------
Laboratory Quality Control
1635 Volumetric glassware is calibrated either "to contain" (TC) or "to deliver" (TD). Glassware
1636 designated as "to contain" requires the complete emptying of the vessel to yield the specified
1637 volume. "To deliver" glassware does not require complete emptying. Specified volumes for this
1638 type of apparatus do not include the residual left from surface adhesion and capillary action. TD
1639 glassware will perform with accuracy only when the inner surface is so scrupulously clean that
1640 the water wets it immediately and forms a uniform film when emptying.
1641 18.7 References
1642 18.7.1 Cited Sources
1643 American National Standards Institute/International Standards Organization/American Society
1644 for Quality Control (ANSI/ISO/ASQC) A3 534-2. Statistics- Vocabulary and Symbols-
1645 Statistical Quality Control.
1646 American National Standards Institute/American Society for Quality Control (ANSI/ASQC) E4.
1647 1994. Specifications and Guidelines for Quality Systems for Environmental Data Collection
1648 and Environmental Technology Programs.
1649 American National Standards Institute (ANSI) Nl. 1. American Nuclear Standard Glossary of
1650 Terms in Nuclear Science and Technology, 1976.
1651 American National Standards Institute (ANSI) N15.37. Guide to the Automation of
1652 Nondestructive Assay Systems for Nuclear Material Control. 1981.
1653 American National Standards Institute (ANSI) N42.12. American National Standard. Calibration
1654 and Usage of Thallium-Activated Sodium Iodide Detector Systems for Assay of
1655 Radionuclides.
1656 American National Standard Institute (ANSI) N42.23. Measurement and Associated
1657 Instrumentation Quality Assurance for Radioassay Laboratories. 1996.
1658
1659 American Society for Testing and Materials (ASTM) D3648, Standard Practices for the
1660 Measurement of Radioactivity., 1995.
1661 American Society for Testing and Materials (ASTM) D6299, Standard Practice for Applying
1662 Statistical Quality Assurance Techniques to Evaluate Analytical Measurement System
1663 Performance, 2000
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 18-55 DO NOT CITE OR QUOTE
-------
Laboratory Quality Control
1664 American Society for Testing and Materials (ASTM) E542, Standard Practice for Calibration of
1665 Laboratory Volumetric Apparatus, 2000.
1666 American Society for Testing and Materials (ASTM) E617, Standard Specification for
1667 Laboratory Weights And Precision Mass Standards, 1997.
1668 American Society for Testing and Materials (ASTM) El 81, Standard Test Methods for Detector
1669 Calibration and Analysis ofRadionuclides.
1670 American Society for Testing and Materials (ASTM) E882, Standard Guide for Accountability
1671 and Quality Control in the Chemical Analysis Laboratory.
1672 American Society for Testing and Materials (ASTM) MNL 7, Manual on Presentation of Data
1673 and Control Chart Analysis ASTM Manual Series, 6th Edition, 1990.
1674 Friedlander, G., Kennedy, J.W., Macias, E.S., and Miller, J.N. 1981. Nuclear and
1675 Radiochemistry. 3rd Edition, John Wiley and Sons, New York.
1676 General Electric Company. 1984. Chart of the Nuclides, Thirteenth Edition.
1677 Gilmore, G. and Hemingway, J.D. 1995. Practical Gamma-Ray Spectrometry. Wiley, Chi Chester,
1678 England.
1679 International Standards Organization (ISO) 5725-1. Accuracy (Trueness and Precision) of
1680 Measurement Methods and Results—Part 1: General Principles and Definitions.
1681 International Standards Organization (ISO) 7870. Control Charts - General Guide and
1682 Introduction.
1683 International Standards Organization (ISO) 7873. Control Charts for Arithmetic Average With
1684 Warning Limits.
1685 International Standards Organization (ISO) 7966. Acceptance Control Charts.
1686 International Standards Organization (ISO) 8258. Shewhart Control Charts.
1687 International Standards Organization/International Electrotechnical Commission (ISO/IEC)
1688 17025. General Requirements for the Competence of Testing and Calibration Laboratories.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 18-56 DRAFT FOR PUBLIC COMMENT
-------
Laboratory Quality Control
1689 December 1999, 26 pp.
1690 Kirby, H.W. 1954. Decay and Growth Tables for the Naturally Occurring Radioactive Series.
1691 Anal. Chew. 26:6, p. 1063-1071.
1692 Lin, Z., K. G. W. Inn, and J. J. Fiilben. 2001. An alternative statistical approach for
1693 interlaboratory comparison data evaluation. Journal of Radioanalytical and Nuclear
1694 Chemistry, 248:1, 163-173.
1695 National Council on Radiation Protection and Measurements (NCRP) 58: A Handbook of
1696 Radioactivity Measurement Procedures, Second Edition. Bethesda, MD. February 1985.
1697 (Supersedes First Edition, November 1978.)
1698 National Environmental Laboratory Accreditation Conference (NELAC). 2000. Quality Systems
1699 Appendix D, Essential Quality Control Requirements. Revision 14. June 29. Available at
1700 http://www.epa.gov/ttn/nelac/2000standards.html.
1701 National Bureau of Standards (NBS). 1964. Handbook of Mathematical Functions. M.
1702 Abramowitz and Stegun, I, Editors.
1703 U.S. Environmental Protection Agency (EPA). 1977. Handbook for Analytical Quality Control
1704 in Radioanalytical Laboratories. EPA-600-7-77-088.
1705 U.S. Environmental Protection Agency (EPA). 1980. Prescribed Procedures for Measurement of
1706 Radioactivity in Drinking Water—Procedure 904.0, Determination of Radium-228 in
1707 Drinking Water. EPA 600-4-80-032.
1708 U.S. Environmental Protection Agency (EPA). 1980. Prescribed Procedures for Measurement of
1709 Radioactivity in Drinking Water—Procedure 908.1 for Total Uranium in Drinking Water.
1710 EPA 600-4-80-032.
1711 18.7.2 Other Sources
1712 American National Standards Institute (ANSI) N42.22. American National Standard.
1713 Traceability of Radioactive Sources to the National Institute of Standards and Technology
1714 (NIST) and Associated Instrument Quality Control.
1715 Chase, G.D. and Rabinowitz, J.L. 1969. Principles of Radioisotope Methodology. 3rd Edition,
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 18-57 DO NOT CITE OR QUOTE
-------
Laboratory Quality Control
1716 Burgess Publishing Co., Minneapolis, MN.
1717 Kanipe, L.G. 1977. Handbook for Analytical Quality Control in Radioanalytical Laboratories.
1718 EPA-600/7-77-088.
1719 U.S. Environmental Protection Agency (EPA). 1995. Guidance for the Preparation of Standard
1720 Operating Procedures (SOPs) for Quality-related Documents. QA/G-6. EPA 600-R-96-027.
1721 Available at http://www.epa.gov/oerrpage/superfund/programs/clp/download/epaqag6.pdf.
1722 Zeigler, L.H. and Hunt, H.M. 1977. Quality Control for Environmental Measurements Using
1723 Gamma-Ray SpectrometryEPK 600-7-77-144.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 18-58 DRAFT FOR PUBLIC COMMENT
-------
Laboratory Quality Control
1724 Attachment ISA: Control Charts
1725 18A.1 Introduction
1726 This attachment provides statistical details to augment Section 18.3.2. The term "statistical
1727 quality control" refers to QC based on statistical principles. Generally, statistical QC in the
1728 laboratory applies the principles of hypothesis testing, with varying degrees of rigor, to make
1729 inferences about a measurement system or process. The primary tool for statistical QC is the
1730 control chart.
1731 The most important purpose for statistical QC in the laboratory is to ensure that measurement
1732 uncertainties are properly estimated. The uncertainty estimate that accompanies a measured value
1733 may be misleading unless the measurement process is in a state of statistical control. Statistical
1734 control implies that the distribution of measured results is stable and predictable. It exists when
1735 all the observed variability in the process is the result of random causes that are inherent in the
1736 process. The existence of variability due to "assignable" causes, including instrumental and
1737 procedural failures and human blunders, which are not inherent in the process, implies that the
1738 process is unpredictable and hence "out of control."
1739 Statistical QC procedures are designed to detect variability due to assignable causes. When such
1740 variability is detected, specific corrective action is required to determine the cause and bring the
1741 measurement process back into a state of statistical control. Laboratory QC procedures should be
1742 strict enough to detect variations in the measurement system that could have a significant impact
1743 on measurement uncertainties.
1744 Statistical QC also may be used in the laboratory to monitor method performance parameters,
1745 such as chemical yield, to ensure that the measurement system is performing as expected. How-
1746 ever, the need for corrective action in the case of a low yield may not be as urgent as in the case
1747 of a malfunctioning radiation counter, since the latter is much more likely to cause underestima-
1748 tion of measurement uncertainties.
1749 The following sections describe the various types of control charts introduced in Section 18.3.2,
1750 including the A'chart, X chart, R chart, and variants of the c chart and u chart for Poisson data.
1751 18A.2 XCharts
1752 Procedure 18.1, shown below, may be used to determine the central line, control limits, and
1753 warning limits for an Xchart. Ideally, the data distribution should be approximately normal,
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 18-59 DO NOT CITE OR QUOTE
-------
Laboratory Quality Control
1754 although the A'chart is often used with other types of distributions. (The data may be tested for
1755 normality using the procedure described in Attachment 19F.)
1756 In order to use Procedure 18.1, an unbiased estimate of the standard deviation of the measured
1757 values X^X2, ..., Xn is required. Although the experimental variance s2 of the data is an unbiased
1758 estimate of the true variance a2, taking the square root of s2 generates a bias . The experimental
1759 standard deviation s is given by the equation
1760
s =
\
n-l T=
(1)
1761
1762
1763
1764
1765
1766
If the data are (approximately) normally distributed, s should then be divided by the value of c4
shown in Table 18A-1 below for the number of degrees of freedom v = n - 1. Thus, a is esti-
mated by si c4. The factor c4 is equal to
C4 =
n- 1
where F denotes the gamma function (NBS 1964 ), but it is well approximated by c4
large n the value of c4 is approximately 1.
(2)
-. For
TABLE 18A-1 — Bias-correction factor for the experimental standard deviation
v = n- 1
1
2
3
4
5
6
7
8
9
10
C4
0.79788
0.88623
0.92132
0.93999
0.95153
0.95937
0.96503
0.96931
0.97266
0.97535
v
11
12
13
14
15
16
17
18
19
20
C4
0.97756
0.97941
0.98097
0.98232
0.98348
0.98451
0.98541
0.98621
0.98693
0.98758
v
21
22
23
24
25
26
27
28
29
30
C4
0.98817
0.98870
0.98919
0.98964
0.99005
0.99043
0.99079
0.99111
0.99142
0.99170
v
31
32
33
34
35
36
37
38
39
40
C4
0.99197
0.99222
0.99245
0.99268
0.99288
0.99308
0.99327
0.99344
0.99361
0.99377
MARLAP
DO NOT CITE OR QUOTE
18-60
JULY 2001
DRAFT FOR PUBLIC COMMENT
-------
Laboratory Quality Control
1767 An alternative method of estimating the standard deviation is based on the average value of the
1768 moving range (ASTM D6299, ASTM E882). The moving range (MR) is the absolute value of
1769 the difference between consecutive measured values Xt and X1+l. If the data are normally distrib-
1770 uted, the expected value of the moving range is
— « 1.128 a (3)
1771 which may be estimated by
-. n-l
n - 1 j=
1772 So, a is estimated by MR /1.128. The moving-range estimate of a may be preferred because it is
1773 less sensitive to outliers in the data. Furthermore, when consecutive values of Xt are correlated, as
1774 for example when a trend is present, the moving-range estimate may produce narrower control
1775 limits, which will tend to lead to earlier corrective action.
1776 Procedure 18.1 (A'chart). Determine the central line, control limits, and warning limits for an X
llll chart based on a series of n independent measurements, which produce the measured values
1778 X^ X2, ..., Xn, during a period when the measurement process is in a state of statistical control.
1779 At least 2 measurements wustbe used. Ideally, at least 20 measurements should be used.
1780 Procedure:
1781 1. Calculate the sum T."=1X..
1782 2. Calculate the arithmetic mean A'using the formula
1783 x=
n 1=1
1784 3. Calculate an unbiased estimatea of the standard deviation (e.g., si c4 or MR/1.128).
1785 4. Define the central line, control limits, and warning limits as follows:
-^
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 18-61 DO NOT CITE OR QUOTE
-------
Laboratory Quality Control
1786
1787
1788
1789
1790
1791
1792
1793
1794
1795
1796
1797
1798
1799
If n is less than 20, a higher rate of false warnings and failures may occur because of the
increased uncertainties of the estimates X and a. So, fewer than 20 measured values should be
used only if 20 values cannot be obtained; and the limits should be recalculated when 20 values
become available.
EXAMPLE
Problem: Suppose a series of 20 observations of a parameter yield the following normally
distributed values.
1,118.9 1,110.5 1,118.3 1,091.0 1,099.8 1,113.7 1,114.4 1,075.1 1,112.
1,120.5 1,104.0 1,125.7 1,117.6 1,097.6 1,099.8 1,102.3 1,119.9 1,107.
Determine the central line and warning and control limits for future measurements.
1,103.7
1,114.9
1800
Solution:
Step 1 Calculate£X. = 22,168.3 .
Step 2
Step 3
Step 4
Calculate the mean ^=22,168.3/20 = 1,108.415
Calculate the experimental standard deviation
s =
\
1
20
i~ H08.415)2 = 12.044
which is based on v = 19 degrees of freedom. Find c4 = 0.98693 for v = 19 in
Table 18.1 (or estimate c.
4/I'
=2L = 0.9870), and calculate
- 3 77
12.044
= 12.2037
0.98693
H
Define the central line, control limits, and warning limits as follows:
CL = 1,108.415
UCL = 1,108.415 + 3(12.2037) = 1,145.0
LCL = 1,108.415 - 3(12.2037) = 1,071.8
UWL = 1,108.415 + 2(12.2037) = 1,132.8
LWL = 1,108.415 - 2(12.2037) = 1,084.0
MARLAP
DO NOT CITE OR QUOTE
18-62
JULY 2001
DRAFT FOR PUBLIC COMMENT
-------
Laboratory Quality Control
1801
18A.3 A'Charts
1802 When subgroup averages are plotted on a control chart, Steps 1 and 2 of Procedure 18.1 may be
1803 used to determine the arithmetic mean X and the standard deviation a of a prior set of data
1804 X^Xj., ..., Xn. If .£ denotes the size of the subgroup, the central line, control limits, and warning
1805 limits for the subgroup average are calculated using the formulas
1806
1807
1808
1809
1810
1811
1812
1813
1814
1815
If n is less than about 20, a higher rate of false warnings and failures may occur because of the
increased uncertainties of the estimates A'and a. For this reason fewer than 20 measured values
should be used only if 20 values cannot be obtained.
EXAMPLE
Problem: Use the data from the preceding example to determine warning and control limits
for subgroup averages when the subgroup size is k= 5.
Solution:
Step 1 Calculate£X. = 22,168.3 .
Step 2 Calculate the mean ~X= 22,168.3 720 = 1,108.415
Step 3 Calculate the experimental standard deviation
s =
\
1
20
- 1108.415)2 = 12.044
which is based on v = 19 degrees of freedom. Find c4 = 0.98693 for v = 19 in
Table 18.1 (or estimate c, ~
V 4
^ = — = 0.9870), and calculate
3 77 '
- S 12 044
CT = _! = J£^±L = 12.2037
c4 0.98693
JULY 2001
DRAFT FOR PUBLIC COMMENT
18-63
MARLAP
DO NOT CITE OR QUOTE
-------
1816
Laboratory Quality Control
Step 4 Define the central line, control limits, and warning limits as follows:
= 1,108.415
LCL^= 1,108.415 - 3(12.2037) / fi = 1,092.0
= 1,108.415 + 3(12.2037)/^/5 = 1,124.8
= 1,108.415 - 2(12.2037) / fi = 1,097.5
= 1,108.415 + 2(12.2037) / fi = 1,119.3
1817
18A.4 R Charts
1818 The range of a set of values is the difference between the largest value and the smallest. Plotting
1819 ranges on a range chart or R chart is used to monitor within group variability because R charts
1820 detect changes in variability more easily. Duplicate measurements for any radiochemistry indi-
1821 cator are made and the difference between the duplicates are used to construct the central line
1822 (the mean range), and the control and warning limits in a similar fashion as in the X chart.
1823 Procedure 18.2 may be used to determine the parameters of the R chart.
1824 Procedure 18.2 (R chart). Determine the central line and control limits for a R chart based on a
1825 series of n independent sets of duplicate measurements, which produce the values R^ R2, ... ,Rn,
1826 during a period when the measurement process is in a state of statistical control.
1827 Procedure:
1828 1. Calculate the range, Rh of each pair of duplicate measurements,
1829
1830
2. Calculate the mean range, R, using the formula
1831
n 1=1
1832
3. Calculate the upper control limit as UCL = 3.267 R.
MARLAP
DO NOT CITE OR QUOTE
18-64
JULY 2001
DRAFT FOR PUBLIC COMMENT
-------
Laboratory Quality Control
1833 This approach may also be used for the moving range of a series of individual results.
1834
1835
1836
1837
1838
1839
1840
1841
1842
1843
1844
1845
1846
1847
1848
1849
1850
1851
1852
1853
1854
1855
The factor 3.267 is called "Z?4" in references on statistical quality control. The value of D4 is
smaller when the range of a larger group is monitored. When the group size is at least seven,
there is also a factor called D3, which may be used to calculate a lower control limit for the range.
Values for D3 and D4 are tabulated in Manual on Presentation of Data and Control Chart
Analysis (ASTM MNL7), as well as many other references.
EXAMPLE
Problem: Suppose a series of 20 duplicate
pairs of values.
(0
(0
(0
(0
Determine
501,0.491)
510,0.488)
523,0.516)
506, 0.508)
(0.490,
(0.505,
(0.500,
(0.485,
0.490)
0.500)
0.512)
0.503)
(0
(0
(0
(0
observations of a parameter yield the following
479, 0.482)
475, 0.493)
513,0.503)
484, 0.487)
(o.
(o.
(o.
(o.
520
500
512
512
the central line and upper control limit for the range
,0.512)
,0.515)
, 0.497)
, 0.495)
of future
(0.500, 0
(0.498, 0
(0.502, 0
(0.509, 0
pairs of
490)
501)
500)
500)
measurements.
Solution:
Stepl
Step 2
Step 3
Calculate the range
C.cu.a.e,
0.010
0.022
0.007
0.002
he mean
Calculate the upper
of each of the 20 pairs .
0.000
0.005
0.012
0.018
range R
0.003
0.018
0.010
0.003
-. 20
1 V^ z? -
20£f '"
control limit: UCL = 3
0.
0.
0.
0.
008
015
015
017
0.189 _
20
.267
0.010
0.003
0.002
0.009
0.00945
# = (3.267)(0.00945)
= 0.0309
ISA.5 Control Charts for Instrument Response
A radioactive check source should be used to monitor the efficiency of every radiation counting
instrument. MARLAP recommends that the activity and count time for the source be chosen to
give no more than 1 percent Poisson counting uncertainty (ANSI N42.23). In other words, at
JULY 2001
DRAFT FOR PUBLIC COMMENT
18-65
MARLAP
DO NOT CITE OR QUOTE
-------
Laboratory Quality Control
1856 least 10,000 counts should be obtained in each measurement of the source.
1857 There may be cases when placing a high-activity source in a detector is undesirable, and
1858 obtaining 10,000 counts is therefore impractical. The instrument response may not have a
1859 Poisson distribution. In this case, if the check source is long-lived, an A'or A'chart based on
1860 replicate measurements should be set up. For example, an XorX chart is the appropriate
1861 efficiency chart for a high-purity germanium detector when the area of a specific photopeak is
1862 monitored, since the calculated size of the photopeak may have significant sources of uncertainty
1863 in addition to counting uncertainty. An A'or X chart may be used even if the response is truly
1864 Poisson, since the Poisson distribution in this case is approximated well by a normal distribution,
1865 but slightly better warning and control limits are obtained by using the unique properties of the
1866 Poisson distribution.
1867 Standard guidance documents recommend two types of control charts for Poisson data. A "c
1868 chart" typically is used in industrial quality control to monitor the number of manufacturing
1869 defects per item. A "u chart" is used to monitor the number of defects per unit "area of
1870 opportunity," when the area of opportunity may vary. Thus, the values plotted on a c chart are
1871 counts and those plotted on a u chart are count rates. The same two types of charts may be
1872 adapted for monitoring counts and count rates produced by a radioactive check source. When a u
1873 chart is used, the "area of opportunity" equals the product of the count time and the source decay
1874 factor. In radiation laboratories a variant of the u chart is more often used when the count time
1875 remains fixed but the decay factor changes during the time when the chart is in use.
1876 Before using control limits derived from the Poisson model, one should use Procedure El,
1877 described in Section 18B.2 of Attachment 18B, to confirm experimentally that the Poisson
1878 approximation is adequate and that any excess variance is relatively small at the expected count
1879 rate. Factors such as source position that may vary during routine QC measurements should be
1880 varied to the same degree during the experiment.
1881 Calculation of warning and control limits using the Poisson model requires only a precise meas-
1882 urement of the source at a time when the instrument is operating properly, preferably near the
1883 time of calibration. The precision can be improved either by counting the source longer or by
1884 averaging several measurements. In principle both approaches should provide equally good esti-
1885 mates of the count rate; however, an advantage of the latter approach is that it can provide the
1886 data needed to detect excess variance (using Procedure El).
1887 Procedures 18.2 and 18.3, listed below, may be used to determine warning and control limits for
1888 measurements of a radioactive check source when the total count follows the Poisson model.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 18-66 DRAFT FOR PUBLIC COMMENT
-------
Laboratory Quality Control
1889 Procedure 18.2 should be used only when the expected count in each measurement is the same,
1890 for example when the source is long-lived and all count durations are equal. Procedure 18.3,
1891 which implements an alternative to the u chart, may be used in all other cases.
1892 Procedure 18.2 (Control chart for Poisson efficiency check data with constant mean) A
1893 check source is counted n times on an instrument, producing the measured counts 7%, N2, ..., Nn.
1894 (Ideally, n is at least 20.) Determine control limits and warning limits for future measurements of
1895 the source count on the same instrument.
1896 Procedure:
1897 1. Estimate the central line by
i n
n 1=1 '
1898 and the standard deviation by
1899 NOTE: The estimate s is biased, but the bias is negligible for the large number of counts typically
1900 obtained from a check source.
1901 2. Define the control limits and warning limits (in counts) as follows:
UCL = CL + 3s UWL = CL+2s
LCL = CL - 3s LWL = CL - 2s
1903 If n is less than 20, a higher rate of false warnings and failures may occur because of the
1904 uncertainty in the estimate of the mean. So, fewer than 20 measurements should be used only if
1905 20 measured values are not available.
1906 Procedure 18.3 (Control chart for Poisson efficiency check data with variable mean) A
1907 check source is counted n times (n > 1) on an instrument, producing the measured counts 7Vl3 N2,
1908 ..., Nn. (It is assumed that the background level is negligible when compared to the source count
1909 rate.) Let tt denote the duration of the /h measurement and dt the decay factor (for example,
1910 exp(-^(At + 0.5 ^))). Determine control limits and warning limits for a future measurement of the
1911 source count on the same instrument when the counting period is Tand the decay factor is D.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 18-67 DO NOT CITE OR QUOTE
-------
Laboratory Quality Control
1912 Procedure:
1913 1. Compute the sums^"=1AA. and Y."=lt1d1.
1914 2. Estimate the mean decay-corrected count rate by
1915
3. Estimate the central line by
CL = rTD
1916
and the standard deviation s by
1917
1918
4. Define the control limits and warning limits as follows:
UCL = CL + 3s
LCL = CL - 3s
UWL = CL + 2s
LWL = CL - 2s
1919
1920
1921
1922
1923
1924
1925
1926
1927
1928
1929
If ^t.d.<20 TD, a higher rate of false warnings and failures may occur because of increased
uncertainty in the estimate of the count rate r.
EXAMPLE
Problem: A source containing 90Sr and 90Y in equilibrium is used for efficiency checks on a
proportional counter. Near the time of calibration, a series of twenty 600-s measurements are
made. The observed counts are as follows:
12,262 12,561 12,606 12,381 12,394 12,518 12,399 12,556 12,565 12,444
12,432 12,723 12,514 12,389 12,383 12,492 12,521 12,619 12,397 12,562
Assume all twenty measurements are made approximately at time 0, so the ten decay factors dt
are all equal to 1. Use Procedure 18.3 to calculate lower and upper control limits for a 600-s
measurement of the same source at a time exactly 1 year later.
MARLAP
DO NOT CITE OR QUOTE
18-68
JULY 2001
DRAFT FOR PUBLIC COMMENT
-------
Laboratory Quality Control
1930
1931
1932
1933
1934
Solution:
Step 1 Compute the sumsEAA. = 249,718 and Et.d. = 12,000.
EN.
Step 2
Step3
Calculate r =
249,718
Et.d. 12,000
= 20.80983.
The decay time for the final measurement is 1 y = 31,557,600 s. The
corresponding decay factor is D= 0.976055. The count time is T= 600 s. So,
compute
CL = (20.80983)(600)(0.976055) = 12,187
and
s = ^/12,187 = 110.39
Step 4 The control limits and warning limits are
UCL = 12,187 + 3 x 110.39 = 12,518
LCL = 12,187 - 3 x 110.39 = 11,856
UWL = 12,187 + 2 x 110.39 = 12,408
LWL = 12,187 -2x 110.39 = 11,966
1935
1936
1937
1938
1939
1940
If substantial excess (non-Poisson) variance is present in the data, the simple Poisson charts
described above should not be used. The c chart may be replaced by an Xchart or X chart, but a
new type of chart is needed to replace the u chart. To determine warning and control limits for
this chart, one must determine the relative excess variance of the data ^2. A value of ^2 may be
assumed or it may be estimated using procedures described in Attachment 18B. Then Procedure
18.3 may be replaced by the Procedure 18.4, shown below.
1941 Procedure 18.4 (Control chart for Poisson efficiency check data with excess variance) A
1942 check source is counted n times on an instrument, producing the measured counts Nly N2, ..., Nn.
1943 Let tj denote the duration of the /h measurement and dt the decay factor. Let the data follow an
1944 approximately Poisson distribution with relative excess variance ^2. Determine control limits and
1945 warning limits for a future measurement of the source count on the same instrument when the
1946 counting period is Tand the decay factor is D.
1947 Procedure:
1948 1. Compute the sums ^"=1N. and E"= l t.d..
1949 2. Estimate the mean decay-corrected count rater by
JULY 2001
DRAFT FOR PUBLIC COMMENT
18-69
MARLAP
DO NOT CITE OR QUOTE
-------
Laboratory Quality Control
/v 7 ™ 1 0 7 7*^ 1
r = where rn =
„ U
y; i
1950 3. Estimate the central line by
CL = fTD
1951 and the standard deviation s by
1=1
1952 4. Define the control limits and warning limits as follows:
UCL = CL + 3s UWL = CL + 2s
LCL = CL - 3s LWL = CL - 2s
1954 18A.6 References
1955 American National Standard Institute (ANSI) N42.23. Measurement and Associated Instru-
1956 mentation Quality Assurance for Radioassay Laboratories. 1996.
1957
1958 American Society for Testing and Materials (ASTM) D6299, Standard Practice for Applying
1959 Statistical Quality Assurance Techniques to Evaluate Analytical Measurement System
1960 Performance, 2000
1961 American Society for Testing and Materials (ASTM) E882, Standard Guide for Accountability
1962 and Quality Control in the Chemical Analysis Laboratory.
1963 American Society for Testing and Materials (ASTM) MNL 7, Manual on Presentation of Data
1964 and Control Chart Analysis ASTM Manual Series, 6th Edition, 1990.
1965 National Bureau of Standards (NBS). 1964. Handbook of Mathematical Functions. M.
1966 Abramowitz and Stegun, I, Editors.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 18-70 DRAFT FOR PUBLIC COMMENT
-------
Laboratory Quality Control
196? Attachment 18B: Statistical Tests for QC Results
1968 18B.1 Introduction
1969 Attachment ISA describes several types of control charts that may be used for statistical quality
1970 control in the laboratory. This attachment describes additional statistical methods that may be
1971 used, where appropriate, to test the performance of measurement results from blank, replicate,
1972 LCS, spikes, CRM, yield-monitor, background, efficiency, calibration, or peak resolution results,
1973 with special emphasis on instrumentation results.
1974 18B.2 Tests for Excess Variance in the Instrument Response
1975 As noted in Chapter 19, the counting uncertainty given by the Poisson approximation does not
1976 describe the total variability in a counting measurement. A number of factors may generate a
1977 small excess component of variance. When a large number of counts are obtained in the meas-
1978 urement, the relative magnitude of the Poisson variance is small; so, the excess component may
1979 dominate.
1980 Regardless of whether replication or the Poisson approximation is used to estimate counting
1981 uncertainties, MARLAP recommends that a series of check source measurements be made on
1982 each instrument periodically to test for excess variance. Procedure El, which is presented below,
1983 may be used to evaluate the measurement results. To check the stability of the instrument itself,
1984 one should perform the measurements while holding constant any controllable factors, such as
1985 source position, that might increase the variance. To check the variance when such factors are not
1986 constant, one may use Procedure El but vary the factors randomly for each measurement.
1987 Assume n measurements of the source produce the counts 7%, N2, . . . , Nn. If the expected count
1988 for each measurement is at least 20, so that the Poisson distribution is approximated by a normal
1989 distribution, and if the average decay-corrected count rate r is determined with adequate
1990 precision, then the quantity
1991 where tt and dt are the count time and source decay factor for the /h measurement, respectively,
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 18-71 DO NOT CITE OR QUOTE
-------
Laboratory Quality Control
1992 should be distributed approximately as chi-square with n - 1 degrees of freedom.5 The precision
1993 of the estimate r should be adequate for the test as long as the expected count for each measure-
1994 ment is at least 20. Since a check source is involved, the expected count is usually much greater
1995 than 20.
1996 Procedure El. Determine whether a series of measurements of a check source provide evidence
1997 of variance in excess of the Poisson counting variance. Let Nt denote the count observed in the /h
1998 measurement. Let wt = ttd^ where tt denotes the count time and dt denotes the source decay factor
1999 (if relevant). If all the values wt are equal, one may use wt=\ instead for all i. It is assumed either
2000 that the background count rate is negligible or that the decay factors are all nearly equal, so that
2001 the expected count in each measurement is proportional to w}6 The procedure tests the null
2002 hypothesis that the total measurement variance is the Poisson counting variance.
2003 Procedure:
2004 1 . Choose the significance level a.
2005 2. Calculate the sums T?j=lN. and ^"=lw..
2006 3 . Estimate the mean decay-corrected count rate by
En
i=\Wi
2007 4. Calculate the chi-square statistic as follows:
(2)
^ '
r = -2^\—-r\ wi
(3)
2008 5. Determine the quantile y^^a(n - 1) (see Table G. 1 in Appendix G). Reject the null
If r denotes the true mean decay-corrected count rate, then under the null hypothesis each measured count rate
Nt / tidi is approximately normal with mean rand variance rl ttdt, and the least-squares estimator for r is
r = E7V,/ 'Et1dr So, the sum E(A^./ ttdt - r)2 / (rl ttd) is approximately chi-square with n- I degrees of freedom.
If r is determined accurately, the true mean count rate r may be replaced in the formula by its estimated value r to
obtain the formulathat_appears in the text. If all the products tjdj are equal, they cancel out of the sum, which
becomes £(7V. - TV)2 /TV, as described by Evans (1955), Goldin (1984), and Knoll (1989).
6 The expected gross count for the /h measurement equals RB tt + rwt, where r is the mean net count rate at time 0.
The expected count is proportional to w, if RB = 0, or if all the decay factors are equal so that tt« w,.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 18-72 DRAFT FOR PUBLIC COMMENT
-------
Laboratory Quality Control
2009 hypothesis if and only if the calculated value of ^2 is greater than^a(/7 - 1). In this case
2010 conclude that the variance is greater than predicted by the Poisson model.
2011
2012
2013
2014
2015
2016
2017
2018
2019
2020
2021
2022
2023
2024
2025
2026
2027
EXAMPLE
Problem: A long-lived source is counted n = 20 times in a gross radiation detector and the
duration of each measurement is 300 s. The following total counts are measured:
11,189 11,105 11,183 10,910 10,998 11,137 11,144 10,751 11,128 11,037
11,205 11,040 11,257 11,176 10,976 10,998 11,023 11,199 11,078 11,149
Are these data consistent with the assumption that the measurement variance is no greater than
predicted by the Poisson model? Use 5 percent as the significance level.
Solution:
Step 1 The significance level is specified to be a = 0.05.
Step 2
Step 3
Step 4
Since the source is long-lived and all the count times are equal, let wt=\ for
each i. Calculate EN. = 221,683 and Ew. = 20.
1 ' 1
Calculate the mean count rate r = 221,683 / 20 = 11,084.15.
Calculate the chi-square statistic
2
=
r 1=1
5
w.
l
20
11,084.15 7~i
. - 11,084.15)2 = 24.87
StepS
The number of degrees of freedom is 20 - 1 = 19. According to Table G.I, the
0.95-quantile for a chi-square distribution with 19 degrees of freedom is 30.14.
Since 24.87 < 30.14, do not reject the null hypothesis. The data are consistent
with the assumption of Poisson counting statistics at the 5 percent significance
level.
A two-sided version of Procedure El may also be used to test whether the measurement variance
is either greater than or less than predicted by the Poisson model. Step 5 must be changed so that
the null hypothesis is rejected if the value of the test statistic ^2 does not lie between the two
quantiles £/2(n - 1) and i^a/2(n - 1).
JULY 2001
DRAFT FOR PUBLIC COMMENT
18-73
MARLAP
DO NOT CITE OR QUOTE
-------
Laboratory Quality Control
2028 A chi-square test may require many measurements or long count times to detect a small excess
2029 variance component. When all measurements have the same expected count |i, the detection limit
2030 for the relative excess variance, or its minimum detectable value, is equal to
(4)
2031 where p is the specified probability of a type n error (failure to detect) (Currie 1972). Note that
2032 since L,D represents a relative variance, its square root L,D represents a relative standard deviation.
2033
2034
2035
2036
2037
2038
2039
2040
2041
2042
2043
EXAMPLE: A long-lived source is counted 20 times, and each
duration. The average of the measured counts is 10,816. If a = p
detectable value of the relative excess variance is estimated by
^i 1 [ %.95(19) J 1 ( 30.14 \
^ 10,816
which corresponds to <
percent.
i relative stand
10,816^ 10.12 'J
ard deviation ^D = ^1.829 x
measurement has the same
= 0.05 , the minimum
1-978 _loonvlo-4
10,816
10^4 = 0.01352, or about 1.35
If (1) the relative excess variance in a measurement is not affected by count time, (2) a fixed total
count time is available, and (3) all measurements have the same expected count (e.g., when all
count times are equal and the source is long-lived), then it is possible to determine the number of
measurements that minimizes ^D (Currie 1972). The optimal number is the number nthat
minimizes the quantity
F(ri) = n
(5)
2044 The solution may be found by computing F(ri) for n = 2, 3, 4, ..., until the computed value
2045 begins to increase. When a = p = 0.05, the optimal number of measurements is n = 15, although
2046 the improvement as n increases from 6 to 15 is slight. If n is increased further, the detection limit
2047 ^ worsens unless the total count time is also increased.
2048 A chi-square test may also be used to test whether the total source measurement variance consists
MARLAP
DO NOT CITE OR QUOTE
18-74
JULY 2001
DRAFT FOR PUBLIC COMMENT
-------
Laboratory Quality Control
2049 of a Poisson component and a specified excess component (Currie 1972). Procedure E2,
2050 described below, implements this test. If the specified component is zero, Procedure E2 is
2051 equivalent to El.
2052 Procedure E2. Determine whether a series of measurements of a check source provide evidence
2053 that the measurement variance is greater than the Poisson component plus a specified excess
2054 component. (Refer to the notation used in Procedure El.) Let ^2 denote the value of the relative
2055 excess variance under the null hypothesis H0.
2056 Procedure:
2057 1. Choose the significance level a.
2058 2. Calculate the sums Yfi= jTV. and Yfi= lwp where 7Vl3 N2, ..., Nn are the measured values.
2059 3. Estimate the mean decay-corrected count rate r in two steps by
E" N. " N. i" w.
r»-^ and '=£- '—JZ- >— (6)
E"1=lWl -i 1 + r0w£2/ 1=1 1 +r0^2
2060 (If wl = w2 = — = wn or ^2 = 0, then r = rQ.)
2061 4. Calculate the chi-square statistic as follows:7
"(NJw.-f)2
5C2 = E- ^ (7)
1=1 rl w. + r t,
2062 5. Determine the quantile ^ _a(n - 1) (see Table G. 1). Reject the null hypothesis if and only
2063 if the calculated value of ^2 is greater than ^ _a(n - 1). In this case conclude that the
2064 relative excess variance is greater than ^2.
2065 Procedure E2, like El, can easily be converted to a two-sided test by changing Step 5.
7 In Currie (1972), the variance of TV, is estimated by TV, + £2TV,2. The estimated variance used here is calculated by
pooling the counting data to reduce any small bias caused by the correlation between TV, and TV, + ^2 TV, .
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 18-75 DO NOT CITE OR QUOTE
-------
Laboratory Quality Control
2066
2067
2068
2069
2070
2071
2072
The excess component may be estimated by solving Equations 18.6 and 18.7 for the value of ^
that gives ^2 = n - 1. An iterative computer algorithm, such as bisection, which repeatedly tries
values of £ and computes ^2 can be used.8 An approximate confidence interval for the relative
excess variance may similarly be found by solving for values of ^ which give j^ = %n±J\j2(n ~ 1
where y is the desired confidence coefficient (Currie, 1972).
Ifwl = w2 = --- = wn, the iterative algorithm is unnecessary. In this case the value of £ may be
estimated directly using the formula
(8)
2073 or by £ = 0 if the preceding formula gives a negative result. Similarly, the approximate lower
2074 confidence limit is given by the formula
,2 = J_
'lower — -
N2
(9)
2075 and the approximate upper confidence limit is given by
t2 =J_
Cupper — ~
N2
(AA. - N)2 - N
(10)
2076
2077
2078
2079
2080
EXAMPLE
Problem: A long-lived efficiency check source is counted once a day for 20 days, and each
measurement has the same duration. Suppose the measured counts (N) are:
14,454 15,140 15,242 14,728 14,756 15,040 14,768 15,128 15,150 14,872
14,845 15,511 15,032 14,746 14,731 14,982 15,047 15,272 14,765 15,143
Newton's method, which converges more rapidly, can also be used, but its use is more practical if one replaces r
by r0 in the denominator of each term of Equation 18.7.
MARLAP
DO NOT CITE OR QUOTE
18-76
JULY 2001
DRAFT FOR PUBLIC COMMENT
-------
Laboratory Quality Control
2081
2082
2083
2084
2085
2086
2087
2088
2089
2090
2091
2092
Use these data to estimate c; and determine a 95 percent two-sided confidence interval for its
value.
Solution: Since the source is long-lived and all the measurements have the same duration,
wl = w2 = ---= w20 and Equations 18.8 through 18.10 may be used. So, calculate
A. = 299,352 and 7?= 299,352/20 = 14,967.6. Then the value of £ is estimated as
$ =
14,967.6
£ (Nj - 14,967.6)2 - 14,967.6 = 0.014463
The 95 percent confidence limits are calculated as follows:
lower
20
-N
1
14,
= 0.0096334
32.852 ^t
1, - 14.
- 14,967.6
upper
N\
Xo025(20 - 1) '=
-N
\
= 0.022846
20
8.9065
(N, - 14,
- 14,967.6
For most practical purposes the excess variance may be considered negligible in a counting
measurement if the total count AAis less than 1 /lOc;2, since, in this case, the excess variance
increases the standard deviation of the measured count by less then 5 percent. Similarly, the
counting variance may be considered negligible if N> 10 / d;2.
JULY 2001
DRAFT FOR PUBLIC COMMENT
18-77
MARLAP
DO NOT CITE OR QUOTE
-------
Laboratory Quality Control
2093
2094
2095
2096
2097
2098
2099
2100
2101
2102
2103
2104
2105
2106
2107
2108
2109
EXAMPLE: Suppose N= 1,000 counts observed in a measurement and £ has been estimated
to be 0.01. Then N= 1 /10^2. The standard uncertainty of Ms evaluated as
u(N) = JN+Z,2N2 = ^1,000
1.05/AA
If N= 100,000, then N= 10 / ^ and
So, i/(7V)
u(N) =
N< 1,000, and i/(7V)
= ^1,100,000 « 1
N> 100,000.
18B.3 Instrument Background Measurements
This section presents statistical tests related to measurements of instrument background levels.
The tests are intended for single-channel detectors but may be applied to multichannel systems if
wide spectral regions are integrated. Tests are described for comparing background levels to
preset limits, for detecting changes in background levels between measurements, and for
detecting the presence of variability in excess of that predicted by the Poisson model.
18B.3.1 Detection of Background Variability
The chi-square test (Procedure El) used to detect excess variance in measurements of a check
source may be adapted for background measurements. Procedure Bl implements a chi-square test
for backgrounds. This test is one-sided, although Step 6 can be modified to implement a two-
sided test.
2110 Procedure Bl. Determine whether a series of measurements of an instrument's background
2111 provide evidence of variance in excess of the Poisson counting variance. Let Nt denote the count
2112 observed in the /h measurement, and let tt denote the count time.
2113 Procedure:
2114 1. Determine the significance level a.
2115 2. Calculate the sums T,"=1N. and ^"=lt..
MARLAP
DO NOT CITE OR QUOTE
18-78
JULY 2001
DRAFT FOR PUBLIC COMMENT
-------
Laboratory Quality Control
2116
3. Estimate the mean background count rate by
(11)
2117
2118
2119
Let tmin be the smallest value of tt. If rtmin > 20, go to Step 5. Otherwise, discard all
measured values N} for which rt.<20. Impossible, restart the test at Step 2; if not, stop.
Calculate the chi-square statistic as follows:
N
(12)
/
2120 6. Determine the quantile %\ -a(# ~ 1) (see Table G. 1 in Appendix G). Reject the null
2121 hypothesis if and only if the calculated value of ^2 is greater than Xi-a(° - 1). In this case,
2122 conclude that the instrument background does not follow the Poisson model.
2123
2124
2125
2126
2127
2128
2129
2130
2131
2132
2133
2134
2135
EXAMPLE
Problem: Twenty overnight background measurements are performed on a proportional
counter. The duration of each measurement is 60,000 s, and the following alpha counts are
measured:
14 23 23 25 28 22 19 26 20 27
30 21 34 32 24 27 25 19 19 25
Are these data consistent with the assumption that the measurement variance is attributable to
Poisson counting statistics? Use 5 percent as the significance level.
Solution:
Step 1 The significance level is specified to be a = 0.05.
Step 2 Calculate EN, = 483 and £ tt = 20 x 60,000 = 1,200,000.
Step 3 Calculate the mean count rate r = 483 /1,200,000 = 0.0004025.
Step 4 Since t . = 60,000, ft . =24.15. Since 24.15 > 20, go to Step 5.
JULY 2001
DRAFT FOR PUBLIC COMMENT
18-79
MARLAP
DO NOT CITE OR QUOTE
-------
Laboratory Quality Control
2136
2137
Step 5 Calculate the chi-square statistic
1 ™( N. Y
Y — - 0.0004025 60,000 = 18.49
0.0004025 £? \, 60,000 )
Step 6 The number of degrees of freedom is 20 - 1 = 19. According to Table G.I, the
0.95-quantile for a chi-square distribution with 19 degrees of freedom is 30.14.
Since 18.49 < 30.14, do not reject the null hypothesis. The data are consistent with
the Poisson model.
2138 All the background tests described below are based on the assumption of Poisson counting
2139 statistics. If Procedure Bl indicates the Poisson assumption is invalid, each test requires
2140 modification or replacement. In most cases, unless the observed background counts are very low,
2141 standard statistical tests for normally distributed data may be used instead (e.g., NBS, 1963;
2142 EPA, 1998).
2143 18B.3.2 Comparing a Single Observation to Preset Limits
2144 High background levels on an instrument degrade detection capabilities and may indicate the
2145 presence of contamination. Unusually low levels on certain types of instruments may indicate
2146 instrument failure. When these issues are of concern, one or both of the two statistical tests
2147 described below may be performed to determine whether the true background level is outside of
2148 its desired range.
2149 The result of the background measurement in counts is assumed to have a Poisson distribution. In
2150 both of the following tests, f denotes the count time, and r denotes the preset lower or upper limit
2151 for the true mean background count rate RB. Given an observed count NB, Procedure B2
2152 determines whether RB > r and B3 determines whether RB < r.
2153 Procedure B2 should be used when ris an upper limit and B3 should be used when ris a lower
2154 limit. Thus, the background level is assumed to be within its acceptable limits unless there is
2155 statistical evidence to the contrary. The alternative approach, which changes the burden of proof,
2156 may be used if /tis large enough.
2157 If rt is extremely large (e.g., if rt > 2,500), there is probably no justification for a statistical test.
2158 Instead, the observed count rate may be compared directly to r.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 18-80 DRAFT FOR PUBLIC COMMENT
-------
Laboratory Quality Control
2159 Procedure B2. Determine whether the mean background count rate RB is greater than r. Test the
2160 null hypothesis H0: RB < r against the alternative hypothesis H^ RB > r.
2161 Procedure:
2162 1. Choose the significance level a.
2163 2. If NB < rt, conclude that there is insufficient evidence to reject the null hypothesis, and
2164 stop. Otherwise, if rt < 20, go to Step 6. If rt > 20, go to Step 3 .
2165 3. Calculate
0.5 + NR - rt
(14)
2177
2178
2179
2180
2181
2166 4. Determine zl_a, the (1 -a)-quantile of the standard normal distribution (see Table G.I in
2167 Appendix G).
2168 5. Reject the null hypothesis if and only if Z> zl_a. Stop.
2169 NOTE: If the background count time t is always the same, a fixed upper control limit may be
2170 calculated using the formula
2171 UCL = round(r? + zl _Jrt)
2172 where round denotes the function that rounds its argument to the nearest integer. Then Steps
2173 3-5 are effectively performed by comparing the observed value NB to UCL.
2
2174 6. Determine ^(ITVg), the a-quantile of the chi-square distribution with 2NB degrees of
2175 freedom (see Table G.I in Appendix G), and calculate Q = 0.5
2176 7. Reject the null hypothesis if and only if Q> rt.
EXAMPLE
Problem: To ensure adequate detection capabilities, a laboratory establishes an upper limit of
0.02 cps for beta backgrounds on a proportional counter. A 6,000-s background measurement
is performed, during which 125 beta counts are observed. Determine whether this
measurement result gives 95 percent confidence that the background is greater than 0.02 cps.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 18-81 DO NOT CITE OR QUOTE
-------
Laboratory Quality Control
2182
2183
2184
2185
2186
2187
2188
2189
2190
2191
2192
2193
2194
2195
2196
2197
Solution: The values of the variables are NB= 125, t= 6,000 and r= 0.02.
Step 1 The significance level a is 1 - 0.95 = 0.05.
Step 2 Since NB > rt= 120 and rt > 20, go to Step 3.
Calculate Z= (0.5 + 125 - 120)//120 = 0.5021.
Table G.I shows that z095 = 1.645.
Step3
Step 4
StepS
Since 0.5021 < 1.645, do not reject the null hypothesis. There is insufficient
evidence to conclude that the beta background exceeds 0.02 cps.
EXAMPLE
Problem: The same laboratory establishes an upper limit of 0.002 cps for alpha backgrounds
on the same counter. A 6,000-s background measurement is performed, during which 19 alpha
counts are observed. Determine whether this measurement result gives 95 percent confidence
that the background is greater than 0.002 cps.
Solution: The values of the variables are NB= 19, t= 6,000 and r= 0.002.
Step 1 The significance level a is 1 - 0.95 = 0.05.
Step 2 Since NB > rt= 12 and rt< 20, go to Step 6.
Step 6 Table G.I shows that %.05(38) = 24-88- So, Q= 0.5 • 24.88 = 12.44.
Step 7 Since 12.44 > 12, reject the null hypothesis. The data give 95 percent confidence
that the alpha background is greater than 0.002 cps.
2198 Procedure B3. Determine whether the mean background count rate RB is less than r. Test the
2199 null hypothesis H0: RB > r against the alternative hypothesis H^ RB < r.
2200 Procedure:
2201 1. Choose the significance level a.
2202 2. If NB > rt, conclude that there is insufficient evidence to reject the null hypothesis, and
2203 stop. Otherwise, if rt < 20, go to Step 6. If rt > 20, go to Step 3.
MARLAP
DO NOT CITE OR QUOTE
18-82
JULY 2001
DRAFT FOR PUBLIC COMMENT
-------
Laboratory Quality Control
2204
3. Calculate
z=
0.5 + ND - rt
(15)
2205
2206
2207
2208
2209
2210
2211
2212
2213
2214
Determine zl_a, the (1 - a)-quantile of the standard normal distribution (see Table G.I in
Appendix G).
Reject the null hypothesis if and only if Z< -zl_a. Stop.
NOTE: If the background count time t is always the same, a lower control limit may be calculated
using the formula
LCL = round(rf - zl _0\
Steps 3-5 are then effectively performed by comparing NB to LCL.
6. Determine ^ ^a(2NB + 2), the (1 - a)-quantile of the chi-square distribution with 2NB+ 2
degrees of freedom (see Table G.I), and calculate Q = 0.5 il_<£lNB + 2).
7. Rej ect the null hypothesis if and only if Q < rt.
2215
2216
2217
2218
2219
2220
2221
2222
2223
2224
2225
EXAMPLE
Problem: A laboratory establishes a lower limit of 0.01 cps for beta backgrounds on a
proportional counter. A 6,000-s background measurement is performed, during which 50 beta
counts are observed. Determine whether this measurement result gives 95 percent confidence
that the background is less than 0.01 cps.
Solution: The values of the variables are NB= 50, t= 6,000 and r= 0.01.
Step 1 The significance level a is 1 - 0.95 = 0.05.
Step 2 Since NB < rt= 60 and rt > 20, go to Step 3.
Step 3 Calculate Z= (0.5 + 50 - 60)7^/60 = -1.226.
Step 4 Table G. 1 shows that z0 95 = 1.645.
Step 5 Since -1.226 > -1.645, do not reject the null hypothesis.
JULY 2001
DRAFT FOR PUBLIC COMMENT
18-83
MARLAP
DO NOT CITE OR QUOTE
-------
Laboratory Quality Control
2226 18B.3.3 Comparing the Results of Consecutive Measurements
2227 If consecutive measurements of the background level on an instrument give significantly differ-
2228 ent values, one should be concerned about the accuracy of any laboratory sample measurements
2229 made between the two background measurements. If the background has increased, the labora-
2230 tory sample activities may have been overestimated. If the background has decreased, the activi-
2231 ties may have been underestimated.
2232 Let A\ and N2 denote the counts observed in two independent background measurements on the
2233 same instrument, and assume they represent Poisson distributions with unknown means. Let ^
2234 and t2 denote the corresponding count times. The following two procedures may be used to
2235 determine whether the difference between the two observed values is significantly larger than
2236 would be expected on the basis of the Poisson model. Procedure B4 determines whether the
2237 second value is significantly greater than the first. Procedure B5 determines whether there is a
2238 significant difference between the two values.
2239
2240
Procedure B4. Determine whether the second mean background count rate R2 is higher than the
first Rl . Test the null hypothesis H0: Rl > R2 against the alternative hypothesis H^ ^ < R2.
2241 Procedure:
2242 1. Choose the significance level a.
2243
2244
2245 Step 6.
2. If 7% / fj > N2 1 12, conclude that there is insufficient evidence to reject the null hypothesis,
and stop. Otherwise, if 7% > 20 and N2 > 20, go to Step 3. If 7% < 20 or N2 < 20, go to
2246 3. Calculate
Z =
N2 A;
N^N2
t,L
(16)
2247 4. Determine z1_a, the (1 - a) -quantile of the standard normal distribution.
2248 5. Rej ect the null hypothesis if and only if Z> zl _ a. Stop.
2249 6. Let p=t1/(t1 + t2) and q=t2/(t1 + £,). If N^
-------
Laboratory Quality Control
TV
(17)
2250
If 7% > A^, calculate S more efficiently using the formula
(18)
2251
7. Reject the null hypothesis if and only if S < a.
2252
2253
2254
2255
2256
2257
2258
2259
2260
2261
2262
EXAMPLE
Problem: A 60,000-s background measurement is performed on an alpha spectrometer and
15 total counts are observed in a particular region of interest. After a test source is counted, a
6,000-s background measurement is performed and 3 counts are observed. Assuming Poisson
counting statistics, is the second measured count rate (0.0005 cps) significantly higher than the
first (0.00025 cps) at the 5 percent significance level?
Solution: The variables are /% = 15, ^ = 60,000, 7V2 = 3, and t2 = 6,000.
Step 1 The significance level a is specified to be 0.05.
Step 2 Since /% / ^ = 0.00025 < 0.0005 = 7V2 / 12, /% < 20, and N2 < 20, go to Step 6.
Step 6
Step 7
=an
66,000 11 66,000
= —. Since TVj > N2, calculate S using the second
formula.
S= 1 - i^-l
17/\11
+/18\/10\18MO
\18/\H/ \H
= 1 -0.7788 =0.2212.
Since S > a, there is not enough evidence to reject the null hypothesis. The second
measured count rate is not significantly higher than the first.
JULY 2001
DRAFT FOR PUBLIC COMMENT
18-85
MARLAP
DO NOT CITE OR QUOTE
-------
Laboratory Quality Control
2263 Procedure B5. Determine whether the mean background count rates are different. Test the null
2264 hypothesis H0: Rl = R2 against the alternative hypothesis H^ Rl * R2.
2265 Procedure:
2266 1. Choose the significance level a.
2267 2. If 7% / fj = N21t2, conclude that there is insufficient evidence to reject the null hypothesis,
2268 and stop. Otherwise, if 7% < 20 or N2 < 20, go to Step 6. If 7% > 20 and N2 > 20, go to
2269 Step 3.
2270 3. Calculate Zusing Equation 18.17.
2271 4. Determine zl_a/2, the (1 - a/2)-quantile of the standard normal distribution.
2272 5. Reject the null hypothesis if and only if |Z| > zl _ /2. Stop.
2273 6. If TVj / fj < N2112, use Procedure B4 with significance level a / 2 to determine whether
2274 Ri N21t2, use Procedure B4 with significance level a / 2 and with the
2275 observations reversed to determine whether R2 < Rl.
2276 18B.4 Negative Activities
2277 When the measured count rate for a test source is less than that of the corresponding instrument
2278 background, giving a negative value for the source activity, Procedure B4 may be used to deter-
2279 mine whether the difference between the two count rates is significantly more than should be
2280 expected on the basis of the Poisson model and the assumption that the source is a blank. (Let A\
2281 and fj be the source count and counting time and let N2 and t2 be the background count and count-
2282 ing time.). If a significant difference is found, it may indicate that the background measurement
2283 was biased, the true background is variable or non-Poisson, or the instrument is unstable.
2284 18B.5 References
2285 Currie, Lloyd A. 1972. The Limit of Precision in Nuclear and Analytical Chemistry. Nuclear
2286 Instruments and Methods 100(3): 387-395.
2287 Environmental Protection Agency (EPA). 1998. Guidance for Data Quality Assessment:
2288 Practical Methods for Data Analysis. EPA QA/G-9, QA97 Version. EPA/600/R-96/084,
MARLAP JULY 2001
DO NOT CITE OR QUOTE 18-86 DRAFT FOR PUBLIC COMMENT
-------
Laboratory Quality Control
2289 EPA, Quality Assurance Division, Washington, DC.
2290 Evans, Robley D. 1955. The Atomic Nucleus. McGraw-Hill, New York, NY.
2291 Goldin, Abraham S. 1984. Evaluation of Internal Control Measurements in Radioassay. Health
2292 Physics 47(3): 361-374.
2293 Knoll, Glenn F. 1989. Radiation Detection and Measurement., 2nd ed. John Wiley and Sons, New
2294 York, NY.
2295 National Bureau of Standards (NBS). 1963. Experimental Statistics. NBS Handbook 91, National
2296 Bureau of Standards, Gaithersburg, MD.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 18-87 DO NOT CITE OR QUOTE
-------
19 MEASUREMENT STATISTICS
2 19.1 Overview
3 This chapter discusses statistical principles and methods applicable to radioanalytical measure-
4 ments, calibrations, data interpretation, and quality control.
5 Laboratory measurements always involve uncertainty, which must be considered when analytical
6 results are used as part of a basis for making decisions. Every measured value obtained by a
7 radioanalytical procedure should be accompanied by an explicit uncertainty estimate. One
8 purpose of this chapter is to give users of radioanalytical data an understanding of the causes of
9 measurement uncertainly and of the meaning of uncertainty statements in laboratory reports. The
10 chapter also describes procedures which laboratory personnel use to estimate uncertainties.
11 The uncertainty associated with laboratory measurements is only a part of the total uncertainty
12 that a data user must consider. Field sampling introduces other types of uncertainty, which are
13 beyond the scope of this chapter.
14 Environmental radioactivity measurements may involve material containing very small amounts
15 of the radionuclide of interest. Measurement uncertainty often makes it difficult to distinguish
16 such small amounts from zero. An important performance characteristic of an analytical proce-
17 dure is therefore its detection capability, which is usually expressed as the smallest concentration
18 of analyte that can be reliably distinguished from zero. Effective project planning requires
19 knowledge of the detection capabilities of the analytical procedures which will be or could be
20 used. This chapter explains the performance measure, called the "minimum detectable concentra-
21 tion," or in certain cases the "minimum detectable amount," that is used to describe radio-
22 analytical detection capabilities, as well as some proper and improper uses for it. The chapter
23 also gives laboratory personnel methods for calculating the minimum detectable concentration.
24 Project planners also need to know the quantification capability of an analytical procedure, or its
25 capability for precise measurement. The quantification capability is expressed as the smallest
26 concentration of analyte that can be measured with a specified relative standard deviation. This
27 chapter explains a performance measure called the "minimum quantifiable concentration," which
28 may be used to describe quantification capabilities.
29 The material in the chapter is arranged so that general information is presented first and the more
30 technical information intended primarily for laboratory personnel is presented last. The general
31 discussion in Sections 19.2 through 19.4 requires little previous knowledge of statistics on the
32 part of the reader and involves no mathematical formulas. Section 19.2 in particular may be
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 19-1 DO NOT CITE OR QUOTE
-------
Measurement Statistics
33 skipped by those familiar with basic statistical concepts. The technical discussion in Sections
34 19.5 through 19.7 requires an understanding of basic algebra and at least some familiarity with
35 the fundamental concepts of probability and statistics. Attachments 19B-G are intended for tech-
36 nical specialists with stronger mathematical backgrounds. The footnotes also contain information
37 which may be skipped by most readers.
38 19.2 Statistical Concepts and Terms
39 19.2.1 Basic Concepts
40 Every laboratory measurement involves a measurement error. Methods for analyzing measure-
41 ment error are generally based on the theory of random variables. A random variable may be
42 thought of as the numerical outcome of an experiment, such as a laboratory measurement, which
43 produces varying results when repeated. In this document a random variable will most often be
44 the result of a measurement. Random variables will usually be denoted by upper-case letters.
45 Of primary importance in almost any discussion of a random variable is its distribution. The
46 distribution of a random variable X describes the possible values of X and their probabilities.
47 Although the word "distribution" has a precise meaning in probability theory, the term will be
48 used loosely in this document. Attachment 19A describes several types of distributions, including
49 the following:
50 • Normal (Gaussian) distributions
51 • Log-normal distributions
52 • Chi-square distributions
53 • Student's ^-distributions
54 • Rectangular, or uniform, distributions
55 • Trapezoidal distributions
56 • Exponential distributions
57 • Binomial distributions
58 • Poisson distributions
59 Normal distributions are particularly important because they appear often in measurement
60 processes. The other types listed are also important in this chapter, but only the exponential,
61 binomial, and Poisson distributions are described in the text.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 19-2 DRAFT FOR PUBLIC COMMENT
-------
Measurement Statistics
62 The distribution of X is uniquely determined by its distribution function., defined by F(x) =
63 Pr[A" < x], where Pr[X < x] denotes the probability that X is less than or equal to x. If there is a
64 function fix) such that the probability of any event a < X < b is equal to \baf(x) dx (i.e., the area
65 under the curve y =f(x) between x = a and x = b), then X is a continuous random variable andyjx)
66 is & probability density function (pdf) forX. When X is continuous, the pdf uniquely describes its
67 distribution. A plot of the pdf is the most often used graphical illustration of the distribution (e.g.,
68 see Figures 19.1 and 19.2), because the height of the graph over a point x indicates the probabil-
69 ity that the value of X will be near x.
70 Two useful numerical characteristics of the distribution of a random variable are its mean and
71 variance. The mean is also called the expectation or the expected value and may be denoted by
72 \ix or E(X). The mean of a distribution is conceptually similar to the center of mass of a physical
73 object. It is essentially a weighted average of all the possible values of X., where the weight of a
74 value is determined by its probability. The variance of X, denoted by o|, Var(JQ, or V(X), is a
75 measure of the variability of X, or the dispersion of its values, and is defined as the expected
76 value of (X -
77 The standard deviation ofX, denoted by cx is defined as the positive square root of the variance.
78 Although the variance appears often in statistical formulas, the standard deviation is a more intui-
79 tive measure of dispersion. If X represents a physical quantity, then cx has the same physical
80 dimensions as X. The variance o|, on the other hand, has the dimensions of X squared.
81 Any numerical characteristic of a distribution, such as the mean or standard deviation, may also
82 be thought of as a characteristic of the random variables having that distribution.
83 The mean and standard deviation of a distribution may be estimated from a random sample of
84 observations of the distribution. The estimates calculated from observed values are sometimes
85 called the sample mean and sample standard deviation. Since the word "sample" here denotes a
86 statistical sample of observations, not a physical sample in the laboratory, metrologists often use
87 the terms arithmetic mean, or average, and experimental standard deviation to avoid confusion.
88 The mean is only one measure of the center of a distribution. Two others are the median and the
89 mode. The median of X is a value x05 that splits the range of X into upper and lower portions
90 which are equally likely, or, more correctly, a value x0 5 such that the probability that X < x05 and
91 the probability that X > x05 are both at least 0.5. The mode of X is its most likely value. Figure
92 19.1 shows the probability density function of a symmetric distribution, whose mean, median,
93 and mode coincide, and Figure 19.2 shows the pdf of an asymmetric distribution, whose mean,
94 median, and mode are distinct.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 19-3 DO NOT CITE OR QUOTE
-------
Measurement Statistics
\i = Mean = Median = Mode
o = Standard deviation
FIGURE 19.1 — A symmetric distribution
Mode
,- - Median
- - • Mean
\i = Mean
o = Standard deviation
FIGURE 19.2 — An asymmetric distribution
95 For some distributions, the median or mode may not be unique. If there is a unique mode, the dis-
96 tribution is called unimodal; otherwise, it is called multimodal.
97 The median of X is also called a quantile of order 0.5, or a 0.5-quantile. In general, ifp is a num-
98 ber between 0 and 1, a/>-quantile of Xis a number xp such that the probability that X< xp is at
99 mostp and the probability thatX < xp is at least/?. A/?-quantile is often called a \00pthpercentile.
100 Sometimes the standard deviation of a nonnegative quantity is more meaningful when expressed
101 as a fraction of the mean. The coefficient of variation, or CV, is defined for this reason as the
102 standard deviation divided by the mean. The coefficient of variation is a dimensionless number,
103 which may be converted to a percentage. The term "relative standard deviation," or RSD, is also
MARLAP
DO NOT CITE OR QUOTE
19-4
JULY 2001
DRAFT FOR PUBLIC COMMENT
-------
Measurement Statistics
104 used. The term "relative variance" is sometimes used to mean the square of the relative standard
105 deviation.
106 The results of two analytical measurements may be correlated when they have measurement
107 errors in common. This happens, for example, if laboratory samples are analyzed using the same
108 instrument without repeating the instrument calibration. Any error in the calibration parameters
109 affects all results obtained from the instrument. This type of association between two quantities X
110 and Y is measured by their covariance, which is denoted by GXJ or Cov(X,Y). The covariance of X
111 and 7 is defined as the expected value of the product (X - |ix) (7 - |i7).
112 Covariance, like variance, is somewhat nonintuitive because of its physical dimensions. Further-
113 more, a large value for the covariance of two variables X and 7 does not necessarily indicate a
114 strong correlation between them. A measure of correlation must take into account not only the
115 covariance oxr, but also the standard deviations GX and o7. The correlation coefficient, denoted
116 by pxj, is therefore defined as GXJ divided by the product of GX and o7. It is a dimensionless
117 number between -1 and +1. The quantities X and 7 are said to be strongly correlated when the
118 absolute value of their correlation coefficient is close to 1.
119 Statistical formulas are generally simpler when expressed in terms of variances and covariances,
120 but the results of statistical analyses of data are more easily understood when presented in terms
121 of standard deviations and correlation coefficients.
122 The lack of a correlation between two quantities X and 7 is not a sufficient condition to guarantee
123 that two valuesX^O and g(Y) calculated from them will also be uncorrelated. A stronger condi-
124 tion called independence is required. For most practical purposes, to say that two quantities are
125 "independent" is to say that their random components are completely unrelated. To be more
126 rigorous, X and 7 are independent if and only if Pr[Jf e / and 7 e J] = Pr[Jf e / ] • Pr[7 e J] for
127 any intervals / and Jin the real line, where the symbol e denotes set membership.
128 When the value of a random variable Xis used to estimate the value of an unknown parameter/?,
129 then Xis called an estimator for p. The bias of X is the difference between the mean \\.x and the
130 actual value/?. If the bias is zero, then X is said to be unbiased, otherwise, Xis biased.
131 19.2.2 Summary of Terms
132 arithmetic mean: The term "arithmetic mean" denotes the estimate of the expectation of a distri-
133 bution calculated by dividing the sum of a set of observed values by the number of values. It is
134 also called the "average."
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 19-5 DO NOT CITE OR QUOTE
-------
Measurement Statistics
135 bias: If X is an estimator for a parameter/?, then the bias of X is \ix - p.
136 coefficient of variation: The coefficient of variation of a nonnegative distribution is the ratio of
137 its standard deviation to its mean.
138 correlated: Two random variables are correlated if their covariance is nonzero.
139 correlation coefficient: The correlation coefficient of two random variables is equal to their
140 covariance divided by the product of their standard deviations.
141 covariance: The covariance of two random variables X and 7, denoted by Cov(X,Y) or GX7, is a
142 measure of the association between them, and is defined as E[(X- |ix)(7 - |i7)].
143 distribution: The distribution of a random variable is a mathematical description of its possible
144 values and their probabilities. The distribution is uniquely determined by its distribution function.
145 distribution function: The distribution function, or cumulative distribution function, of a ran-
146 dom variable X is the function F defined by F(x) = Pr[X < x].
147 estimator: A random variable whose value is used to estimate an unknown parameter/? is called
148 an estimator for p.
149 expectation: The expectation of a random variable X, denoted by E(X) or \ix, is a measure of the
150 center of its distribution and is defined as a probability- weighted average of the possible numer-
151 ical values.
152 expected value: See expectation.
153 independent: A collection of random variables X^ X2, ...,Xnis independent if PrjJ^ e 7l3 X2 e 72,
154 ...,Xne !„] = PrjJfj e /! ] • Pr[X, e 72] • • • Pr[Xn e 7J for all intervals 4 12, ...,!„ in the real line.
155 mean: See expectation.
156 median: A median of a distribution is any number that splits the range of possible values into
157 two equally likely portions, or, to be more rigorous, a 0.5-quantile.
158 mode: The mode of a distribution is its most probable value.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 19-6 DRAFT FOR PUBLIC COMMENT
-------
Measurement Statistics
159 percentile: A lOOp percentile of Xis the same as a/>-quantile of X.
160 probability density function (pdf): ^probability density function for a random variable X is a
161 function f(x) such that the probability of any event a < X < b is equal to the value of the integral
162 J*XX) dx- The pdf, when it exists, equals the derivative of the distribution function.
163 quantile: Kp-quantile of a random variable Xis any value xp such that the probability that X< xp
164 is at most/? and the probability that X
-------
Measurement Statistics
185 of the measurement and should therefore be considered a random variable. The difference
186 between the measured result and the actual value of the measurand is the error of the measure-
187 ment, which is also a random variable.
188 Measurement error may be caused by random effects or systematic effects in the measurement
189 process. Random effects cause the measured result to vary randomly when the measurement is
190 repeated. Systematic effects cause the result to tend to differ from the value of the measurand by
191 a constant absolute or relative amount, or to vary in a nonrandom manner. Generally, both
192 random and systematic effects are present in a measurement process.
193 A measurement error produced by a random effect is a random error, and an error produced by a
194 systematic effect is a systematic error. The distinction between random and systematic errors
195 depends on the specification of the measurement process, since a random error in one measure-
196 ment process may appear systematic in another. For example, a random error in the measurement
197 of the concentration of a radioactive standard solution may be systematic from the point of view
198 of a laboratory that purchases the solution and uses it to calibrate instruments.
199 Measurement errors may also be spurious errors, such as those caused by human blunders and
200 instrument malfunctions. Blunders and other spurious errors are not taken into account in the
201 statistical evaluation of measurement uncertainty. They should be avoided, if possible, by the use
202 of good laboratory practices, or at least detected and corrected by appropriate quality assurance
203 and quality control activities.
204 The error of a measurement is primarily a theoretical concept, because its value is unknowable.
205 The uncertainty of a measurement, however, is a concept with practical uses. According to the
206 GUM, the term "uncertainty of measurement" denotes a "parameter, associated with the result of
207 a measurement, that characterizes the dispersion of the values that could reasonably be attributed
208 to the measurand." The uncertainty of a measured value thus gives a bound for the likely size of
209 the measurement error. In practice, there is seldom a need to refer to the error of a measurement,
210 but an estimate of the uncertainty is required for every measured result.
211 19.3.2 The Measurement Process
212 The first step in defining a measurement process is to define the measurand clearly. The specifi-
213 cation of the measurand is always ambiguous to some extent, but it should be as clear as neces-
MARLAP JULY 2001
DO NOT CITE OR QUOTE 19-8 DRAFT FOR PUBLIC COMMENT
-------
Measurement Statistics
214 sary for the intended purpose of the data.1 For example, when measuring the concentration of a
215 radionuclide in a laboratory sample, it is generally necessary to specify the concentration as of a
216 certain date and time and whether the entire sample or only a certain fraction is of interest. For
217 very accurate work, it may be necessary to specify other conditions, such as temperature (e.g.,
218 concentration per unit volume of liquid at 20°C).
219 Often the measurand is not measured directly but instead an estimate is calculated from the meas-
220 ured values of other input quantities, which have a known mathematical relationship to the
221 measurand. For example, input quantities in a measurement of radioactivity may include the
222 gross count, instrument background count, counting efficiency, and test portion size. The second
223 step in defining the measurement process is therefore to determine the mathematical model for
224 the relationship between the measurand Y and measurable input quantities Xt on which its value
225 depends. The relationship may be a simple functional relationship, expressible as 7 =
226 J(Xl,X2,... Jffj), or it may happen that Y is most conveniently expressed as the simultaneous
227 solution of a set of equations.
228 The mathematical model for a radioactivity measurement often has the general form
Y_ (Gross Instrument Signal) - (Blank Signal + Estimated Interferences)
Sensitivity
230 Each of the quantities shown here may actually be a more complicated expression. For example,
231 the sensitivity (the ratio of the net signal to the concentration) may be the product of factors such
232 as the mass of the test portion, the chemical yield, and the instrument counting efficiency.
233 When the measurement is performed, a value xt is estimated for each input quantity, Xt, and an
234 estimated value y of the measurand is calculated using the relationship^ =f[xl,x2,.. .,%).2 Since
235 there is an uncertainty in each input estimate, xt, there is also an uncertainty in the output
236 estimate, y. In order to obtain a complete estimate of the uncertainty ofy, all input quantities that
237 could have a potentially significant effect ony should be included in the model.
1 Because of the unavoidable ambiguity in the specification of the measurand, one should, to be precise, speak of
"a value" of the measurand and not "the value."
2 In accordance with the GUM, an uppercase Roman letter is used here to denote both the input or output quantity
and the random variable associated with its measurement, while a lowercase letter is used for the estimated value of
the quantity. For simplicity, in most of the later examples this convention will be abandoned. Only one symbol will
be used for the quantity, the random variable, and the estimated value of the quantity.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 19-9 DO NOT CITE OR QUOTE
-------
Measurement Statistics
238 19.3.3 Analysis of Measurement Uncertainty
239 Determining the uncertainty of the output estimate y requires that the uncertainties of all the input
240 estimates xt be determined and expressed in comparable forms. The uncertainty of xt is expressed
241 in the form of a standard deviation, called the standard uncertainty and denoted by w(x,), or in the
242 form of a variance, denoted by w2(x;), which is the square of the standard uncertainty. A standard
243 uncertainty is sometimes informally called a "one-sigma" uncertainty. The ratio u(x,) I xt is called
244 the relative standard uncertainty of xt. If the input estimates are potentially correlated, covariance
245 estimates u(xt,x^) must also be determined. The covariance u(xt,Xj) is often recorded and presented
246 in the form of an estimated correlation coefficient, r(xt,Xj), which is defined as the quotient
247 ufaXj) I u(x^)u(x^). The standard uncertainties and estimated covariances are combined to obtain
248 the combined standard uncertainty of_y, denoted by uc(y). (The term "total propagated uncertain-
249 ty," or TPU, has been used for the same concept; however, MARLAP recommends the ISO
250 terminology.) The square of the combined standard uncertainty, denoted by u2c(y\ is called the
251 combined variance.
252 The process of combining the standard uncertainties of the input estimates xt to obtain the com-
253 bined standard uncertainty of the output estimate^ is called "uncertainty propagation." Mathe-
254 matical methods for propagating uncertainty and for evaluating the standard uncertainties of the
255 input estimates are described in Section 19.5.
256 Methods for evaluating the standard uncertainties u(x,) are classified as either Type A or Type B.
257 A Type A evaluation of a standard uncertainty u(x,) may be performed by making a series of inde-
258 pendent measurements of the quantity xt and calculating the arithmetic mean and experimental
259 standard deviation of the mean. The arithmetic mean is used as the input estimate xt and the
260 experimental standard deviation of the mean is used as the standard uncertainty u(x^). There are
261 other Type A methods, but all are based on repeated measurements. Any evaluation of standard
262 uncertainty that is not a Type A evaluation is a Type B evaluation.
263 Sometimes a Type B evaluation of uncertainty involves making a best guess based on all avail-
264 able information and professional judgment. Laboratory workers may be reluctant to make this
265 kind of evaluation, but it is better to make an informed guess about an uncertainty component
266 than to ignore it completely.
267 A standard uncertainty u(x,) may be called a "Type A" or "Type B" standard uncertainty, depend-
268 ing on its method of evaluation, but no distinction is made between the two types for the
269 purposes of uncertainty propagation.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 19-10 DRAFT FOR PUBLIC COMMENT
-------
Measurement Statistics
270 19.3.4 Corrections for Systematic Effects
271 When a systematic effect in the measurement process has been identified and quantified, a quan-
272 tity should be included in the mathematical measurement model to correct for it. The quantity,
273 called a correction (additive) or correction factor (multiplicative), will have an uncertainty which
274 should be evaluated and propagated.
275 Whenever a previously unrecognized systematic effect is detected, the effect should be investi-
276 gated and either eliminated procedurally or corrected mathematically.
277 19.3.5 Counting Uncertainty
278 The counting uncertainty of a radiation measurement (historically called "counting error") is the
279 component of uncertainty caused by the random nature of radioactive decay and radiation count-
280 ing. Radioactive decay is inherently random in the sense that two atoms of a radionuclide will
281 generally decay at different times, even if they are identical in every discernible way. Radiation
282 counting is also inherently random unless the efficiency of the counting instrument is 100%.
283 In many cases the counting uncertainty in a single gross radiation counting measurement can be
284 estimated by the square root of the observed counts. The Poisson counting model, which is the
285 mathematical basis for this rule, is discussed in Section 19.6. Note that the use of this approxi-
286 mation is a Type B evaluation of uncertainty.
287 Historically many radiochemistry laboratories reported only the counting uncertainties of their
288 measured results. MARLAP recommends that a laboratory consider all possible sources of meas-
289 urement uncertainty and evaluate and propagate the uncertainties for all sources believed to be
290 potentially significant in the final result.
291 19.3.6 Expanded Uncertainty
292 The laboratory may report the combined standard uncertainty, uc(y), or it may multiply uc(y) by a
293 factor k, called a coverage factor, to produce an expanded uncertainty, denoted by U, such that
294 the interval from;; - Utoy+ C/has a specified high probability/? of containing the value of the
295 measurand. The specified probability, p, is called the level of confidence or the coverage proba-
296 bility and is generally only an approximation of the true probability of coverage.
297 When the distribution of the measured result is approximately normal, the coverage factor is
298 often chosen to be k = 2 for a coverage probability of approximately 95%. An expanded uncer-
JULY2001 MARLAP
DRAFT FOR PUBLIC COMMENT 19-11 DO NOT CITE OR QUOTE
-------
Measurement Statistics
299 tainty calculated with k = 2 or 3 is sometimes informally called a "two-sigma" or "three-sigma"
300 uncertainty. In general, if the desired coverage probability is y and the combined standard uncer-
301 tainty is determined accurately, the coverage factor for a normally distributed result is k = z(1+y)/2,
302 which can be found in a table of quantiles of the standard normal distribution (see Table G.I in
303 Appendix G).
304 The GUM recommends the use of coverage factors in the range 2-3 when the combined standard
305 uncertainty is determined accurately. Attachment 19C describes a more general procedure for
306 calculating the coverage factor kp that gives a desired coverage probability/? when there is sub-
307 stantial uncertainty in the estimate of uc(y).
308 19.3.7 Significant Figures
309 The number of significant figures that should be reported for the result of a measurement
310 depends on the uncertainty of the result. A common convention is to round the uncertainty
311 (standard uncertainty or expanded uncertainty) to either one or two significant figures and to
312 report both the measured value and the uncertainty to the resulting number of decimal places
313 (ISO 1995, Bevington 1992, EPA 1980). MARLAP recommends this convention and suggests
314 that uncertainties be rounded to two figures. The following examples demonstrate the application
315 of the rule.
316
317
318
319
320
321
322
323
324
325
326
327
328
329
EXAMPLES
MEASURED
VALUE
00
0.8961
0.8961
0.8961
0.8961
0.8961
EXPANDED
UNCERTAINTY
U = kufy)
0.0234
0.2342
2.3419
23.4194
234.1944
REPORTED
RESULT
0.896 ± 0.023
0.90 ± 0.23
0.9 ±2.3
1±23
0±230
Only final results should be rounded in this manner. Intermediate results in a series of calculation
steps should be carried through all steps with additional figures to prevent unnecessary roundoff
errors. Additional figures are also recommended when the data are stored electronically. Round-
ing should be performed only when the result is reported. (See Section 19.6.10 for a discussion of
the measurement uncertainty associated with rounding.)
MARLAP
DO NOT CITE OR QUOTE
19-12
JULY 2001
DRAFT FOR PUBLIC COMMENT
-------
Measurement Statistics
330 19.3.8 Reporting the Measurement Uncertainty
331 When a measured value y is reported, its uncertainty should always be stated. The laboratory may
332 report either the combined standard uncertainty uc(y) or the expanded uncertainty U.
333 The measured value y and its expanded uncertainty f/may be reported in the format^ ± U or
334 y +- U.
335 The plus-minus format may be used to report an expanded uncertainty, but it generally should be
336 avoided when reporting a standard uncertainty, because readers are likely to interpret it as a con-
337 fidence interval. A commonly used shorthand format for reporting a result with its standard
338 uncertainty places the one or two digits of the standard uncertainty in parentheses immediately
339 after the corresponding final digits of the rounded result. For example, if the rounded result of the
340 measurement is 1.92 and the standard uncertainty is 0.14, the result and uncertainty may be
341 shown together as 1.92(14). One may also report the standard uncertainty explicitly.
342 Since laboratories may calculate uncertainties using different methods and report them using
343 different coverage factors, it is a bad practice to report an uncertainty without explaining what it
344 represents. Any analytical report, even one consisting of only a table of results, should state
345 whether the uncertainty is the combined standard uncertainty or an expanded uncertainty, and in
346 the latter case it should also state the coverage factor used and the approximate coverage prob-
347 ability. A complete report should also describe the methods used to calculate the uncertainties.
348 The uncertainties for environmental radioactivity measurements should be reported in the same
349 units as the results. Relative uncertainties (i.e., uncertainties expressed as percentages) may also
350 be reported, but the reporting of relative uncertainties alone is not recommended when the
351 measured value may be zero, because the relative uncertainty in this case is undefined. A partic-
352 ularly bad practice, sometimes implemented in software, is to compute the relative uncertainty
353 first and multiply it by the measured value to obtain the absolute uncertainty. When the measured
354 value is zero, the uncertainty is reported incorrectly as zero. Reporting of relative uncertainties
355 without absolute uncertainties for measurements of spiked samples or standards generally
356 presents no problems, because the probability of a negative or zero result is negligible.
357 It is possible to calculate radioanalytical results that are less than zero, although negative radio-
358 activity is physically impossible. Laboratories sometimes choose not to report negative results or
359 results that are near zero. Such censoring of results is not recommended. All results, whether
360 positive, negative, or zero, should be reported as obtained, together with their uncertainties.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 19-13 DO NOT CITE OR QUOTE
-------
Measurement Statistics
361 The preceding statement must be qualified, because a measured value y may be so far below zero
362 that it indicates a possible blunder, procedural failure, or other quality control problem. Usually,
363 if y + 3 uc(y) < 0, the result should be considered invalid, although the accuracy of the uncertainty
364 estimate uc(y) must be considered, especially in cases where only few counts are observed during
365 the measurement and counting uncertainty is the dominant component of uc(y). (See Chapter 18,
366 Laboratory Quality Control, and Attachment 19C of this chapter.)
367 19.3.9 Recommendations
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
MARLAP makes the following recommendations.
• All radioanalytical laboratories should adopt the terminology and methods of the Guide
to the Expression of Uncertainty in Measurement (ISO 1995) for evaluating and
reporting measurement uncertainty.
• Each measured value should be reported with either its combined standard uncertainty
or its expanded uncertainty.
• The reported measurement uncertainties should be clearly explained. In particular, the
coverage factor and approximate coverage probability should be stated whenever an
expanded uncertainty is reported.
• A laboratory should consider all possible sources of measurement uncertainty and
evaluate and propagate the uncertainties for all sources believed to be potentially
significant in the final result.
• Each uncertainty should be rounded to two significant figures, and the measured value
should be rounded to the same number of decimal places as its uncertainty.
• All results, whether positive, negative, or zero, should be reported as obtained, together
with their uncertainties.
19.3.10 Summary of Terms
blunder: mistake made by a person performing a measurement.
MARLAP
DO NOT CITE OR QUOTE
19-14
JULY 2001
DRAFT FOR PUBLIC COMMENT
-------
Measurement Statistics
386 combined standard uncertainty: standard uncertainty of an output estimate calculated by
387 combining the standard uncertainties of the input estimates. The combined standard uncertainty
388 of y is denoted by uc(y).
389 combined variance: the square of the combined standard uncertainty. The combined variance of
390 y is denoted by u2c(y).
391 counting error: See counting uncertainty. MARLAP uses the term "counting uncertainty" to
392 maintain a clear distinction between the concepts of measurement error and uncertainty.
393 counting uncertainty: component of measurement uncertainty caused by the random nature of
394 radioactive decay and radiation counting.
395 coverage factor: value k multiplied by the combined standard uncertainty uc(y) to give the
396 expanded uncertainty U.
397 coverage probability: approximate probability that the reported interval will contain the value of
398 the measurand.
399 error (of measurement): difference between a measured result and the value of the measurand
400 (cf uncertainty of measurement).
401 expanded uncertainty: product U of the combined standard uncertainty of a measured value y
402 and a coverage factor k chosen so that the interval from_y - Utoy+ C/has a desired high proba-
403 bility of containing the value of the measurand Y.
404 GUM: abbreviation used in this chapter for the Guide to the Expression of Uncertainty in
405 Measurement (ISO 1995).
406 input estimate: measured value of an input quantity.
407 input quantity: any of the quantities in a mathematical measurement model whose values are
408 measured and used to calculate the value of another quantity, called the output quantity.
409 level of confidence: See coverage probability.
410 measurand: quantity subject to measurement.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 19-15 DO NOT CITE OR QUOTE
-------
Measurement Statistics
411 output estimate: calculated value of an output quantity.
412 output quantity: the quantity in a mathematical measurement model whose value is calculated
413 from the measured values of other quantities in the model.
414 random effect: any effect in a measurement process which causes the measured result to vary
415 randomly when the measurement is repeated.
416 random error: a measurement error which varies randomly when the measurement is repeated
417 — caused by random effects.
418 relative standard uncertainty: the ratio of the standard uncertainty of a measured result to the
419 result itself. The relative standard uncertainty of x may be denoted by ur(x).
420 sigma (o): The term "sigma" is sometimes used informally to mean "standard uncertainty," and
421 "&-sigma" is used to mean an expanded uncertainty calculated using the coverage factor k. The
422 symbol o and the term "sigma" are more properly used to denote a true standard deviation.
423 spurious error: a measurement error caused by a human blunder, instrument malfunction, or
424 other unexpected or abnormal event
425 standard uncertainty: uncertainty of a measured value expressed as a standard deviation —
426 often called a "1-sigma" uncertainty. The standard uncertainty of x is denoted by u(x).
427 systematic effect: any effect in a measurement process which does not vary randomly when the
428 measurement is repeated.
429 systematic error: a measurement error which does not vary randomly when the measurement is
430 repeated — caused by systematic effects.
431 total propagated uncertainty (TPU): See combined standard uncertainty, which is the
432 preferred term.
433 Type A evaluation: experimental evaluation of a standard uncertainty or covariance using
434 repeated measurements.
435 Type B evaluation: evaluation of a standard uncertainty or covariance by a method that is not a
436 Type A method.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 19-16 DRAFT FOR PUBLIC COMMENT
-------
Measurement Statistics
437 uncertainty (of measurement): "parameter, associated with the result of a measurement, that
438 characterizes the dispersion of the values that could reasonably be attributed to the measurand"
439 (ISO 1993a).
440 uncertainty propagation: mathematical technique for combining the standard uncertainties of
441 the input estimates for a mathematical model to obtain the combined standard uncertainty of the
442 output estimate.
443 19.4 Detection and Quantification Capability
444 19.4.1 Analyte Detection Decisions
445 An obvious question to be answered following the analysis of a laboratory sample is: "Does the
446 sample contain a positive amount of the analyte?" Uncertainty in the measured value often makes
447 the question difficult to answer. There are different methods for making a detection decision, but
448 the methods most often used in radiochemistry involve the principles of statistical hypothesis
449 testing.
450 Hypothesis testing has been used for analyte detection in radiochemistry since at least 1962. Two
451 influential early publications on the subject were Altshuler and Pasternack 1963 and Currie 1968.
452 Other important but perhaps less well-known documents were Nicholson 1963 and 1966. Most
453 approaches to the detection problem have been similar in principle, but there has been inadequate
454 standardization of terminology and methodology. However, there has been recent progress. In
455 1995 the International Union of Pure and Applied Chemistry (IUPAC) published "Nomenclature
456 in Evaluation of Analytical Methods Including Detection and Quantification Capabilities"
457 (IUPAC 1995), which recommends a uniform approach to defining various performance char-
458 acteristics of any chemical measurement process, including detection and quantification limits;
459 and in 1997 the International Organization for Standardization (ISO) issued the first part of ISO
460 11843 "Capability of Detection," a two-part standard which deals with issues of detection in an
461 even more general context of measurement (ISO 1997). Part 1 of ISO 11843 includes terms and
462 definitions. Part 2, which is not available at the time of this writing, will deal with methodology.
463 Although members of the IUPAC and ISO working groups collaborated during the development
464 of their guidelines, substantial differences between the final documents remain. MARLAP
465 follows both the ISO and IUPAC guidelines where they agree but prefers the definitions of ISO
466 11843-1 for the critical value and minimum detectable value, relating them to the terminology
467 and methodology already familiar to most radiochemists.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 19-17 DO NOT CITE OR QUOTE
-------
Measurement Statistics
468 In July 2000, ISO also published the first three parts of ISO 11929 "Determination of the Detec-
469 tion Limit and Decision Threshold for Ionizing Radiation Measurements" (ISO 2000a-c). Unfor-
470 tunately, ISO 11929 is not completely consistent with either the earlier ISO standard or the
471 IUPAC recommendations.
472 In the terminology of ISO 11843-1, the analyte concentration of a laboratory sample is the state
473 variable, denoted by Z, which represents the state of the material being analyzed. Blank material
474 is said to be in the basic state. The state variable cannot be observed directly, but it is related to
475 an observable response variable, denoted by 7, through a calibration function F, the mathemat-
476 ical relationship being written as Y = F(Z). In radiochemistry the response variable Y is most
477 often an instrument signal, such as the number of counts observed. The difference between the
478 state variable Z and its value in the basic state is called the net state variable, which is denoted
479 by X. In radiochemistry there generally is no difference between the state variable and the net
480 state variable, because the basic state is represented by material whose analyte concentration is
481 zero. (In principle the basic state might correspond to a positive concentration, but MARLAP
482 does not address this scenario.)
483 A detection decision requires a choice between two hypotheses about the material being ana-
484 lyzed. The first hypothesis is the "null hypothesis" H0: The analyte concentration of the material
485 is no greater than that of the blank (i.e., the material is in the basic state). The second hypothesis
486 is the "alternative hypothesis" H^ The analyte concentration of the material is greater than that of
487 the blank. The choice between the two hypotheses is based on the observed value of the response
488 variable Y. The value of Y must exceed a certain threshold value to justify rejection of the null
489 hypothesis. This threshold is called the critical value of the response variable and is denoted
490 by yc. The calculation of yc requires the choice of a significance level for the test. The signifi-
491 cance level is the probability a that the null hypothesis will be rejected in a situation where it is
492 in fact true (i.e., a "type I error," or "false positive"). The significance level a is usually chosen to
493 be 0.05. This means that when a blank sample is analyzed, there is a 5% probability of incor-
494 rectly deciding that the analyte is present. A smaller value of a makes type I errors less likely, but
495 also makes type U errors ("false negatives") more likely when the laboratory sample concentra-
496 tion is near the blank concentration.
497 The term "blank" here may mean any of several types of blanks, including instrument blanks (or
498 backgrounds) and reagent blanks. The blank is chosen to provide an estimate of the mean signal
499 produced by an actual sample that contains none of the analyte, whether the signal is produced by
500 the instrument background, contaminated reagents, or other causes.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 19-18 DRAFT FOR PUBLIC COMMENT
-------
Measurement Statistics
501 The inverse F~l of the calibration function is sometimes called the evaluation function (IUPAC
502 1995). The evaluation function, which gives the value of the net concentration in terms of the
503 response variable, is closely related to the mathematical model described in Section 19.3.2.
504 The critical value of the analyte concentration xc, according to the ISO definition, is the value
505 obtained by applying the evaluation function F~l to the critical value of the response variable ^
506 Thus, xc = F~l(yc~). In radiochemistry this formula typically involves division by the counting
507 efficiency, test portion size, chemical yield, decay factor, and possibly other factors. In ANSI
508 N42.23, the same value xc is called the decision level concentration., or DLC (ANSI 1996b).
509 According to ISO 11843-1, a detection decision involves the critical value of the response
510 variable, or gross instrument signal, which, in a radioactivity measurement, is typically a total
511 count or count rate. However, it has become standard practice in radioanalysis to use instead the
512 critical value of the net instrument signal, which is calculated from the gross signal by subtract-
513 ing the estimated blank value and any interferences. This practice is consistent with the recom-
514 mendations of IUPAC (1995), where the critical value of the net instrument signal S is denoted
515 by Sc. In principle, either approach should lead to the same detection decision.
516 Since the term "critical value" alone is ambiguous, one should specify the variable to which the
517 term refers. For example, one may discuss the critical (value of the) analyte concentration, the
518 critical (value of the) net count, or the critical (value of the) gross count.
519 Section 19.7.1 and Section 19D.2 of Attachment 19D provide more information on the calcula-
520 tion of critical values.
521 19.4.2 The Minimum Detectable Concentration
522 The minimum detectable concentration is the concentration of analyte that must be present in a
523 laboratory sample to give a specified probability 1 - P of detection. Then p is the probability of
524 failing to reject the null hypothesis when it is false (i.e., a "type II error," or "false negative").
525 The minimum detectable concentration is often abbreviated as MDC. In the ISO terminology the
526 MDC is called the minimum detectable value of the net state variable, denoted by XD, which is
527 defined as the smallest (true) value of the net state variable that gives a specified high probability
528 1 - P that the value of the response variable will exceed its critical value, thus leading one to
529 conclude correctly that the material analyzed is not in the basic state (i.e., the material is not
530 blank). The relationship between the critical value and the minimum detectable value of the net
531 state variable is shown in Figure 19.3.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 19-19 DO NOT CITE OR QUOTE
-------
Measurement Statistics
0 x c x D
FIGURE 19.3 — The critical value xc and minimum detectable value XD
of the net state variable
532 Sections 19.7.2 and 19D.3 provide more information about the calculation of the minimum
533 detectable concentration.
534 When the quantity being measured is the total amount of analyte in an item and not an analyte
535 concentration, the minimum detectable value is sometimes called the minimum detectable
536 amount, which may be abbreviated as MDA. This chapter focuses on the MDC, but with few
537 changes the guidance is also applicable to the MDA.
538 While project planners and laboratories have some flexibility in choosing the significance level a
539 used for detection decisions, the MDC is usually calculated with a = P = 0.05. The use of stan-
540 dard values for a and P allows meaningful comparison of analytical procedures.
541 The MDC concept has generated controversy among radiochemists for years and has frequently
542 been misinterpreted and misapplied. The term must be carefully and precisely defined to prevent
543 confusion. The MDC is by definition the true concentration of analyte required to give a speci-
544 fied high probability that the measured response will be greater than the critical value. Thus, the
545 common practice of comparing a measured concentration to the MDC to make a detection
546 decision is not defensible.
547 There are still disagreements about the proper uses of the MDC concept. Some define the MDC
548 strictly as an estimate of the nominal detection capability of a measurement process. Those in
549 this camp consider it invalid to compute an MDC for each measurement using sample-specific
550 information such as test portion size, chemical yield, and decay factors (e.g., ANSI N42.23). The
551 opposing view is that the "sample-specific" MDC is a useful measure of the detection capability
MARLAP
DO NOT CITE OR QUOTE
19-20
JULY 2001
DRAFT FOR PUBLIC COMMENT
-------
Measurement Statistics
552 of the measurement process, not just in theory, but as it actually performed. The sample-specific
553 MDC may be used, for example, to determine whether an analysis that has failed to detect the
554 analyte of interest should be repeated because it did not have the required or promised detection
555 capability.
556 Neither version of the MDC can legitimately be used as a threshold value for a detection deci-
557 sion. The definition of the MDC presupposes that an appropriate detection threshold (i.e., the
558 critical value) has already been defined.
559 Many experts strongly discourage the reporting of a sample-specific MDC because of its limited
560 usefulness and the likelihood of its misuse. Nevertheless, this practice has become firmly estab-
561 lished at many laboratories and is expected by many users of radioanalytical data. Furthermore,
562 NUREG/CR-4007 states plainly that "the critical (decision) level and detection limit [MDC]
563 really do vary with the nature of the sample" and that "proper assessment of these quantities
564 demands relevant information on each sample, unless the variations among samples (e.g., inter-
565 ference levels) are quite trivial" (NRC 1984).
566 Since a sample-specific MDC is calculated from measured values of input quantities such as the
567 chemical yield, counting efficiency, test portion size, and background level, the MDC estimate
568 has a combined standard uncertainty, which in principle can be obtained by uncertainty propa-
569 gation.
570 In the calculation of a sample-specific MDC, the treatment of any randomly varying but precisely
571 measured quantities, such as the chemical yield, is important and may not be identical at all lab-
572 oratories. The most common approach to this calculation uses the measured value and ignores
573 the variability of the quantity. For example, if the chemical yield routinely varies between 0.85
574 and 0.95, but for a particular analysis the yield happens to be 0.928, the MDC for that analysis
575 would be calculated using the value 0.928 with no consideration of the typical range of yields. A
576 consequence of this approach is that the MDC varies randomly when the measurement is
577 repeated under similar conditions; or, in other words, the sample-specific MDC with this
578 approach is a random variable. The nominal MDC for the measurement process is a constant —
579 not a random variable.
580 If sample-specific MDCs are reported, it must be clear that no measured value should ever be
581 compared to an MDC to make a detection decision. In certain cases it may be valid to compare
582 the sample-specific MDC to a required detection limit to determine whether the laboratory has
583 met contractual or regulatory requirements (remembering to consider the uncertainty of the MDC
584 estimate), and in general it may be informative to both laboratory personnel and data users to
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 19-21 DO NOT CITE OR QUOTE
-------
Measurement Statistics
585 compare sample-specific MDCs to nominal estimates, but other valid uses for the sample-
586 specific MDC are rare.
587 19.4.3 Differences between the ISO and IUPAC Definitions
588 The ISO and IUPAC guidance documents give different definitions for some of the terms listed
589 above and promote somewhat different concepts. In general, the IUPAC approach is to define the
590 "critical value" and "minimum detectable value" separately for the signal and concentration
591 domains. A detection decision may be made in either domain, but the outcome of the decision
592 may depend on which domain is chosen. With the ISO approach the outcome does not depend on
593 the domain. Either domain may be chosen, but in effect all detection decisions are made in the
594 signal domain.
595 The IUPAC and ISO approaches to detection in the signal domain, although expressed differ-
596 ently, are effectively equivalent. (IUPAC bases detection decisions on the net signal S, whereas
597 ISO bases detection decisions on the gross signal 7.) The more important differences are in the
598 concentration domain (X). For example, according to IUPAC, the critical analyte concentration
599 xc is determined from the distribution of the measured concentration X, taking into account its
600 overall measurement uncertainty. According to ISO, xc is simply a function of yc, the critical
601 value of the response variable. Since xc is related toyc in the same way that Xis related to 7, it
602 makes no difference whether detection decisions are based on Xor 7— the outcome is the same.
603 The IUPAC guidance defines the minimum detectable concentration XD as the smallest concentra-
604 tion that gives a specified high probability of obtaining a measured concentration greater than xc,
605 which is inconsistent with the ISO guidance because of the differing definitions ofxc.
606 One consequence of the IUPAC definitions is that the measurement variances of sensitivity fac-
607 tors such as the test portion size, counting efficiency, and chemical yield increase the values of xc
608 and XD because they increase the variance of X. According to the ISO definitions, these variances
609 do not increase the values of xc and XD, although they generate uncertainties in the estimates of xc
610 and XD. In principle, the ISO definitions imply that variability in the true values of these sensitiv-
611 ity factors does increase XD, although the draft implementation guidance in ISO 11843-2 appar-
612 ently does not deal with the issue.
613 As stated above, MARLAP adopts the ISO definitions but also follows the IUPAC guidance
614 where it does not contradict the definitions of ISO 11843-1. The draft implementation guidance
615 in ISO 11843-2 appears not to be designed for typical radioanalytical measurement processes.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 19-22 DRAFT FOR PUBLIC COMMENT
-------
Measurement Statistics
616 19.4.4 Other Detection Terminologies
617 Another term frequently used for a measure of detection capability is the "lower limit of detec-
618 tion," or LLD (Altshuler 1963, EPA 1980, NRC 1984). Unfortunately this term has been used
619 with more than one meaning. In Upgrading Environmental Radiation Data (EPA 1980), the LLD
620 is defined as a measure of the detection capability of an instrument and is expressed as an activ-
621 ity. However, the Nuclear Regulatory Commission defines the LLD to be identical to the MDC
622 when a = P = 0.05 (see, for example, NUREG/CR-4007). It is thus a measure of the detection
623 capability of a measurement process and is expressed as an activity concentration.
624 The term "detection limit" is often used as a synonym for "MDC" or for "minimum detectable
625 value" of any other measured quantity.
626 Many other terms have been used to describe detection capabilities of measurement procedures.
627 Most of them will not be listed here, but one term deserves attention because of the possibility of
628 its confusion with the MDC. The method detection limit, or MDL, is a measure of detection
629 capability used routinely in the context of analyzing samples for chemical contaminants.
630 The term "method detection limit" is defined in the Code of Federal Regulations. In Title 40
631 CFR Part 136, Appendix B, the following definition appears:
632 The method detection limit (MDL) is defined as the minimum concentration of a
633 substance that can be measured and reported with 99% confidence that the analyte
634 concentration is greater than zero and is determined from analysis of a sample in a
635 given matrix containing the analyte.
636 The definition is later clarified somewhat by a statement that the MDL "is used to judge the sig-
637 nificance of a single measurement of a future sample." Thus, the MDL serves as a critical value;
638 however, it is also used as a measure of detection capability, like an MDC. Note that, in
639 MARLAP's usage, the "method detection limit" is not truly a detection limit.
640 The similarity between the abbreviations MDC and MDL tends to produce confusion. The term
641 "method detection limit" is seldom used in the context of radioanalysis except when the analyt-
642 ical method is one that is commonly used to measure stable elements (e.g., ICP/MS methods), or
643 when the term is misused by those who are more familiar with the terminology of hazardous
644 chemical analysis. The confusion is made worse by the fact that "MDL" is sometimes interpreted
645 by radiochemists as an abbreviation for nonstandard terms such as "minimum detectable level"
646 and "minimum detectable limit," the use of which MARLAP strongly discourages.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 19-23 DO NOT CITE OR QUOTE
-------
Measurement Statistics
647 19.4.5 The Minimum Quantifiable Concentration
648 The minimum quantifiable concentration., or the minimum quantifiable value of the analyte con-
649 centration, is defined as the concentration of analyte in a laboratory sample at which the measure-
650 ment process gives results with a specified relative standard deviation.3 A relative standard devi-
651 ation of 10% is usually specified, although other values are possible (see for example MARLAP
652 Appendix C). Since ISO 11843 addresses detection capability but not quantification capability,
653 MARLAP follows IUPAC guidance in defining "minimum quantifiable value" (IUPAC 1995).
654 IUPAC defines both the minimum quantifiable instrument signal and the minimum quantifiable
655 concentration, although MARLAP considers only the latter. In this document the minimum quan-
656 tifiable concentration will be abbreviated as MQC and denoted in equations by XQ.
657 The term "quantification limit" may be used as a synonym for "minimum quantifiable concentra-
658 tion" or for "minimum quantifiable value" of any other measured quantity.
659 Section 19.7.3 provides more information about the calculation of the minimum quantifiable
660 concentration.
661 Historically much attention has been given to the detection capabilities of radioanalytical meas-
662 urement processes, but less attention has been given to quantification capabilities, although for
663 some analytical projects, quantification capability may be a more relevant issue. For example,
664 suppose the purpose of a project is to determine whether the 226Ra concentration in soil from a
665 site is below an action level. Since 226Ra occurs naturally in almost any type of soil, the analyte
666 may be assumed to be present in every sample, making detection decisions irrelevant. The MDC
667 of the measurement process obviously should be less than the action level, but a more important
668 question is whether the MQC is less than the action level (see also Chapter 3 and Appendix C).
3 The MQC is defined in terms of the relative standard deviation of the estimator — not the relative standard
uncertainty of the measured result. The standard uncertainty is generally an estimate of the standard deviation.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 19-24 DRAFT FOR PUBLIC COMMENT
-------
Measurement Statistics
669 19.4.6 Recommendations
670
671
672
673
674
MARLAP makes the following recommendation.
• A measurement result should not be compared to the minimum detectable concentra-
tion to make an analyte detection decision. A detection decision may be made by
comparing the gross signal, net signal, or measured analyte concentration to its
corresponding critical value.
675 19.4.7 Summary of Terms
676 basic state: in radiochemistry, the chemical composition of blank material.
677 critical level: See critical value.
678 critical value: in the context of analyte detection, the minimum value of the response variable
679 (or the measured analyte concentration) required to give confidence that a positive amount of
680 analyte is present in the material analyzed.
681 decision level: See critical value.
682 detection limit: See minimum detectable value.
683 false negative: See type I decision error. This chapter avoids the terms "false negative" and
684 "false positive," because they may be confusing in some contexts.
685 false positive: See type II decision error.
686 lower limit of detection (LLD): (1) "the smallest concentration of radioactive material in a
687 sample that will yield a net count, above the measurement process (MP) blank, that will be
688 detected with at least 95% probability with no greater than a 5% probability of falsely concluding
689 that a blank observation represents a 'real' signal" (NRC 1984); (2) "an estimated detection limit
690 that is related to the characteristics of the counting instrument" (EPA 1980).
691 method detection limit (MDL): "the minimum concentration of a substance that can be meas-
692 ured and reported with 99% confidence that the analyte concentration is greater than zero ...
693 determined from analysis of a sample in a given matrix containing the analyte" (40 CFR 136,
694 Appendix B).
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 19-25 DO NOT CITE OR QUOTE
-------
Measurement Statistics
695 minimum detectable amount (MDA): the minimum detectable value of the total amount of
696 analyte in the sample being analyzed.
697 minimum detectable concentration (MDC): the minimum detectable value of the analyte con-
698 centration in a laboratory sample.
699 minimum detectable value: the smallest value of the net state variable (amount or concentration
700 of analyte) that ensures a specified high probability 1 - P of detection.
701 minimum quantifiable concentration (MQC): the minimum quantifiable value of the analyte
702 concentration in a laboratory sample.
703 minimum quantifiable value: the smallest value of the net state variable (analyte amount or
704 concentration) that ensures the relative standard deviation of the measurement is not greater than
705 a specified value, usually 10%.
706 net state variable (X): the difference between the state variable Z and its value in the basic state
707 —in radiochemistry, usually equal to Z, because the value of Z in the basic state is zero.
708 quantification limit: See minimum quantifiable value.
709 response variable (Y): the variable that gives the observable result of a measurement—in radio-
710 chemistry, typically a gross count or count rate.
711 significance level (a): in a hypothesis test, the probability of a type I decision error.
712 state variable (Z): the quantity that describes the state of the material analyzed—in radiochem-
713 istry, usually the analyte activity concentration.
714 type I decision error: in a hypothesis test, the error made by rejecting the null hypothesis when
715 it is true (a "false positive").
716 type II decision error: in a hypothesis test, the error made by failing to reject the null hypothesis
717 when it is false (a "false negative").
MARLAP JULY 2001
DO NOT CITE OR QUOTE 19-26 DRAFT FOR PUBLIC COMMENT
-------
Measurement Statistics
718 19.5 Procedures for Estimating Uncertainty
719 The steps for evaluating and reporting the uncertainty of a radioactivity measurement may be
720 summarized as follows (adapted from Chapter 8 of the GUM):
721 1. Identify the measurand Y and all the input quantities Xt for the mathematical model.
722 Include all quantities whose variability or uncertainty could have a potentially significant
723 effect on the result. Express the mathematical relationship Y =f(Xl,X2,... ,XN) between the
724 measurand and the input quantities.
725 2. Determine an estimate xt of the value of each input quantity Xt (an "input estimate," as
726 defined in Sections 19.3.2 and 19.3.9).
727 3. Evaluate the standard uncertainty M(X;) for each input estimate xt, using either a Type A or
728 Type B method of evaluation (see Section 19.5.2).
729 4. Evaluate the covariances u(xt,x^) for all pairs of input estimates with potentially
730 significant correlations.
731 5. Calculate the estimate^ of the measurand from the relationship y =f(xl,x2,...,%), where/
732 is the function determined in Step 1.
733 6. Determine the combined standard uncertainty uc(y) of the estimate^ (see Section 19.5.3).
734 7. Multiply uc(y) by a coverage factor k to obtain the expanded uncertainty U such that the
735 interval \y - U,y+ U] can be expected to contain the value of the measurand with a
736 specified probability (see Section 19.3.6 and Attachment 19C).
737 8. Report the result as_y ± C/with the unit of measure, and, at a minimum, state the coverage
738 factor used to compute C/and the estimated coverage probability.
739 19.5.1 Identifying Sources of Uncertainty
740 The procedure for assessing the uncertainty of a measurement begins with listing all conceivable
741 sources of uncertainty in the measurement process. Even if a mathematical model has been iden-
742 tified, further thought may lead to the inclusion of more quantities in the model. Some sources of
743 uncertainty will be more significant than others, but all should be listed.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 19-27 DO NOT CITE OR QUOTE
-------
Measurement Statistics
744 After all conceivable sources of uncertainty are listed, they should be categorized as either poten-
745 tially significant or negligible. Each uncertainty that is potentially significant should be evaluated
746 quantitatively. In particular, counting uncertainty, pipetting and weighing uncertainties, and
747 uncertainties in standard concentrations should always be evaluated. Other possible causes of
748 uncertainty include source geometry and placement, variable instrument backgrounds and effi-
749 ciencies, time measurements used in decay and ingrowth calculations, instrument dead-time
750 corrections, approximation errors in simplified mathematical models, impurities in reagents, and
751 uncertainties in the published values for half-lives and radiation emission probabilities.
752 19.5.2 Evaluation of Standard Uncertainties
753 Calculating the combined standard uncertainty of an output estimate y =f(xl,x2, . . . ,%) requires
754 the estimation of the standard uncertainty of each input estimate xt. As stated earlier, methods for
755 evaluating standard uncertainties are classified as either "Type A" or "Type B." A Type A eval-
756 uation of an uncertainty uses a series of measurements to estimate the standard deviation empiri-
757 cally. Any other method of evaluating an uncertainly is a Type B method.
758 19.5.2.1 Type A Evaluations
759 Suppose Xt is an input quantity in the mathematical model. If a series of n independent observa-
760 tions of Xt are made under the same measurement conditions, yielding the results X. VX. 2, ...,X. n,
761 the appropriate value for the input estimate xt is the arithmetic mean, or average, X., defined as
n k=i
762 The experimental variance of the observed values is defined as
(19.2)
n - k=i
763 and the experimental standard deviation, s(Xt^, is the square root of s\Xih). The experimental
764 standard deviation of the mean, s(Xt), is obtained by dividing s(Xt k) by Jn .
(19.3)
MARLAP JULY 2001
DO NOT CITE OR QUOTE 1 9-28 DRAFT FOR PUBLIC COMMENT
-------
Measurement Statistics
765 The experimental standard deviation of the mean is also commonly called the "standard error of
766 the mean."
767 The Type A standard uncertainty of the input estimate xt = X. is defined to be the experimental
768 standard deviation of the mean. Combining the preceding formulas gives the following equation
769 for the standard uncertainty of xt:
u(x) =
\
n(n -
(19.4)
770 When the input estimate xi and standard uncertainty u(x^) are evaluated as described above, the
771 number of degrees of freedom for the evaluation is equal to n - 1, or one less than the number of
772 independent measurements of the quantity Xt. In general, the number of degrees of freedom for a
773 statistical determination of a set of quantities equals the number of independent observations
774 minus the number of quantities estimated. The number of degrees of freedom for each evaluation
775 of standard uncertainty is needed to implement the procedure for calculating coverage factors
776 described in Attachment 19C.
777 In some cases there may be accumulated data for a measurement system, such as a balance or
778 pipet, which can be used in a Type A evaluation of uncertainty for future measurements,
779 assuming the measurement process remains in control. In fact, the use of recent historical data is
780 advisable in such cases, because it enlarges the pool of data available for uncertainty evaluation
781 and increases the number of degrees of freedom. This type of uncertainty evaluation can be
782 linked closely to the measurement system's routine quality control.
783
784
785
786
787
EXAMPLE: Ten independent measurements of a quantity^ are made, yielding the values
12.132 12.139 12.128 12.133 12.132
12.135 12.130 12.129 12.134 12.136
The estimated value xt is the arithmetic mean of the values Xik.
c = x. = - £ X., = 12L328 = 12.1328
' ' n£i a 10
JULY 2001
DRAFT FOR PUBLIC COMMENT
19-29
MARLAP
DO NOT CITE OR QUOTE
-------
Measurement Statistics
788
789
The standard uncertainty of xt is
u(x) = s(X) =
10
10(10-1) k=
L12888x 1Q-6 =0.0011
790
791
792
793
If Xt and Xj are two input quantities and estimates of their values are correlated, a Type A evalua-
tion of covariance may be performed by making n independent pairs of simultaneous observa-
tions of Xt andXj and calculating the experimental covariance of the means. If the observed pairs
are (X. VX. j), (X. 2,X. 2),..., (X. n,X. n), the experimental covariance of the values is
E
k ~ X)
(19.5)
794
and the experimental covariance of the means X. and X. is
__ s(X ,,X ,)
^ >-* ^
n
(19.6)
795 So, the Type A covariance of the input estimates xt = X. and Xj = X is
(19.7)
796 An evaluation of variances and covariances of parameters determined by the method of least
797 squares may also be a Type A evaluation (see Attachment 19B).
MARLAP
DO NOT CITE OR QUOTE
19-30
JULY 2001
DRAFT FOR PUBLIC COMMENT
-------
Measurement Statistics
798
799
800
801
802
803
804
805
806
807
808
809
810
19.5.2.2 Type B Evaluations
There are many ways to perform Type B evaluations of standard uncertainty. This section
describes some common Type B evaluations but is not meant to be exhaustive.
One example of a Type B method already given is the estimation of counting uncertainty using
the square root of the observed counts. If the observed count is «, when the Poisson counting
model is used, the standard uncertainty of n may be evaluated as u(n) = \Jn. When n may be very
small or even zero, MARLAP recommends the use of the equation u(n) =
-------
Measurement Statistics
821 If the value of x is believed to lie between a lower bound a_ and an upper bound a+, but values
822 near these two bounds are considered less likely than those near the midpoint, then a symmetric
823 trapezoidal distribution may be used to obtain the input estimate and its standard uncertainty (see
824 Section 19A.7 in Attachment 19A). If the ratio of the width of the trapezoid at its top to the width
825 at its base is P, where 0 < P < 1, then the input estimate is the midpoint x = (a_ + a+) 1 2, and its
826 standard uncertainty is
/ \
u(x) =
A
1 +P2
(19.9)
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
As P approaches zero, the trapezoidal distribution becomes triangular. As P approaches one, the
trapezoidal distribution becomes rectangular.
EXAMPLE: Extreme bounds for a quantity Xare estimated to be 34.3 and 34.5, with values
between 34.35 and 34.45 considered most likely. Using the trapezoidal distribution with
a_ = 34.3, a+ = 34.5, and P = (34.45 - 34.35) / (34.5 - 34.3) = 0.5, one obtains the input esti-
mate x = 34.4 and the standard uncertainty u(x) =
34.5 -34.3
"\
1 +0.52
= 0.046.
When the estimate of an input quantity is taken from an external source, such as a book or a
calibration certificate, which states the uncertainty as a multiple of the standard deviation s, the
standard uncertainty is obtained by dividing the stated uncertainty by the stated multiplier of s.
EXAMPLE: The uncertainty for a measured concentration x is stated to be 0.015 Bq g"1 and the
stated multiplier is 2. So, the standard uncertainty of x is u(x) = 0.015 12 = 0.0075 Bq g"1.
If the estimate is provided by a source which gives a bound c for the error such that the interval
from x - c to x + c contains the true value with 100y% confidence (0 < y < 1) but no other infor-
mation about the distribution is given, the measured result may be assumed to have a normal
distribution, and the standard uncertainty may therefore be evaluated as
u(x) =
(19.10)
MARLAP
DO NOT CITE OR QUOTE
19-32
JULY 2001
DRAFT FOR PUBLIC COMMENT
-------
Measurement Statistics
842 The value of z(1+y)/2 may be found in a table of quantiles of the standard normal distribution (see
843 Table G. 1 in Appendix G).
844
845
846
EXAMPLE: The activity concentration x of a commercial standard solution is stated to lie
within the interval 4530 ± 64 Bq g"1 with 95% confidence. The standard uncertainty may
therefore be evaluated as u(x) = 64 / z0975 = 64 /1.96 = 33 Bq g"1.
847 19.5.3 Combined Standard Uncertainty
848 Consider the mathematical model Y =f(Xl,X2, . . . ,XN). Ifx^ x2, . . . , % are measured values of the
849 input quantities Xt and_y =fixl,x2, ...,%) is the calculated value of the measurand 7, the variance
850 of y is estimated using the following formula.
N-l N 2 2
y; JLJLM(X.JX.)
-J / -J -, -, \ i> ;/
,-=i;=,-+i a*,, a*, (1911)
The Uncertainty Propagation Formula
851 Here w2(x,) denotes the estimated variance of xt, or the square of its standard uncertainty; u(xt,x^)
852 denotes the estimated covariance of xt and x,.; 3y / 3x; (or a// ax,) denotes the partial derivative of
853 7 with respect to Xt evaluated at the measured values xl3 x2, . . . , xw; and wc(y) denotes the com-
854 bined standard uncertainty of y. The partial derivatives dy I dxi are called sensitivity coefficients.
855 The preceding formula, called the "law of propagation of uncertainty" in the GUM, will be called
856 the "uncertainty propagation formula" in this document.
857 If the input estimates xl3 x2, ...,XN are uncorrelated, the uncertainty propagation formula reduces
858 to
N
859 Equation 19.12 is only valid when the input estimates are uncorrelated. Although this case occurs
860 frequently in practice, there are notable exceptions. When input estimates are obtained using the
861 same measuring devices or the same standard solutions, or when they are calculated from the
862 same data, there is a potential for correlation. For example, instrument calibration parameters
863 determined by least-squares analysis may be strongly correlated. Fortunately, the method of least
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 19-33 DO NOT CITE OR QUOTE
-------
Measurement Statistics
TABLE 19.1 — Applications of the uncertainty propagation formula
SUMS AND
DIFFERENCES
If a and b are constants, then
uc(ax±by)=a2u2(x) + b2u2(y)±2ab • u(x,y)
PRODUCTS
If x and>- are measured values, then
uc (xy) = u 2(x)y 2 +x2u 2(y) + 2xy • u(x,y)
When x andy are nonzero, the formula may be rewritten as
2, . 2 i( u2(x) u2(y) 2u(x,y)\
uc(xy) = x y \ —^- + —^ + —±^LL|
QUOTIENTS
If x andy are measured values, then
„*(*}=«&
2x• u(x,y)
y) y~ y^ y
When x is nonzero, the variance formula may be rewritten as
y x
EXPONENTIALS
If a is a constant, then
Mc2(em)=«2e2mM2W
If n is a positive integral constant, then
uc(x") = n2x2n~2u2(x)
If Dispositive, then
y2u2(x) ,, ,2 2/ x 2y(\nx)u(x.y)'
±i v > -i- /^In-v-^z^ (v) +
uc(x
LOGARITHMS
If a is a constant and ax is positive, then
1n , U2(X) , 2,, - U2(X) U2(X)
uc (in ax) = —Y. and uc (Iog10 ax) = ^— » ±L-
x2 (InlO)2*2 5.302 • x'
864 squares provides covariance estimates with almost no additional effort (see Attachment 19B). In
865 general, ignoring correlations between the input estimates may lead to overestimation or under-
866 estimation of the combined standard uncertainty.
867 Table 19.1 shows how to propagate uncertainties in some common cases.
868 The product of | dy I dxt and the standard uncertainty u(x^) is called the component of the
869 combined standard uncertainty uc(y) generated by the standard uncertainty of xt, and may be
MARLAP
DO NOT CITE OR QUOTE
19-34
JULY 2001
DRAFT FOR PUBLIC COMMENT
-------
Measurement Statistics
870 denoted by u,(y). When all the input estimates are uncorrelated, the combined standard uncer-
871 tainty may be written in terms of its components as follows.
(19.13)
2 = 1
872 Since u2c(y) is the sum of the squares of the components u,(y), the combined standard uncertainty
873 tends to be determined primarily by its largest components.
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
EXAMPLE
Problem: A 6000-s gross alpha measurement is performed on a test source prepared by evap-
orating water on a stainless steel planchet. The measurement produces 120 alpha counts. The
preceding background measurement on the instrument had a duration of 6000 s and produced
42 alpha counts. The estimated alpha counting efficiency is 0.223 with a standard uncertainty
of 0.015. The sample volume analyzed is 0.05000 L, with a standard uncertainty of 0.00019 L.
The alpha emission rate per unit volume is described by the mathematical model
A =
Nslts-NBltB
where
1B
£
V
is the source count (Ns = 120)
is the background count (NB = 42)
is the source count time (ts = 6000)
is the background count time (tB = 6000)
is the counting efficiency (e = 0.223)
is the volume analyzed (V= 0.0500)
What is the output estimated and what is its combined standard uncertainty, uc(A)l
Solution: First compute the output estimate A (alphas per second per liter).
A =
120/6000 - 42/6000
(0.223)(0.05000)
1.17
JULY 2001
DRAFT FOR PUBLIC COMMENT
19-35
MARLAP
DO NOT CITE OR QUOTE
-------
Measurement Statistics
892
893
894
895
896
897
898
899
900
901
902
903
904
Then compute the combined standard uncertainty uc(A). The only uncertainties included in the
model will be those associated with the counts Ns and NB, the efficiency £, and the volume V.
There is no reason to suspect correlations between the measured values; so, the uncertainty
propagation formula becomes
dA\2
an
dN,
The partial derivatives are evaluated as follows:
dA _ 1 dA _ -1 QA
dNn t~eV dNR tBzV de
= 0.0149477
= -0.0149477
= -5.22834
Nslts-NBltB
= -23.3184
The Poisson model is used for the standard uncertainties of the counts Ns and NB. So,
u2(Ns) = NS=120 and u\NB) = NB = 42
Recall from the statement of the problem that w(e) = 0.015 and u(V) = 0.00019. When the
values of all these expressions are substituted into the uncertainty propagation formula, the
combined variance is u2c(A) = 0.0424; so, the combined standard uncertainty is uc(A) =
J0.0424 « 0.21.
905
906
907
It is helpful to remember certain special forms of the uncertainty propagation formula. For
example, if the values xl3 x2, ...,xn and zl3 z2, ..., zm are uncorrelated and nonzero, the combined
standard uncertainty of y =
may be calculated from the formula
- -,,2
(19.14)
f[x x x )
908 As another example, suppose y = —" 2'"" " , where/is some specified function of xl3 x2, ...x
ZlZ2'"Zm
909 all the zt are nonzero, and all the input estimates are uncorrelated. Then
MARLAP
DO NOT CITE OR QUOTE
19-36
JULY 2001
DRAFT FOR PUBLIC COMMENT
-------
Measurement Statistics
910
911
912
913
914
915
916
917
/
n (\*\ — + v +
) u2(zS
-i- ... -i- /i r\ i r\
U°V) 2 2 2 } 2 2 ' ' 2 Vy-LJ>
Z\ Z1 Zm \ Z\ Z1 Zm t
Equation 19.15 is particularly useful in radioanalysis, where flx^ x2,. . .,*„) might be a net count
rate and z^z2- • -zm might be the product of the test portion size, chemical yield, counting effi-
ciency, decay factor, and other sensitivity factors.
EXAMPLE: Consider the preceding gross-alpha example. Equation 19.15 implies the following
equation for the combined variance of A.
1/2(4\ c S S B B+/l2\u(£) + u (/)
c 22 I 2 2 I
OF \ O Y /
u\Ns)lt2s+u\NB)lt2B A2(u2(e) u\V)\
82F2
Then, since u2(Ns) = Ns and u\NB) = NB,
1,2(/t\ S S B B + A 2 \ U
1 82 F2 )
c(} W ( e2 F2 )
918 19.5.4 The Estimated Covariance of Two Output Estimates
919 Measured values obtained from two measurement processes may be correlated if some of the
920 same input estimates are used to calculate output estimates in both models. If the two measured
921 values are to be used as input quantities in a third model, their covariance must be estimated.
922 Suppose the combined set of input quantities in two mathematical models consists of X±, X2,
923 XN. Then the models can be expressed as Y =f(Xl,X2,... ,XN) and Z = g(Xl:,X2,... ,X^, where each
924 of the measurands may actually depend on only a subset of the combined list of input quantities.
925 If the input estimates are xlyx2, ..., % and the output estimates are y =f(xl,x2,..., %) and z =
926 g(xi,*2, • • • >%)>tne covariance of y and z is estimated by
JULY 2001
DRAFT FOR PUBLIC COMMENT
19-37
MARLAP
DO NOT CITE OR QUOTE
-------
Measurement Statistics
N N n n
x _ v-> yX qy cz
u\y->z) ~ / -/ / -/ ~r—r—M(-"-,-3•"•;•) (19.16)
2 = 17 = i ox. dx. J
927 Since z/(y,.y) = u2c(y), the preceding equation may be considered a generalization of the uncertainty
928 propagation formula.4
929 19.5.5 Nonlinear Models
930 19.5.5.1 Uncertainty Propagation
931 The uncertainty propagation formula tends to give better variance estimates when the function/
932 is linear, because the formula is derived from a linear approximation of/(i.e., a first-order Taylor
933 polynomial). Generally, obtaining a reliable estimate of u2c(y) using the uncertainty propagation
934 formula requires (at least) that whenever/is nonlinear in one of the input quantities Xt, the rela-
935 tive uncertainty of the input estimate xt must be small.5 In radiochemistry this rule applies, for
936 example, to the uncertainty of an instrument calibration factor, chemical yield, or test portion
937 size.
938 If all the input estimates xt are uncorrelated and distributed symmetrically about their means, a
939 better approximation of u2c(y) may be made by including higher-order terms in the uncertainty
940 propagation formula, as shown below.
4 The uncertainty propagation formula may also be generalized using the matrix notation of Attachment 19B. If
=f(x), where x and y are column vectors and/is a vector-valued function, then
This formula describes how the variances and covariances of the vector components of y are related to the variances
and covariances of the vector components of x. Whenj has only one component, the formula here is equivalent to
the uncertainty propagation formula.
5 The uncertainty propagation formula also provides finite estimates of variance in cases where, strictly speaking,
the true variance is infinite or undefined. For example, if x has a normal or Poisson distribution, the variance of 1 / x
is undefined, although the formula provides a finite estimate of it. On the other hand, if the relative standard uncer-
tainty of x is small, the combined variance u](l I x) will almost always be consistent with observation, making the
estimate useful in practice.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 19-38 DRAFT FOR PUBLIC COMMENT
-------
Measurement Statistics
N
a*.
(19.17)
941 See also Section 5.1.2 of the GUM.
942
943
944
945
946
947
948
949
950
951
952
953
954
EXAMPLE: Suppose x and_y are independent estimates of input quantities X and 7, respec-
tively. Then the combined variance of the product/* = xy according to the (first-order)
uncertainty propagation formula is
For example, suppose x = 5, with u(x) = 0.5, and_y =10, with u(y) = 1. Then/? = 50, and the
first-order formula gives the combined standard uncertainty
uc(p) =
When higher-order terms are included,
= 7.07
U2(p) =y2u\x) + x2u2(y) + 0 • u\x) + -u\x)u2(y) + -u2(y)u\x) + 0 • z/4(y)
_ ,,2,. 2f
= y u (x) +x u (y) + u (x)u (y)
With numbers,
uc(p) = V/1020.52 + 5212 + 0.5212 = 7.09
The combined variance of the quotient q = x I y according to the first-order formula is
2, N u 2(x) , 2 u 2(y)
uc(q)= —r^ + tf —^
JULY 2001
DRAFT FOR PUBLIC COMMENT
19-39
MARLAP
DO NOT CITE OR QUOTE
-------
Measurement Statistics
955
956
957
958
959
960
961
962
Using the same values for x and_y again, q = 0.5 and the first-order formula gives
05^
102
102
= 0.0707
When the higher-order terms are included,
With numbers,
dx
J_
y
dq_ _ _ x
dy y2
d2g _
dxdy
a!i = 0
dx2
82q _ 2x
dy2 y3
83q _ 2
dx3
83q _ _6x
dy3 y4
83q
dxdy2 y3 8ydx:
= 0
11 i
- ^_^_
9
y
y)\y
J_
"2
i r J 2,, 2,, i 4x2
\ +0 \u2(y)u\x)+ \- —
y ) ) (2( y
UA 4, s
-—II11 oo
y
0.52
To2
1+3
+ 0.52— 1 +8— =0.0726
102
102
19.5.5.2 Bias
963 If/is nonlinear, its nonlinearity may also tend to bias the output estimate^. The bias may be esti-
964 mated, if necessary, by the formula
MARLAP
DO NOT CITE OR QUOTE
19-40
JULY 2001
DRAFT FOR PUBLIC COMMENT
-------
Measurement Statistics
Bias(y) « -
u(x x)
J
(19.18)
965 which, in practice, is equivalent to
»=i
E
ax. ax.
(19.19)
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
This bias is usually negligible in comparison to the combined standard uncertainty ufy) if the
relative standard uncertainty of each input estimate is small.
EXAMPLE: If x is an estimate of a positive quantity^7, the bias ofy = 1 /x as an estimate of
1 / Xmay be approximated using Equation 19.19. Since y is a function of only one variable,
the partial derivatives of y are the same as ordinary derivatives. The first derivative is dyldx =
-x'
and the second derivative is cfy/dx2 = 2x~3. So the bias due to nonlinearity can be esti-
mated as Bias(y) ~ (1 /2)(2x~3)w2(x) = w2(x)/x3. The combined variance of^ given by the
uncertainty propagation formula is u2c(y) = (-x ~2)2w2(x) = w2(x)/x4. So, the ratio of the bias to
the combined standard uncertainty can be estimated as (u 2(x) / x3) / (w(x) / x2) = u(x) / x,
which is approximately the same as the relative standard uncertainty of x. Therefore, the size
of the relative standard uncertainty gives an indication of the practical significance of the bias.
EXAMPLE: If x andy are uncorrelated estimates of quantities X and 7, respectively, the bias of
the product z = xy as an estimate of XY is given approximately by
Bias(z)«-
&L
dy2
u2fy)
which equals zero, since 32z/ dx2 = 32z I dy 2 = 0
JULY 2001
DRAFT FOR PUBLIC COMMENT
19-41
MARLAP
DO NOT CITE OR QUOTE
-------
Measurement Statistics
981
982
983
984
985
986
987
988
989
990
991
EXAMPLE: If tis an estimate of the decay time Tfor a radionuclide whose decay constant is X
(assumed to have negligible uncertainty), the bias of the estimated decay factor d= e^J is given
approximately by
and the relative bias is J^u\f) 12. For example, suppose the radionuclide is 228Ac, which has a
half-life of tll2 = 6.15 h, and the decay time has a standard uncertainty of u(f) = 2 h. Then the
decay constant X equals In 2 / 6.15 = 0.112707 h"1. The bias equation above implies that the
relative bias of the decay factor d due to the uncertainty of t is approximately
-(0.112707)2(2)2 = 0.025
or 2.5%. Note that the relative bias of dis small if u2(t) I tl/2 is small.
19.5.5.3 Nominal Values
992 Sometimes an input estimate xt is a nominal value and not the result of a measurement. This may
993 be true for example when an analyst uses a pipet to dispense a predetermined amount of tracer
994 into a sample. In this case the input estimate xt is the predetermined volume. Since xt never
995 varies, its variance is zero, but the volume of liquid dispensed varies each time the measurement
996 is repeated. So, the final result does have a variance component associated with the pipet. If the
997 tracer is used to measure the yield for a chemical separation, the value xt appears as a factor in the
998 denominator of a mathematical expression, but the variable factor in that expression is actually
999 the count rate produced by the tracer, which appears in the numerator. The variance of this count
1000 rate is increased by the variability of the tracer volume. The first-order uncertainty propagation
1001 formula gives the same result for the uncertainty of the yield regardless of whether the nominal
1002 value or the true value is assumed to be variable, but the higher-order formula may not.
1003 When nominal values appear in the calculation, one must also be careful when applying the bias
1004 formula. For example, the quotient x/y may by biased if y is the result of a measurement, but it
1005 is not inherently biased if y is a nominal value.
MARLAP
DO NOT CITE OR QUOTE
19-42
JULY 2001
DRAFT FOR PUBLIC COMMENT
-------
Measurement Statistics
1006
1007
1008
1009
1010
1011
1012
1013
1014
1015
1016
1017
1018
1019
1020
1021
1022
1023
EXAMPLE: Suppose the measurement model is
X =
Y-B
a
where 7 is the gross signal, B is the blank signal, and a is the nominal value for a randomly
varying sensitivity factor^, whose true value is always unknown. Suppose 7 can be written in
the form 7 = xA + b + £7; where x is the true value of the measurand; b is the true blank level;
and s7 denotes the measurement error of 7 If all the measured (and nominal) values are
unbiased (i.e., if E(A) = a, E(B) = b, and E(eY) = 0), then the mean of X is given by
E(X) =
0) - 6
a
a
So, X is an unbiased estimator for x. If one treats a as a random variable, this chapter' s bias-
approximation formula gives the incorrect value Xu2(a) I a2 for the bias of X.
Assume A, B, and s7 are uncorrelated. Then the variance of 7 is the sum of two components
GE and x2o^, which may be estimated by w2(s7) and X2u2(a), respectively, where u\d) is
actually an estimate of the variance of A. The combined variance of X is given by
u\B) _ w2(er) +X2u2(a) + u\B)
- --- - -
a
a
The expression on the right may be obtained from the first-order uncertainty propagation
formula even if one incorrectly treats a as a random variable and A as a constant, so that
u2(Y) = w2(s7). If the higher-order approximation is used, the same expression is obtained only
if one correctly treats a as the constant and A as the random variable.
1024 19.6 Radiation Measurement Uncertainty
1025 19.6.1 Radioactive Decay
1026 Although it is impossible to know when an unstable nucleus will decay, it is possible to calculate
1027 the probability of decay during a specified time interval. The lifetime of the nucleus has an
JULY 2001
DRAFT FOR PUBLIC COMMENT
19-43
MARLAP
DO NOT CITE OR QUOTE
-------
Measurement Statistics
1028 exponential distribution., which is a model for the life of any object whose expected remaining
1029 life does not change with age.
1030 The exponential distribution is described by one parameter X, which measures the expected firac-
1031 tional decay rate. This parameter X is called the decay constant and equals In 2 / tll2, or approx-
1032 imately 0.693 / tl/2, where tll2 is the half-life of the radionuclide (sometimes denoted by Tl/2). The
1033 half-life is the same as the median of the exponential distribution.
1034 The probability that an atom will survive until time t without decaying is equal to e~u. Thus the
1035 probability of survival decreases exponentially with time. Consequently, when a large number of
1036 atoms of the same radionuclide are considered, the expected number of surviving atoms also
1037 decreases exponentially with time, as shown in Figure 19.4.
o.oo
FIGURE 19.4 — Expected fraction of atoms remaining at time t
1038 Since the probability that an atom survives until time t is equal to e~^, it follows that the
1039 probability of decay during this time is 1 - e~^.
1040 19.6.2 Radiation Counting
1041 Undoubtedly the best-known rule of radiation measurement statistics is the fact that the counting
1042 uncertainty for a gross radioactivity measurement can be evaluated as the square root of the
1043 observed counts. The square-root rule is useful, because it permits the estimation of a potentially
1044 significant uncertainty component without replicate measurements. Although the rule is usually
1045 valid as an approximation, for reasons which are discussed below, there are limits to its applica-
1046 bility. It is also important to remember that the counting uncertainty is only one component of the
1047 total measurement uncertainty.
MARLAP
DO NOT CITE OR QUOTE
19-44
JULY 2001
DRAFT FOR PUBLIC COMMENT
-------
Measurement Statistics
1048 When a source containing a radionuclide is placed in a detector, the probability that a particular
1049 atom of the radionuclide will produce a count is the product of three factors: the probability of
1050 decay (nuclear transformation), the probability of emission of the radiation being measured, and
1051 the probability of detection. According to the exponential decay model, the probability of decay
1052 is equal to 1 - e~^, where X is the decay constant and t is the counting time. The probability of
1053 radiation emission, denoted here by F, is a characteristic of the radionuclide. The probability of
1054 detection is the same as the counting efficiency £. Then the probability that an atom will generate
1055 a count isp = (1 - e~^)Fz.
1056 If the source initially contains n atoms of the radionuclide, the instrument is stable, and its back-
1057 ground is negligible, the number of observed counts N has a binomial distribution withparame-
1058 ters n and p. In general, if an experiment has only two possible outcomes, which may be called
1059 "success" and "failure," and the probability of success isp, then the number of successes
1060 observed when the experiment is repeated in n independent trials has a binomial distribution with
1061 parameters n and p.
1062 Actually the probability/? is a random variable, because the counting efficiency for an instrument
1063 and source can vary for a number of reasons, such as source placement, dead time, and other
1064 instrument characteristics. These variations generate measurement uncertainty, but their effects
1065 are not included in the "counting uncertainty." The counting uncertainty is the standard deviation
1066 of the theoretical distribution of counts observed in a fixed time period when the efficiency is
1067 held constant. Thus, the actual variability observed in repeated measurements of a single radio-
1068 active source may be greater than the theoretical counting uncertainty.
1069 The mean and variance of the binomial distribution are np and np(\ - p), respectively. In radia-
1070 tion counting, the value of p is usually small enough that the factor 1 - p in the variance can be
1071 ignored. When this is true, the binomial distribution can be approximated by a Poisson distri-
1072 button with mean |i = np. The variance of a Poisson distribution equals the mean; so, both can be
1073 estimated by the same measured result N, and the standard deviation can be estimated by
6 In the rare cases when the Poisson counting model is inadequate and the binomial model is required, if the
instrument background level is negligible, the standard deviation of the source count Ns can be estimated by
J(l -p)Ns. If a Poisson background is measured for time tB and NB counts are observed, the standard deviation of
Ns should be estimated instead by
These two expressions are appropriate only when the source counts are generated by a single radionuclide or by one
radionuclide plus the instrument background.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 19-45 DO NOT CITE OR QUOTE
-------
Measurement Statistics
1074 When |i is large, \[N is an excellent estimator for the standard deviation, but the estimate may be
1075 poor when |i is small. For example, if |i = 100, the coefficient of variation of \fN is only about
1076 5% and its bias is negligible. If |i = 10, the coefficient of variation is more than 16% and there is
1077 a negative bias of more than 1%. If ji = 1, the coefficient of variation is more than 63% and the
1078 negative bias is more than 22%. Furthermore, when |i is small, it is possible to observe zero
1079 counts, so that ^N = 0. MARLAP recommends that JN be replaced by ^N + 1 when extremely
1080 low counts are possible (see also Attachment 19C).7
1081 A sum of independent Poisson quantities also has a Poisson distribution. So, when the Poisson
1082 approximation is valid for all the sources of counts in a counting measurement, the total count
1083 obeys Poisson counting statistics as well.
1084 If a short-lived radionuclide (large X) is counted in a high-efficiency detector (large e), the prob-
1085 ability/* that an atom placed in the detector will produce a count may be so large that the Poisson
1086 approximation is invalid. In this case the Poisson approximation overestimates the counting
1087 uncertainty, but it is important to consider that the statistical model described thus far represents
1088 only the process of counting. In most cases previous steps in the measurement process decrease
1089 the probability that one of the atoms of interest initially present in the test portion will produce a
1090 count. If a correction for decay before counting is performed, the decay factor must be included
1091 in p. If the measured activity of a (single) decay product is used to estimate the activity of a
1092 parent, p must include both ingrowth and decay factors. If a chemical extraction is performed, the
1093 recovery factor must be considered. When these factors are included, the Poisson counting model
1094 is usually valid. Note, however, that these factors must be measured and their standard uncertain-
1095 ties evaluated and propagated, increasing the total measurement uncertainty even further.8
1096 Both the binomial and Poisson models may be invalid if one atom can produce more than one
1097 count during the measurement. This situation occurs when the activity of a parent is estimated
1098 from the total count produced by a series of short-lived progeny (Lucas and Woodward 1964,
1099 Colle and Kishorel997). For example, when 222Rn is measured by counting the emissions of its
7 The negative bias of fif is largely eliminated if one replaces it by JN + 0.25. MARLAP recommends the
estimator JN + 1 although it is positively biased.
8 It is possible to evaluate the uncertainties associated with the decay and ingrowth of a small number of short-
lived atoms before counting using the binomial model, but under the stated conditions, the assumption of Poisson
counting statistics simplifies the calculation. A more complete evaluation of uncertainty may be necessary if the
same source is counted more than once.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 19-46 DRAFT FOR PUBLIC COMMENT
-------
Measurement Statistics
1100 progeny, an atom of 222Rn may produce several counts as it decays through the short-lived series
1101 218Po, 214Pb, 214Bi, and 214Po, to the longer-lived 210Pb.
1102 Both counting models may also be invalid if the total dead time of the measurement is significant
1103 (see Section 19.6.3.1).
1104 Instrument background measurements are usually assumed to follow the Poisson model. This
1105 assumption is reasonable if the background counts are produced by low levels of relatively long-
1106 lived radionuclides. However, the true background may vary between measurements (e.g.,
1107 cosmic background). Furthermore, the measured background may include spurious instrument-
1108 generated counts, which do not follow a Poisson distribution. Generally, the variance of the
1109 observed background is somewhat greater than the Poisson counting variance, although it may be
1110 less for certain types of instruments, such as those that use parallel coincidence counters to com-
1111 pensate for background instability (Currie et al. 1998). Departures from the Poisson model may
1112 be detected using the chi-square test described in Section 18B.2 of Attachment 18B; however,
1113 deviations from the model over short time periods may be small and difficult to measure.
1114 19.6.3 Count Rate
1115 Suppose a radiation counting measurement of duration t is made for the purpose of estimating a
1116 mean count rate R, assumed to be constant, and the result of the measurement N (in counts) has a
1117 distribution that is approximately Poisson with mean Rt. If tis known precisely, the best estimate
1118 of R given a single observation N = n is the measured count rate r = nl t, and the best estimate of
1119 the variance of the measured rate is u\r) = n 112 = r 11. Under the Poisson assumption, even if
1120 repeated measurements are made, the best estimates of r and its variance are obtained by pooling
1121 the counts and count times and using the same formulas.
1122 In fact the count time t is known imperfectly; so, a more complete estimate of the variance of r is
it \ n n2 if \
u(r} = - + -u(t} (19-2°)
1123 The uncertainty of t may be ignored if u(t) / t<^l / Jn, that is, if the relative standard uncertainly
1124 of t is much less than 1 over the square root of the count.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 19-47 DO NOT CITE OR QUOTE
-------
Measurement Statistics
1125
1126
1127
1128
1 1 ?Q
1130
EXAMPLE: A source is counted for t = 100 s, where t has standard uncertainty u(f)
n = 96l
rate r is
is
counts observed. When u(f) is ignored, the combined standard uncertainty
uc(r) =
-------
Measurement Statistics
1145 behavior of an actual counting system tends to fall between the two extremes. At low count rates,
1146 however, both models give essentially the same predictions.
1147 At low count rates the observed count rate nit may be corrected for dead time by dividing by the
1148 factor \-mlt. Many counting instruments perform the correction automatically by extending
1149 the real time t of the measurement to achieve a desired live time tL. Since tL = t- m, the correct-
1150 ed count rate is simply n I tL. When the dead time rate for the measurement is low, the variance
1151 of the corrected count rate may be estimated as n I tL . Thus, the Poisson model remains adequate
1152 if the "count time" is equated with the live time. When the dead time rate is high (above 20%),
1153 the same estimate may not be adequate (NCRP 1985). In this case the measurement should be
1154 repeated, if possible, in a manner that reduces the dead time rate.
1155 Dead time effects may be evaluated experimentally to confirm that they do not invalidate the
1156 Poisson model at the count rates expected for typical measurements. The chi-square test
1157 described in Section 18B.2 of Attachment 18B can be used for this purpose.
1158 19.6.3.2 A Confidence Interval for the Count Rate
1159 When the Poisson counting model is valid, lower and upper confidence limits for the mean count
1160 rate R given an observation of n counts in time t may be calculated as follows:10
1161 Here y is the desired confidence coefficient, or the minimum probability of coverage, and j^(ti)
1162 denotes the/>-quantile of the chi-square distribution with n degrees of freedom (see Table G.3 in
1163 Appendix G). If n = 0, the chi-square distribution -£(n) is degenerate. For our purposes
1164 should be considered to be 0.
10 The chi-square distribution is a special case of a gamma distribution, whose relationship to the Poisson distribu-
tion is described by Hoel et al. (1971) and Stapleton (1995). This relationship is the basis for the two formulas in
Equation 19.21. The relationship is such that if X is chi-square with In degrees of freedom and 7 is Poisson with
mean u, then Pr[Jf < 2u] = Pr[7 > n].
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 19-49 DO NOT CITE OR QUOTE
-------
Measurement Statistics
1165
1166
1167
1168
1169
1170
1171
1172
1173
1174
1175
1176
1177
1178
1179
1180
1181
1182
1183
1184
1185
1186
EXAMPLE: Suppose 10 counts are observed during a 600-second instrument background
measurement. Then the 95% confidence limits for the background count rate are
R,
lower
R
Xo.Q25(2°) _ 9.59078
(2)(600) 1200
fa.97s(22) _ 36.7807
(2)(600) " 1200
= 0.00799 cps
= 0.03065 cps
EXAMPLE: Suppose 0 counts are observed during a 600-second measurement. Then the 95%
confidence limits for the count rate are
P
R,
,
lower
X0.02S(°)
-
(2)(600)
„
R
n
= 0 Cps
7.3778
upper
(2)(600) 1200
nnnA1-
= 0.00615 cps
19.6.4 Instrument Background
As noted above, single-channel background measurements are usually assumed to follow the
Poisson model, although there may be effects which increase the variance beyond what the model
predicts. For example, the cosmic radiation and other natural sources of instrument background
may vary between measurements, the composition of source holders and containers may vary, the
instrument may become contaminated by sources, or the instrument may be unstable. For certain
types of instruments, the Poisson model may overestimate the background variance (Currie et al.
1998). If the background does not closely follow the Poisson model, its variance should be esti-
mated by repeated measurements.
The "instrument background," or "instrument blank," is usually measured with source holders or
containers in place, since the presence of the container may affect the count rate. In many cases,
perhaps most, it is not feasible to use the same container during both the background and test
source measurements, but nearly identical containers should be used. Variations in container
composition may affect the background count rate. If test sources contain enough mass to atten-
uate background radiation, then it is best to use a similar amount of blank material during the
background measurement.
MARLAP
DO NOT CITE OR QUOTE
19-50
JULY 2001
DRAFT FOR PUBLIC COMMENT
-------
Measurement Statistics
1187 If repeated measurements demonstrate that the background level is stable, then the average x of
1188 the results of many similar measurements performed over a period of time may give the best
1189 estimate of the background. In this case, if all measurements have the same duration, the experi-
1190 mental standard deviation of the mean s(x) is also a good estimate of the measurement uncer-
1191 tainty. Given the Poisson assumption, the best estimate of the uncertainty is still the Poisson
1192 estimate, which equals the square root of the summed counts, divided by the number of measure-
1193 ments, but the experimental standard deviation may be used when the Poisson assumption is
1194 false.
1195 If the background drifts or varies nonrandomly over time (i.e., is nonstationary), it is important to
1196 minimize the consequences of the drift by performing frequent blank measurements.
1197 If the background variance includes a small non-Poisson component, that component can be esti-
1198 mated from historical background data and added to the calculated Poisson component. A chi-
1199 square statistic may be used to detect and quantify non-Poisson background variance (Currie
1200 1972; see also Section 18B.3 of Attachment 18B), but chi-square provides an unbiased estimate
1201 of the additional variance only if the background remains stationary while the data are being
1202 collected. If the observed background counts, in order, are 7Vl3 N2, ...,Nn and the corresponding
1203 counting intervals are ^, t2, ...,tn, then the quantity
2= 1
^!/J
H-l
(19.22)
1204 may be used to estimate the non-Poisson variance of a net count rate due to background even if
1205 the background is not stationary. The distribution of ^ is not simple, and ^ may even assume
1206 negative values, which are clearly unrealistic. So, if this estimator is used, it should be calculated
1207 for several data sets and for more than one instrument, if possible, to give an indication of its
1208 reliability. Although replicate measurements are involved, this type of evaluation of uncertainty
1209 should be considered a Type B method.
1210 If background and test source measurements are performed under different conditions, the back-
1211 ground measurement may be biased. Such a bias may occur, for example, if test sources are
1212 counted in containers or on planchets which are not present during background measurements. A
1213 situation of this kind should be avoided if possible.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 19-51 DO NOT CITE OR QUOTE
-------
Measurement Statistics
1214 When instrument background levels are low or when count times are short, it is possible that too
1215 few counts will be observed to provide an accurate estimate of the measurement uncertainty.
1216 Attachment 19C describes a method for choosing an appropriate coverage factor when only few
1217 counts are observed.
1218 19.6.5 Counting Efficiency
1219 The counting efficiency for a measurement of radioactivity may depend on many factors, includ-
1220 ing source geometry, placement, composition, density, activity, radiation type and energy, and
1221 other instrument-specific factors. The estimated efficiency is sometimes calculated explicitly as a
1222 function of such variables (in gamma spectrometry, for example). In other cases a single meas-
1223 ured value is used (e.g., alpha spectrometry). If an efficiency function is used, the uncertainties of
1224 the input estimates, including those for both calibration parameters and sample-specific quanti-
1225 ties, must be propagated to obtain the combined standard uncertainly of the estimated efficiency.
1226 Calibration parameters tend to be correlated; so, estimated covariances must also be included. If
1227 a single value is used instead of a function, the standard uncertainty of the value is determined
1228 when the value is measured.
1229
1230
1231
1232
1233
1234
1235
1236
1237
1238
1239
1240
1241
1242
1243
EXAMPLE
Several sources with the same geometry are prepared and used to calibrate a radiation counter.
One blank
count rate,
efficiency.
C
M
n
NB
ts
IB
Ns,,
R,
R
£
measurement is made. Each source is counted once to obtain an estimate of the
the estimates are averaged, and the average is used to calculate the counting
The sources are long-lived and all source count times are equal. Let
= concentration of standard solution (C = 1500, u(C) = 20 Bq g"1)
= mean mass of solution added to each source (0.09980 g, added by a 0. 1-mL
= number of sources (15)
= blank count (90)
= source count time (300 s)
= blank count time (6000 s)
= gross count observed during the measurement of the /'th source
= gross count rate observed in the /th source measurement
= arithmetic mean of the gross count rates, Rt
= estimated counting efficiency
pipet)
MARLAP
DO NOT CITE OR QUOTE
19-52
JULY 2001
DRAFT FOR PUBLIC COMMENT
-------
Measurement Statistics
Then the following equations may be used to calculate the mean efficiency and its standard
uncertainty:
n(n -
£ =
CM
s\R)+NBlt2B +
C2M2
( C2
u\M)\
M2 1
11(8) =
The source-to-source variability of the mass Mis not explicitly evaluated, because it is
included in the observed variability of the count rates, Rt. So, the standard uncertainty u(M)
represents only the uncertainty of the mean mass added by the pipet. This uncertainty arises
from uncertainty in the capacity of the pipet, the density of the solution, temperature effects,
and the analyst's technique. Assume for this example that u(M) is 0.00050 g (about 0.5%).
Note that the uncertainty of the blank count, NB, is negligible in this example and could have
been ignored. It was included only for completeness.
Assume the observed source counts,
15,708
15,924
16,120
15,946
15,844
15,902
15,953
16,020
16,211
are as follows:
16,012
15,877
16,181
16,066
16,061
15,984
Then the observed gross count rates, Rt, are:
52.360
53.080
53.733
53.153
52.813
53.007
53.177
53.400
54.037
53.373
52.923
53.937
53.553
53.537
53.280
JULY 2001
DRAFT FOR PUBLIC COMMENT
19-53
MARLAP
DO NOT CITE OR QUOTE
-------
Measurement Statistics
1262
1263
1264
1265
1266
1267
1268
1269
1270
1271
1272
1273
1274
1275
1276
1277
1278
1279
1280
1281
1282
The average of the gross count rates is calculated as follows.
15
=53.2909
The experimental variance of R is
S2(R) =
15
(R - 53.2909)2 = 0.012876
'
Then the estimated counting efficiency is
53.2909 -90/6000
(1500)(0.09980)
£ =
= 0.355884
and the standard uncertainty of £ is given by
11(8) =
0.012876 + 90 / 60002 + „ ^
(1500)2(0.09980)2
«ei2( 202 ^ O.OOOS2^
( 15002 0.099802J
In fact, the standard uncertainty of £ calculated in the preceding example may be incomplete. The
true counting efficiency may vary from source to source because of variations in geometry, posi-
tion, and other influence quantities not explicitly included in the model. So, the standard uncer-
tainty of £ should include not only the standard uncertainty of the estimated mean, as calculated
in the example, but also a second component of uncertainty due to variations of the true effi-
ciency during subsequent measurements. The second component may be written as £2(p2, where (p
is an estimate of the coefficient of variation of the true efficiency. Then the standard uncertainty
of £ equals the square root of the sum of the squares of the two components.
In the example above, the experimental variance of the count rates, s2(Ri), might be used to esti-
mate (p2. Procedure E2, which is described in Section 18B.2 of Attachment 18B, is a step-by-step
procedure for estimating such "excess" variance in a series of measurements. However, if the
procedure were applied to the series of measurements made in the example, the estimated vari-
ance might be inflated by errors in the pipetting of the standard solution. The resulting estimate
MARLAP
DO NOT CITE OR QUOTE
19-54
JULY 2001
DRAFT FOR PUBLIC COMMENT
-------
Measurement Statistics
1283 would therefore tend to be an upper bound. A lower bound for the excess variance could be esti-
1284 mated by making replicate measurements of only one source, thus eliminating the effects of
1285 pipetting errors but also unfortunately eliminating the effects of variable source geometry. A
1286 better approach is to weigh the amount of standard solution added to each source, use the results,
1287 Mt, to calculate 15 individual estimates of the counting efficiency, £,, and estimate the excess
1288 variance of the values £,,
1289 Variations in counting efficiency due to source placement should be reduced as much as possible
1290 through the use of positioning devices that ensure a source with a given geometry is always
1291 placed in the same location relative to the detector. If such devices are not used, variations in
1292 source position may significantly increase the measurement uncertainty.
1293 Calibrating an instrument under conditions different from the conditions under which test sources
1294 are counted may lead to large uncertainties in the sample activity measurements. Source geome-
1295 try in particular tends to be an important factor for many types of radiation counting instruments.
1296 Generally, calibration sources should be prepared with the sizes and shapes of test sources and
1297 counted in the same positions, although in some cases it may be possible to calculate correction
1298 factors which allow one calibration to be used for different geometries. When correction factors
1299 are used, their uncertainties should be evaluated and propagated.
1300 If the efficiency £ is calculated from a model that includes one of the quantities X{ appearing else-
1301 where in the sample activity model, there is a correlation between the measured values of £
1302 and Xj, which should not be ignored. It is often simpler to include the entire expression for £ in
1303 the expression for the laboratory sample activity before applying the uncertainty propagation
1304 formula.
1305
1306
1307
1308
1309
1310
1311
1312
EXAMPLE: Suppose the counting efficiency for a measurement is modeled by the equation
£ = A exp(-BMs), where A and B are calibration parameters and Ms is the source mass; and
suppose the chemical yield 7 is modeled byMs/Mc, where Mc is the expected mass at 100%
recovery. Then the estimated values of the counting efficiency and the yield are correlated,
because both are calculated from the same measured value of the source mass. When the com-
bined standard uncertainty of the sample activity is calculated, the covariance w(£,F) may be
included in the uncertainty propagation formula, or the variables £ and 7 in the model may be
replaced by the expressions A exp(-BMs) and Ms I Mc, respectively.
1313 In some cases the estimated value of the counting efficiency has no effect on the output estimate
1314 of laboratory sample activity. This happens often in alpha spectrometry, for example, when iso-
1315 topic tracers are used. The efficiency estimate is needed to obtain an estimate of the yield of the
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 19-55 DO NOT CITE OR QUOTE
-------
Measurement Statistics
1316 chemistry procedure, but the efficiency usually cancels out of the mathematical model for the
1317 laboratory sample activity and its uncertainty is not propagated when determining the combined
1318 standard uncertainty of the activity estimate.
1319 19.6.6 Radionuclide Half-life
1320 The component of combined standard uncertainty associated with the half-life of a radionuclide
1321 is often negligible in measurements performed by typical radioanalytical laboratories, since the
1322 half-lives of most radionuclides of interest have been measured very accurately and in many
1323 cases decay times are short relative to the half-life (so that the sensitivity coefficient is small).
1324 However, this uncertainty component is also one of the most easily obtained components, since
1325 radionuclide half-lives and their standard uncertainties are evaluated and published by the
1326 National Nuclear Data Center (NNDC) at Brookhaven National Laboratory. The data may be
1327 obtained from the NNDC website (www.nndc.bnl.doe.gov).
1328 19.6.7 Gamma Spectrometry
1329 There are a number of sources of measurement uncertainty in gamma spectrometry, including:
1330 • Poisson counting uncertainty
1331 • Compton baseline determination
1332 • Background peak subtraction
1333 • Multiplets and interference corrections
1334 • Peak-fitting model errors
1335 • Efficiency calibration model error
1336 • Summing
1337 • Density correction factors
1338 • Dead time
1339 See Chapter 17 for further discussion of measurement models and uncertainty analysis for
1340 gamma spectrometry.
1341 19.6.8 Balances
1342 The uncertainty of a balance measurement tends to be small, even negligible, when the balance is
1343 used properly and the mass being measured is much larger than the balance's readability. How-
1344 ever, the uncertainty may also be difficult to evaluate unless the balance is well maintained and
1345 operated in a controlled environment that protects it from external influences. In particular, drafts
MARLAP JULY 2001
DO NOT CITE OR QUOTE 19-56 DRAFT FOR PUBLIC COMMENT
-------
Measurement Statistics
1346 or sudden changes in pressure, temperature or humidity (e.g., opening doors or dishwashers) may
1347 produce spurious errors.
1348 The uncertainty of the result of a balance measurement generally has components associated with
1349 balance calibration, linearity, repeatability, day-to-day variability due to environmental factors,
1350 and air buoyancy. Other sources of uncertainty may include leveling errors and off-center errors,
1351 which should be controlled. Static electrical charges may also have an effect. For some materials,
1352 gain or loss of mass before or after weighing (e.g., by absorption or evaporation of water) may be
1353 significant. Attachment 19G of this chapter describes several of these uncertainty components in
1354 more detail.
1355 Balance manufacturers provide specifications for repeatability and linearity, which are usually of
1356 the same order of magnitude as the balance's readability, but tests of repeatability and linearity
1357 should also be included in the routine quality control for the balance.
1358 Repeatability is expressed as a standard deviation and is typically assumed to be independent of
1359 the load. It represents the variability of the result of zeroing the balance, loading and centering a
1360 mass on the pan, and reading the final balance indication.
1361 The linearity tolerance of a balance, aL, should be specified by the manufacturer as the maximum
1362 deviation of the balance indication from the value that would be obtained by linear interpolation
1363 between the calibration points. Different methods may be used to convert this tolerance to a
1364 standard uncertainty, depending on the form the linearity error is assumed to take. One method,
1365 which is recommended by the Eurachem/CITAC Guide: Quantifying Uncertainty in Analytical
1366 Measurement, is to treat the tolerance, aL, as the half-width of a rectangular distribution and
1367 divide aL by ^3 to obtain the standard uncertainty (Eurachem 2000). Another method, suggested
1368 in Attachment 19G of this chapter, is to treat aL as the amplitude of a sinusoidal function. This
1369 model requires that aL be divided by ^2 to obtain the standard uncertainty. The latter method is
1370 used below.
1371 Procedures for evaluating the relative standard uncertainties due to calibration and environmental
1372 factors and for calculating the buoyancy correction factor and its standard uncertainty are
1373 described in Attachment 19G.
1374 A typical mass measurement in the laboratory involves separate measurements of a gross mass
1375 and a tare mass. The net mass, m, is determined by subtracting the balance indication for the tare
1376 mass, /Tare, from the indication for the gross mass, TQ^, and multiplying the difference, /Net, by
1377 the buoyancy correction factor, B. That is,
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 19-57 DO NOT CITE OR QUOTE
-------
Measurement Statistics
(19.23)
1378 The standard uncertainty of m is given by
q>) + <* + ^r) + X0) (19.24)
1379 where
1380 m is the buoyancy-corrected net mass
1381 7Net is the net balance indication (/G,.OSS - /Tare)
1382 /Tare is the balance indication for the tare mass
1383 /Q^ is the balance indication for the gross mass
1384 B is the buoyancy correction factor
1385 Attachment 19G describes uncertainty equations for use in other circumstances.
1386 19.6.9 Pipets and Other Volumetric Apparatus
1387 Generally, a pipet or volumetric flask is used not to measure an existing volume of liquid, but to
1388 obtain a volume of a predetermined nominal size. The nominal value is treated as if it were a
1389 measured value, although it is known before the "measurement." The true volume is the variable
1390 quantity. Since a volumetric "measurement" of this type cannot be repeated, pipets and flasks are
1391 good examples of measurement systems for which historical data are important for Type A eval-
1392 uations of standard uncertainty.
1393 The density of a liquid depends on its temperature. For this reason, when a volume is being
1394 measured, one should determine whether the volume of interest is the volume at the current room
1395 temperature, the long-term mean room temperature, or some other temperature, such as 20°C.
1396 One should also determine whether the effect of temperature is significant for the measurement.
1397 Often it is not, but in some cases a correction for thermal expansion may be necessary.
1398 The standard uncertainty for a volumetric measurement includes components associated with the
1399 capacity of the measuring device, temperature effects, repeatability, and the analyst's bias in
1400 using the device (e.g., reading a meniscus).
1401 The capacity of a volumetric pipet or flask (at 20°C) is generally specified with a tolerance a,
1402 which may be assumed to represent the half-width of a triangular distribution (e.g., see ASTM
MARLAP JULY 2001
DO NOT CITE OR QUOTE 19-58 DRAFT FOR PUBLIC COMMENT
-------
Measurement Statistics
1403 1994 and ASTM 1995). Assuming a triangular distribution, one evaluates the uncertainty com-
1404 ponent of the volume associated with the capacity as a I ^6.
1405 The relative standard uncertainty due to temperature variations is typically a Type B standard
1406 uncertainty, which may be derived from a temperature range, T± 57, and the liquid's coefficient
1407 of thermal expansion, P, at the center of the range. Assuming a rectangular distribution for the
1408 temperature with half-width 57, the relative standard uncertainty component due to temperature
1409 variations is 10157Y/3.
1410 The nominal capacity of any volumetric glassware is usually specified at 20°C. If the glassware
1411 is used at a different temperature, the capacity is slightly different. Temperature effects on the
1412 capacity are generally very small (much smaller than the effects on the density of the liquid) and
1413 for this reason one may usually ignore them. The relationship between the capacity and the
1414 temperature is given approximately by
Fr=F20(l+a(r-20)) (19.25)
1415 where
1416 T is the temperature (°C)
1417 VT is the capacity at temperature T
1418 F20 is the capacity at 20°C
1419 a is the glassware's coefficient of thermal cubical expansion ("CT1)
1420 The value of a for ASTM Type I, Class A, borosilicate glassware is approximately 0.00001 °C"1;
1421 so, the capacity increases by only about 0.001% for each degree Celsius of temperature increase.
1422 An analyst may calibrate a pipet gravimetrically using an analytical balance. The balance, to be
1423 useful, must provide better accuracy than the pipet. In particular, the balance's repeatability and
1424 linearity tolerance should be small relative to the tolerances for the pipet. The calibration pro-
1425 vides an estimate of the pipet's capacity, the standard uncertainty of the capacity, and the var-
1426 lability to be expected during use. The procedure involves dispensing a series of n pipet volumes
1427 of a specified liquid into a container and weighing the container and zeroing the balance after
1428 each volume is added. Usually the container must have a small mouth to reduce evaporation. The
1429 temperature of the room, the liquid, and the apparatus involved should be specified, equilibrated,
1430 and controlled during the experiment.
1431 The procedure produces a set of balance indications, /,, which are averaged to obtain the arith-
1432 metic mean /. To obtain the estimated mean pipet volume, v, the mean balance indication, /, is
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 19-59 DO NOT CITE OR QUOTE
-------
Measurement Statistics
1433
1434
1435
1436
1437
1438
1439
1440
multiplied by a factor Z, which equals the quotient of the buoyancy correction factor divided by
the density of the liquid at room temperature. A correction factor for thermal expansion of the
pipet may also be included, if necessary.
v=/Z
where
z =
1 "
PM - PA,M
(19.26)
and where
pM is the density of the liquid
pAM is the density of the air at the time the liquid is weighed
pc is the density of the calibration mass standard for the balance
pA:C is the density of the air at the time of the balance calibration
1441 The calibration is most often performed using water.
1442 ASTM E542, "Standard Practice for Calibration of Laboratory Volumetric Apparatus," provides
1443 additional information about the procedure, including tables of values of Z for various conditions
1444 (ASTM 2000). Table 19.2, which is taken from ASTM E542, shows the density of air-free water
1445 at various temperatures. Attachment 19G of this chapter describes an equation to calculate the
1446 density of air as a function of temperature, pressure, and humidity.
TABLE 19.2 — Density of air-free water
Temperature, °C
15
16
17
18
19
20
21
22
23
24
25
Density, g/cm3
0.999098
0.998941
0.998773
0.998593
0.998403
0.998202
0.997990
0.997768
0.997536
0.997294
0.997043
Temperature, °C
26
27
28
29
30
31
32
33
34
35
Density, g/cm3
0.996782
0.996511
0.996232
0.995943
0.995645
0.995339
0.995024
0.994701
0.994369
0.994030
MARLAP
DO NOT CITE OR QUOTE
19-60
JULY 2001
DRAFT FOR PUBLIC COMMENT
-------
Measurement Statistics
1447 The volume, v, estimated by the calibration may be substituted for the pipet's nominal capacity
1448 when the pipet is used later in an analytical measurement. The uncertainty of v as an estimate of
1449 the mean volume may be calculated as follows.
u(IZ) =
A
z^
n
V(Z)
(19.27)
2 2 p2sr2
9cal + 9Env + -^T-
1450 where (pCal and (pEnv denote the relative standard uncertainties of mass measurements associated
1451 with balance calibration and environmental factors, respectively (see Section 19.6.8). Note that
1452 the uncertainty of the buoyancy correction factor has been ignored here and the standard uncer-
1453 tainty of Z has been equated with the component due to thermal expansion of the liquid, which is
1454 assumed to be dominant. Also note that the correlation between Z and / induced by temperature
1455 effects on both the liquid density and the balance sensitivity is unknown and has been ignored.
1456 The uncertainty of v as a predictor of the true volume that will be dispensed during a subsequent
1457 measurement includes additional components for repeatability and temperature variability.
(19.28)
1458 Note that if a different analyst performs the measurement, there may be an additional uncertainty
1459 component associated with the difference in individual techniques.
1460 If the mean volume is within specified tolerances, a slightly simpler approach is possible. The
1461 pipet's nominal capacity may be used as the volume v and the tolerance a may be used in a Type
1462 B evaluation of standard uncertainty. In this case, the standard uncertainty of v is evaluated as
1463 shown below.
A
2o2?T^2
V P 87
(19.29)
JULY 2001
DRAFT FOR PUBLIC COMMENT
19-61
MARLAP
DO NOT CITE OR QUOTE
-------
Measurement Statistics
1464 The experimental procedure outlined above may also be adapted for other volume measuring
1465 devices, including flasks and graduated cylinders.
1466 The manufacturers of certain types of automatic pipetting devices (e.g., Eppendorf® pipettors)
1467 provide specifications for bias and imprecision. For these devices the manufacturer's specifica-
1468 tions for bias and imprecision may be assumed. In this case the Type B standard uncertainty of a
1469 pipetted volume v is
1483
1484
a2 2 v2B25r2 no-jrvv
— +s +— (19.30)
1470 where a is the manufacturer's stated bias tolerance, assumed to represent the half-width of a tri-
1471 angular distribution, and s is the stated standard deviation. This approach has the advantage of
1472 simplicity; however, since many analysts may not achieve the same accuracy as the manufac-
1473 turer, the standard uncertainty given by Equation 19.30 may be unrealistic.
1474 19.6.10 Digital Displays and Rounding
1475 If a measuring device, such as an analytical balance, has a digital display with resolution 5, the
1476 standard uncertainty of a measured value is at least 5 / 2/3. This uncertainty component exists
1477 even if the instrument is completely stable.
1478 A similar Type B method may be used to evaluate the standard uncertainty due to computer
1479 roundoff error. When a value x is rounded to the nearest multiple of 10", the component of uncer-
1480 tainty generated by roundoff error is 10" / 2/3. When rounding is performed properly and x is
1481 printed with an adequate number of figures, this component of uncertainty should be negligible
1482 in comparison to the total uncertainty of x.
EXAMPLE: The readability of a digital balance is 0.1 mg. Therefore, the minimum standard
uncertainty of a measured mass is 0.1 / 2/3 = 0.029 mg.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 19-62 DRAFT FOR PUBLIC COMMENT
-------
Measurement Statistics
1485
1486
1487
1488
1489
1490
1491
1492
1493
1494
1495
1496
1497
1498
1499
1500
1501
1502
1503
1504
1505
1506
1507
1508
1509
1510
1511
1512
1513
EXAMPLE: A computer printout shows the result x of a measurement as
3.40E+01 +- 9.2E-02
where the expanded uncertainty is calculated using a coverage factor of 2. The measured value
is rounded to the nearest multiple of 0.1. So, the standard uncertainty of x is
«(*) =
^
0.092
0.1
2 ) (
= 0.054.
19.6.11 Subsampling
Appendix F of this manual discusses laboratory subsampling. The subsampling of heterogeneous
materials for laboratory analysis increases the variability of the measurement result and thus adds
a component of measurement uncertainty, which is usually difficult to quantify without replicate
measurements. Appendix F summarizes important aspects of the statistical theory of particulate
sampling and applies the theory to subsampling in the radiation laboratory (see also Gy 1992 and
Pitard 1993). The mathematical estimates obtained using the theory often require unproven
assumptions about the material analyzed and rough estimates of unmeasurable parameters. How-
ever, in some cases the theory can be used to suggest how subsampling errors may be affected by
either changing the subsample size or grinding the material before subsampling. Of course, the
total measurement uncertainty, including components contributed by subsampling, may always
be evaluated by repeated subsampling and analysis.
If subsampling is not repeated, its effects may be represented in the mathematical measurement
model by including an input quantity Fs whose value is the ratio of the analyte concentration of
the subsample to that of the total sample. This ratio, which will be called the subsampling factor
(a MARLAP term), appears in the model as a divisor of the net instrument signal and thus is
similar to the chemical yield, counting efficiency, and other sensitivity factors. The value of Fs is
estimated as 1, but the value has a standard uncertainty which increases the combined standard
uncertainty of the result. (Since its value is always 1, the factor Fs is an example of a "nominal
value," as discussed in Section 19.5.5.) The uncertainty of Fs also increases the MDC and the
MQC.
Although the component of uncertainty caused by the subsampling of heterogeneous solid matter
may be difficult to estimate, it should not be ignored, since it may be relatively large and in some
cases may even dominate all other components. One may use previous experience with similar
JULY 2001
DRAFT FOR PUBLIC COMMENT
19-63
MARLAP
DO NOT CITE OR QUOTE
-------
Measurement Statistics
1514 materials to estimate the uncertainty, possibly with the aid of the information and methods pre-
1515 sented in Appendix F. By default, if "hot particles" are not suspected, and if reasonable precau-
1516 tions are taken to homogenize (mix) the material and to obtain a sufficient number of particles in
1517 an unbiased subsample, one may simply assume a nominal relative standard uncertainty compo-
1518 nent of 5% for solid materials.
1519 19.6.12 The Standard Uncertainty for a Hypothetical Measurement
1520 MARLAP's recommended method selection criteria in Chapter 3 require that a laboratory esti-
1521 mate the standard uncertainty for the measured concentration of a hypothetical laboratory sample
1522 with a specified concentration (i.e., the "method uncertainty," as defined by MARLAP). To
1523 estimate the combined standard uncertainty of the measured concentration, one must obtain esti-
1524 mates for all the input quantities and their standard uncertainties. All quantities except the gross
1525 instrument signal may be measured and the standard uncertainties evaluated by routine Type A
1526 and Type B methods. Alternatively, the values and their standard uncertainties may be deter-
1527 mined from historical data. The estimate of the gross signal and its standard uncertainty must be
1528 obtained by other means, since the laboratory sample is only hypothetical. The predicted value of
1529 the gross count Ns is calculated by rearranging the equation or equations in the model and solving
1530 for Ns. The standard uncertainty of the measured value may then be evaluated either from theory
1531 (e.g., Poisson counting statistics), historical data, or experimentation.
1532
1 533
_L ,J J J
1534
1535
1536
1537
1538
1539
1540
1541
1542
1543
1544
EXAMPLE:
where
X
Ns
NB
tB
tD
Ms
Y
£
X
Suppose the mathematical model for a radioactivity measurement is
T/" iJ iJ D D
is the activity concentration (Bq kg"1)
is the test source count
is the blank count
is the source count time (s)
is the blank count time (s)
is the decay time (s)
is the size of the test portion (kg)
is the chemical yield
is the counting efficiency
is the decay constant (s"1)
MARLAP
DO NOT CITE OR QUOTE
19-64
JULY 2001
DRAFT FOR PUBLIC COMMENT
-------
Measurement Statistics
1545
1546
1547
1548
1549
1550
1551
With specified values for the concentration X, test portion size Ms, blank count NB, count
times ts, tB, and tD, efficiency £, and yield 7,
the source count Ns can be
dieted value is Ns = ts(XMsYe exp(-Xfe + ts / 2) ) + NB 1 tB).
measured value, its estimated variance according to Poisson
When this
statistics is
predicted. The pre-
value is treated like a
^(NS) = NS. So,
assuming negligible uncertainties in the times ts, tB, and tD, the uncertainty propagation for-
mula gives the combined variance of the output estimate X as
2fkT \ 2 2fkT \ 2
n2(Y\ - S B + Y2
M2Y2£2e~2HtD+ts/2)
(XMsY£e D+ s +NB/tB)/
A/r2V2 2 ~2^Sr\X=0] = a
(19.31)
1560 where Pr[S > Sc \ X= 0] denotes the probability that the observed net signal S exceeds its critical
1561 value Sc when the true analyte concentration Xis zero, and a denotes the significance level, or
1562 the specified probability of a type I error. When the signal assumes only discrete values (e.g.,
1563 numbers of counts), there may be no value Sc that satisfies Equation 19.31 exactly. The critical
1564 value in this case is defined as the smallest value Sc such that Pr[^ > Sc \ X = 0] < a.
JULY 2001
DRAFT FOR PUBLIC COMMENT
19-65
MARLAP
DO NOT CITE OR QUOTE
-------
Measurement Statistics
1565 Determining a value of Sc which satisfies the definition requires knowledge of the distribution of
1566 the net signal S under the assumption that the analyte concentration is zero (the null hypothesis).
1567 The measured net signal may be written as S = Y - B, where Y denotes the measured gross
1568 signal and B denotes the estimated value of the gross signal under the null hypothesis H0. In the
1569 absence of interferences, the value of B is usually estimated by measuring one or more blanks
1570 using the same procedure used to measure the test sample, and the distribution of Y under H0 is
1571 determined from that of B. In other cases, however, the value of B includes estimated baseline
1572 and other interferences that are present only during the measurement of the sample and cannot be
1573 determined from the blank.
1574 Since Sc, not_yc, has traditionally been used for analyte detection decisions in radioanalysis, the
1575 following presentation focuses primarily on Sc. However, conversion of either of these values to
1576 the other is simple, becauseyc = Sc + B.
1577 19.7.1.1 Normally Distributed Signals
1578 If the distribution of the net signal S under H0 is approximately normal with a well-known
1579 standard deviation o0, the critical value of S is
SC = Zl-a°0 (19.32)
1580 where z1_a denotes the (1 - a)-quantile of the standard normal distribution. Table G.I in Appen-
1581 dix G shows that zl_a~ 1.645 when a = 0.05. Attachment 19D describes the calculation of Sc
1582 when the standard deviation is not well-known.
1583 The blank signal B and its standard deviation OB may be estimated by replicate blank measure-
1584 ments, but at least 20 measurements are generally needed to ensure that the experimental stan-
1585 dard deviation SB is an accurate estimate of CB. (If fewer than 20 measurements are made, see
1586 Attachment 19D.) Given OB, the standard deviation o0 of the net signal S under the null hypothe-
1587 sis is given equal to
G0 - G5A
(19.33)
n
MARLAP JULY 2001
DO NOT CITE OR QUOTE 19-66 DRAFT FOR PUBLIC COMMENT
-------
Measurement Statistics
1588 19.7.1.2 Poisson Counting
1589 Radionuclide analyses typically involve radiation counting measurements. Although radiation
1590 counting data never follow the Poisson model exactly, the model may be a useful approximation
1591 in some situations, especially those where the mean count is extremely low and the observed
1592 count therefore does not follow a normal distribution. At somewhat higher count levels, features
1593 from both models are often used, since the Poisson distribution may be approximated by a
1594 normal distribution. In this case, the Poisson model allows one to estimate o0 without replication,
1595 because one blank measurement provides an estimate of OB.
1596 When a test source is analyzed in a radiation counting measurement, either the gross count or the
1597 gross count rate may be considered the instrument signal Y. In this section, it is assumed that the
1598 instrument signal is the gross count. Therefore,
Y = NS B = \-^+Rl\ts (19.34)
V IB
1599 and the net instrument signal is the net count, defined as
(19.35)
V 1B I
1600 where
1601 Ns is the gross count (source count)
1602 NB is the blank count
1603 Rj is the estimated count rate due to interferences
1604 ts is the count time for the test source
1605 tB is the count time for the blank
1606 The net signal is always assumed to have zero mean.
1607 THE POISSON-NORMAL APPROXIMATION
1608 When Poisson counting statistics are assumed (possibly with additional variance components)
1609 and the instrument background remains stable at a level where the Poisson distribution is approx-
1610 imately normal, the critical net count is given approximately by the equation
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 19-67 DO NOT CITE OR QUOTE
-------
Measurement Statistics
^C ~ Zl-atS
.
R+R
I
(19.36)
1611 where RB denotes the (true) mean count rate of the blank, Rt denotes the mean interference count
1612 rate, ^ denotes non-Poisson variance in the blank (count rate) correction (see Section 19.6.4),
1613 and c2(Rf) denotes the variance of the estimator for R{. When there are no interferences and no
1614 non-Poisson blank variance, this equation becomes
(19.37)
1615 The preceding formula is equivalent to "Currie's equation" Lc = 2.33 J\iB when tB = ts,a = 0.05,
1616 and the symbols Lc and \IB are identified with Sc and RBts, respectively (Currie 1968).
1617 In Equation 19.37, RB denotes the true mean blank count rate, which can only be estimated. In
1618 practice, one must substitute an estimated value RB for RB, as shown in the following equation.
(19.38)
1619 Equation 19.38 resembles Equation 19.37 (Currie's equation) but involves the estimated count
1620 rate RB, which varies with repeated measurements. The value of RB is usually estimated from the
1621 same blank value NB used to calculate the net instrument signal. (See Attachment 19D for other
1622 possible estimators.)
~ NB
RB = — (19.39)
1623 The resulting formula, shown below, is equivalent to equations published by several authors
1624 (Currie 1968, Lochamy 1976, Strom and Stansbury 1992, ANSI 1996a).
l-a
^
(19.40)
MARLAP
DO NOT CITE OR QUOTE
19-68
JULY 2001
DRAFT FOR PUBLIC COMMENT
-------
Measurement Statistics
1625 If a = 0.05 and tB = ts, Equation 19.40 leads to the well-known expression 2.33 JNB for the
1626 critical net count.
1627 When the blank count is high (e.g., 100 or more), Equation 19.40 works well. At lower blank
1628 levels, it can produce a high rate of type I errors. For example, if the true mean blank count is
1629 0.693, there is a 25% chance of observing 0 blank counts and a positive number of test source
1630 counts in paired measurements of equal duration. In this case, a critical value calculated by Equa-
1631 tion 19.40 produces type I errors more than 25% of the time regardless of the chosen significance
1632 level a. Attachment 19D describes several expressions for Sc that have been proposed for use in
1633 situations where the mean blank count is less than 100.
1634
1635
1636
1637
1638
1639
EXAMPLE
Problem: A 6000-s blank measurement is performed on a proportional counter and 108 beta
counts are observed. A test source is to be counted for 3000 s. Estimate the critical value of the
net count when a = 0.05.
Solution:
"1-a,
N
= 1.645
\
108
3000
6000;
1 +
3000
6000,
= 14.8 counts.
1640
1641
1642
EXAMPLE
Problem: Repeat the same problem assuming the blank correction, expressed as a count rate,
has a non-Poisson uncertainty component of ^g = 0.001 cps (see Section 19.6.4).
JULY 2001
DRAFT FOR PUBLIC COMMENT
19-69
MARLAP
DO NOT CITE OR QUOTE
-------
Measurement Statistics
1643
1644
1645
Solution:
,2,2
= 1.645
\
(0001)2(6000)2
= 15.6 counts.
So, 15.6 may be a slightly more realistic value for the critical net count.
1646 19.7.1.3 Reagent Blanks
1647 Equation 19.40 is derived with the assumption that a detection decision is based on counts
1648 obtained from a single radiation counter. When laboratory samples are analyzed in batches, it is
1649 common to analyze a single reagent blank per batch, so that the measurement conditions for the
1650 blank may differ somewhat from those of the samples. In particular, the counts for the laboratory
1651 samples and the blank may be measured using different instruments. If detection in a laboratory
1652 sample is defined relative to a reagent blank counted on a different instrument, Equation 19.40 is
1653 inappropriate. Even if a single instrument is used, the presence of positive amounts of analyte in
1654 the reagents probably invalidates the Poisson assumption. In principle, B should be estimated by
1655 converting the total analyte activity of the reagent blank Z^ to an estimated gross count on the
1656 instrument used to measure the laboratory sample. Thus,
1657
1658
1659
1660
1661
1662
where
F
(19.41)
is the calibration function for the laboratory sample measurement, whose parameters
include the instrument background, counting efficiency, chemical yield, and any
estimated interferences
ZRB is the estimated total activity of the reagent blank
Then the net count is S = Y - B, whose critical value is
1663
(19.42)
where
MARLAP
DO NOT CITE OR QUOTE
19-70
JULY 2001
DRAFT FOR PUBLIC COMMENT
-------
Measurement Statistics
1664 o2( YQ) is the variance of the gross count Y in the test source measurement when all of the
1665 analyte in the source is derived from reagents
^ /*, /*,
1666 ^B is the variance of the estimator B
1667 If Poisson counting statistics are assumed, then o2(70) may be estimated by B (assuming B > 0),
1668 but estimating c\B) still requires a more complicated expression, which may be based on uncer-
1669 tainty propagation or replication. The variance of B may be difficult to estimate if positive blank
1670 values are caused not by the presence of the analyte in reagents but by contaminated glassware or
1671 instruments, which may represent a loss of statistical control of the analytical process.
1672 19.7.2 Calculation of the Minimum Detectable Concentration
1673 The minimum detectable concentration (MDC) is defined as the concentration of analyte XD that
1674 must be present in a laboratory sample to give a probability 1 - P of obtaining a measured
1675 response greater than its critical value, leading one to conclude correctly that the analyte concen-
1676 tration is positive. In other words, the MDC is the analyte concentration at which the type II error
1677 rate is p.
1678 The MDC may also be defined as the analyte concentration XD that satisfies the relation
Pr[S
-------
Measurement Statistics
1691 The MDC is also commonly used in radiochemistry to describe the detection capability of the
1692 analytical process as implemented in a particular instance. In this case, the need for conservative
1693 choices is reduced. Instead, the measured values of the variable quantities may be used. How-
1694 ever, since the measured values have uncertainties, their uncertainties contribute to a combined
1695 standard uncertainty in the calculated value ofxD. For purposes of regulatory compliance, an
1696 uncertainty interval or conservative upper bound for XD may still be needed (see NRC 1984).
1697 19.7.2. 1 The Minimum Detectable Net Instrument Signal
1698 The traditional method for calculating the MDC involves first calculating the minimum detect-
1699 able value of the net instrument signal and then converting the result to a concentration using the
1700 mathematical measurement model. The minimum detectable value of the net instrument signal,
1701 denoted by SD, is defined as the mean value of the net signal that gives a specified probability
1702 1 - p of yielding an observed signal greater than its critical value Sc. Thus,
= SD]=P (19.44)
1703 where S denotes the true mean net signal.
1704 19.7.2.2 Normally Distributed Signals
1705 If the net signal S is normally distributed and its estimated standard deviation q> under H0 is well-
1706 known, the critical value of S is
SC=Zl-a°0 (19.45)
1707 as previously noted. Then, the minimum detectable net signal SD is determined implicitly by the
1708 equation
SD = Sc+zl_J(S\S = SJ (19.46)
1709 where o2(S S = SD) denotes the variance of the measured signal S when the true mean signal S
1710 equals SD. If the function c2(S S = SD) is constant, Equation 19.46 gives the value of SD immedi-
1711 ately, but typically o2(S | S = SD) is an increasing function of SD.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 19-72 DRAFT FOR PUBLIC COMMENT
-------
Measurement Statistics
1712 If the function G2(-S' | S = SD) has a simple form, it may be possible to transform Equation 19.46
1713 by algebraic manipulation into an explicit formula for SD. For example, the variance of S often
1714 has the form
c\S)=aS2 + bS + c (19.47)
1715 where S denotes the true mean net signal and the constants a, b, and c do not depend on S (see
1716 Section 19.7.2.3, "Poisson Counting"). In this case, the minimum detectable net signal is given
1717 approximately by
/ _ \
2 2 2
(19.48)
\ i
1718 where/p= 1 -
2
c 2
+ zi-P\
Z2^
>2 2
c 4 — c -r.
1719
1720
1721 below
r f
If Equation 19.46 cannot be transformed algebraically, an iterative procedure, such as fixed-point
iteration, may be used to solve the equation for SD. An outline of fixed-point iteration is shown
U ~1 „„, 11
1722 1. Set SD = Sc + zj _p/o2(^ | S = Sc)
1723 2 repeat
1724 3. Seth = SD
1725 4. Set SD = SC-
1726 5.
1727
until \SD- h\is sufficiently small
6. output the solution SD
1728 In many cases, one iteration of the loop (Lines 2-5) provides an adequate approximation of SD. In
1729 almost all cases, repeated iteration produces an increasing sequence of approximations
11 Fixed-point iteration, or functional iteration, is the term for a general technique for solving an equation of the
formx=J(x). The iteration produces a sequence x0, xbx2,..., where xn+l =j[xn). Under certain conditions, the
sequence converges to a fixed point off, where fix) = x. Newton's Method for finding a zero of a function g(x) is
one example of the technique.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 19-73 DO NOT CITE OR QUOTE
-------
Measurement Statistics
1730 converging upward to the solution; so, the stopping condition at Line 5 may be replaced by
1731 "until SD < h" to obtain full machine precision in the result.
1732 19.7.2.3 Poisson Counting
1733 If Sc is calculated using the Poisson model and the blank is measured with a sufficiently large
1734 number of counts, and if a = P, the minimum detectable net signal SD is given by the following
1735 simple equation.12
SD = z?_p+2Sc (19.49)
1736 In the special case when ts = tB and a = P = 0.05, Equation 19.49 becomes
SD = 2.7l+2Sc (19.50)
1737 In the general case, SD is determined from Equation 19.48 using the following values for a, b,
1738 and c.
1739 a = 0 b=\ c = Rntv
D o
1740 The resulting formula for SD is
I c
1+^1
1741 As previously noted, counting data never follow the Poisson model exactly. Variable factors such
1742 as counting efficiency, and source geometry and placement tend to increase a, while interferences
1743 and background instability tend to increase c. For example, if the counting efficiency has a 2%
12 Some references use the value 3 instead of Zj_p in this formula. A straightforward derivation gives the value
Zj_p, which is approximately 2.71 when (3 = 0.05, but replacing this value by -In (3 (approximately 3 when (3 = 0.05)
accounts for the fact that when the mean count is low, a Poisson distribution is only imperfectly approximated by a
normal distribution. The value - In (3 is the exact value of SD when the mean blank count rate is zero, because in this
case Sc = 0, and Pr[S = 0] < (3 if and only if S > - In (3. Note also that the equation in the text is valid only if a = (3.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 19-74 DRAFT FOR PUBLIC COMMENT
-------
Measurement Statistics
1744 coefficient of variation and background instability contributes a non-Poisson standard deviation
1745 of 0.001 cps to the blank correction, then one might use Equation 19.48 with the values
1746 a = (0.02)2 b = l c=RRt
1747 19.7.2.4 TheMDC
1748 Traditionally the minimum detectable net signal SD has been converted directly to the minimum
1749 detectable concentration XD using the same measurement model used to convert an observed
1750 value of the signal S to a concentration f . In a typical model, the net count is divided by the
1751 sensitivity A, which is the product of factors such as the count time, test portion size, counting
1752 efficiency, chemical yield, and decay factor. The sensitivity may also include the subsampling
1753 factor, denoted by Fs, which was defined in Section 19.6. 1 1 as the ratio of the analyte concentra-
1754 tion of a subsample to that of the original sample. This factor is always estimated to be 1 and is
1755 included only for its contribution to the measurement uncertainty.
1756 If the sensitivity does not vary substantially from measurement to measurement, the MDC is
1757 given by
SD
XD = (19.52)
1758 If the variance of A is not negligible, it increases the value ofxD. Recall that when the variance of
1759 the net count S has the form c\S) = aS2 + bS + c, the minimum detectable net instrument signal
1760 may be approximated by Equation 19.48. If the sensitivity is normally distributed, the effect of its
1761 variance on the detection limit may be accounted for (approximately) by increasing the value of
1762 the constant a in Equation 19.48 by an amount equal to (p^(l + a), where q>A denotes the relative
1763 standard deviation of A13 For example, in the Poisson-counting scenario, where the value of a
1764 would otherwise be zero, a becomes <$A. Then the MDC is given by
13 The word "approximately" is used here because the signal is only approximately normal when its conditional
distribution depends on the sensitivity in the manner described.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 19-75 DO NOT CITE OR QUOTE
-------
Measurement Statistics
XD =
S,
i-
.
*
(19.53)
1765
where /= 1 -
anda=
1766 Often the distribution of A may not be well-known or may not be approximately normal. In this
1767 case, one may replace A in the formula by a somewhat low value, such as the p-quantile ap of its
1768 distribution, and ignore its variance. Thus, assuming Poisson counting statistics, one may use
1769 Equation 19.53 with a = 0 and A = ap. Alternatively, if the subsampling error is thought to be
1770 approximately normal, one may increase a by (psamp(l + aX where (pgamp denotes the relative sub-
1771 sampling variance, and ignore the subsampling error when estimating the quantile ap (the
1772 approach used in Attachment 19E). If (pgamp i§ negligible, the MDC may be obtained directly from
1773 the minimum detectable net count SD using the following formula.
S
D
aa
(19.53)
1774 When a "sample-specific" MDC is calculated, the measured value of the sensitivity A may be
1775 substituted for A in the equation for XD and the variance of A may be ignored. Then, if the sub-
1776 sampling variance (psamp is also negligible, the MDC is estimated by
S
D
A
(19.54)
1777 However, it should be remembered that the resulting value for the MDC has an uncertainty gen-
1778 erated by the measurement uncertainties of the input estimates from which it is calculated. It may
1779 also be variable because of the variability of the true sensitivity factors (e.g., chemical yield).
1780 19.7.2.5 Regulatory Requirements
1781 More conservative (higher) estimates of the MDC may be obtained by following the recommen-
1782 dations of NUREG/CR-4007, in which formulas for MDC (LLD) include estimated bounds for
1783 relative systematic error in the blank determination (AB) and the sensitivity (&A~). The critical net
1784 count Sc is increased by AB^, and the minimum detectable net count SD is increased by 2 ARB.
MARLAP
DO NOT CITE OR QUOTE
19-76
JULY 2001
DRAFT FOR PUBLIC COMMENT
-------
Measurement Statistics
1785 The MDC is then calculated by dividing SD by the sensitivity and multiplying the result by
1786 1 + &A. The approach of NUREG/CR-4007, which deals with detection limits, differs fundamen-
1787 tally from that of the GUM, which considers only measurement uncertainty. The NUREG's
1788 conservative approach treats random errors and systematic errors differently to ensure that the
1789 MDC for a measurement process is unlikely to be consistently underestimated, which is an
1790 important consideration if the laboratory is required by regulation or contract to achieve a speci-
1791 fiedMDC.
1792 19.7.2.6 Testing the MDC
1793 To ensure that the MDC has been estimated properly, one may test the estimate experimentally
1794 by analyzing n identical control samples spiked with an analyte concentration equal to XD. If the
1795 MDC has been determined properly (the null hypothesis), the probability of failing to detect the
1796 analyte in each control sample is at most p. Then the number of nondetectable results in the
1797 experiment may be assumed to have a binomial distribution with parameters n and p. If A: non-
1798 detectable results are actually obtained, one calculates the cumulative binomial probability
^ = E(W) P'O-Pr7 or l-£Hp/'(l-pr-/ (19.55)
j=k\JJ j=0\J)
1799 and rejects the null hypothesis ifP is smaller than the chosen significance level for the test
1800 (which may differ from the significance level for the analyte detection test).
1801 To make the test realistic, one should ensure that the physical and chemical characteristics of the
1802 control samples, including potential interferences, are representative of laboratory samples
1803 encountered in practice.
1804
1805
1806
1807
EXAMPLE
Problem: Assume XD is estimated with p = 0.05. As a check, 10 control samples spiked with
concentration XD are analyzed and 3 of the 10 produce nondetectable results. Does XD appear to
have been underestimated (at the 2% level of significance)?
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 19-77 DO NOT CITE OR QUOTE
-------
Measurement Statistics
1808
1809
1810
Solution: The variables are n = 10, p = 0.05, and k=3. Calculate the P-value
(0.05y(0.95)10^ = 1-0.9885 = 0.01 15
Since P < 0.02, reject the null hypothesis and conclude that the MDC was underestimated.
1811 19.7.3 Calculation of the Minimum Quantifiable Concentration
1812 The minimum quantifiable concentration (MQC), or the minimum quantifiable value of the con-
1813 centration, was defined in Section 19.4.5 as the analyte concentration in a laboratory sample that
1814 gives measured results with a specified relative standard deviation 1 / kQ, where kQ is usually
1815 chosen to be 10.
1816 Calculation of the MQC requires that one be able to estimate the standard deviation for the result
1817 of a hypothetical measurement performed on a laboratory sample with a specified analyte con-
1818 centration. Section 19.6.12 discusses the procedure for calculating the standard deviation for such
1819 a hypothetical measurement.
1820 The MQC is defined symbolically as the value XQ that satisfies the relation
Zj (19.56)
1821 where c2(X \ X= XQ) denotes the variance of the estimator X when the true concentration X
1822 equals XQ. If the function c2(X \ X= XQ) has a simple form, it may be possible to solve Equation
1823 19.56 for XQ using only algebraic manipulation. Otherwise, fixed-point iteration, which was
1824 introduced in Section 19.7.2, may be used. The use of fixed-point iteration for this purpose is
1825 shown below.
1826 1. Set xe =
1827 2 repeat
1828 3.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 19-78 DRAFT FOR PUBLIC COMMENT
-------
Measurement Statistics
1829
1830
1831
Setxe=
5. until XQ - h\ is sufficiently small
6. output the solution XQ
1832 The sequence of values generated by the algorithm typically converges upward to the solution.
1833 When Poisson counting statistics are assumed, possibly with excess variance components, and
1834 the mathematical model for the analyte concentration is X = SIAFS, where S is the net count, A
1835 denotes the overall sensitivity of the measurement, and Fs is the subsampling factor, Equation
1836 19.56 may be solved for XQ to obtain the formula
1837
1838
1839
1840
1841
1842
1843
1844
1845
1846
where
9s,
X/-1 =
^ 2 A I.
amp
o
1 +
\
(19.57)
is the count time for the test source
is the count time for the blank
is the mean blank count rate
is the non-Poisson variance component of the blank count rate correction
is the mean interference count rate
is the standard deviation of the measured interference count rate
is the relative variance of the measured sensitivity, A
is the relative subsampling variance
is equal to 1 - k2Q ((p2, + (p2, )
1847
1848
1849
1850
1851
1852
1853
If the true sensitivity^ may vary, then a conservative value, such as the 0.05-quantile a005,
should be substituted for^4 in the formula. Note that 0. If IQ < 0, the MQC is defined to be
infinite, because there is no concentration at which the relative standard deviation of X fails to
exceed 1 / kQ. In particular, if the relative standard deviation of the measured sensitivity A or the
subsampling standard deviation (pSamp exceeds 1 / kQ, then IQ < 0 and the MQC is infinite.
JULY 2001
DRAFT FOR PUBLIC COMMENT
19-79
MARLAP
DO NOT CITE OR QUOTE
-------
Measurement Statistics
1854 More generally, if the variance of the measured concentration X can be expressed in the form
1855 G\X) = aX2 + bX+ c, where a, b, and c do not depend on X, then the MQC is given by the
1856 formula
2(l-k2Qa}
(19.58)
1857 For example, if pure Poisson counting statistics are assumed and there are no interferences, then
1858 a =
-------
Measurement Statistics
1875 American Society for Testing and Materials (ASTM). 1995. Standard Specification for Glass
1876 Volumetric (Transfer) Pipets, E 969. ASTM, West Conshohocken, PA.
1877 American Society for Testing and Materials (ASTM). 2000. Standard Practice for Calibration of
1878 Laboratory Volumetric Glassware., E 542. ASTM, West Conshohocken, PA.
1879 Bevington, Philip R., and D. Keith Robinson. 1992. Data Reduction and Error Analysis for the
1880 Physical Sciences, 2nd ed. McGraw-Hill, New York, NY.
1881 Brodsky, Allen. 1992. Exact Calculation of Probabilities of False Positives and False Negatives
1882 for Low Background Counting. Health Physics 63(2): 198-204.
1883 Burden, Richard L., and J. Douglas Faires. 1993. Numerical Analysis, 5th ed. PWS Publishing
1884 Company, Boston, MA.
1885 CFR 136. 1999. U.S. Environmental Protection Agency. "Guidelines Establishing Test
1886 Procedures for the Analysis of Pollutants."
1887 Colle, R., and Raj Kishore. 1997. An Update on the NIST Radon-in-water Standard Generator:
1888 Its Performance Efficacy and Long-term Stability. Nuclear Instruments and Methods in
1889 Physics Research A 391: 511-528.
1890 Currie, Lloyd A. 1968. Limits for Qualitative Detection and Quantitative Determination:
1891 Application to Radiochemistry. Analytical Chemistry 40(3): 586-593.
1892 Currie, Lloyd A. 1972. The Limit of Precision in Nuclear and Analytical Chemistry. Nuclear
1893 Instruments and Methods 100(3): 387-395.
1894 Currie, L.A. 1997. Detection: International Update, and Some Emerging Di-lemmas Involving
1895 Calibration, the Blank, and Multiple Detection Decisions. Chemometrics and Intelligent
1896 Laboratory Systems 37: 151-181.
1897 Currie, L.A., E.M. Eijgenhuijsen, and G.A. Klouda. 1998. On the Validity of the Poisson
1898 Hypothesis for Low-Level Counting: Investigation of the Distributional Characteristics of
1899 Background Radiation with the NIST Individual Pulse Counting System. Radiocarbon 40(1):
1900 113-127.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 19-81 DO NOT CITE OR QUOTE
-------
Measurement Statistics
1901 Environmental Protection Agency (EPA). 1980. Upgrading Environmental Radiation Data,
1902 Health Physics Society Committee Report HPSR-1, EPA, 520/1-80-012, EPA, Office of
1903 Radiation Programs, Washington, DC.
1904 Environmental Protection Agency (EPA). 1998. Guidance for Data Quality Assessment:
1905 Practical Methods for Data Analysis. EPA QA/G-9, QA97 Version. EPA/600/R-96/084,
1906 EPA, Quality Assurance Division, Washington, DC.
1907 Eurachem. 2000. Eurachem/CITAC Guide: Quantifying Uncertainty in Analytical Measurement.,
1908 2nd ed. Eurachem.
1909 Filliben, James J. 1975. The Probability Plot Correlation Coefficient Test for Normality.
1910 Technometrics 17(1): 111-117.
1911 Friedlander, Gerhart, et al. 1981. Nuclear andRadiochemistry 3rd ed. John Wiley and Sons, New
1912 York, NY.
1913 Fuller, Wayne A. 1987. Measurement Error Models. John Wiley and Sons, New York, NY.
1914 Gy, Pierre M. 1992. Sampling of Heterogeneous and Dynamic Material Systems: Theories of
1915 Heterogeneity, Sampling and Homogenizing. Elsevier Science Publishers, Amsterdam, The
1916 Netherlands.
1917 Hoel, Paul G., Sidney C. Port, and Charles J. Stone. 1971. Introduction to Probability Theory.
1918 Houghton-Mifflin, Boston, MA.
1919 International Organization for Standardization (ISO). 1993a. International Vocabulary of Basic
1920 and General Terms in Metrology. ISO, Geneva, Switzerland.
1921 International Organization for Standardization (ISO). 1993b. Statistics - Vocabulary and
1922 Symbols - Part 1: Probability and General Statistical Terms. ISO 3534-1. ISO, Geneva,
1923 Switzerland.
1924 International Organization for Standardization (ISO). 1995. Guide to the Expression of
1925 Uncertainty in Measurement. ISO, Geneva, Switzerland.
1926 International Organization for Standardization (ISO). 1997. Capability of Detection - Part 1:
1927 Terms and Definitions. ISO 11843-1. ISO, Geneva, Switzerland.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 19-82 DRAFT FOR PUBLIC COMMENT
-------
Measurement Statistics
1928 International Organization for Standardization (ISO). 2000a. Determination of the Detection
1929 Limit and Decision Threshold for Ionizing Radiation Measurements - Part 1: Fundamentals
1930 and Application to Counting Measurements without the Influence of Sample Treatment. ISO
1931 11929-1. ISO, Geneva, Switzerland.
1932 International Organization for Standardization (ISO). 2000b. Determination of the Detection
1933 Limit and Decision Threshold for Ionizing Radiation Measurements - Part 2: Fundamentals
1934 and Application to Counting Measurements with the Influence of Sample Treatment. ISO
1935 11929-2. ISO, Geneva, Switzerland.
1936 International Organization for Standardization (ISO). 2000c. Determination of the Detection
1937 Limit and Decision Threshold for Ionizing Radiation Measurements - Part 3: Fundamentals
1938 and Application to Counting Measurements by High-resolution Gamma Spectrometry,
1939 without the Influence of Sample Treatment. ISO 11929-2. ISO, Geneva, Switzerland.
1940 International Union of Pure and Applied Chemistry (IUPAC). 1995. Nomenclature in Evaluation
1941 of Analytical Methods Including Detection and Quantification Capabilities. Pure and Applied
1942 Chemistry 67(10): 1699-1723.
1943 Knoll, Glenn F. 1989. Radiation Detection and Measurement, 2nd ed. John Wiley and Sons, New
1944 York, NY.
1945 Lawson, Charles L., and Richard J. Hanson. 1974. Solving Least Squares Problems. Prentice-
1946 Hall, Englewood Cliffs, NJ.
1947 Lochamy, Joseph C. 1976. The Minimum Detectable Activity Concept. NBS Report No. NBS-
1948 SP456, National Bureau of Standards, Gaithersburg, MD.
1949 Lucas, H.F., Jr., and D.A. Woodward. 1964. Journal of Applied Physics 35: 452.
1950 Marquardt, D.W. 1963. An Algorithm for Least-Squares Estimation of Nonlinear Parameters.
1951 Journal of the Society for Industrial and Applied Mathematics 11(2): 431-441.
1952 National Bureau of Standards (NBS). 1963. Experimental Statistics. NBS Handbook 91, National
1953 Bureau of Standards, Gaithersburg, MD.
1954 National Bureau of Standards (NBS). 1964. Handbook of Mathematical Functions. Applied
1955 Mathematics Series 55, National Bureau of Standards, Gaithersburg, MD.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 19-83 DO NOT CITE OR QUOTE
-------
Measurement Statistics
1956 National Council on Radiation Protection and Measurements (NCRP). 1985. Handbook of
1957 Radioactivity Measurement Procedures. NCRP Report 58, 2nd ed., NCRP, Bethesda, MD.
1958 Nicholson, W.L. 1963. Fixed Time Estimation of Counting Rates with Background Corrections.
1959 AEC Research and Development Report HW-76279.
1960 Nicholson, W.L. 1966. Statistics of Net-counting-rate Estimation with Dominant Background
1961 Corrections. Nucleonics 24(8): 118-121.
1962 Nuclear Regulatory Commission (NRC). 1984. Lower Limit of Detection: Definition and
1963 Elaboration of a Proposed Position for Radiological Effluent and Environmental
1964 Measurements. NUREG/CR-4007. NRC, Washington, DC.
1965 Pitard, Francis F. 1993. Pierre Gy 's Sampling Theory and Sampling Practice: Heterogeneity,
1966 Sampling Correctness, and Statistical Process Control., 2nd ed. CRC Press, Boca Raton, FL.
1967 Press, William H., et al. 1992. Numerical Recipes in C: The Art of Scientific Computing, 2nd ed.
1968 Cambridge University Press, New York, NY.
1969 Stapleton, James H. 1995. Linear Statistical Models. John Wiley and Sons, New York, NY.
1970 Stapleton, James H. 1999. Personal correspondence. Department of Statistics and Probability,
1971 Michigan State University.
1972 Strom, Daniel J., and Paul S. Stansbury. 1992. Minimum Detectable Activity When Background
1973 Is Counted Longer than the Sample. Health Physics 63(3): 360-361.
1974 Turner, James E. 1995. Atoms, Radiation, and Radiation Protection, 2nd ed. John Wiley and
1975 Sons, New York, NY.
1976 19.8.2 Other Sources
1977 American Chemical Society (ACS). 1988. Detection in Analytical Chemistry: Importance,
1978 Theory, andPractice. ACS Symposium Series 361, ACS, Washington, DC.
1979 National Institute of Standards and Technology (NIST). 1994. Guidelines for Evaluating and
1980 Expressing the Uncertainty of NIST Measurement Results. NIST Technical Note 1297, NIST,
1981 Gaithersburg, MD.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 19-84 DRAFT FOR PUBLIC COMMENT
-------
1982 ATTACHMENT 19A
1983 Distributions
1984 19A.1 Introduction
1985 This attachment briefly describes the probability distributions used in Chapter 19.
1986 Distributions may be classified according to their mathematical properties. Distributions in the
1987 same class or family are described by the same mathematical formulas. The formulas involve
1988 numerical parameters which distinguish one member of the class from another.
1989 Two important kinds of distributions are the normal and log-normal, which are observed often in
1990 nature. Other types of distributions important in radioanalysis include the rectangular, binomial,
1991 Poisson, Student's t, chi-square, and exponential distributions. Poisson distributions in particular
1992 are important in radiation counting measurements and are described in Section 19.6.2.
1993 19A.2 Normal Distributions
1994 Many quantities encountered in nature and in the laboratory have distributions which can be
1995 described by the "bell curve." This type of distribution, called a normal, or Gaussian, distribu-
1996 tion, is usually a reasonably good model for the result of a radioanalytical measurement. A num-
1997 ber of commonly used methods for evaluating data sets depend on their having an approximately
1998 normal distribution. The probability density function (pdf) for a normal distribution is shown in
1999 Figure 19.5.
-3-2-10 1 2 3
FIGURE 19.5 — A normal distribution
2000 A normal distribution is uniquely specified by its mean |i and variance a2. The normal distribu-
2001 tion with mean 0 and variance 1 is called the standard normal distribution. If Xis normally dis-
2002 tributed with mean |i and variance a2, then (X - |i) / a has the standard normal distribution.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 19-85 DO NOT CITE OR QUOTE
-------
Measurement Statistics
2003 The sum of a large number of independent random variables has an approximately normal distri-
2004 bution, even if the individual variables themselves are not normally distributed, so long as the
2005 variance of each term is much smaller than the variance of the sum.14 This is one reason why the
2006 normal distribution occurs often in nature. When a quantity is the result of additive processes
2007 involving many small random variations, the quantity tends to be normally distributed. It is also
2008 true that many other distributions, such as the binomial, Poisson, Student's t, and chi-square, can
2009 be approximated by normal distributions under certain conditions.
2010 The mean value of a normal distribution is also its mode, or most likely value, which corresponds
2011 to the location of the peak of the curve shown in Figure 19.5. Since the distribution is symmetric
2012 about this point, the mean is also the median, or the value that splits the range into equally likely
2013 portions.
2014 The value of a normally distributed quantity will be within one standard deviation of the mean
2015 about 68% of the time. It will be within two standard deviations about 95% of the time and
2016 within three standard deviations more than 99% of the time. It is important to remember that
2017 these percentages apply only to normal distributions.
2018 19A.3 Log-normal Distributions
2019 The concentration of a contaminant in the environment may not be normally distributed. Instead
2020 it often tends to be log-normally distributed, as shown in Figure 19.6.
A*)
0 M n u.
&
FIGURE 19.6 — A log-normal distribution
14 The number of quantities required to obtain a sum that is approximately normal depends on the distribution of
the quantities. If the distribution is already symmetric and mound-shaped like the bell curve, the number may be
rather small. Other distributions such as the log-normal distribution, which is asymmetric, may require a much larger
number.
MARLAP
DO NOT CITE OR QUOTE
19-86
JULY 2001
DRAFT FOR PUBLIC COMMENT
-------
Measurement Statistics
2021 By definition, a quantity X has a log-normal distribution if the logarithm of X is normally distrib-
2022 uted. The product of a large number of independent positive random variables with similar var-
2023 iances is approximately log-normal, because the logarithm of the product is a sum of independent
2024 random variables, and the sum is approximately normal. The concentration of a contaminant in
2025 the environment tends to be log-normal because it is the result of processes of concentration and
2026 dilution, which are multiplicative.
2027 The distribution of a log-normal quantity X can be uniquely specified by the mean |ilnX and
2028 variance olnX of In X, but more commonly used descriptors are the geometric mean |ig =
2029 exp(|ilnX) and the geometric standard deviation og = exp(olnjr). The geometric mean and geomet-
2030 ric standard deviation are defined so that, if & is a positive number, the probability that X will fall
2031 between \ig I o* and \igcg is the same as the probability that InX, which is normally distributed,
2032 will fall between \iinX - kclnX and \iinX + kclnX. For example, the value of X will be between
2033 \ng I c2g and \ng
-------
Measurement Statistics
5 10 15 20
FIGURE 19.7 — Chi-square distributions
2049 Chi-square distributions are used frequently in hypothesis testing, especially for tests of hypothe-
2050 ses about the variances of normally distributed data. Chi-square distributions also appear in least-
2051 squares analysis (see Attachment 19B).
2052 A sum of independent chi-square random variables is also chi-square. Specifically, if X and 7 are
2053 independent chi-square random variables with vl and v2 degrees of freedom, respectively, then
2054 X+ 7 has a chi-square distribution with vl + v2 degrees of freedom.
2055 The mean of a chi-square distribution equals the number of degrees of freedom v, and the vari-
2056 ance equals 2v. The mode equals zero if v < 2 and equals v - 2 otherwise. The median does not
2057 have a simple formula.
2058 19A.5 T-Distributions
2059 If Zis standard normal, Xis chi-square with v degrees of freedom, and Z and X are independent,
2060 then Z / ^X/v has a Student's t-distribution with v degrees of freedom. A ^-distribution is sym-
2061 metric and mound-shaped like a normal distribution and includes both positive and negative
2062 values. Figure 19.8 shows the pdf for a ^-distribution with 3 degrees of freedom. A dotted stan-
2063 dard normal curve is also shown for comparison.
MARLAP
DO NOT CITE OR QUOTE
19-88
JULY 2001
DRAFT FOR PUBLIC COMMENT
-------
Measurement Statistics
Normal
-4-3
FIGURE 19.8
-2-10 1 2 3 4
The ^-distribution with 3 degrees of freedom
2064 When v is large, the ^-distribution is virtually identical to the standard normal distribution.
2065 The median and mode of a ^-distribution are both zero. The mean is also zero if v > 1 but is
2066 undefined for v = 1 . The variance equals v / (v - 2) if v > 2 and is undefined otherwise.
2067 T-distributions are often used in tests of hypotheses about the means of normally distributed data
2068 and are important in statistical quality control, /"-distributions are also used in the procedure
2069 described in Attachment 19C for calculating measurement coverage factors.
2070 If X±, X2, ...,Xn are independent and normally distributed with the same mean ji and the same
2071 variance, then the quantity
2072 where X is the arithmetic mean and sx is the experimental standard deviation, has a ^-distribution
2073 with n - 1 degrees of freedom.
2074 IfX^ X2, ...,Xn, Yare independent and normally distributed with the same mean and variance,
2075 then the quantity
Y-X
+ II n
JULY 2001
DRAFT FOR PUBLIC COMMENT
19-89
MARLAP
DO NOT CITE OR QUOTE
-------
Measurement Statistics
2076 where X is the arithmetic mean of the Xi and sx is the experimental standard deviation, has a t-
2077 distribution with n - 1 degrees of freedom.
2078 If Z is standard normal, X is chi-square with v degrees of freedom, Z and X are independent, and
2079 5 is a constant, then (Z + 5) / \/XI v has the non-central t-distribution with v degrees of freedom
2080 and non-centrality parameter 5. When the (central) ^-distribution is used to test the null hypothe-
2081 sis that two normal distributions have the same mean, a non-central ^-distribution describes the
2082 distribution of the test statistic if the null hypothesis is false. For example, if X^ X2, ...,Xn, Y are
2083 independent and normally distributed with the same variance a2, and X^ X2, ...,Xn have the same
2084 mean \ix, then the statistic
2085
l/n
2086 where X is the arithmetic mean of the Xi and sx is the experimental standard deviation, has a t-
2087 distribution with n - 1 degrees of freedom if \ix = \IY, but it has a non-central ^-distribution with
2088 non-centrality parameter
2089 6 = Y X
2090 if \ix * |I7.
2091 The non-central ^-distribution is useful in the theory of detection limits and appears in Section
2092 19D.3.2 of Attachment 19D.
2093 19A.6 Rectangular Distributions
2094 If Xonly assumes values between a_ and a+ and all such values are equally likely, the distribution
2095 of Xis called a rectangular distribution, or a uniform distribution (see Figure 19.9).
MARLAP JULY 2001
DO NOT CITE OR QUOTE 19-90 DRAFT FOR PUBLIC COMMENT
-------
Measurement Statistics
FIGURE 19.9 — A rectangular distribution
2096 The mean and median of the rectangular distribution equal the midrange (a_ + a+) 12, and the
2097 standard deviation is (a+ - a_) I 2/3. The rectangular distribution is multimodal.
2098 Rectangular distributions are frequently used for Type B evaluations of standard uncertainty (see
2099 Sections 19.5.2.2 and 19.6.10).
2100 19A.7 Trapezoidal and Triangular Distributions
2101 Another type of bounded distribution used for Type B evaluations of standard uncertainty is a
2102 trapezoidal distribution, which is described in Section 19.5.2.2. If Xhas a trapezoidal distribu-
2103 tion, it only assumes values between two numbers a_ and a+, but values near the midrange
2104 (a_ + a+) 12 are more likely than those near the extremes. The pdf for a symmetric trapezoidal
2105 distribution is shown in Figure 19.10. Asymmetric trapezoidal distributions are not considered
2106 here.
2a
FIGURE 19.10 — A trapezoidal distribution
JULY 2001
DRAFT FOR PUBLIC COMMENT
19-91
MARLAP
DO NOT CITE OR QUOTE
-------
Measurement Statistics
2107
2108
2109
2110
2111
2112
2113
2114
2115
2116
2117
2118
2119
2120
2121
2122
The mean and median of this distribution are both equal to the midrange. If the width of the trap-
ezoid at its base is la and the width at the top is 2«p, where 0 < p < 1, then the standard deviation
is ay (1 + p2) / 6 . As P approaches 0, the trapezoidal distribution approaches a triangular distri-
bution, whose standard deviation is a/, the number of successes
2126 has a binomial distribution with parameters n and p. Important facts about the binomial distribu-
2127 tion include the following:
MARLAP
DO NOT CITE OR QUOTE
19-92
JULY 2001
DRAFT FOR PUBLIC COMMENT
-------
Measurement Statistics
2128 • The distribution is discrete; its only possible values are 0, 1,2, ..., n.
2129 • The mean of the distribution is np.
2130 • The variance is np(\ - /?).
2131 • If n is large and/? is not close to 0 or 1, the distribution is well approximated by a normal
2132 distribution.
2133 If Xis binomial with parameters n and/?, then for k = 0, 1,2, ..., n, the probability that X = k is
2134 given by the equati on
( n\
Vr[X=k\=\ \p\l-pT (19.61)
(k)
2135 19A.10 Poisson Distributions
2136 As explained in Section 19.6.2, the Poisson distribution arises naturally as an approximation to
2137 the binomial distribution when n is large and/? is small. Even if n is not large, the variance of the
2138 binomial distribution can be approximated using the Poisson model if/? is small. Other important
2139 facts about a Poisson distribution include the following:
2140 • The distribution is discrete; its only possible values are the nonnegative integers
2141 0,1,2,....
2142 • The mean and variance of the distribution are equal.
2143 • If the mean is large, the distribution is well approximated by a normal distribution.
2144 • A sum of independent Poisson random variables is also Poisson.
2145 If Xhas a Poisson distribution with mean |i, then for any nonnegative integer w, the probability
2146 that X=n is given by
ri\=e-» (19.62)
2147 The Poisson distribution is related to the chi-square distribution, since
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 19-93 DO NOT CITE OR QUOTE
-------
Measurement Statistics
TABLE 19.3 — 95% confidence interval for a Poisson mean
n
0
1
2
3
4
5
1 2
f^lower ~~ "Xo.CCSV^W)
0.000
0.025
0.242
0.619
1.090
1.623
Hupper= 7^.975 (2«+ 2)
3.689
5.572
7.225
8.767
10.242
11.668
Pr[X < n] = Pr[x2(2« + 2) > 2|i] and Pr[X > n] = Pr[x2(2«) < 2|i]
(19.63)
2148
2149
2150
2151
2152
2153
2154
where %2(v) denotes a chi-square random variable with v degrees of freedom. This fact allows one
to use quantiles of a chi-square distribution to construct a confidence interval for ji based on a
single observation X= n. Table 19.3 lists 95% two-sided confidence intervals for ji some small
values of n. For larger values of w, the quantiles ^(2n) and ^(2n + 2) may be approximated
using the Wilson-Hilferty formula (NBS 1964):
A
9v
(19.64)
As noted above, when the mean ji is large, the Poisson distribution may be approximated by a
normal distribution. Specifically,
Pr[X 20.
MARLAP
DO NOT CITE OR QUOTE
19-94
JULY 2001
DRAFT FOR PUBLIC COMMENT
-------
Measurement Statistics
2157 19 A. 11 References
2158 National Bureau of Standards (NBS). 1964. Handbook of Mathematical Functions. Applied
2159 Mathematics Series 55, National Bureau of Standards, Gaithersburg, MD.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 19-95 DO NOT CITE OR QUOTE
-------
Measurement Statistics
MARLAP JULY 2001
DO NOT CITE OR QUOTE 19-96 DRAFT FOR PUBLIC COMMENT
-------
2160 ATTACHMENT 19B
2161 Multicomponent Analyses
2162 19B.1 Matrix Equations
2163 A multicomponent mathematical model may require the simultaneous solution of a system of
2164 equations formulated in terms of vector and matrix operations, which are implemented in soft-
2165 ware. For example, one procedure for radiostrontium analysis involves the precipitation of stron-
2166 tium from a sample, followed by multiple beta measurements of the precipitate over a period of
2167 time. Both 89Sr and 90Sr are beta emitters, and 90Sr decays to 90Y, another beta emitter. The half-
2168 life of 90Y is short enough (64 h) that significant ingrowth occurs over a period of several days,
2169 allowing the activities of 89Sr and 90Sr to be determined from the changing count rate.
2170 The net beta count _y, for a measurement of duration tt at time A^ after precipitation has an
2171 expected value given by
(19.66)
2172 where
2173 xl is the 89Sr activity in the precipitate
2174 x2 is the 90Sr activity in the precipitate
2175 atl is a function of tt, A^, and the 89Sr counting efficiency and half-life
2176 aa is a function of tt, A^, and the 90Sr and 90Y counting efficiencies and half-lives
2177 If m measurements are performed, Equation 19.66 is repeated for each measurement, giving a
2178 system of m equations. After replacing E(y,) by the measured value yt, one can rewrite the
2179 equations as approximations in the form
/7V + /7 V ~ V
UHA1 U12 2 -^ 1
(19.67)
2180 or in matrix form as Ax ~ y. Ifm > 2, the system of equations can be solved simultaneously for xl
2181 and x2. If there are exactly two measurements (m = 2), the system can be solved easily without
2182 matrix operations, but if additional measurements are made (m > 2), a least-squares solution,
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 19-97 DO NOT CITE OR QUOTE
-------
Measurement Statistics
2183 which typically involves matrix algebra, is required. The use of matrix algebra can make uncer-
2184 tainty propagation more tedious.
2185 19B.2 Random Vectors and Matrices
2186 Uncertainty propagation in matrix equations is best described in terms of random vectors and
2187 random matrices. A useful exposition of matrix theory in this manual is impractical; so, some
2188 familiarity with the basic concepts must be assumed. These basic concepts will be extended to
2189 incorporate randomness.
2190 A random vector is a vector whose components are random variables. Similarly, a random
2191 matrix is a matrix whose components are random variables.
2192 Vectors are usually denoted by bold lower-case letters and matrices by bold upper-case letters.
2193 The /'th component of a vector v is denoted by vi. The if1 component of a matrix A is usually
2194 denoted by atj. The transpose of a matrix A will be denoted here by A'. If A is square and
2195 invertible, the inverse is denoted by A'1. The length of a vector v is denoted by \\v\\.
2196 The expected value of a random vector x is defined as the vector E(x) whose /th component
2197 is E(xi). The expected value of a random matrix Fis similarly defined as the matrix E(Y) whose
2198 /yth component is E(yt). The covariance matrix of a column vector x and a column vector y is
2199 defined by
Cov(*,j) = 4* - E(x))(y - E(y))f] (19.68)
2200 The covariance matrix of a random column vector x (or the variance-covariance matrix) is
2201 defined by
V(x) = Cov(x,x) (19.69)
2202 The covariance matrix gets its name from the fact that the if* component of CGV(JC,J) equals the
2203 covariance Cov(xi,yj).16 When* and y are vectors of measured values, the estimated covariance
2204 matrices will be denoted here by «(*,j) and u2(x).
16 In the literature, one often sees the covariance matrix for x and y denoted by S^ and the variance-covariance
matrix for x denoted by Ex.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 19-98 DRAFT FOR PUBLIC COMMENT
-------
Measurement Statistics
2205 19B.3 Linear Least Squares
2206 Assume y^y2, •••-,ym are independent, normally distributed measured results and F(y,) = o^ for
2207 each /'. Let xl3 x2, ...,xn denote unknown quantities on which the^ depend and whose values one
2208 needs to determine. Assume the means E(y,) are related to the quantities Xj by the following
2209 system of equations.
anxi
= E(y2)
(19.70)
2210 For example the^ might be measured beta counts of a sample and the Xj could represent the
221 1 unknown activities of 89Sr and 90Sr in the sample at the time of collection.
2212 The linear system 19.70 can be represented using matrix notation as
Ax=E(y) (19.71)
2213 Typically E(y) is unknown and must be replaced in Equation 19.71 by the measured vector y, but
2214 there may be no vector A: for which Ax exactly equals y. So, it is necessary to find an approximate
2215 solution x such thatAv; is close to y in some sense. The components of the difference Ax - y are
2216 called residuals, and when Ax is close to y, the residuals should be small. If o, = 1 for all /', the
2217 method of least squares finds a vector x that minimizes the sum of the squares of the residuals
2218 SSRES = || Ax - y \\2 . If o * 1 for some /', then both sides of equation /' should be divided by o,
2219 before applying the least-squares method. So, if W denotes the m x m diagonal matrix whose /'th
2220 diagonal element is 1 / cr), then SSRES = (Ax-y)'W(Ax-y). In practice, the standard devia-
2221 tions G; are usually replaced by the standard uncertainties w(
2222 A least-squares solution always exists. If rank^4 < w, there may be more than one solution, but
2223 this case only occurs if the measurement process is inadequate even in principle for determining
2224 the unknown quantities. So, in practice rank^4 = n. (The rank of A is the number of linearly
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 19-99 DO NOT CITE OR QUOTE
-------
Measurement Statistics
2225 independent columns or rows.) Under this assumption the unique least-squares solution is given
2226 by Equation 19.72.17
x = (A'WAYlA'Wy (19.72)
2227 When quantities such as the test portion size Fand chemical yield Y can be factored out of the
2228 matrix A, it is generally better to do so. The presence of such variables increases the variance of
2229 the least-squares solution Jc, making critical values unnecessarily large when they are calculated
2230 as described in Section 19B.6. When quantities such as Fand Y are factored out, the components
2231 of the least-squares solution x must be divided by the missing factors to obtain activity concen-
2232 trations, and the uncertainties in the factors must be propagated.
2233 Approximating the standard deviations o, in the weight matrix Wby the standard uncertainties
2234 w(y;) may bias the least-squares solution slightly ifyt and w(y;) are correlated, which happens, for
2235 example, whenjA is a measured count and z/(y,) is the Poisson counting uncertainty calculated
2236 from a single measurement. This bias can be virtually eliminated by using the initial least-squares
2237 solution to refine the values of the standard uncertainties and then repeating the least-squares
2238 procedure using the refined estimates.
2239 The solution i: is a random vector, because it is a function of the random vector y. The covariance
2240 matrix for Jc is
u2(jc) = (A'WA)-1 (19.73)
2241 The diagonal elements of this matrix are the variances of the components of Jc, and the off-
2242 diagonal elements are the covariances. This expression for the covariance matrix is complete
2243 only when there are no uncertainties in the coefficient matrix A. A more general formula for the
2244 covariance matrix is presented in Section 19B.5.
2245 In some cases, the variance of each yt may be unknown, although all components of y are
2246 believed to have the same variance. When this is true, the solution Jc may be computed by
jc = (A'A^A'y (19.74)
17 For some least-squares problems, a direct calculation of the solution x using Equation 19.72 can be computa-
tionally unstable. Singular value decomposition of the matrix A gives a more stable method for obtaining x but is
beyond the scope of this document. The S VD method also allows one to find a least-squares solution (not unique)
when rank A < n. See Lawson 1974 or Press et al. 1992 for more details.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 19-100 DRAFT FOR PUBLIC COMMENT
-------
Measurement Statistics
2247 and the variance of the components^, may be estimated by
~-^- (19.75)
m-n
2248 (The use of Equation 19.75 is a Type A evaluation of uncertainty with m-n degrees of
2249 freedom.) When this equation is used, the covariance matrix for x is
(19.76)
2250 19B.4 General Least Squares
225 1 The general least-squares problem arises when there is a set of measured values _yl3 y2, ...,ym,
2252 whose expected values are functions of an w-dimensional vector* of unknown quantities, as
2253 indicated by the following system of equations.
fl(x)=E(yl)
f2(x)=E(y2)
. (19.77)
/„(*)= *W
2254 The system of equations can be written in matrix form as/(jc) = E(y) The method of least squares
2255 finds a vector A: that minimizes the sum of the squares of the residuals
SSRES = £ _ = (fW -y}'W(f(x) - y) (19.78)
2256 When f(x) can be written as Ax for some matrix A, the problem is linear least squares, whose
2257 solution was presented in the preceding section. When the functions/ are nonlinear but differen-
2258 liable, the solution can be obtained by iterative approximation methods. The most commonly
2259 used algorithm for nonlinear least squares is the Levenberg-Marquardt algorithm (Press et al.
2260 1992). Whatever algorithm is used, it should compute the covariance matrix u2(x), described in
2261 the next section. For more details on nonlinear least-squares problems, see Marquardt 1963,
2262 Press et al. 1992, or Bevington 1992.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 19-101 DO NOT CITE OR QUOTE
-------
Measurement Statistics
2263 19B.5 The Covariance Matrix for a Least-Squares Solution
2264 Letv4 = dfl dx denote the m x n matrix whose if1 component is dft I dxj.18 Then the covariance
2265 matrix for the least-squares solution x is approximately equal to (A'WA)~l.
2266 It often happens that the function/depends on variables other than jc, whose values, like the
2267 components of y, are measured before the least-squares method is applied. In the strontium
2268 analysis described at the beginning of this attachment, the measured counting efficiencies for
2269 89Sr, 90Sr, and 90Y are good examples. Measurement uncertainties in these variables contribute to
2270 the uncertainties in the solution A:, although the least-squares covariance matrix (A'WAy1
2271 accounts only for uncertainties in the measurement of y. Better estimates of the variances and
2272 covariances of the components of A; require that the expression for the covariance matrix be
2273 expanded.
2274 Let the additional measured quantities be written as a vector z with components zl3 z2, ..., z,, and
2275 write f(x;z) to indicate that/depends on both x and z. Assume the components of z are measured
2276 independently of y, and the covariance matrix u\z) is known. If the method of least squares is
2277 applied to find the unique solution x that minimizes SSRES, and if the uncertainties in the com-
2278 ponents z{ are small, the covariance matrix for the solution is
' (19.79)
2279 where dx I dz denotes the n x r matrix whose ifh component is dxi I dz.. The/h column of dx I dz
2280 may be calculated using the formula
dx i
— = (A'WA)-i\ ^-w\y-J($;z))-A'W^-\ (19.80)
2281 If the uncertainties in the components z; are not small, another method of solution may be needed
2282 (e.g., see Fuller 1987).
2283 When the least-squares problem is linear, they* column of dfl dz is given by the formula
18
The matrix/1 is the Jacobian matrix of the component functions/b/2, ...,fm.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 19-102 DRAFT FOR PUBLIC COMMENT
-------
Measurement Statistics
8f 8A
2284 and the ifh component is given by
%i Y- da* ^
-T = E -^xk (19.82)
dzj * = i 3z7
2285 When the problem is nonlinear, the components dft I dzj are calculated by other means.
2286 19B.6 Critical Values
2287 The general approach to the determination of critical values even in the case of nonlinear least
2288 squares is conceptually no different from that outlined in Section 19.7. 1 . The standard uncer-
2289 tainty of a signal or response variable is determined under the null hypothesis H0 and then multi-
2290 plied by an appropriate factor, such as the normal quantile zl_a. The response variable for a
2291 component Xj may be taken to be the corresponding component x. of the least-squares solution
2292 vector. Let x* denote the value of the vector x under H0. It will be assumed here that x* = 0, but
2293 note that the null hypothesis must give values not only to Xj but to all the components of jc,
2294 because the value of one component generally affects the measurement uncertainties of the other
2295 components of the solution vector. Generally, for this purpose one must use the measured values
2296 of all the components of x except xp although these values may not be known accurately.
2297 To determine the critical value, first calculate the vector y* =f(x*), which is the expected value of
2298 y under H0. If the least-squares problem is linear, then y* =Ax*. Next calculate the diagonal
2299 weight matrix W, whose /'th diagonal element is the inverse 1 / u2(y,) of the estimated variance of
2300 yt under the null hypothesis. For example, if the problem is the strontium problem described in
2301 Section 19B. 1, in which yt denotes a net count, then u2(y,) might be the counting variance given
2302 by
u2(y) = anx* + or.2x2* + RBj t\\+-i-\ (19.83)
2303 where RB t is the blank count rate and tB^ is the corresponding count time. Finally, evaluate the
2304 covariance matrix C for the solution of the least-squares problem f(x) = y* , as described in
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 19-103 DO NOT CITE OR QUOTE
-------
Measurement Statistics
2305 Section 19B.5. (The solution vector x here equals x* because of the method by which y* was
2306 constructed.) Then the critical value of the/h component x is zl _ aJc~., where zl _ a is the
2307 (1 - a)-quantile of the standard normal distribution.
2308 19B.7 Detection and Quantification Limits
2309 Computing the minimum detectable value of a component Xj requires one to find the value d such
2310 that d = zl _ a/F(0) + z\ - ^V(d), where Vfy) denotes the variance of the estimator x. as a func-
2311 tion of the true value Xj. The value of V(x^) is they* diagonal element of the covariance matrix C
2312 determined under the assumption that the true value of the/h component is Xj. Solving for d pre-
23 13 cisely generally requires an iterative algorithm, which generates a sequence of values converging
2314 to d. Given that V(x^) and its derivative can be calculated, the equation may be solved by Newton-
2315 Raphson iteration. A simpler version of fixed-point iteration, which does not involve the deriv-
2316 ative, may also be used. The use of fixed-point iteration for this purpose is described in Section
2317 19.7.
2318 The problem of determining the minimum quantifiable value of a concentration estimated by the
2319 least-squares methods is similar to that of finding the minimum detectable value and generally
2320 requires an iterative algorithm (e.g., see Section 19.7).
2321 19B.8 References
2322 Bevington, Philip R., and D. Keith Robinson. 1992. Data Reduction and Error Analysis for the
2323 Physical Sciences, 2nd ed. McGraw-Hill, New York, NY.
2324 Fuller, Wayne A. 1987. Measurement Error Models. John Wiley and Sons, New York, NY.
2325 Lawson, Charles L., and Richard J. Hanson. 1974. Solving Least Squares Problems. Prentice-
2326 Hall, Englewood Cliffs, NJ.
2327 Marquardt, D.W. 1963. An Algorithm for Least-Squares Estimation of Nonlinear Parameters.
2328 Journal of the Society for Industrial and Applied Mathematics 11(2): 431-441.
2329 Press, William H., et al. 1992. Numerical Recipes in C: The Art of Scientific Computing, 2nd ed.
2330 Cambridge University Press, New York, NY.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 19-104 DRAFT FOR PUBLIC COMMENT
-------
2331 ATTACHMENT 19C
2332 Estimation of Coverage Factors
2333 19C.1 Introduction
2334 Although it is common for laboratories to use a fixed coverage factor such as 2 or 3 when deter-
2335 mining an expanded uncertainty for a measured value, the true coverage probability for the resul-
2336 ting interval may be lower than expected if the standard uncertainties of the input estimates are
2337 determined from evaluations with too few degrees of freedom. This attachment summarizes a
2338 general method presented in Annex G of the GUM for determining appropriate coverage factors
2339 in these circumstances (ISO 1995). Section 19C.3 applies the method to Poisson counting
2340 uncertainties.
2341 19C.2 Procedure
2342 Assume the mathematical model for a measurement is Y =f[Xl,X2,... ,XN\ the input estimates
2343 Xj, x2, ..., % are independent, and the output estimate isy =J(xl,x2,...,%). Also assume that the
2344 combined standard uncertainty of y is not dominated by one component determined from a Type
2345 A evaluation with only a few degrees of freedom or from a Type B evaluation based on a distri-
2346 bution very different from a normal distribution. Then the distribution of the output estimate^
2347 should be approximately normal, and the following procedure may be used to obtain a coverage
2348 factor kp for the expanded uncertainty of y that gives a desired coverage probability/?.
2349 First compute the effective degrees of freedom veff of the measurement using the Welch-
2350 Satterthwaite formula
v
6ff " «fa) (19.84)
1 ~ 1 z
2351 Here w,(y) = \dy I dxt\ u(x^) is the component of the combined standard uncertainty generated by
2352 u(x,). If w(x;) is evaluated by a Type A method, then v, is the number of degrees of freedom for
2353 that evaluation. If w(x;) is evaluated instead by a Type B method, then v, is defined to be
_ 1 u\x) _ i
V' 2 rtfu(x\\ 2
-2
(19.85)
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 19-105 DO NOT CITE OR QUOTE
-------
Measurement Statistics
2354 where A.u(x^) is the estimated standard deviation of the standard uncertainty u(x^). Estimation of
2355 Aw(x;) often requires professional judgment.
2356 In some cases, one may consider the value of Aw(x;) for a Type B standard uncertainty to be zero
2357 or negligible, as for example when evaluating the uncertainty associated with rounding a number
2358 (Section 19.6.10). In such cases, one may assume v; = °°; so, the /th term of the sum appearing in
2359 the denominator of the Welch-Satterthwaite formula vanishes.
2360 The coverage factor kp is defined to be the (1 +p) 12-quantile t(l+p)i2(yeS) of a /-distribution with
2361 veff degrees of freedom.19 Since the calculated value of veff will generally not be an integer, it must
2362 be truncated to an integer, or else an interpolated /-factor should be used. That is, if
2363 n < veff < n + 1, then use either kp = /(i+/,)/2(veff) or
kP = (n + l ~ veff) '(i +,)/2(w) + Kff - ») '(i +P)/2(n + !) (19.86)
2364 The expanded uncertainty Up = kpuc(y) is estimated to have a coverage probability approximately
2365 equal top.
2366 19C.3 Poisson Counting Uncertainty
2367 As stated in Section 19.5.2.2, the standard uncertainty in the number of counts n observed during
2368 a radiation measurement may often be estimated by u(n) = Jn, according to the Poisson counting
2369 model. This method of evaluating the standard uncertainty is a Type B method; so, the effective
2370 degrees of freedom v for the evaluation should be determined from A.u(n). The standard deviation
2371 of Jn is always less than 0.65.20 If n is greater than about 10, the standard deviation of Jn is
19 The GUM uses the notation tp(v) to denote the (1 + p) /2-quantile of a/-distribution with v degrees of freedom
(ISO 1995), but the same notation in most statistical literature denotes the^-quantile (e.g., ISO 1993). MARLAP
follows the latter convention.
20 Taking the square root of a Poisson random variable is a common variance-stabilizing transformation, as
described in Chapter 20 of Experimental Statistics (NBS 1963). The stated (slightly conservative) upper bound for
the standard deviation of \fn is based on calculations performed at the EPA's National Air and Radiation Environ-
mental Laboratory, although the same approximate value may be determined by inspecting Figure 20-2 of NBS
1963. The precise calculation maximizes a functional) whose value is the variance of the square root of a Poisson
random variable with mean I. The first derivative of/is positive, decreasing, and convex between 1 = 0 and the
location of the maximum of the function at I = 1.31895; so, Newton's Method converges to the solutionfrom
below. The maximum value of/is found to be (0.642256)2.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 19-106 DRAFT FOR PUBLIC COMMENT
-------
Measurement Statistics
2372 approximately equal to 0.5, and, in this case, Equation 19.85 gives the estimate v ~ In. For
2373 smaller values of w, the same approximation is inadequate.
2374 MARLAP recommends that the standard uncertainty u(n) and degrees of freedom v for a Poisson
2375 measured value n be estimated by
u(n) = Jn and v = In (19.87)
2376 or, if very low counts are possible, by
u(n) = V^TTT and v = 2(n + 1) (19.88)
2377 If the expected count is greater than about 10, these formulas tend to give a coverage probability
2378 near the desired probability p. When the expected count is small, the coverage probability tends
2379 to be greater than p.
2380 Although the estimate u(n) = ^n + 1 may be derived by the Bayesian approach to counting statis-
2381 tics assuming a flat prior distribution for the mean count (Friedlander et al. 1981), the recom-
2382 mended expressions for u(n) and v in Equation 19.88 have been chosen for the purely practical
2383 reason that they are simple and seem to give satisfactory results. When the count is low, the
2384 assumptions underlying the Welch-Satterthwaite formula are usually violated, because the com-
2385 bined standard uncertainty is dominated by counting uncertainty, and the distribution of the count
2386 is not normal. However, even in this case, if the formula is used, the recommended expressions
2387 for u(n) and v tend to give conservative results.
2388 19C.4 References
2389 Friedlander, Gerhart, et al. 1981. Nuclear and Radiochemistry 3rd ed. John Wiley and Sons, New
2390 York, NY.
2391 International Organization for Standardization (ISO). 1993. Statistics - Vocabulary and Sym-
2392 bols - Part 1: Probability and General Statistical Terms. ISO 3534-1 . ISO, Geneva,
2393 Switzerland.
2394 International Organization for Standardization (ISO). 1995. Guide to the Expression ofUncer-
2395 tainty in Measurement . ISO, Geneva, Switzerland.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 19-107 DO NOT CITE OR QUOTE
-------
Measurement Statistics
2396 National Bureau of Standards (NBS). 1963. Experimental Statistics. NBS Handbook 91, National
2397 Bureau of Standards, Gaithersburg, MD.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 19-108 DRAFT FOR PUBLIC COMMENT
-------
2398 ATTACHMENT 19D
2399 Low-Background Detection Limits
2400 19D.1 Overview
2401 This attachment describes methods for determining critical values and minimum detectable con-
2402 centrations (MDCs) when the standard deviation of the blank signal is not known precisely,
2403 which occurs for example when the blank is measured by low-background Poisson counting or
2404 when the standard deviation is estimated from a small number of replicate measurements.
2405 19D.2 Calculation of the Critical Value
2406 The critical value of the net signal Sc was defined earlier by the relation
Pr[S>Sc X=0]=a (19.89)
2407 When the signal assumes only discrete values (e.g., numbers of counts), there may be no value Sc
2408 that satisfies Equation 19.89 exactly. The critical value in this case is defined as the smallest
2409 value Sc such that Pr[S > Sc X = 0] < a.
2410 19D.2.1 Normally Distributed Signals
2411 If the distribution of the net signal S under H0 is approximately normal with a well-known stan-
2412 dard deviation, o0, the critical value of S is
SC=Zl-a?0 (19.90)
2413 where z1_a denotes the (1 - a)-quantile of the standard normal distribution. Typically the stan-
2414 dard deviation o0 is not well-known and must therefore be replaced by an estimate, 60. If 60 is
2415 determined by a statistical evaluation with v degrees of freedom, the multiplier zl _ a should be
2416 replaced by ^_a(v), the (1 - a)-quantile of the ^-distribution with v degrees of freedom (cf. Type
2417 A evaluation of standard uncertainty in Section 19.5.2.1). Thus,
^c = 'i-a(v)°o (19.91)
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 19-109 DO NOT CITE OR QUOTE
-------
Measurement Statistics
2418 Table G.2 in Appendix G lists values of ^_a(v). In general, ^_a(v) is greater than z1_a, but the two
2419 values are approximately equal if v is large.
/*,
2420 When B is estimated by the average of n replicate blank measurements (assuming no interfer-
2421 ences), the standard deviation 60 of the net signal S under the null hypothesis may be estimated
2422 from the experimental standard deviation of the measured blank values, SB. Specifically,
G0 =
1+1
n
(19.92)
2423
The number of degrees of freedom, v, in this case equals n - 1; so, the critical value of S is
(19.93)
n
2424 19D.2.2 Poisson Counting
2425 It is assumed here, as in Section 19.7, that the instrument is a radiation counter and the instru-
2426 ment signal is the gross count. Therefore,
= N
(19.94)
2427 and the net instrument signal is the net count, defined as
2428
2429
2430
243 1
2432
2433
where
Ns
N
(19.95)
is the gross count (source count)
is the blank count
is the estimated count rate due to interferences
is the count time for the test source
is the count time for the blank
MARLAP
DO NOT CITE OR QUOTE
19-110
JULY 2001
DRAFT FOR PUBLIC COMMENT
-------
Measurement Statistics
2434 If the mean blank count rate, RB, is well-known and there are no interferences, then according to
2435 the Poisson model, the critical gross count, yc, equals the smallest nonnegative integer n such that
k\
> 1 -a
(19.96)
2436 Then Sc, the critical net count, equals^ - NBts I tB. Table 19.4 shows critical gross counts for
2437 a = 0.05 for small values ofRBts (adapted from NRC 1984).21 To use the table, one calculates the
2438 value ofRBts, finds the appropriate line in the table, and compares the observed gross count Ns to
2439 the value ofyc read from the table. The analyte is considered detected if and only ifNs>yc.
2440 When RBts is greater than about 20, yc may be approximated by
(19.97)
2441 where zl_a denotes the (1 - a)-quantile of the standard normal distribution, and \x] denotes the
2442 largest integer not greater than x.
TABLE 19.4 — Critical gross count (well-known blank)
RB*S
0.000-0.051
0.051-0.355
0.355-0.818
0.818-1.366
1.366-1.970
1.970-2.613
2.613-3.285
3.285-3.981
3.981-4.695
4.695-5.425
Jc
0
1
2
3
4
5
6
7
8
9
RB*S
5.425-6.169
6.169-6.924
6.924-7.690
7.690-8.464
8.464-9.246
9.246-10.036
10.036-10.832
10.832-11.634
11.634-12.442
12.442-13.255
Jc
10
11
12
13
14
15
16
17
18
19
RB*S
13.255-14.072
14.072-14.894
14.894-15.719
15.719-16.549
16.549-17.382
17.382-18.219
18.219-19.058
19.058-19.901
19.901-20.746
20.746-21.594
Jc
20
21
22
23
24
25
26
27
28
29
21 The breaks in the table occur at RBts = 0.5 XQ 05 (2>c) and 0.5 Xo os (2yc + 2).
JULY 2001
DRAFT FOR PUBLIC COMMENT
19-111
MARLAP
DO NOT CITE OR QUOTE
-------
Measurement Statistics
2443 When the blank count rate RB is low, which is often true for alpha counting, measuring its value
2444 with good relative precision tends to be difficult, especially if the instrument background tends to
2445 drift. However, a conservative bound, such as a 1 - a upper confidence limit, may be used if one
2446 wishes to limit type I error rates and is willing to tolerate the resulting higher detection limits.
2447 More commonly used methods for calculating the critical value are described below.
2448 THE POISSON-NORMAL APPROXIMATION
2449 As stated in Section 19.7.1.2, when Poisson counting statistics are assumed (possibly with
2450 additional variance components) and the instrument background remains stable between meas-
2451 urements at a level where the Poisson distribution is approximately normal, the critical net count
2452 is given approximately by the equation
Z\-atS
\
RV + RT Rf> 9 0 ~
/ . 5 . C2 . _2/^^ (19.98)
t
s
2453 where RB denotes the (true) mean count rate of the blank, Rj denotes the mean interference count
2454 rate, ^ denotes non-Poisson variance in the blank (count rate) correction, and o2^) denotes the
2455 variance of the estimator for Rj. When there are no interferences and no non-Poisson blank
2456 variance, this equation becomes
^
(19.99)
2457 Low mean blank levels cause the Poisson distribution to deviate from the normal model. Figure
2458 19.12 shows the effects of these deviations on the type I error rates for the Poisson-normal
2459 approximation when tB = ts and a = 0.05. The graph has discontinuities because of the discrete
2460 nature of the Poisson distribution, but the type I error rate is approximately correct (equal to 0.05)
2461 when the mean blank count is 10 or more.22
! Probabilities on the curve are calculated using the equation
where \a denotes the (true) mean blank count. Terms of the infinite sum are accumulated until the cumulative
Poisson probability, e ~M£"=0 u! / /'!, approaches 1. The calculated values agree with those listed in Table 1 of
MARLAP JULY 2001
DO NOT CITE OR QUOTE 19-112 DRAFT FOR PUBLIC COMMENT
-------
Measurement Statistics
P
0.20-
0.15-
0.10-
0.05-
0.00
0 5 10 15 20 E(NB)
FIGURE 19.12 — Type I error rate for the Poisson-normal approximation (tB = ts)
2462
2463
2464
2465
2466
2467
2468
2469
2470
In Equation 19.99, RB denotes the true mean blank count rate, which can only be estimated. In
practice, one must substitute an estimated value, RB, as shown in the following equation.
l\
(19.100)
The most frequently used expressions for Sc may be derived from Equation 19.100 using an
estimator RB that equals a weighted average of the measured blank count rate NBI tB and the
measured source count rate Ns I ts. A weighted average of both measured rates may be used here
to estimate the true blank level for the purpose of the hypothesis test, because, under the null
hypothesis of zero net source activity, both measured rates are unbiased estimates of the true
blank count rate. Given nonnegative weights ws and WB such that ws + WB = 1, the mean blank
count rate is estimated by
4=ws-T+w5-T (19.101)
Brodsky 1992. The discontinuities occur at \a = 1? 12.332 for k = 1, 2, 3, ....
JULY 2001
DRAFT FOR PUBLIC COMMENT 19-113
MARLAP
DO NOT CITE OR QUOTE
-------
Measurement Statistics
2471 This estimate RB is always unbiased under the null hypothesis of zero net activity and no inter-
2472 ferences, but the choice of weights affects the variance of the estimator. (When interferences are
2473 present, this weighted average is inappropriate.)23
2474 This attachment will use the notation Sc, which is nonstandard, to denote any version of the
2475 critical value that depends on the gross signal Ns (or 7).
2476 It is often convenient to eliminate Ns from the expression for Sc (e.g., when calculating the
2477 MDC). When the same measured value ofNB is used to calculate both the critical value Sr and
/*, '—
2478 the net signal S, elimination ofNs from Equation 19.100 produces the following formula for an
2479 alternative critical value Sr24
Zl-aWS
1 +-
•1-a
\
l-aWS
1 +JL +N^\ 1 +-i
(19.102)
2480 It is not generally true that Sc = Sc unless ws = 0, but either critical value may be used to imple-
2481 ment the same test for analyte detection, because S > Sc if and only if S > Sc.
2482 If there is additional non-Poisson variance associated with the blank correction, an extra term
2483 may be included under the radical (e.g., ^t^, where ^ is as in Equation 19.98), although at very
2484 low background levels the Poisson variance tends to dominate this excess component.
2485 FORMULA A
2486 The most commonly used approach for calculating Sc is given by Formula A (shown below).
23 The common practice of using the same Poisson measurement data to calculate both the net signal S and its
critical value tends to produce a correlation between the two variables. This correlation does not exist when the
critical value is determined by a statistical evaluation of normally distributed data as described earlier in the
attachment.
24 The critical value Sc may be written as a function ftS) of the observed net signal S and the blank count NB.
Then S exceeds Sc if and only if it exceeds the fixed point off, which is the value Sc where J(SC) = Sc. The fixed
point is a function ofNB but not ofNs.
MARLAP
DO NOT CITE OR QUOTE
19-114
JULY 2001
DRAFT FOR PUBLIC COMMENT
-------
Measurement Statistics
*c = -
Formula A
(19.103)
2487
2488
2489
2490
2491
2492
2493
2494
If a = 0.05 and tB = ts, Formula A leads to the well-known expression 2.33 JWB for the critical net
count (e.g., see Currie 1968).
Formula A may be derived from the standard approximation by using the blank measurement
alone to estimate the true blank count rate — i.e., by using the weights ws = 0 and WB = 1.
As noted in Section 19.7.1.2, when the blank count is high (e.g., 100 or more), Formula A works
well, but at lower blank levels, it can produce a high rate of type I errors. Figure 19.13 shows
type I error rates for Formula A as a function of the mean blank count for count time ratios
tR/tv=\ and 5 when a = 0.05
25
0.00
0 5 10 15 20 E(NB)
FIGURE 19.13 — Type I error rates for Formula A
' Probabilities on the two curves are calculated using the equation
= 1 -e
k\
where yc(n) = n(ts/tB) + 1.645Jn (ts I tB) (1 + ts I tB) and \i denotes the mean blank count. The same equation with
different expressions for>-c(w) is used to calculate the type I error rates shown in Figures 19.14-17.
JULY 2001
DRAFT FOR PUBLIC COMMENT
19-115
MARLAP
DO NOT CITE OR QUOTE
-------
Measurement Statistics
2495
FORMULAE
2496 Another published formula for the critical value is (equivalent to) the following (Nicholson
2497 1966).
\-a
(19.104)
2498 The critical value calculated by Equation 19.104 equals zl_a times the combined standard uncer-
2499 tainty of the net count. This fact is the basis for the original derivation of the formula, but the
2500 formula may also be derived from Equation 19.100 using the weights ws = tB I (ts + tB) and WB =
2501 ts I (ts + tB) to estimate RB. When Ns is eliminated from Equation 19.104, one obtains Formula B
2502 (below), which is equivalent to the equation for the critical value given in Atoms, Radiation, and
2503 Radiation Protection (Turner 1995).
•l-a
H
2 /
Zl -a AT 1S 1 !
4 + sr
H '5 V
Formula B
^s 1
IB)
(19.105)
2504 Type I error rates for Formula B are shown in Figure 19.14.
P
0.15-
0.10-
0.05
= 5t<,
5 10 15 20 E(NB)
FIGURE 19.14 — Type I error rates for Formula B
MARLAP
DO NOT CITE OR QUOTE
19-116
JULY 2001
DRAFT FOR PUBLIC COMMENT
-------
Measurement Statistics
2505 Formula B appears natural and intuitive when it is derived in terms of the combined standard
2506 uncertainty of the net count, and it gives excellent results when tB = ts and the pure Poisson
2507 model is valid. However, when the formula is derived using the weights ws and WB, as described
2508 above, the expression seems much less natural, because the weights clearly are not optimal when
2509 tB* ts. Notice that when tB > ts, the type I error rate tends to be less than a.
2510 FORMULA C
2511 If the pure Poisson model is valid, then under the null hypothesis, the weights ws=tsl (ts + tB)
2512 and WB = tB I (ts + tg) provide the minimum-variance unbiased estimator RB for the mean blank
2513 count rate and lead to the following formula for the critical net count (Nicholson 1963, 1966).26
l\
t
B
2514 Elimination of Ns from Equation 19.106 produces Formula C, shown below.
2tB ^ 4tj tB{ tB) (19.107)
Formula C
2515 Formula C is equivalent to the equation for the "decision threshold" given in Table 1 of ISO
2516 11929-1 (ISO 2000a) for the case of fixed-time counting. Figure 19.15 shows type I error rates
2517 for Formulae.
26 The approach here is conceptually similar to that of a two-sample /-test, which employs a pooled estimate of
variance in the comparison of two normal populations.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 19-117 DO NOT CITE OR QUOTE
-------
Measurement Statistics
0 5 10 15 20 E(NB)
FIGURE 19.15 — Type I error rates for Formula C
2518 If the blank correction involves additional non-Poisson variance, an extra term may be included
2519 under the radical in Formula C; however, the weights ws and WB used to derive the formula are
2520 not necessarily optimal in this case. (See ISO 2000b for another approach.)
2521 Note that Formulas B and C are equivalent when tB = ts, because both assign equal weights to the
2522 blank measurement and the source measurement. In this case, both formulas are also equivalent
2523 to the formula given by Altshuler and Pasternack (1963).
2524 THE STAPLETON APPROXIMATION
2525
2526
2527
2528
When the mean counts are low and tB * ts, another approximation formula for Sc appears to out-
perform all of the approximations described above. For small values of the constant d, the
statistic
Z = 2
Ns+d
\
\
(19.108)
2529 which involves variance-stabilizing transformations of the Poisson counts Ns and NB, has a distri-
2530 bution that is approximately standard normal under the null hypothesis (Stapleton 1999). So, the
2531 critical value of Zis zl_(V the (1 - a)-quantile of the standard normal distribution. From these
2532 facts one may derive the following expression for the critical net count as a function ofNB.
MARLAP
DO NOT CITE OR QUOTE
19-118
JULY 2001
DRAFT FOR PUBLIC COMMENT
-------
Measurement Statistics
Sc = d\-*--l\ + ^
(*B ) 4
•l-a.
(19.109)
The Stapleton Approximation
2533 When a = 0.05, the value d= 0.4 appears to be a near-optimal choice. Then for tB = ts, the
2534 Stapleton approximation gives the equation
Sc= 1.35 + 2.33
0.4
(19.110)
2535 Figure 19.16 shows the type I error rates for the Stapleton approximation when a = 0.05 and
2536 d= 0.4. This approximation gives type I error rates almost identical to those of Formulas B and C
2537 when tB = ts, but it has an advantage when tB* ts.
P
0.15-
0.10-
0.05-----;.
0.00
= 5t<,
0 5 10 15 20 E(NB)
FIGURE 19.16 — Type I error rates for the Stapleton approximation
2538 When a * 0.05, the value d= zl_a14.112 appears to give good results (4.112 = z095 / 0.4).
2539 When the blank correction involves a small non-Poisson variance component, a term (Z^tj) may
2540 be included under the radical in Equation 19.109 to account for it.
JULY 2001
DRAFT FOR PUBLIC COMMENT
19-119
MARLAP
DO NOT CITE OR QUOTE
-------
Measurement Statistics
2541
THE EXACT TEST
2542 Poisson counting statistics also permit an "exact" test for analyte detection, whose type I error
2543 rate is guaranteed to be no greater than the chosen value of a, although it may be less. A random-
2544 ized version of the test can provide a type I error rate exactly equal to a (Nicholson 1963), but
2545 only the nonrandomized version will be considered here, since its outcome is always based solely
2546 on the data and not on a random number generator. The test is implemented by rejecting H0 if and
2547 only if the following inequality is true.27
2548 Nicholson presents the test as a comparison of the gross count Ns to a critical value. The critical
2549 value yc is the smallest nonnegative integer n such that28
*l-a
(19.112)
2550 The same (nonrandomized) test is implemented by calculating a critical gross count yc equal to
2551 the smallest nonnegative integer n such that
27 The left-hand side of the inequality is a cumulative binomial probability (see Attachment 19A). It also equals
where Ix(a,b) denotes the incomplete beta function (NBS 1964, Press et al. 1992).
28 To implement the randomized test, calculate the critical value yc, and, if Ns > yc, reject H0, as in the non-
randomized test. If Ns = yc, calculate a rejection probability P by subtracting 1 - a from the sum on the left-hand
side of the inequality (with n = Ns) and dividing the difference by the summation's last term
Then reject H0 with probability P.
MARLAP
DO NOT CITE OR QUOTE
19-120
JULY 2001
DRAFT FOR PUBLIC COMMENT
-------
Measurement Statistics
2552 Then the critical net count Sc equals yc - NB (ts I tB). (Note that Inequality 19.113 is intended for
2553 use when NB is small.) Table G.4 in Appendix G lists critical values ^ for a = 0.01 and 0.05 and
2554 for integral values of the count time ratio tB I ts ranging from 1 to 5.
2555 Figure 19.17 shows the type I error rates for the nonrandomized exact test. (The type I error rate
2556 for the randomized version of the test equals 0.05 everywhere.)
P
0.15 H
0.10-
0.05--
0.00
0 5 10 15 20 E(NB)
FIGURE 19.17 — Type I error rates for the nonrandomized exact test
2557
2558
2559
2560
2561
2562
2563
EXAMPLE
Problem: A 6000-s blank measurement is performed on a proportional counter and 108 beta
counts are observed. A test source is to be counted for 3000 s. Estimate the critical value of the
net count when a = 0.05.
Solution: Formula A gives the result
= 1.645
\
108
= 14.8 counts.
Formula B is not recommended.
3000
6000;
1 +
3000
6000,
JULY 2001
DRAFT FOR PUBLIC COMMENT
19-121
MARLAP
DO NOT CITE OR QUOTE
-------
Measurement Statistics
2564
2565
2566
2567
2568
2569
2570
2571
Formula C gives the result
2t
zl
-a
^
*lVs tS
2 -° /
(i + 's)
V 'SJ
_1.6452(3000)+1615
2(6000)
15.5 counts.
1.6452(3000)2 1ft_|3000
+ 10o
4(6000^
6000
The Stapleton approximation (with d= 0.4) gives the result
= 0.4 M222-1
^ 6000
= 15.6 counts.
(108 +0.4)
3000
6000,
1 +
3000
6000;
The exact test gives the result yc = 70 counts (the entry in Table G.4 for a = 0.05, tBl ts = 2,
and NB = 108), which implies that
SC = 1Q- (108)(3000 / 6000) = 16 counts.
COMPARISONS
2572 Although Formula A gives the highest type I error rates of all the formulas described above in the
2573 pure Poisson counting scenario, it is the formula that can be adapted most easily for dealing with
2574 interferences. It can also be modified to reduce the very high type I error rates at low blank levels
2575 (by adding 1 or 2 to the number of blank counts NB under the radical). Formula B cannot be
2576 recommended. When the pure Poisson model is valid, Formula C gives better results than either
2577 A or B, but the Stapleton approximation appears to give the most predictable type I error rates of
2578 all. Nicholson's exact test is the most complicated of the tests and requires either software or
2579 lookup tables to be practical, but it is the only one of the tests whose type I error rate is guaran-
2580 teed not to exceed the chosen significance level. Achieving the chosen significance level exactly
2581 appears to require the randomized version of Nicholson's test.
MARLAP
DO NOT CITE OR QUOTE
19-122
JULY 2001
DRAFT FOR PUBLIC COMMENT
-------
Measurement Statistics
2582 19D.3 Calculation of the Minimum Detectable Concentration
2583 The minimum detectable concentration, or MDC, was defined earlier as the concentration of
2584 analyte, XD, that must be present in a laboratory sample to give a probability 1 - P of obtaining a
2585 measured response greater than its critical value. Equivalently, the MDC is defined as the analyte
2586 concentration XD that satisfies the relation
= xD\=$ (19.114)
2587 where the expression Prf^ < Sc \ X = XD] may be read as "the probability that the net signal S
2588 does not exceed its critical value Sc when the true concentration Xis equal to XD"
2589 19D.3.1 The Minimum Detectable Net Instrument Signal
2590 The MDC may be estimated by calculating the minimum detectable value of the net instrument
2591 signal, SD, and converting the result to a concentration. The minimum detectable value of the net
2592 instrument signal is defined as the mean value of the net signal that gives a specified probability
2593 1 - P of yielding an observed signal greater than its critical value Sc. Thus,
= ^]=P (19.115)
2594 where S denotes the true mean net signal.
2595 19D.3.2 Normally Distributed Signals
2596 If the net signal S is normally distributed and its estimated standard deviation 60 under H0 is
2597 determined from a statistical evaluation with v degrees of freedom (e.g., n = v + 1 replicate blank
2598 measurements), then the critical value of S is
SC = tl-a(V^0 (19.116)
___ /v
2599 Then, if the variance of S is constant at all concentrations, the minimum detectable value of the
2600 signal is given by
^ = 5o,p,vGo (19.117)
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 19-123 DO NOT CITE OR QUOTE
-------
Measurement Statistics
2601 where 5^p v denotes the non-centrality parameter of the non-central ^-distribution with v degrees
2602 of freedom. The parameter 5^p v is such that
'p'(vAv) = 'i-a(v) (19.H8)
2603 where ^(^5^) denotes the p-quantile of the non-central ^-distribution. The non-centrality
2604 parameter 5^p v may be approximated by
2605 which is based on an approximation for the non-central t distribution function (NBS 1964).
2606 When a = P = 0.05 and v > 4, the non-centrality parameter is also approximated adequately by
2607 f0-95(v) x 8v / (4v + 1) (Currie 1997).
2608 Conceptually the standard deviation 60 used to calculate the critical value Sc is only an estimate
2609 and therefore can be considered a random variable. If it were the true standard deviation, the cor-
2610 rect multiplier used to calculate Sc would be zl _ a, not tl _ a(v). However, the standard deviation
2611 used to calculate SD is, conceptually at least, the true standard deviation o0, even if its value is not
2612 known exactly. The true standard deviation may be estimated by 60, but since the estimator 60 is
2613 biased, a correction factor should be used for v less than about 20. An unbiased estimator for o0 is
2614 G0/c4, where
Wv + i\ r
2
- (19.120)
v
2615 and where F denotes the gamma function (NBS 1964). The gamma function is easily computed
2616 in software (Press et al. 1992), but c4 is also approximated well by 4v / (4v + 1), and values of c4
2617 are commonly tabulated in references for statistical quality control (whence the notation c4 is
2618 borrowed). Then SD is estimated by
c s 6o
^ = 5o,Bv— (19.121)
*p,
MARLAP JULY 2001
DO NOT CITE OR QUOTE 19-124 DRAFT FOR PUBLIC COMMENT
-------
Measurement Statistics
TABLE 19.5 — Bias factor for the experimental standard deviation
V
1
2
3
4
5
6
7
8
9
10
C4
0.79788
0.88623
0.92132
0.93999
0.95153
0.95937
0.96503
0.96931
0.97266
0.97535
V
11
12
13
14
15
16
17
18
19
20
C4
0.97756
0.97941
0.98097
0.98232
0.98348
0.98451
0.98541
0.98621
0.98693
0.98758
V
21
22
23
24
25
26
27
28
29
30
C4
0.98817
0.98870
0.98919
0.98964
0.99005
0.99043
0.99079
0.99111
0.99142
0.99170
V
31
32
33
34
35
36
37
38
39
40
C4
0.99197
0.99222
0.99245
0.99268
0.99288
0.99308
0.99327
0.99344
0.99361
0.99377
2619 which is approximately 2 tOS5(v)60, or 2SC, when a = P = 0.05 and v > 4. Values of c4 for v = 1 to
2620 40 are listed in Table 19.5.
2621 Lower and upper confidence limits for SD may be calculated using the equations
2622
2623
2624
2625
, lower "o,p,v
and
Cupper = 5o,P,v
(19.122)
where ^(v) denotes the/?-quantile of the chi-square distribution with v degrees of freedom and y
denotes the desired confidence coefficient (see Table G.3 in Appendix G).
/*,
If the variance of S is not constant but increases with the mean signal S, the minimum detectable
net signal is determined implicitly by the equation
S
D
JD>
(19.123)
JD
2626 where CD denotes the standard deviation of S when S = SD. An iterative algorithm, such as the
2627 one shown below, may be needed to solve the equation for SD.
2628
Set GO =
JULY 2001
DRAFT FOR PUBLIC COMMENT
19-125
MARLAP
DO NOT CITE OR QUOTE
-------
Measurement Statistics
2629
2630
2631
2632
2633
2634
2635
2636
2.
3.
4.
5.
6.
7.
8.
9.
Se, SD
repeat
Set c
Find
= WVK
/ __2/ o 1 o o \
D ~ \ \ 1 ~ jD-'
the value of 5 such that fy(v, 5) = ^ _a(v) o0 / CD
Set h = SD
Sett
until \l
output
1D = dcD
>D - h \ is sufficiently small
the solution SD
2637 The value of the non-centrality parameter 5 in Step 5 may be approximated by
t' = t ,re(v) — (19.124)
i-LUz.
4V;
2v
2638 When G0 is determined by any means other than a statistical evaluation, SD must be calculated
2639 differently.
2640 19D.3.3 Poisson Counting
2641 Another equation for SD, which was described in Section 19.7.2.2, is
77^n) (19.125)
2642 where Sc = zl_ ao0 and c2(S \S = SD) denotes the variance of the measured signal S when the true
2643 mean signal S equals SD. This equation is the basis for formulas that are commonly used for SD
2644 when the Poisson-normal approximation is assumed. Regardless of whether the signal follows
2645 the pure Poisson model or has non-Poisson variance, the function o2(^ | S = SD) can often be
2646 expressed in the form
c\S) = aS2 + bS + c (19.126)
MARLAP JULY 2001
DO NOT CITE OR QUOTE 19-126 DRAFT FOR PUBLIC COMMENT
-------
Measurement Statistics
2647 where S denotes the true mean net signal and the constants a, b, and c do not depend on S. In this
2648 case, the minimum detectable net signal is given approximately by
bS
aS
(19.127)
9
where /„ = 1 - zl_»a.
2649
2650 Equation 19.125 is often used even when Sc is calculated using one of the formulas presented
2651 above for low-background Poisson counting, with RB tB substituted for the blank count NB, but in
2652 this case SD may be underestimated because of the fact that the calculated value of Sc varies from
2653 measurement to measurement. One option for obtaining a more conservative estimate of SD is to
2654 substitute a conservative value of Sc, which will be denoted here by [Sc]. For Poisson counting,
2655 one method of obtaining [Sc] is to use the value of Sc calculated from the largest blank count NB
2656 likely to be observed, given the assumed mean blank count rate RB (e.g., use Table 19.4 with RB tB
2657 replacing RB ts and NB replacing_yc in the column headings). To calculate SD, one may substitute
2658 [Sc] for Sc in Equation 19.127.
2659 Note that [Sc] is not used to make detection decisions. It is used only to calculate SD.
2660 For example, suppose a = P = 0.05, the assumed mean blank count rate is RB = 8 x icr4 cps, and
2661 the blank count time is tB = 6000 s. Then RBtB = 4.8 counts. Using Table 19.4, one finds 4.8 in
2662 the first column between 4.695 and 5.425, and reads the value 9 from the second column. So, 9 is
2663 the largest value ofNB likely to be observed when measuring a blank. Now, if Stapleton's
2664 approximation is used to calculate Sc when making a detection decision, the value of [Sc] used
2665 to calculate SD is given by the following equation.
2666
[Sc\ = 0.4
^-1
1.6452
+ 1.645
^
(9+0.4)-!
tn
1+-£|
(19.128)
2667 So, if ts = tB, then [Sc] = 8.48 counts. lfRBtB (4.8 counts) were used as the blank count instead,
2668 [Sc] would be only 6.66 counts.
JULY 2001
DRAFT FOR PUBLIC COMMENT
19-127
MARLAP
DO NOT CITE OR QUOTE
-------
2669
Measurement Statistics
PURE POISSON COUNTING
2670 When the pure Poisson model is assumed and Formula A is used for the critical value, if the
2671 critical value, Sc, is determined from a sufficiently large total number of counts and if a = P, the
2672 minimum detectable net signal SD is given by the following simple equation.
SD=Zl-&+2S<
C
(19.129)
2673
2674
2675
More generally, if Formula A or C is used to calculate the critical net count Sc, then SD may be
determined from Equation 19. 127 using the following values for a, b, and c.
2676 The resulting formula for SD is
SD = SC
2
1 P
2
+ZIA
zi%+ +p,d.^
4 + c +
sn j
(19.130)
2677 As previously noted, counting data never follow the Poisson model exactly. Variable factors such
2678 as source geometry and placement, counting efficiency, and subsampling variance tend to
2679 increase a, while interferences and background instability tend to increase c.
2680 THE STAPLETON APPROXIMATION
2681 When the Stapleton approximation is used for Sc, the minimum detectable net count SD may be
2682 calculated using Equation 19.130, but when the Poisson model is valid, a better estimate is given
2683 by the formula
(19.131)
MARLAP
DO NOT CITE OR QUOTE
19-128
JULY 2001
DRAFT FOR PUBLIC COMMENT
-------
Measurement Statistics
2684 Equation 19.131 also gives a better approximation of SD even when Formula C is used for the
2685 critical value as long as the ratio of count times tB I ts is not too far from 1 (see Table 19.6). It is
2686 recommended by ISO 11929-1 (ISO 2000a) in a slightly different but equivalent form.
2687 When a = P = 0.05 and tB = ts, the preceding equation becomes
SD = 5.41 + 4.65jR^Ts (19.132)
2688 The Stapleton approximations for Sc and SD give very predictable type I and type II errors when
2689 the only measurement variance is Poisson.
2690 When the Poisson model is incomplete because of excess relative variance (a > 0), one can use
2691 Equation 19.127 with appropriate values for a, b, and c. However, a somewhat better estimate of
2692 SD can be obtained. The calculation is more involved.
2a/2
2693 where
-2aV + W^_4oV_^ (19m)
2694 a' = 1 -
2695
2
-1-f
1 + —
2
t*
2696 c' = Rf. + -^ L£ !+_£
ts}
1 + —
2697 PRECISE CALCULATION OF SD
2698 When the Poisson model is valid, the mean blank count rate RB and the analyte detection criteria
2699 completely determine SD. So, in principle, a computer program can be written to calculate SD
2700 precisely. The calculation is most easily described when the critical net count is expressed in
2701 terms of NB but not Ns (e.g., Sc as defined by Formulas A-C, the Stapleton approximation, and
2702 the exact test). Then, at any specified value S of the mean net signal, the power of the detection
2703 test can be computed using the expression:
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 19-129 DO NOT CITE OR QUOTE
-------
Measurement Statistics
BBB^
Power = 1 - ^(-RB(ts + tB}-S) E ^~ E (19-134)
2704 where yc(n) denotes the value of yc (or Sc + NB ts I tB) when NB = n. Terms of the infinite sum
2705 must be accumulated only until the cumulative Poisson probability, e B BY."m=0(RBtB)m /ml,
2706 approaches 1. Given a software procedure to compute Equation 19.134, the value of SD may be
2707 determined using an iterative algorithm, such as Newton's method or bisection, which calculates
2708 the power at various trial values of S until the correct value is found where the power equals
2709 1 - p (e.g. see Burden and Faires 1993).
2710 A procedure of the type described above generated the true values of SD for Table 19.6, which
2711 shows both the estimated and true values of SD obtained when Formulas A and C and the
2712 Stapleton approximation are used for the critical value. The estimated values of SD in this table
2713 are based on values of Sc calculated using the true mean net count, not the upper bound [NB]. The
2714 use of [NB] would produce larger estimates.
2715 PRECISE CALCULATION OF XD
2716 Suppose the analyte concentration X is calculated by dividing the net signal S by the sensitivity A,
2717 where A varies considerably or there is considerable subsampling variance, but the signal is
2718 otherwise adequately described by the Poisson model. If one can assume that A has a particular
2719 distribution, such as a rectangular or triangular distribution, then it is possible to calculate XD pre-
2720 cisely in software, although the mathematics is less straightforward than that needed to calculate
2721 SD in the preceding section. At any specified concentration x, the detection power equals
oo (R t }" M")J
Power = l-RBts + (Mx + 8)x) -Pfc+l,RBts + (^ - 8)x)
2724 where P(-, •) denotes the incomplete gamma function. Other combinations of the incomplete
2725 gamma function appear when different polygonal distributions are assumed (e.g., triangular).
MARLAP JULY 2001
DO NOT CITE OR QUOTE 19-130 DRAFT FOR PUBLIC COMMENT
-------
Measurement Statistics
TABLE 19.6 — Estimated and true values of SD (tB = ts)
Mean Blank
Count
0
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
Formula A
Estimated True
2.706
7.358
9.285
10.764
12.010
13.109
14.101
15.015
15.864
16.663
17.418
18.136
18.822
19.480
20.113
20.724
21.315
21.888
22.444
22.985
23.511
2.996
8.351
10.344
11.793
13.021
14.091
15.076
16.028
16.945
17.804
18.595
19.324
20.002
20.642
21.257
21.854
22.438
23.010
23.569
24.116
24.649
Formula C
Estimated True
7.083
9.660
11.355
12.719
13.894
14.942
15.897
16.780
17.605
18.383
19.120
19.823
20.496
21.142
21.764
22.366
22.948
23.513
24.062
24.596
25.116
6.296
10.095
12.010
13.551
14.826
15.930
16.902
17.785
18.614
19.406
20.170
20.903
21.602
22.267
22.900
23.506
24.091
24.657
25.206
25.738
26.252
Stapleton
Estimated True
5.411
10.063
11.991
13.469
14.716
15.814
16.807
17.720
18.570
19.368
20.123
20.841
21.527
22.185
22.819
23.430
24.020
24.593
25.149
25.690
26.217
6.296
10.095
12.010
13.551
14.826
15.930
16.902
17.785
18.614
19.406
20.170
20.903
21.602
22.267
22.900
23.506
24.091
24.657
25.206
25.738
26.252
2726 A precise power calculation of this type was performed to evaluate the results derived in the
2727 example in Attachment 19E assuming an approximately normal distribution for the subsampling
2728 error. The assumption of a normal distribution is nonsensical unless the relative standard devia-
2729 tion of A is small (because^ is positive), and in the latter case, the assumption of a triangular
2730 distribution, or even a rectangular distribution, gives approximately the same result.
JULY 2001
DRAFT FOR PUBLIC COMMENT
19-131
MARLAP
DO NOT CITE OR QUOTE
-------
Measurement Statistics
2731 19D.4 References
2732 Altshuler, Bernard, and Bernard Pasternack. 1963. Statistical Measures of the Lower Limit of
2733 Detection of a Radioactivity Counter. Health Physics 9: 293-298.
2734 Brodsky, Allen. 1992. Exact Calculation of Probabilities of False Positives and False Negatives
2735 for Low Background Counting. Health Physics 63(2): 198-204.
2736 Burden, Richard L., and J. Douglas Faires. 1993. Numerical Analysis, 5th ed. PWS Publishing
2737 Company, Boston, MA.
2738 Currie, Lloyd A. 1968. Limits for Qualitative Detection and Quantitative Determination:
2739 Application to Radiochemistry. Analytical Chemistry 40(3): 586-593.
2740 Currie, L.A. 1997. Detection: International Update, and Some Emerging Di-lemmas Involving
2741 Calibration, the Blank, and Multiple Detection Decisions. Chemometrics and Intelligent
2742 Laboratory Systems 37: 151-181.
2743 International Organization for Standardization (ISO). 1997. Capability of Detection - Part 1:
2744 Terms and Definitions. ISO 11843-1. ISO, Geneva, Switzerland.
2745 International Organization for Standardization (ISO). 2000a. Determination of the Detection
2746 Limit and Decision Threshold for Ionizing Radiation Measurements - Part 1: Fundamentals
2747 and Application to Counting Measurements without the Influence of Sample Treatment. ISO
2748 11929-1. ISO, Geneva, Switzerland.
2749 International Organization for Standardization (ISO). 2000b. Determination of the Detection
2750 Limit and Decision Threshold for Ionizing Radiation Measurements - Part 2: Fundamentals
2751 and Application to Counting Measurements with the Influence of Sample Treatment. ISO
2752 11929-2. ISO, Geneva, Switzerland.
2753 International Union of Pure and Applied Chemistry (IUPAC). 1995. Nomenclature in Evaluation
2754 of Analytical Methods Including Detection and Quantification Capabilities. Pure and Applied
2755 Chemistry 67(10): 1699-1723.
2756 Lochamy, Joseph C. 1976. The Minimum Detectable Activity Concept. NBS Report No. NBS-
2757 SP456, National Bureau of Standards, Gaithersburg, MD.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 19-132 DRAFT FOR PUBLIC COMMENT
-------
Measurement Statistics
2758 National Bureau of Standards (NBS). 1964. Handbook of Mathematical Functions. Applied
2759 Mathematics Series 55, National Bureau of Standards, Gaithersburg, MD.
2760 Nicholson, W.L. 1963. Fixed Time Estimation of Counting Rates with Background Corrections.
2761 AEC Research and Development Report HW-76279.
2762 Nicholson, W.L. 1966. Statistics of Net-counting-rate Estimation with Dominant Background
2763 Corrections. Nucleonics 24(8): 118-121.
2764 Nuclear Regulatory Commission (NRC). 1984. Lower Limit of Detection: Definition and
2765 Elaboration of a Proposed Position for Radiological Effluent and Environmental
2766 Measurements. NUREG/CR-4007. NRC, Washington, DC.
2767 Press, William H., et al. 1992. Numerical Recipes in C: The Art of Scientific Computing, 2nd ed.
2768 Cambridge University Press, New York, NY.
2769 Stapleton, James H. 1999. Personal correspondence. Department of Statistics and Probability,
2770 Michigan State University.
2771 Strom, Daniel J., and Paul S. Stansbury. 1992. Minimum Detectable Activity When Background
2772 Is Counted Longer than the Sample. Health Physics 63(3): 360-361.
2773 Turner, James E. 1995. Atoms, Radiation, and Radiation Protection., 2nd ed. John Wiley and
2774 Sons, New York, NY.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 19-133 DO NOT CITE OR QUOTE
-------
Measurement Statistics
MARLAP JULY 2001
DO NOT CITE OR QUOTE 19-134 DRAFT FOR PUBLIC COMMENT
-------
2775 ATTACHMENT 19E
2776 Example Calculations
2777 19E.1 Overview
2778 The following example shows how to calculate the combined standard uncertainty, critical net
2779 signal, minimum detectable concentration (MDC), and minimum quantifiable concentration
2780 (MQC) for a typical radioanalytical measurement.
2781 19E.2 Sample Collection and Analysis
2782 A soil sample is analyzed for 239/240pu and 238Pu by alpha spectrometry.
2783 • The sample is collected on July 10, 1999, at 11:17 am EDT, and shipped to a laboratory
2784 for analysis.
2785 • The entire laboratory sample is dried, weighed, and ground to a maximum particle size of
2786 0.2 mm. The dry weight is approximately 2 kg.
2787 • The prepared sample is homogenized, and a test portion is removed by increments. The
2788 documented procedure requires a test portion of approximately 0.5 g.
2789 • The test portion is weighed and the mass is found to be 0.5017 g. The standard
2790 uncertainty of the mass, including contributions from repeatability, linearity, day-to-day
2791 variability, and the balance calibration, is estimated to be 2.2 x 10~4 g.
2792 • A 1-mL aliquant of 242Pu tracer is added to the test portion. The concentration of the
2793 tracer solution has previously been measured as 0.0705 Bq ml/1 with a standard
2794 uncertainty of 0.0020 Bq ml/1 on June 30, 1999, at 11:00 am CDT. The aliquant is
2795 dispensed by a pipet, whose dispensed volume has a combined standard uncertainty
2796 previously determined to be 0.0057 mL.
2797 • After fusion, dissolution, chemical purification, and coprecipitation, a test source on a
2798 stainless steel planchet is prepared for counting in an alpha spectrometer.
2799 • The efficiency of the spectrometer for the chosen geometry, which is assumed to be con-
2800 stant over the range of alpha energies of interest, has previously been measured as 0.2805
2801 with a standard uncertainty of 0.0045.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 19-135 DO NOT CITE OR QUOTE
-------
Measurement Statistics
2802 • A blank source is counted in the spectrometer for 60,000 s. The blank consists of a filter
2803 mounted on a planchet in the same geometry as the test source. In the 242Pu region of
2804 interest, 2 counts are measured; and in the 238Pu region of interest, 0 counts are measured.
2805 Historical data for this and similar spectrometers at the laboratory indicate that the back-
2806 ground is stable between measurements.
2807 • The test source is placed in the spectrometer and counted for 60,000 s, beginning on
2808 August 24, 1999, at 4:47 pm CDT. In the 242Pu region of interest, 967 counts are meas-
2809 ured; and in the 238Pu region of interest, 75 counts are measured.
2810 • It is assumed that there is no detectable plutonium in the reagents; however, a method
2811 blank is analyzed simultaneously using a different spectrometer to check for contamina-
2812 tion of reagents and glassware.)
2813 In this example the measurand will be the mean activity concentration, or massic activity, of
2814 238Pu in the 2-kg sample (dry weight) at the time of collection.
2815 19E.3 The Measurement Model
2816 The following notation will be used:
2817 Ms is the mass of the test portion (0.5017 g)
2818 T is the tracer activity concentration (0.1205 Bq ml/1)
2819 Vt is the tracer aliquant volume (1 mL)
2820 tB is the blank count time (60,000 s)
2821 ts is the count time for the test source (60,000 s)
2822 Ns is the total count in a region of interest when the source is counted (238Pu or 242Pu)
2823 NB is the count in a region of interest when the blank is counted (238Pu or 242Pu)
2824 R is the fraction of alphas with measured energy in the region of interest (238Pu or 242Pu)
2825 D is the decay-correction factor (238Pu or 242Pu)
2826 £ is the alpha counting efficiency
2827 Y is the plutonium chemical yield fraction
2828 Fs is the subsampling factor (estimated as 1.00 with a Type B standard uncertainty of
2829 0.05)
2830 X is the 238Pu activity concentration in the dried laboratory sample, decay-corrected to
2831 the time of collection
2832 Subscripts will be used to distinguish between quantities associated with particular regions of
2833 interest (238Pu or 242Pu).
MARLAP JULY 2001
DO NOT CITE OR QUOTE 19-136 DRAFT FOR PUBLIC COMMENT
-------
Measurement Statistics
2834 The decay-correction factor for either isotope is calculated as follows:
.. ~ S
2835 D = e ^ID ^—^ -
2836 where X is the decay constant (s"1) and tD is the time between collection and the start of the
2837 counting measurement (3,91 1,400 s). Since \ts is small for both isotopes in this example, D may
2838 be approximated accurately by
2839
238 242
2840 The half-lives of 238Pu and 242Pu are 87.75 y and 375,800 y, respectively. So,
2841 £>,,, = exp — 3,911,400 + ^^ = 0.9990
238 ( 87.75 -365.25 -86,400 ( 2 JJ
2842 and £>242 = 1.000.
2843 Dead time is negligible in this example; so, no distinction is made between the real time and the
2844 live time. If the real time were greater than the live time, the correction for decay during the
2845 counting period would be based on the real time.
2846 The fraction of alphas of each isotope actually measured in the nominal region of interest is esti-
2847 mated to lie between 0.96 and 1.00. A rectangular distribution is assumed, with center at 0.98
2848 and half-width equal to 0.02. Then the Type B standard uncertainties of R23S and R242 are
2849 «(&>«) = ^949) = -^ = 0.01155
ZW v 2.0,2.' r-
V3
2850 The chemical yield of plutonium is calculated using the model
l-T? 0,10 / '? -L V D o/)o / I-r.
- * c
2851 Y = S'
TVtzR242D242
2852 Then the following model is used to estimate the measurand.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 19-137 DO NOT CITE OR QUOTE
-------
Measurement Statistics
2853
2854 When numerical values are inserted,
967 / 60,000 - 2 / 60,000
7 =
= 0.82990
X =
0.0705 • 1 • 0.2805 • 0.98 • 1
75/60,000 -0/60,000
0.5017 • 0.82990 • 0.2805 • 0.98 • 0.9990 • 1.00
(or 10.932 Bq kg"1)
= 0.010932 Bqg
-i
2855 19E.4 The Combined Standard Uncertainty
2856 The efficiency £ effectively cancels out of the equation for X, because it is multiplied by the yield
2857 7 and also appears as a factor in the denominator of the expression for 7 (see also Section
2858 19.6.5). Therefore, the uncertainty of £ has no effect on the uncertainty of X. When using the
2859 uncertainty propagation formula to calculate the combined standard uncertainty of X, one might
2860 include a covariance term for w(7,£) to account for the relationship between the measured values
2861 of 7 and £, but it is simpler to treat 7£ as one variable. Application of the uncertainty propagation
2862 formula (Section 19.5.3) to the equations above then gives the following:
2863
WE) =
2) /tS+U
242
)
1 2. 2, 2,
T Vt R242D2A
242
(7£)2
U2(p> U\V) u\R242)
T2
R,
242
2864
u'(X) =
M
u\R
238^
Rr
38
2865 All other input estimates are assumed to be uncorrelated.
2866 Note that u2(Fs) is the subsampling variance associated with taking a small test portion
2867 (0.5017 g) from a much larger sample (2 kg). A default value is used here for this variance
2868 component. However, Appendix F provides more information about subsampling errors and
2869 methods for estimating their variances.
MARLAP
DO NOT CITE OR QUOTE
19-138
JULY 2001
DRAFT FOR PUBLIC COMMENT
-------
Measurement Statistics
2870 Since extremely low counts are possible, each Poisson counting variance in this example will be
2871 estimated by the number of observed counts plus one (see Section 19.5.2.2 and Section 19C.3 of
2872 Attachment 19C). So, for example, ^(TVg 238) equals one, not zero.
2873 Table 19.7 summarizes the input estimates and their standard uncertainties.
TABLE 19.7 — Input estimates and standard uncertainties
INPUT
QUANTITY
Ms
T
v,
ts
ts
^,238
^,242
Ng, 238
NS,242
^ 238; ^242
8
Fs
^238
Dw
INPUT
ESTIMATE
0.5017
0.0705
1.0000
60,000
60,000
0
2
75
967
0.98
0.2805
1.00
0.9990
1.0000
STANDARD
UNCERTAINTY
2.2 x 10~4
0.0020
0.0057
Negligible
Negligible
1
1.73
8.72
31.1
0.01155
0.0045
0.05
Negligible
Negligible
MEASUREMENT
UNIT
g
BqmL '
mL
s
s
counts
counts
counts
counts
none
none
none
none
none
TYPE OF
EVALUATION
Combined
Combined
Combined
B
B
B
B
B
B
B
Combined
B
B
B
2874 Other possible sources of uncertainty in alpha spectrometry measurements include the following:
2875 • uncertainties in half-lives and decay times
2876 • spillover and baseline interferences caused by poor peak resolution
2877 • incomplete equilibration of tracer and analyte before chemical separation
2878 • changing instrument background
2879 • dependence of counting efficiency on alpha energy
2880 These uncertainties are evaluated as negligible in this example. Uncertainties associated with
2881 half-lives and decay times are negligible, because the decay times in the example are much
2882 shorter than the half-lives; but in practice one should confirm that any other uncertainties are
2883 small enough to be neglected.
JULY 2001
DRAFT FOR PUBLIC COMMENT
19-139
MARLAP
DO NOT CITE OR QUOTE
-------
Measurement Statistics
2884 When numerical values are inserted into the formulas
- 968 760.000^ 3/60.000*
2805)2
0.07052- 12-0.982- I2 0.0705 I 0.98
= 0.0001094007 = 0.010462
2886 and
2887 u 2(X) = 76 ' 60'00°2 + l ' 60>00°2
0.50172 • (0.82990 • 0.2805)2 • 0.982 • 0.99902
.
0.50172 0.829902 • 0.28052 0.982 l.OO2
= 2.1926 x 10"6 = 0.00148082
2888 So, uc(X) = 0.00148 Bq g"1 or 1 .48 Bq kg"1. If the concentration is to be reported with an expan-
2889 ded uncertainty calculated from the combined standard uncertainty uc(X) and a coverage factor
2890 k = 2, the result should appear (in SI units) as 10.9 ± 3.0 Bq kg"1 (dry weight).
2891 19E.5 The Critical Net Count
2892 Chapter 19 discusses several methods for estimating the critical net count Sc. In this example, the
2893 observed blank count is zero; so, the mean blank count is obviously very low, and nonnormal
2894 Poisson counting statistics may be assumed. Sections 19E.5.1 through 19E.5.4 below show how
2895 to apply the formulas discussed in Section 19D.2.2 for Poisson counting measurements,
2896 assuming a significance level of a = 0.05.
2897 19E.5.1 Formula A
2898 Formula A is not recommended when the blank count is extremely low, as in this example. How-
2899 ever, if Formula A is used, it gives the following estimate of the critical value of the net count.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 19-140 DRAFT FOR PUBLIC COMMENT
-------
2900
2904
2905
2906
Measurement Statistics
•l-a.
'
~
J. l
= 1.645V(0)(1)(2)
= 0 counts
2901 Since the net count 75 exceeds the critical net count 0, the analyte 238Pu is considered "detected."
2902 19E.5.2 Formulae
2903 Using Formula C, one obtains
2
— + T
0, l-a
2tB ^
- L6452m + i<
2
= 2.71 counts
72 f2 f ( f\
^•1 n ' 2.71, the analyte is considered detected.
19E.5.3 The Stapleton Approximation
Using the Stapleton approximation, the critical net count is calculated as follows.
Sr = 0.4 — - 1
•l-a
•l-a,
\
VS,238
^ I ^ 1
0.4)— 1 + —
tD\ tD
= 0.4(0)
i
. _
(2)+ 1.645^(0 +0.4)(1)(2)
= 2.82 counts
JULY 2001
DRAFT FOR PUBLIC COMMENT
19-141
MARLAP
DO NOT CITE OR QUOTE
-------
Measurement Statistics
2907 Since 75 > 2.82, the analyte is considered detected.
2908 19E.5.4 Exact Test
2909 When the exact test is used, the critical value of the source count NSt23S is the smallest nonnega-
2910 tive integer j/c such that
(19.144)
2911 First the right-hand side is calculated:
2912
t \ ^B 238 + 1
— '
I 73 /
=(0.95)(2)0 + 1 = 1.90
2913
2914
2915
2916
2917
2918
2919
2920
2921
2922
2923
2924
2925
Then, terms of the sum on the left-hand side are accumulated until the total is at least 1.90. The
iteration stops at k = 4, when the sum reaches 1.9375 (illustrated below).
k
0
1
2
3
4
kth Term
1
0.5
0.25
0.125
0.0625
Sum
1
1.5
1.75
1.875
1.9375
Thus, the critical value of the total count is_yc = 4, which may also be found in Table G.4 in
Appendix G. Since the observed count 7V5j238 = 75 exceeds the critical count, one concludes that
the sample contains a positive amount of 238Pu.
The critical net count Sc in this case is also 4, because the blank count is zero. Note that this
value of Sc is the most conservative of the critical values calculated in this example.
MARLAP
DO NOT CITE OR QUOTE
19-142
JULY 2001
DRAFT FOR PUBLIC COMMENT
-------
Measurement Statistics
2926 19E.6 The Minimum Detectable Concentration
2927 Assume the specified probability of a type n error at the minimum detectable concentration is
2928 p = 0.05. The following describes a conservative approach to the estimation of the nominal
2929 MDC for the analytical process.
2930 Let RB denote the mean blank count rate for the 238Pu region of interest. Suppose a total of 21
2931 counts are accumulated in the 238Pu region of interest during ten 60,000-s blank measurements.
2932 The estimated blank count rate is then
2933 RK = ——— = 3.5xlO~5cps
B 600,000
2934 This estimate has a moderately large relative standard uncertainty (approximately 22%), but
2935 detection decisions are based on the results of shorter measurements (60,000 s, not 600,000 s),
2936 which will vary even more. So, a conservative upper bound [NB] will be used for the blank count,
2937 as suggested in Section 19D.3.2 of Attachment 19D. A method for calculating the critical gross
2938 count can be adapted to calculate the largest value of the blank count that is likely to be observed
2939 given the assumption of a mean blank count rate of 3.5 x 10~5 cps. For the current problem, Table
2940 19.4 will be used, withRBtB replacing RB ts and [NB] replacing^ in the column headings. Since
2941 the value of RB tB is 2.1, which lies between 1.970 and 2.613, Table 19.4 shows that the required
2942 value is [NB] = 5. Therefore, one expects the number of blank counts observed in 60,000 s (tB~) to
2943 be no greater than 5. So, the MDC will be calculated here using a critical value [Sc] based on the
2944 assumption of a blank count [NB] = 5.
2945 The overall sensitivity for the measurement process is the product^ = tsMsYeR23SD23S. Since the
2946 most variable factor in this product by far is the chemical yield 7, a conservative lower bound for
2947 A may be found by estimating the p-quantile (5th percentile) of 7 and multiplying it by estimated
2948 values of the other factors. Assume that historical data show that the 5th percentile of 7 is approx-
2949 imately 0.60. Then with the measured efficiency 0.2805, nominal test portion mass 0.5 g, and
2950 estimated values for the ROI fraction 0.98 and decay factor 0.999, the 5th percentile of A is esti-
2951 mated as
2952 ap = a005 = (60,000)(0.60)(0.2805)(0.5)(0.98)(0.999) = 4943 g s
2953 The approximation formulas given in the chapter will be used and the results will be compared to
2954 the results obtained from a precise power calculation using the value ap for the sensitivity and
2955 with the assumptions that the mean blank count rate is RB = 3.5 x 10~5 cps and that the subsamp-
2956 ling error is approximately normal.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 19-143 DO NOT CITE OR QUOTE
-------
Measurement Statistics
2957 The following values, which appear in several formulas, are calculated first.
2958
c=RBts 1 +— =(3.5 x 1(T5) (60,000) (1 + 1) = 4.2 counts
oo
'p = 1 - *i -pq>,p = 1 - (1.645)2(0.05)2 = 0.993236
/c = (0.993236)(4.2) = 4.172 counts
p
2959
19E.6.1 Formula A
2960 Assuming the net signal is approximately normal at the MDC, the value of the MDC may be
2961 approximated by
D
rj
2963 where [Sc] denotes the critical net count calculated using [NB] as the blank count and (pSamp
2964 denotes the subsampling variance, which also equals u\F^). When Formula A is used, [Sc] is
2965
PC]=*l-a.
[NB]— 1 +— = 1.645^/(5)(1)(1 + 1) = 5.201 counts
2966
and the minimum detectable concentration is
1-
^
(4943) (0.993236) 2
= 0.0024 Bqg-1 or 2.4Bqkg^
^
L645"
5.201 +(0.05)2(5.201)2+4.172
2968 If the calculation is repeated with RBtB = 2.1 substituted for [NB] = 5 as the blank count used to
2969 calculate the critical value, the resulting value of XD is 1.9 Bq kg"1. A precise power calculation
2970 shows that the actual value of XD is 2.1 Bq kg"1.
MARLAP
DO NOT CITE OR QUOTE
19-144
JULY 2001
DRAFT FOR PUBLIC COMMENT
-------
2971
2973
2974
2975
2976
2977
2978
Measurement Statistics
19E.6.2 Formulae
2972 Using Formula C, one obtains
[Sc} =
2
Zl-aTS
2tB
1.6452
2
HZl-a
^
;i) + 1.<
Zl a** tS( tS\
4t2B [*^tB(l J
515 L6452(l)2 + 10
^ 4
= 6.727 counts
Then the minimum detectable concentration is
XD =
1
6.727 + L645 + 1.645,
(4943)(0.993236) 2
= 0.0028 Bqg^1 or 2.8Bqkg^
+ 6.727 +(0.05)2(6.727)2+4.172
If the critical value is calculated using RB tB = 2.l instead of [NB] = 5, the resulting value of XD is
2.3 Bq kg"1. A precise power calculation gives the value XD = 2.5 Bq kg"1.
19E.6.3 The Stapleton Approximation
When the Stapleton approximation is used, the critical net count is
[V]=0.4-1
"1 -a,
\
~ ~
([NB] +0.4)^ 1+^
i
= 0.4(0)
= 6.758 counts
, _
(2)+ 1.645^(5 +0.4)(1)(2)
JULY 2001
DRAFT FOR PUBLIC COMMENT
19-145
MARLAP
DO NOT CITE OR QUOTE
-------
Measurement Statistics
2979 Then the minimum detectable concentration may be approximated by
2980
2981
2982
2984
2985
2986
2987
2988
2989
XD =
Vp
1
6.758
1.645:
(4943)(0.993236) ^ 2
= 0.0028 Bqg"1 or 2.8Bqkg~1
+ 1.645
\
1.645^
6.758 +(0.05)2(6.758)2+4.172
WhenRBtB is substituted for [NB] in the calculation of the critical value, the resulting value of XD
is 2.4 Bq kg"1.
2983 Alternatively, the longer calculation given in Section 19D.3.3 of Attachment 19D may be used.
XD =
an
where
A
1 + _i = 5.2244
c' =
i i
7
-------
Measurement Statistics
2992 19E.6.4 Exact Test
2993 When the exact test for detection is used, the critical gross count [yc] equals the smallest nonneg-
2994 ative integer n such that
2995
2996 The right-hand side of the inequality is found as follows
2997 RHS = (1 -0.05)(1 + 1)5 + 1 =60.
2998 The value of the left-hand side exceeds 60.8 when n equals 12
2999
3000
3001
5 2 5 4
5 4096
= «>.92
Therefore,
3002 So,
\
[yc] = 12 counts and [Sc] = [yc] - [NB]— = 7 counts
3003
3004
XD =
Vp
1
(4943) (0.993236)
= 0.0029 Bqg"1 or 2.9Bqkg'1
The result of the precise calculation is XD = 2.8 Bq kg"1.
2
4
6452
2
+ [Sc] +
+ 1.645
2 9
(Psampl^c] + *$C
1.645 „ ,_
4
\
05)2(7)2+4.172
JULY 2001
DRAFT FOR PUBLIC COMMENT
19-147
MARLAP
DO NOT CITE OR QUOTE
-------
Measurement Statistics
3005 19E.7 The Minimum Quantifiable Concentration
3006 For the purpose of this example, the MQC is defined to be the analyte concentration XQ at which
3007 the relative standard deviation of the measured result is 1 / kQ, where kQ = 10. Calculation of XQ
3008 requires knowledge of the relative standard deviation of the measured sensitivity when the true
3009 sensitivity is A = a005. Assume for this example that the relative standard deviation is
102
^)
= 0.042 Bqg"1 or 42Bqkg
-i
3013 The MQC is substantially increased by the measurement variance of the sensitivity A and the
3014 subsampling variance. Without them the minimum quantifiable concentration would be only
3015 21 Bq kg"1. Note also that if either the relative standard deviation of A or the subsampling stan-
3016 dard deviation were 0.1 or more, the MQC would be infinite.
MARLAP
DO NOT CITE OR QUOTE
19-148
JULY 2001
DRAFT FOR PUBLIC COMMENT
-------
3017 ATTACHMENT 19F
sois Tests for Normality
3019 19F.1 Purpose
3020 Many common statistical hypothesis tests are based on the assumption that data are normally dis-
3021 tributed. Normality is often assumed by default, but, since some tests may not perform well with
3022 data that are not normal, it is often important to check the validity of the assumption. Performing
3023 a test for normality cannot prove that data are normally distributed, but it may produce strong
3024 evidence that they are not.
3025 There are a number of tests for normality. Each test requires a random sample 7l3 72, ..., Yn from
3026 the distribution being checked. Whatever test is used, it is a good idea to plot the data for visual
3027 inspection. The normal probability plot described in Section 19F.2 is useful for this purpose.
3028 One of the most powerful tests for normality is the Shapiro-Wilk test, but it is difficult to imple-
3029 ment manually. EPA QA/G-9 recommends the Shapiro-Wilk test when the sample size n is less
3030 than 50, and either Filliben's statistic or the studentized range test when n > 50 (EPA 1998). In
3031 fact, if software for the Shapiro-Wilk test is not available, then Filliben's statistic may be used in
3032 all cases for which critical values are available. Instructions for computing and using Filliben's
3033 statistic are given in Section 19F.3.
3034 19F.2 Normal Probability Plots
3035 A normal probability plot is a graph of the observed quantiles of a data set against the correspon-
3036 ding quantiles of a standard normal distribution. If the data are normally distributed and the data
3037 set is large enough (more than about 10 values), the plotted points should lie approximately on a
3038 straight line. A preliminary decision about the distribution of the data may be based on inspection
3039 of the graph. Normal probability plots may be produced manually, although software is generally
3040 needed to make plots of large data sets feasible.
3041 Manual construction of a normal probability plot is easier when pre-printed normal probability
3042 paper is available (see Figure 19.18 at the end of this attachment).
3043 To plot a set of data on normal probability paper, perform the following steps (EPA 1998).
3044 1. Arrange the data in ascending order:
3045 7(1) * 7(2) * - * 7(||)
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 19-149 DO NOT CITE OR QUOTE
-------
Measurement Statistics
3046
3047
3048
3049
3050
3051
3052
3053
3054
3055
3056
3057
3058
3059
3060
3061
3062
3063
3064
3065
3066
3067
2. Label the vertical axis to encompass all values between 7(1) (the minimum) and Y(n) (the
maximum).
3. For each /' compute the cumulative frequency Ft of the value 7(!), which is defined as the
number of values in the data set that are less than or equal to Y(f>. (Note that Ft > /'.)
4. Compute the horizontal coordinate Xt = Ft /(«+!)* 100% for each /'.
5. Plot each ordered pair (Xt, 7(0) at the appropriate location on the grid.
To plot a set of data on ordinary graph paper, perform Steps 1-3 above followed by Steps 4'-6'
below.
4'. For each /', determine the quantile X. = ZF/,+I) of the standard normal distribution (see
for example Table G. 1).
5'. Label the horizontal axis to encompass all values between Xl and Xn.
6'. Plot each ordered pair (Xt, 7(0).
The latter version of the procedure can be adapted to construct probability plots for other types of
distributions. Only Step 4' must change, since Xt is required to be a quantile of the appropriate
distribution.
EXAMPLE
Problem: Given the data set
123 122 124 118 118 122 121 117 125 119
construct a normal probability plot using normal probability paper.
Solution:
Step 1 Sort the 10 values:
Step 2
117 118 118 119 121 122 122 123 124 125
Label the vertical axis to encompass the values from 117 to 125.
MARLAP
DO NOT CITE OR QUOTE
19-150
JULY 2001
DRAFT FOR PUBLIC COMMENT
-------
Measurement Statistics
3068
3069
(
(
>tep 3 For each /' compute the cumulative frequency Ft of 7(!) (see the table below).
>tep 4 For each i compute Xi . = (Fi .1 1 1) x 100% and plot (Xp 7(0) .
/
7(0
Fi
x>
1
117
1
9.1%
2
118
o
6
27.3%
3
118
o
5
27.3%
4
119
4
36.4%
5
121
5
45.5%
6
122
7
63.6%
7
122
7
63.6%
8
123
8
72.7%
9
124
9
81.8%
10
125
10
90.9%
The results are shown as a normal probability plot in Figure 19.18.
3070 19F.3 Filliben's Statistic
3071 Filliben's statistic is derived from the concept of the normal probability plot and is often called
3072 the "normal probability plot correlation coefficient." The use of the statistic makes the
3073 interpretation of the probability plot less subjective, although a visual inspection of the plot is
3074 still recommended. The procedure for calculating and using the statistic is given below (Filliben
3075 1975).
3076 1. Choose the significance level a.
3077 2. Arrange the data in ascending order.
3078
< v < ... < v
-'~ -
3079
3080
3081
Compute the quantities Y and S as follows.
4. For / = 1,2, ...,«, compute
A
2=1
JULY 2001
DRAFT FOR PUBLIC COMMENT
19-151
MARLAP
DO NOT CITE OR QUOTE
-------
Measurement Statistics
3082
3083
3084
3085
3086
mt =
1 -0.51/B,
(7 -0.3 175)
0.51/B,
0.365),
/ = 1
7 = 2,3,.
i = n
and letM, be the wrquantile of the standard normal distribution zm . (Table G.I in
Appendix G may be interpolated to obtain approximate values of tnese quantiles.)
5. Compute cn = JE"=1M. .
6. Compute Filliben's statistic r (the normal probability plot correlation coefficient).
3087
r =
3088 7. Determine a critical value from Table G.5. If r is less than the critical value, conclude that
3089 the data are not normally distributed.
3090
3091
3092
3093
3094
3095
3096
EXAMPLE
Problem: Determine whether the values
123 122 124 118 118 122 121 117 125 119
appear to come from a normal distribution. Use the significance level 0.05.
Solution:
Step 1 The significance level is specified to be a = 0.05.
Step 2 Sort the 10 values:
117 118 118 119 121 122 122 123 124 125
MARLAP
DO NOT CITE OR QUOTE
19-152
JULY 2001
DRAFT FOR PUBLIC COMMENT
-------
Measurement Statistics
3097
3098
3099
3100
3101
Step 3 Compute 7 = — £7 = 120.9 and
^ ^ 10 '
Step 4 For each / compute mt and M. = zm (see the table below). (The quantiles Mt in this
example have been computed without using Table G.I.)
Step 5 Con
Step 6 Con
Step 7 Tab!
0.97
/
Y
n
k
1
o 117
(, 0.06697
f. -1.499
ipute cn
= fi]°=1M* = /L515 = 2.752.
OwM. 22.37 _nn,n
cnS (2.752)(8.301)
e G.5 shows that the critical value for n = 10 and a = 0.05 is 0
9 > 0.917, the data appear to be normally distributed.
2
118
0.1623
-0.9849
3
118
0.2588
-0.6470
4
119
0.3553
-0.3711
5
121
0.4518
-0.1212
678
122 122 123
0.5482 0.6447 0.7412
0.1212 0.3711 0.6470
.917. Since
9 10
124 125
0.8377 0.9330
0.9849 1.499
JULY 2001
DRAFT FOR PUBLIC COMMENT
19-153
MARLAP
DO NOT CITE OR QUOTE
-------
Measurement Statistics
MARLAP
DO NOT CITE OR QUOTE
19-154
JULY 2001
DRAFT FOR PUBLIC COMMENT
-------
Measurement Statistics
3102 19F.4 References
3103 Environmental Protection Agency (EPA). 1998. Guidance for Data Quality Assessment:
3104 Practical Methods for Data Analysis. EPA Q A/G-9, Q A97 Version. EP A/600/R-96/084,
3105 EPA, Quality Assurance Division, Washington, DC.
3106 Filliben, James J. 1975. The Probability Plot Correlation Coefficient Test for Normality.
3107 Technometrics 17(1): 111-117.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 19-155 DO NOT CITE OR QUOTE
-------
Measurement Statistics
MARLAP JULY 2001
DO NOT CITE OR QUOTE 19-156 DRAFT FOR PUBLIC COMMENT
-------
3108 ATTACHMENT 19G
3109 Balance Measurement Uncertainty
3110 19G.1 Purpose
3111 This attachment describes methods that may be used to evaluate balance measurement uncer-
3112 tainty. The relative standard uncertainty of a measurement made with a laboratory balance tends
3113 to be small if the balance is used properly, and it may even be considered negligible when com-
3114 pared to other uncertainties associated with radioanalysis (e.g., see Section 19.6.11, "Subsamp-
3115 ling")- However, one needs to know the performance limits of any measuring instrument. For
3116 example, the measurement uncertainty may actually be relatively large if a balance is used to
3117 weigh a mass that is too small for it. Establishing reasonable acceptance criteria for balance qual-
3118 ity control also requires an understanding of the sources of the measurement uncertainly.
3119 19G.2 Considerations
3120 Regardless of the methods used to evaluate balance measurement uncertainty, the results may be
3121 misleading unless the balance is well maintained and protected from external influences, such as
3122 drafts and sudden changes in pressure, temperature and humidity.
3123 The appropriate method for evaluating the standard uncertainty of a mass measured using a bal-
3124 ance depends on the type of balance, including its principles of calibration and operation, but the
3125 uncertainty of the measured result generally has components associated with balance sensitivity,
3126 linearity, repeatability, and air buoyancy. Typically, the component associated with sensitivity
3127 includes the uncertainty of calibration and may include variability caused by changing environ-
3128 mental conditions, such as temperature. Other sources of uncertainty may include leveling errors
3129 and off-center errors, which should be controlled. Static electrical charges may also have an
3130 effect. Changes in mass (e.g., by absorption or evaporation of water) may be very significant for
3131 some materials.
3132 19G.3 Repeatability
3133 The repeatability of a balance is expressed as a standard deviation and is usually assumed to be
3134 independent of the load. It represents the variability of the result of zeroing the balance, loading a
3135 mass on the pan, and reading the indication.
3136 Balance manufacturers provide specifications for repeatability, but a test of repeatability should
3137 also be part of the routine quality control for the balance (see ASTM 1993). The simplest pro-
3138 cedure for evaluating repeatability is to make a series of replicate measurements of a mass
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 19-157 DO NOT CITE OR QUOTE
-------
Measurement Statistics
3139 standard under "repeatability conditions." Repeatability conditions require one balance, one
3140 observer, one measurement location, and repetition during a short time period. For each
3141 measurement, one must zero the balance, load the mass standard, and read the balance indication.
3142 A nested experimental design can also be used to evaluate both the repeatability and the day-to-
3143 day variability due to environmental factors. In this procedure, one makes a series of replicate
3144 measurements with the same mass standard each day for a number of days. Ideally one should
3145 use a mass near the capacity of the balance to obtain the most reliable estimate of day-to-day var-
3146 lability. The repeatability standard deviation is then estimated by
Sr = <
K J
(x -x}2 (19.150)
\*-ki *k> ^ '
K(J-\
3147 where
3148 sr is the estimated repeatability standard deviation
3149 J is the number of repetitions per day
3150 K is the number of days
3151 xkj is they* result obtained on the A* day
3152 xk is the average of all the results on the ^h day
3153 The repeatability standard deviation determined by this method is a Type A standard uncertainty
3154 with K (J - 1) degrees of freedom.
3155 19G.4 Environmental Factors
3156 Given the experimental data from the preceding section, one may estimate the variability due to
3157 environmental factors (day-to-day variability) as follows.29
5Env
K
k = l
3158 where
3159 slnv is the estimated variance due to environmental factors
3160 x is the grand average of all the data (the average of the xk)
29 An F-test may be used to test for the presence of variance due to environmental factors. If this variance is zero,
then the quantity Js? I s?, where a? denotes the experimental variance of the averages x., may be assumed to have
an F-distribution with K- I numerator degrees of freedom and K(J - 1) denominator degrees of freedom.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 19-158 DRAFT FOR PUBLIC COMMENT
-------
3161
3 162
3163
Measurement Statistics
If ^Lv is found to be positive, then sEw is estimated by its square root; otherwise, sEw is assumed
to be zero. One estimates the relative component of standard uncertainty of a measured mass due
to environmental factors by
(PEnv = T- (19.152)
3 164 where Mcheck is the mass of the standard used in the experiment.
3165 19G.5 Calibration
3166
3167
3168
3169
3170
3 171
3172
3173
3174
3175
3176
3 177
3178
3179
3180
3181
3 182
3183
The uncertainty of calibration includes components associated with the mass standard or stan-
dards, repeatability, and variability due to environmental factors.
When a precision mass standard is used for calibration, the standard uncertainty of its mass is
generally negligible. However, the uncertainty may be evaluated if necessary from the specified
mass tolerance. For example, a 100-g ASTM Class-1 mass standard has a tolerance of 0.00025 g,
which may be assumed to represent the half-width of a triangular distribution centered at zero
(ASTM 1991). The standard uncertainty may be found by dividing this tolerance by ^6 and is
approximately 0.00010 g, or 1.0 x 10~6 when expressed in relative terms.
The total relative standard uncertainty of a measured mass due to calibration may be estimated as
follows.
-------
Measurement Statistics
3184 19G.6 Linearity
3185 The linearity of a balance should be specified by the manufacturer as a tolerance, aL, which repre-
3186 sents the maximum deviation of the balance indication from the value that would be obtained by
3187 linear interpolation between the calibration points. Routine quality control should ensure that the
3188 linearity remains within acceptable limits.
3189 The Eurachem/CITAC Guide: Quantifying Uncertainty in Analytical Measurement recommends
3190 that the linearity tolerance aL be treated as the half-width of a rectangular distribution and that aL
3191 therefore be divided by i/3 to obtain the standard uncertainty (Eurachem 2000). However, since
3192 the linearity error is likely to vary as a sinusoidal function of the load, the divisor y/2 may be
3193 more appropriate. So, the standard uncertainty due to linearity for a simple mass measurement
3194 may be evaluated as aL/^2. Whether one uses /3 or the more conservative value \[2 depends
3195 partly on how conservative one believes the estimate of aL to be.
3196 19G.7 Air Buoyancy Corrections
3197 Air buoyancy corrections have not often been performed in radiochemistry laboratories, but they
3198 are necessary for a realistic estimate of the standard uncertainty of a mass measurement,
3199 especially when the material being weighed has a low density. Failure to correct for air buoyancy
3200 when weighing water, for example, introduces a relative error of approximately - 0.1%, which
3201 may be much larger than the standard uncertainty of the uncorrected mass (e.g., when weighing a
3202 gram or more of an aqueous solution on a typical four-place analytical balance).
3203 When a buoyancy correction factor is used, the true mass is estimated as follows.
m=I^B (19.154)
3204 where
1 - pAr/ pr
B = ^C7 (19.155)
I-PA,M!PM
3205 and
3206 m is the corrected value for the mass of the material being weighed
3207 7Net is the net balance indication
3208 B is the buoyancy correction factor
3209 pM is the density of the material being weighed
3210 pAM is the density of the air at the time the material is weighed
MARLAP JULY 2001
DO NOT CITE OR QUOTE 19-160 DRAFT FOR PUBLIC COMMENT
-------
3211
3212
Measurement Statistics
pc is the density of the calibration mass standard
pA:C is the density of the air at the time of calibration
3213 The standard uncertainty of B may be obtained as follows.
u\B) PA,C
B2
9 9
( V
Pc
1
PA^PA
PM _l
, p^M
pi
2
(19.156)
3214
3215
3216
3217
3218
3219
3220
3221
3222
3223
3224
3225
3226
3227
3228
3229
Evaluation of this uncertainty requires estimates of pM, pc, p^^ and p^,c as well as their standard
uncertainties and covariances. The covariance z/(pAC, pc) is usually zero or negligible, and
u($AM, pM) also is usually negligible if the material being weighed is a solid.
The density of air at any time (p^) depends on temperature, pressure, and humidity, as shown in
the following equation.
p -p
A °
P-(0.3783)(RH/100%)(PV )
- 1 - >± - ll^
760
(19.157)
where
pA
p0
T
P
RH
P
Vap
is the density of air
is the density of dry air at 0°C and 760 torr (mm of Hg)
is the temperature (°C)
is the barometric pressure (torr)
is the relative humidity (%)
is the vapor pressure (torr) of water at temperature T
The vapor pressure, PV3p, is a nonlinear function of T, but it can be approximated by a linear
function in the range of temperatures typically encountered in the laboratory. When this approxi-
mation is made, the resulting equation for the air density (g ml/1) may be written as follows.
aP-(KH)(bT-c)
273.15 +r
where
JULY 2001
DRAFT FOR PUBLIC COMMENT
19-161
(19'158>
MARLAP
DO NOT CITE OR QUOTE
-------
3230
3231
3232
Measurement Statistics
a = 4.64746 x 1(T4
b = 2.5211151 x i(r6
c = 2.0590571 x i(r5
3233 Then the standard uncertainty of pA is given by
273.15 +T
(19.159)
3234
3235
3236
3237
3238
3239
The density of the calibration weight (pc) and of the solid or liquid material being weighed (pM)
also depend on temperature somewhat, but these temperature effects can usually be safely
ignored when calculating the uncertainty of the buoyancy correction factor, since temperature
affects the density of air much more than the density of a solid or liquid.
The effect of pressure on the density of the material being weighed can also usually be neglected.
For most practical purposes, the compressibility of a solid or liquid can be considered to be zero.
3240
3241
3242
3243
3244
3245
3246
3247
3248
EXAMPLE
Suppose the density of the weighed material, pM, is 0.5 g ml/1 with a tolerance of 0.2 g ml/1,
assumed to represent the half-width of a triangular distribution. The density of the calibration
mass standard, pc, is 7.850 g ml/1 with a tolerance of 0.025 g ml/1. Instead of measuring tem-
perature, pressure and humidity at the time of each measurement, the laboratory assumes the
following nominal values and tolerances:
Temperature 22.5 ±4 °C
Pressure 750 ±20 ton-
Relative humidity 50 ±20 %
MARLAP
DO NOT CITE OR QUOTE
19-162
JULY 2001
DRAFT FOR PUBLIC COMMENT
-------
Measurement Statistics
Then
PA,C
4,M
273.15 +T
_ (4.64746 x I(r4)(750) - (50)((2.5211151 x !Q-b)(22.5) - 2.0590571
273.15 +22.5
-3 _T -1
= 1.1728x KT'gmL
If each of the tolerances for T, P, and RH represents the half- width of a rectangular
distribution, then
^(7) = ^. = -^, u\P) = ^- = ^-, and M2(RH) =
4 16
— =
33
202 400
So, the standard uncertainties of pAiC and pAM are
U(PA.O> = M(
pA)2u2(T) + (bT- c)2w2(RH)
273.15 +T
_ ^a2(400/3)+ (6(50) + 1.1728 x !Q-3)2(16/3) + (b(22.5) - c)2(400/3)
273.15 +22.5
-5 „ ™T -1
= 2.1 x 10"5gmL
Then the buoyancy correction factor is
B =
1 "P^c/Pc _ 1 - 1.1728x !Q-3/7.85
I-PAM/PM 1 - 1.1728 x 1Q"3 / 0.5
= 1.00220
The tolerances for the densities pc and pM are the half- widths of triangular distributions; so,
°-252 2 0.22
2/ x
u (pr) =
A
and u
JULY 2001
DRAFT FOR PUBLIC COMMENT
19-163
MARLAP
DO NOT CITE OR QUOTE
-------
Measurement Statistics
3260
3261
3262
3263
3264
3265
3266
3267
The covariances u(pAf, pc) and ii(pAM, pM) are zero in this example. So, the standard uncer-
tainty of B is
u(B) = B
'^c)/Pic + w2(Pc)/Pc
\
= 1.00220
(Pc
.-I)2
r-l)2
(2.1 x l(T5)2
(1.1728x 10~3)2
[ 7.85
1 1.1728x 1Q~3
0.252/6
7.852
il2
(2.1 x l(T5)2 +0.22/6
(1.1728x 10~3)2 0.52
( °5 if
I 1.1728x 1Q~3 J
= 3.87x 1Q-4
Thus, the buoyancy correction factor increases the result of the measurement by 0.22% and
generates an uncertainty component of approximately 0.04%. Note that this uncertainty
component is very small and would generally be considered negligible in the final result of a
radiochemistry measurement, but it may represent a significant fraction of the uncertainty of
the mass measurement.
3268 19G.8 Combining the Components
3269 When the balance is used to measure the mass, m, of an object placed on the pan, the mass is
3270 given by m = IB, and its standard uncertainty by
3271
3272
3273
3274
3275
3276
3277
3278
where
m
I
B
-------
Measurement Statistics
3279 Often the balance is used to weigh material in a container. The balance is zeroed with the empty
3280 container on the pan and the container is then filled and weighed without being removed from the
3281 pan. In this case the linearity uncertainty component is counted twice, because the linearity error
3282 is assumed to vary between the two loads. (This assumption tends to be conservative when small
3283 masses are weighed.) Although the buoyancy factor for the tare and gross measurements may be
3284 different because of the different densities of the container and the material inside it, the only
3285 value of B that is used is the buoyancy factor for the material being weighed.
3286 In a third scenario, the empty container is weighed, removed from the pan, and then filled with
3287 material. The balance is zeroed again, and the filled container is weighed. Finally, the net mass is
3288 determined by subtracting the mass of the empty container from the total mass of the container
3289 and material. In this case both the linearity and repeatability components of uncertainty must be
3290 counted twice, because two distinct measurements are made. So, the corrected net mass and its
3291 standard uncertainty are
22
(19.161)
IT2 , 2 2 x 2 0
VNetOPcal +
-------
i 20 WASTE MANAGEMENT IN A RADIOANALYTICAL
2 LABORATORY
3 20.1 Introduction
4 This chapter presents information on the management of radioactive waste generated during
5 analytical processes. Federal, state, and local laws stringently regulate radioactive waste and
6 impose severe consequences for violations. Management of waste in compliance with such
7 regulations is, therefore, critical to the laboratory's sustained operation. Many—but not all—
8 applicable regulations are addressed in this Chapter. A laboratory waste management plan that
9 details procedures for the management of radioactive waste should be implemented before
10 radioactive materials are accepted for processing.
11 The following sections provide background information on managing radioactive waste and
12 identifies issues that should be considered when preparing a laboratory-waste management plan.
13 Sections 20.2 through 20.5 of this chapter provide general guidance for managing waste in a
14 radioanalytical laboratory. Descriptions of the types of wastes that may be produced in a
15 radioanalytical laboratory are provided in Section 20.2. Section 20.3 reviews various approaches
16 that have been used to achieve effective lab oratory-waste management programs. Waste
17 avoidance and waste minimization programs are discussed in Section 20.4. Waste determination
18 and characterization are briefly reviewed in Section 20.5. Some of the specific regulatory
19 requirements that apply to laboratory waste management are provided in Section 20.6. A
20 proposed outline for a waste management plan is provided in Section 20.7, and Section 20.8
21 suggests a number of useful web resources related to the management of laboratory waste.
22 20.2 Types of Laboratory Wastes
23 The types of wastes generated and the waste management issues the laboratory may face are
24 determined by the analytical processes used in the laboratory and the characteristics of the
25 samples analyzed. A laboratory that performs only one or two analytical processes may produce
26 only a few waste streams, whereas a multi-service laboratory that performs a variety of processes
27 may produce many waste streams. Waste streams produced by radioanalytical procedures can
28 include radioactive and non-radioactive wastes. A laboratory waste stream is defined as all
29 wastes that are produced by a given analytical process. Table 20.1 provides a list of wastes that
30 may be generated by a laboratory.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 20-1 DO NOT CITE OR QUOTE
-------
Waste Management in a Radioanalytical Laboratory
TABLE 20.1 — Examples of Laboratory-Generated Wastes
Waste
Dry solid waste
Aqueous waste
Organic solvent waste (used solvents,
analytical processes)
Acidic wastes
Waste Oil
Sample
Sample residue
Reagent chemicals
Sanitary waste
Sludge waste
Sharps
Example of Laboratory Generation
(Not Inclusive)
Gloves, glassware, pipette tips, plastic vials generated through
analytical processes
Solutions from analytical processes (filtrates, supernates, liquid
scintillation fluid)
Used solvents, de-greasers in cleaning operations, liquid
scintillation fluid
Solutions from analytical processes (filtrates, supernates)
Used oil from vacuum pumps
Unused sample from analytical process
Processed sample residue from analytical processes (precipitate,
filters, planchets)
Unused, expired, or surplus reagent chemicals
Sewage
Water treatment
Analytical processes (gas chromatography)
Various metal wastes/Radioactive sources | Laboratory equipment
Biohazardous waste
Toxic Substances Control Act (TSCA) waste
Radioactive waste
Resource Conservation and Recovery Act
(RCRA) hazardous waste
Mixed waste
Fecal, urine, blood-borne pathogen waste, animal carcasses, body
parts, tissues generated from bioassay, tissue or other biological
analyses
Analytical processes on polychlorinated bi-phenyls (PCB),
asbestos, chlorinated dioxin/furans
Analytical processes, radioactive standards, radioactive solutions,
dry waste, aqueous waste
Analytical processes generating characteristic and listed waste as
defined per 40 CFR 261 (Used solvents, reagent chemicals, acidic
waste, etc.)
Analytical processes generating any combination of RCRA waste
and radioactive wastes or TSCA waste and radioactive wastes
31 20.3 Waste Management Program
32 One source of guidance in assisting the laboratory in developing a waste management plan is
33 Profile and Management Options for EPA Laboratory Generated Mixed Waste (EPA, 1996).
34 This report reviews various approaches that have been taken to achieve effective laboratory waste
35 management programs. Much of the EPA report provides a review of articles and books that
MARLAP
DO NOT CITE OR QUOTE
20-2
JULY 2001
DRAFT FOR PUBLIC COMMENT
-------
Waste Management in a Radioanalytical Laboratory
36 detail the experiences of labs that manage radioactive wastes. This section draws significantly
37 from that report.
38 20.3.1 Program Integration
39 Successful waste management programs integrate important components, such as administrative,
40 regulatory requirements, training, record keeping, treatment, waste minimization, and prevention.
41 Individual management options, taken in isolation, may not be as effective as a more comprehen-
42 sive approach to waste management (EPA, 1996). Reviewing all aspects of waste management in
43 the laboratory should reveal the interactions among the component areas, providing insights that
44 allow improvements to the program as a whole without creating unknown negative effects.
45 20.3.2 Staff Involvement
46 All levels of management, scientists, and technicians should be actively involved in developing
47 and implementing the waste management program since each brings a valuable and unique
48 perspective to the waste management issue. Upper management must be committed to
49 maintaining a current and effective waste management plan because of the significant costs of
50 waste management and because of the serious civil and criminal penalties associated with non-
51 compliance. Program and project managers bring insight regarding issues, such as returning
52 samples to a site, waste management cost recovery, and data quality objectives. These managers
53 are also familiar with a full range of waste management alternatives. Laboratory environmental,
54 safety, and health personnel are essential to the process since they typically interface with
55 regulators to ensure that waste management practices are fully compliant. The input from
56 laboratory supervisors, scientists, and technicians is necessary because they generate waste at the
57 bench level and have first-hand process knowledge of how various waste streams are produced.
58 These individuals also have to implement the waste management plan on a daily basis and can
59 provide valuable feedback on improving the waste management system.
60 Waste generation planning is essential to proper waste management. Waste life cycle manage-
61 ment is a concept within the U.S. Department of Energy (DOE) Order 435.1 to reduce the
62 amount of radioactive waste generated. Waste life cycle is described as the life of a waste from
63 generation through storage, treatment, transportation, and disposal. For waste generated from a
64 new project or activity, consideration of the waste begins in the planning stage of the project or
65 activity.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 20-3 DO NOT CITE OR QUOTE
-------
Waste Management in a Radioanalytical Laboratory
66 20.4 Waste Minimization
67 Waste avoidance actively reduces the amount of waste to be managed and is a critical part of a
68 waste management plan. An integrated approach to laboratory waste management necessarily
69 implies pollution prevention. The term pollution prevention has served as an all-encompassing
70 term for any technique, process, or procedure that minimizes waste. Broadly defined, pollution
71 prevention refers to activities that keep pollutants from being created in any media (i.e., control
72 pollution at the source). There are many strong benefits to pollution prevention including safety,
73 waste minimization, efficiency, regulatory compliance, reduction in liability, and cost reduction.
74 Pollution prevention techniques are a critical component of prudent laboratory practices and have
75 been incorporated into many laboratory waste management procedures (EPA, 1996).
76 Management options that address waste avoidance will result in the most substantial cost
77 savings. Two of the primary areas to review when seeking to minimize laboratory waste are the
78 processes and definitions that the laboratory uses to identify and categorize waste. A laboratory
79 may define and manage various categories of wastes and may develop a hierarchy of waste
80 streams similar to the one described in Table 20.1. Properly categorizing waste at the point of
81 production will help to ensure health, safety, and regulatory compliance. This process also will
82 help to avoid unnecessary, costly, and inappropriate treatment, storage, and disposal. However,
83 proper categorization of waste streams can be difficult, requiring knowledge of the chemical and
84 radiological characteristics of the wastes, the production process, and a thorough understanding
85 of all-applicable regulations and regulatory guidance. Waste management regulations were
86 written primarily to regulate industrial production facilities and commercial storage, treatment,
87 and disposal facilities; their application to laboratories may not be readily apparent. The
88 laboratory waste management plan should require that each waste stream be identified prior to
89 production, so that waste minimization steps may be taken and production of unknown wastes
90 avoided.
91 The processes and definitions that a laboratory uses to determine that a waste is radioactive or
92 non-radioactive have a great influence on the amount of radioactive waste that a laboratory must
93 manage. The regulations offer little or no guidance for establishing that a waste is non-
94 radioactive, therefore it may be up to the laboratory to make this determination. Laboratory
95 management should develop clear guidelines to make this determination. The guidelines must
96 comply with requirements specified by the agency that issues the laboratory's license for
97 radioactive materials since waste considered non-radioactive in one state may be considered
98 radioactive in another.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 20-4 DRAFT FOR PUBLIC COMMENT
-------
Waste Management in a Radioanalytical Laboratory
99 Once the waste has been properly categorized (either through 10 CFR Part 61 or DOE O 435.1),
100 the laboratory can prioritize the review of waste streams for elimination, reduction, or
101 modification. A waste stream schematic or flow diagram that lists waste stream characteristics
102 and management pathways can be a useful tool in reviewing waste stream management. Various
103 management options that have been used to achieve waste stream minimization include the
104 following:
105 REGULATORY. Some wastes may be exempted from regulations because of the production
106 process, level of contaminants, volume of waste produced, or management option chosen. For
107 example, some hazardous wastes may be disposed in an industrial wastewater discharge if their
108 contaminants are below established regulatory levels and if the discharge is regulated under the
109 Clean Water Act. Also, a hazardous waste generator that produces less than 100 kg of waste in a
110 month may be considered a conditionally exempt small quantity generator and thus be exempt
111 from many of the requirements of RCRA (40 CFR 261.5). Some radioactive waste may be
112 managed as not-radioactive if the total level of radioactivity is below an exempt or de minimis
113 level, or if the activity for specific radionuclides is below established levels (10 CFR 61
114 20.2005). For certain licensees, radioactive wastes are released into the environment as gaseous
115 and liquid effluents in accordance with 10 CFR Part 61 20.2001(a)(3) and specific license
116 conditions.
117 METHOD SELECTION. The analytical method selected for the analysis of radioactive material
118 determines the type and volume of waste generated. When two methods will achieve the required
119 measurement quality objectives of the project, the laboratory may select the method that
120 produces the most easily managed waste (see Chapter 6, Selection and Application of an
121 Analytical Method).
122 PRODUCT SUBSTITUTION. In an analytical method, it may be possible to replace a hazardous
123 reagent with a non-hazardous reagent and still meet all health, safety, and data quality objectives.
124 In addition, substituting a short-lived radionuclide for a long-lived radionuclide may ultimately
125 result in a reduction of radioactive waste.
126 SAMPLE VOLUME COLLECTED. Excess sample material should not be collected. Personnel should
127 only collect enough sample material for the planned analysis and any reserve needed for re-
128 analysis or potential future use. Reserve volume should be minimized with up-front planning.
129 SAMPLE/REAGENT VOLUME. It may be possible to reduce the amount of sample and/or reagents
130 used in a method. It may also be possible to convert a method to a micro-scale method that uses
131 significantly less sample and reagents than the original method.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 20-5 DO NOT CITE OR QUOTE
-------
Waste Management in a Radioanalytical Laboratory
132 REAGENT PROCUREMENT CONTROLS. Often, the quantities of chemicals purchased by a
133 laboratory are determined by the price discounts available on larger quantities, instead of by the
134 amount of chemical required. The real cost of chemicals should be recognized as the initial
135 purchase price plus any disposal costs (lifetime costs). It should be noted that disposal costs of
136 excess chemicals can easily exceed the initial purchase costs. Procurement procedures for
137 hazardous material should be implemented to determine if a non-hazardous substitute is
138 available. Rotating chemical stock (first in, first out) may help avoid expiration of the chemical
139 shelf life.
140 RE-USE OF MATERIALS. Some materials may be recovered from the analytical process and re-
141 used in subsequent analyses. For example, distillation of certain used organic solvents may purify
142 them sufficiently for reuse.
143 DECAY IN STORAGE. Since the level of radioactivity decreases with time, it may be possible to
144 store a short-lived radionuclide until the natural-decay process reduces the radioactivity to a level
145 at which the waste can be considered non-radioactive for waste management purposes.
146 Laboratory management should be aware that RCRA storage limitations might impact the
147 feasibility of this option.
148 WASTE STREAM SEGREGATION. Segregating wastes by the appropriate category allows them to
149 be managed by the most cost-effective option. Combining highly regulated waste streams with
150 less stringently regulated waste streams usually requires the total waste stream to meet the most
151 stringent waste management requirements. For example:
152 • Non-hazardous waste mixed with hazardous waste must be managed as hazardous waste.
153 • Non-radioactive waste mixed with radioactive waste must be managed as radioactive waste.
154 • Hazardous waste mixed with radioactive waste must be managed in compliance with the
155 requirements of the Atomic Energy Act (AEA), RCRA, and TSCA.
156 20.5 Waste Determinations and Characterization
157 Laboratory wastes should be properly characterized to assure compliance with applicable federal,
158 state, and local regulations, and to determine appropriate means of disposal. Waste container
159 contents should be adequately characterized during waste generation and packaging. Characteri-
160 zations should address the type of material and the physical and chemical characteristics of the
161 waste. Minimum waste characterization criteria may be specified for the radioactive waste
162 generated (DOE M 43 5.1 -1, Ch. IV, Sec. I and NRC criteria specified in 10 CFR Part 61 for
163 commercial low-level radioactive waste sites).
MARLAP JULY 2001
DO NOT CITE OR QUOTE 20-6 DRAFT FOR PUBLIC COMMENT
-------
Waste Management in a Radioanalytical Laboratory
164 Three basic methods of characterization are denoted here: (a) process knowledge; (b) chemical
165 characterization through laboratory analysis; and (c) activities. Factual process knowledge (e.g.,
166 from a process waste assessment) influences the amount of sampling required to correctly
167 characterize waste.
168 A generic laboratory waste management plan should be established to describe the waste life
169 cycle. This plan should focus on characterizing each waste stream and establishing a waste
170 stream profile, so that the waste stream can be properly managed. The profiled waste stream may
171 only require a periodic partial characterization, based on the profile and regulatory status.
172 20.6 Specific Waste Management Requirements
173 This section provides general guidance on the storage, treatment, and disposal of radioactive
174 waste generated within a laboratory. It should not be used as definitive guidance for managing
175 radioactive waste. Laboratory managers are encouraged to review the complete regulatory
176 requirements in developing a waste management plan to fit the compliance and operational needs
177 of the laboratory. Laboratory managers may choose to have an environmental compliance
178 specialist assist with developing the waste management plan since waste management
179 requirements can be complex and contradictory.
180 Radioactive waste is regulated under AEA, administered by the Nuclear Regulatory Commission
181 (NRC). Thirty states are NRC Agreement States and have the authority and the regulatory
182 programs in place to regulate radioactive materials management in accordance with 10 CFR Part
183 61. Some wastes may also be regulated under RCRA , TSCA, or both, administered by EPA.
184 Most states have been granted authority to administer the mixed waste rules under RCRA.
185 Although many of the state hazardous waste laws are very similar to the federal RCRA
186 regulations, important differences may exist. This chapter focuses only on the federal
187 requirements, therefore, to ensure compliance with all applicable regulations, laboratory
188 management is strongly encouraged to review state and local regulations when developing a
189 waste management plan. Wastes that are regulated as radioactive under AEA and as hazardous
190 under RCRA or TSCA are termed "mixed wastes." Laboratories that generate mixed waste must
191 satisfy both NRC, which regulates the radioactive component, and EPA, which regulates the
192 hazardous component. Mixed waste management is a difficult responsibility, due to the complex
193 regulatory framework and the lack of approved treatment and disposal options for these wastes.
194 Other laws, such as the Clean Water Act (CWA) and the Clean Air Act (CAA), are not
195 summarized in this chapter. However, they may also have some impact on the management of
196 radioactive waste.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 20-7 DO NOT CITE OR QUOTE
-------
Waste Management in a Radioanalytical Laboratory
197 Federal regulatory requirements for waste management are found in Title 10 of the Code of
198 Federal Regulations (10 CFR) and Title 40 of the Code of Federal Regulations (40 CFR). The
199 following Federal citations address specific areas that regulate the management of waste
200 generated by a laboratory.
201 NRC REQUIREMENTS FOR RADIOACTIVE WASTE. Title 10 CFR 20, Standards for Protection
202 Against Radiation, and 10 CFR 61, Licensing Requirements for Land Disposal of Radioactive
203 Waste, address issues that may apply to management of radioactive waste in the laboratory.
204 LICENSE. Each laboratory that handles radioactive materials must be licensed by NRC, a NRC
205 Agreement State, or be operating under a site-wide license held by DOE. Radioactive materials
206 license issued by NRC or an Agreement State may provide additional requirements that affect the
207 management of waste. DOE-owned laboratories might be required to comply with DOE orders
208 that regulate the management of radioactive wastes (such as O 435.1 or 5820.2a).
209 DOE REQUIREMENTS FOR RADIOACTIVE WASTE. Any generator of DOE radioactive waste and
210 radioactive recyclable materials shall have a Waste Certification Plan (WCP). This plan provides
211 assurance that appropriate sections of the acceptance criteria of the waste and applicable RCRA
212 waste analysis requirements are met (DOE Order 5820.2A, Radioactive Waste Management).
213 The radioactive waste generator requirements are to ensure the development, review, approval,
214 and implementation of a program for waste generation planning, characterization, certification,
215 and transfer. This program shall address characterization of waste, preparation of waste for
216 transfer, certification that waste meets the receiving facility's radioactive waste acceptance
217 requirements, and transfer of waste (DOE M 435.1-1).
218 RCRA REQUIREMENTS FOR HAZARDOUS WASTE. Laboratories that generate hazardous waste
219 must meet detailed and specific requirements for the storage, treatment, and disposal of that
220 waste. Some of the regulatory requirements vary with the total amount of hazardous waste
221 generated each month, thus it is important that the laboratory understand how to properly
222 categorize its operation (small quantity exempt generator, small quantity generator, or large
223 quantity generator). Generator status is a regulatory issue that may vary among states. RCRA
224 regulations for generators found in 40 CFR 260-262, Hazardous Waste Management System:
225 General, list requirements in the following sections:
226 • 40 CFR 261, Identification and Listing of Hazardous Waste, describes what is, and what is
227 not, hazardous waste and how to determine if a waste is considered hazardous under RCRA.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 20-8 DRAFT FOR PUBLIC COMMENT
-------
Waste Management in a Radioanalytical Laboratory
228 • 40 CFR 262, Standards Applicable to Generators of Hazardous Waste, establishes
229 management requirements for generators of hazardous waste.
230 • 40 CFR 262.34, Accumulation Time, provides specific time and volume limitations on the
231 storage of hazardous waste.
232 • 40 CFR 262.40, Recordkeeping and Reporting, lists requirements a generator must meet in
233 documenting and reporting hazardous waste management activities.
234 TSCA REQUIREMENTS FOR PCB WASTE. The primary TSCA regulations that normally would
235 apply to an analytical laboratory relate to PCB waste. Laboratory waste containing PCBs at
236 concentrations of 50 ppm or greater, or are derived from PCB waste samples with concentrations
237 of 50 ppm or greater, are considered PCBs and are subject to the following regulations:
238 • 40 CFR 761.60, Disposal Requirements, describes requirements for the disposal of PCB
239 waste.
240 • 40 CFR 761.61, Poly chlorinated Biphenyls (PCBs) Manufacturing, Processing, Distribution
241 in Commerce, and Use Prohibitions, establishes prohibitions of, and requirements for, the
242 manufacture, processing, distribution in commerce, use, disposal, storage, and marking of
243 PCBs and PCB items.
244 • 40 CFR 761.65, Storage and Disposal, describes time limits for storage and storage
245 requirements of PCB waste.
246 • 40 CFR 761.64, Disposal of Wastes Generated as a Result of Research and Development
247 Activities ... and Chemical Analysis of PCBs, provides regulatory exclusion for some PCB
248 analytical samples.
249 20.6.1 Sample/Waste Exemptions
250 Laboratory samples and certain mixed wastes may be exempted or excluded from certain
251 regulatory provisions. Management should evaluate those regulations to determine if they affect
252 their waste management practices. Three examples are provided below.
253 RCRA ANALYTICAL SAMPLE/TREAT ABILITY SAMPLE EXCLUSIONS. Under 40 CFR 261.4(d), a
254 sample of solid waste or a sample of water, soil, or air, which is collected for the sole purpose of
255 testing to determine its characteristics or composition, is not subject to certain RCRA regulations
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 20-9 DO NOT CITE OR QUOTE
-------
Waste Management in a Radioanalytical Laboratory
256 if the laboratory is meeting the conditions specified in 40 CFR 261.4. Similarly, samples
257 undergoing treatability studies, and the laboratory or testing facility conducting such treatability
258 studies, are not subject to certain portions of RCRA [40 CFR 261.4(e)]. However, once a
259 material can no longer be considered a sample, it becomes waste and is subject to RCRA
260 requirements.
261 POLYCHLORINATED BiPHENYL (PCB) SAMPLE EXCLUSION. Portions of samples used in a
262 chemical extraction and analysis method for PCBs, and extracted for purposes of determining the
263 presence of PCBs or concentration of PCBs, are unregulated for PCB disposal (40 CFR 761.64).
264 All other PCB wastes from laboratory operations must be disposed in accordance with 40 CFR
265 761.61. Radioactive PCB waste may be exempt from the one year time limit for storage if the
266 waste is managed in accordance with all other applicable federal, state, and local laws and
267 regulations for the management of radioactive material (40 CFR 761.65).
268 MIXED WASTE EXEMPTION. Since August 1991, EPA has maintained a special policy on the
269 enforcement of the storage prohibition of RCRA mixed waste, which applies to generators that
270 are storing mixed wastes for which no viable treatment technology or disposal capacity exists.
271 The policy explains that EPA considers violation of the RCRA storage prohibition in section
272 3004(j) of RCRA to be a relatively low priority item among the Agency's potential civil
273 enforcement actions, as long as the wastes are stored in accordance with a RCRA permit or
274 interim status or in an environmentally sound manner. This policy, which only applies to certain
275 wastes, has been extended to October 2001. However, the policy does not apply to DOE
276 facilities.
277 20.6.2 Storage
278 Regulatory requirements for the storage of radioactive, hazardous, or PCB waste vary by the type
279 of waste, and typically address the waste storage area, type of acceptable waste containers, length
280 of time the waste may be stored, marking the storage area and the containers, and waste
281 monitoring. Significant civil and criminal penalties exist for storing waste improperly or for a
282 longer time period than allowed. The following sections summarize some of these requirements.
283 However, laboratory management is encouraged to review the regulations in depth so they may
284 develop a waste management plan that meets the compliance and operational needs of the
285 laboratory.
286 In the case of DOE analytical contract laboratories, low-level radioactive waste that has an
287 identified path to disposal shall not be stored longer than one year prior to disposal, except for
288 the purpose of radioactive decay. Low-level waste that does not have an identified path to
MARLAP JULY 2001
DO NOT CITE OR QUOTE 20-10 DRAFT FOR PUBLIC COMMENT
-------
Waste Management in a Radioanalytical Laboratory
289 disposal shall be characterized as necessary to meet the data quality objectives and minimum
290 characterization requirements to ensure safe storage and to facilitate disposal (DOE M 435.1-1).
291 20.6.2.1 Container Requirements
292 RADIOACTIVE WASTE. NRC has container requirements for low-level waste. Refer to 10 CFR
293 Part 61 for Class B and C requirements. For disposal, NRC requires the use of a high integrity
294 container approved by NRC.
295 RCRA HAZARDOUS WASTE. 40 CFR 265.170-177 provides requirements for the use and
296 management of containers storing hazardous waste. In summary, this section requires that
297 containers be in good condition, be compatible with the waste stored, be closed at all times
298 except when adding or removing waste, and be inspected weekly, in the case of 90-day
299 accumulation areas, for signs of corrosion or leakage.
300 PCB WASTE. 40 CFR 761.65 details TSCA requirements for the storage of PCB waste, including
301 the physical constraints of the storage area and the type of containers acceptable for storing liquid
302 and non-liquid PCB wastes. Laboratory PCB waste and samples returned to the sample collector
303 or submitted to a disposal facility when sample use is terminated may be exempt from the storage
304 requirements of 40 CFR 761.65.
305 20.6.2.2 Labeling Requirements
306 RADIOACTIVE WASTE. Radioactive waste storage areas should be posted with signs and labeled
307 in accordance with 10 CFR 20.1901 -1906, Precautionary Procedures. This section specifies
308 requirements for caution signs, labeling, signals, controls, and the storage of licensed material in
309 unrestricted areas.
310 RCRA HAZARDOUS WASTE. Hazardous waste containers must be labeled with the words
311 "Hazardous Waste" and, in the case of a 90-day accumulation area, the date upon which the
312 waste accumulation began 40 CFR 262.34(a)(4)(c)(ii).
313 PCB WASTE. 40 CFR 761.40 and 761.45 provides requirements for marking and labeling PCB
314 containers and the PCB storage area (40 CFR 761.50).
315 20.6.2.3 Time Constraints
316 RADIOACTIVE WASTE. NRC regulations in Title 10 of the Code of Federal Regulations do not
317 specifically establish a maximum amount of time that one may store radioactive waste. A
318 facility's NRC or Agreement State radioactive materials license may address this issue.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 20-11 DO NOT CITE OR QUOTE
-------
Waste Management in a Radioanalytical Laboratory
319 RCRA-HAZARDOUS WASTE. A generator may store hazardous waste up to 90 days, 180 days, or
320 270 days depending on its status as defined by the regulations or the distance the generator is
321 from the disposal facility (40 CFR 262.34). A generator may accumulate as much as 55 gallons
322 of hazardous waste or one quart of acutely hazardous waste in containers at or near the point of
323 generation where wastes initially accumulate, which is under the control of the operator of the
324 process generating the waste (40 CFR 262.34). The storage time clock (90, 180, or 270 days)
325 does not begin until the waste volume reaches 55 gallons (or one quart, in the case of acutely
326 hazardous waste), or whenever waste is stored in a 90-day accumulation area.
327 PCB WASTE. Radioactive PCB waste may be exempt from the one-year time limit for PCB
328 storage if the waste is managed in accordance with all other applicable federal, state, and local
329 laws and regulations for the management of radioactive material (40 CFR 761.65). According to
330 40 CFR 761.65(a)10, certain PCB waste containers may be exempt from 40 CFR 761.65 if the
331 containers are disposed within 30 days.
332 20.6.2.4 Monitoring Requirements
333 RADIOACTIVE WASTE. Radioactive waste storage areas should be surveyed and personnel should
334 be monitored in accordance with 10 CFR 20.1901-1906, Precautionary Procedures. These
335 sections specify the requirements for surveys, personnel monitoring, and storage of licensed
336 material in unrestricted areas. 10 CFR 20.1101 and 10 CFR 20.1201 address permissible doses,
337 levels, and concentrations of airborne radiation that would apply to radioactive waste storage
338 areas.
339 RCRA FlAZARDOUS WASTE. The owner or operator of a hazardous waste storage area must
340 inspect areas in which containers are stored, at least weekly, looking for leaks and deterioration
341 caused by corrosion or other factors (40 CFR 265.174). 40 CFR 262.34 address requirements for
342 Prevention and Preparedness, Contingency Plans, and Emergency Procedures that may apply to a
343 laboratory that stores RCRA waste.
344 PCB WASTE. All PCB containers in storage shall be checked for leaks at least once every 30 days
345 [40CFR761.65(c)(5)].
346 20.6.3 Treatment
347 Radioactive and mixed waste may require treatment to meet one or more objectives prior to final
348 disposal. Treatment involves the physical or chemical processes that result in a waste form that is
349 acceptable for disposal or further treatment. Treatment objectives include: (1) producing a waste
350 form acceptable for land disposal; (2) volume/mobility reduction through possible solidification
351 or sizing; (3) producing a waste more amenable for further treatment; or (4) separating radio-
352 active components from RCRA or TSCA components. Another treatment objective is to convert
MARLAP JULY 2001
DO NOT CITE OR QUOTE 20-12 DRAFT FOR PUBLIC COMMENT
-------
Waste Management in a Radioanalytical Laboratory
353 a radioactive RCRA regulated waste to a radioactive non-RCRA waste. Special permits may be
354 required from regulatory agencies prior to the treatment of waste.
355 Radioactive wastes may require treatment to meet the waste characteristics provided in 10 CFR
356 61.56. The following types of treatment have been used to meet those requirements:
357 • Non-solid radioactive waste may be treated with various solidification agents (such as
358 cement, asphalt, or polymers) to immobilize waste or sludge not otherwise acceptable for
359 disposal. Low-level radioactive waste (LLRW) may be absorbed onto a porous material, such
360 as silica, vermiculite, or organic materials to reduce the liquid volume.
361 • Dry radioactive waste may be treated with compaction or super-compaction to reduce the
362 waste volume.
363 • Some radioactive waste items may be decontaminated for unrestricted release by removal of
364 surface radioactivity through chemical or physical means. The residue from the
365 decontamination of a surface may require disposal as a radioactive waste.
366 • The relatively short half-lives of some radionuclides warrant storing the waste for a period of
367 time. Once the levels of radioactivity are undetectable or below an accepted de minimis level,
368 the waste may be disposed as a non-radioactive waste or in accordance with license
369 conditions.
370 20.6.4 Disposal
371 The disposal of radioactive waste is regulated by NRC in accordance with 10 CFR 20.2001,
372 which requires that waste be disposed at a licensed LLRW site. Radioactive waste that is mixed
373 with waste regulated under RCRA or TSCA is also subject to disposal requirements of the
374 respective regulations. Mixed waste must go to a facility that is licensed under both of the
375 appropriate laws. For example, radioactive RCRA waste cannot go to a RCRA landfill that is not
376 licensed under the Low Level Radioactive Waste Policy Act (LLRWPA), nor can it be disposed
377 at a LLRW site that is not licensed under RCRA.
378 In some cases, radioactive material may be disposed in a sanitary-sewage system if the
379 requirements of 10 CFR 20.2003 are met. This section provides specific limits on the quantity of
380 radionuclides that can be discharged into a sewage system. Discharges into a sewage system may
381 also be regulated by the Clean Water Act. For example, media used for liquid scintillation
382 counting, containing tritium (3H) or carbon-14 (14C) in concentration of 0.05 microcuries per
383 gram or less may be disposed as if it were not radioactive. Also, animal tissue containing 3H or
384 14C at levels less than or equal to 0.05 microcuries per gram (1,850 Bq/g) may be disposed
385 without regard to radioactivity (10 CFR 20.2005).
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 20-13 DO NOT CITE OR QUOTE
-------
Waste Management in a Radioanalytical Laboratory
386 The DOE also regulates the disposal of radioactive waste. Under DOE M 435.1-1, all radioactive
387 waste generators must have a waste certification program to ensure that the waste acceptance
388 criteria for the radioactive disposal facility are met. An outline of a waste certification plan is
389 contained in the following section.
390 20.7 Contents of a Laboratory Waste Management Plan/Certification Plan
391 20.7.1 Laboratory Waste Management Plan
392 A laboratory waste management plan will describe the waste generated by the analytical
393 laboratory. Each section of the plan is usually divided into two separate entities B one addressing
394 the needs of the laboratory analyst and the second addressing the needs of the waste management
395 personnel. An outline of a generic plan follows:
396 1. Recyclable Wastes
397 2. Sanitary Wastes/Industrial Wastes
398 3. Radioactive Wastes
399 4. Hazardous and Mixed Wastes
400 • Satellite Accumulation Area operations
401 • 90-day Accumulation Area operations
402 Within each section, the laboratory should delineate the types of waste that fall into each
403 category. Also, within the section for laboratory analysts, the disposal of the waste should be
404 clearly defined (e.g., paper in recyclable waste bin, unknown waste to environmental and/or
405 waste personnel). The waste management section should describe the process used by the waste
406 management personnel to dispose of the waste.
407 20.7.2 Waste Certification Plan/Program
408 The general outline for waste certification plans described below was taken from DOE M 435.1-
409 1 Ch. IV, Sec. J (1-3):
410 CERTIFICATION REQUIREMENTS. The waste certification program shall designate the officials
411 who have the authority to certify and release waste for shipment and to specify the documen-
412 tation required for waste generation, characterization, shipment, and certification. The program
413 shall provide requirements for auditing, retrieving and storing required documentation, including
414 records retention.
415 CERTIFICATION BEFORE TRANSFER. Low-level waste shall be certified as meeting waste
416 acceptance requirements before it is transferred to the facility receiving the waste.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 20-14 DRAFT FOR PUBLIC COMMENT
-------
Waste Management in a Radioanalytical Laboratory
417 MAINTAINING CERTIFICATION. Low-level waste that has been certified as meeting the waste
418 acceptance requirements for transfer to a storage, treatment, or disposal facility shall be managed
419 in a manner that maintains its certification status.
420 A general outline for a laboratory waste certification plan follows:
421 1. FACILITY NAME AND LOCATION. Provide the name and the physical location of the
422 facility.
423 2. ORGANIZATION. Describe the organizational structure for the facility's operation, quality
424 assurance program, and waste management program.
425 3. CONTENTS OF WASTE CERTIFICATION PLAN. Provide a detailed Table of Contents,
426 including list of tables, figures, and appendices as appropriate.
427 4. FACILITY RECYCLABLE AND WASTE MINIMIZATION STRATEGY. Identify the wastes and
428 waste streams the facility has targeted for recycling and waste minimization (i.e., source
429 reduction through product replacement).
430 5. DUTIES AND RESPONSIBILITIES OF MANAGEMENT AND WASTE MANAGEMENT
431 PERSONNEL. Provide a description of the positions at the laboratory, including primary
432 and secondary responsibilities and line of reporting.
43 3 6. QUALIFICATION REQUIREMENT s AND TRAINING OF WASTE MANAGEMENT PERSONNEL .
434 Describe the training and qualification program implemented for the environmental and
435 waste personnel. No specialized certifications (e.g., certified hazardous materials
436 manager, professional engineer) is needed unless specified by the job description or
437 standard operation procedures.
43 8 7. QUALIFICATIONS OF PROCEDURES AND EQUIPMENT USED IN WASTE MANAGEMENT .
439 Describe all equipment used in the waste management processes and procedures.
440 8. RECYCLABLE MATERIAL AND WASTE SEGREGATION CONTROL. Describe the process of
441 segregating various types of waste streams, especially in regards to radioactive and non-
442 radioactive wastes.
443 9. PACKAGING, HANDLING AND STORAGE CONTROL. Describe the process of packaging,
444 handling, and storing waste at the facility. This would include drum inspections, cipher-
445 locked storage, etc. The disposal of the supernates is a third example of a waste stream.
446 These supernates may be disposed in a sewage system, but the pH must be above 2 or
447 below 12 to allow the supernate solutions to be exempt from RCRA regulations.
448 Elementary neutralization is allowed in the laboratory under RCRA, but state regulations
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 20-15 DO NOT CITE OR QUOTE
-------
Waste Management in a Radioanalytical Laboratory
449 may require registration of the laboratory as an elementary neutralization unit before
450 neutralization and disposal take place.
45i 20.8 Useful Web Sites
452 Listed below are useful federal web sites relevant to the management of laboratory waste. Due to
453 the nature of the Internet, these addresses may change in the future.
454 Federal and State Government Regulation and Program References
455 http://www.epa.gov/docs/epacfr40/fmd-aid.info/state/
456 Environmental Laws and Regulations, Full Text (U.S. Code)
457 More than a dozen major statutes or laws form the legal basis for the programs of the
458 Environmental Protection Agency (EPA). The full text of these laws and the U.S. Code
459 Citation for each environmental law can be accessed through the following address.
460 http://www.epa.gov/epahome/lawreg.htm
461 Environmental Regulations in Federal Register
462 Full text of all Federal Register documents issued by EPA, as well as selected documents issued
463 by other Departments and Agencies. Notices, meetings, proposed rules, and regulations are
464 divided into twelve topical categories for easy access (e.g., air, water, pesticides, toxics, and
465 waste).
466 http://www.epa.gov/fedrgstr/
467 State and Federal Agency Contact List for Mixed Waste Regulations
468 http://www.epa.gov/rpdwebOO/mixed-waste/mw_pg6e.htm
469 States and Territories Where EPA Regulates Mixed Waste
470 http://www.epa. gov/rpdwebOO/mixed-waste/mw_pg6a. htm
471 States and Territories With EPA Authorization to Regulate Mixed Waste
472 http://www.epa.gov/rpdwebOO/mixed-waste/mw_pg6b.htm
473 State Solid and Hazardous Waste Web Sites
474 http://www.epa.gov/epaoswer/osw/stateweb.htm
475 RCRA State Authorization, By State and Program Element
476 http://www.epa.gov/epaoswer/hazwaste/state/index.htm
477 NRC Agreement States
478 http ://www.hsrd. ornl .gov/nrc/asframe.htm
MARLAP JULY 2001
DO NOT CITE OR QUOTE 20-16 DRAFT FOR PUBLIC COMMENT
-------
Waste Management in a Radioanalytical Laboratory
479 DOE Mixed Waste Policies
480 http://www.directives.doe.gov/
481 EPA Mixed Waste Home Page
482 http://www.epa.gov/rpdwebOO/mixed-waste/index.html
483 Mixed Waste Glossary
484 http://www.epa.gov/radiati on/mixed-waste/mw_pg5.htm#AEA
485 Guidance on the Definition and Identification of Commercial Mixed Low Level Radioactive and
486 Hazardous Waste
487 http://www.epa.gov/rpdwebOO/mixed-waste/mw_pg25.htm
488 Current Mixed Waste Treatment, Storage, or Disposal Facilities (TSDFs)
489 http://www.epa. gov/rpdwebOO/mixed-waste/mw_pgl la.htm
490 NRC/EPA Draft Storage Guidance
491 http://www.epa.gov/radiati on/mixed-waste/mw_pg27.htm
492 Mixed Waste Shipping and Transportation
493 http://www.epa. gov/rpdwebOO/mixed-waste/mw_pglO. htm
494 Mixed Waste Pollution Prevention
495 http://www.epa. gov/rpdwebOO/mixed-waste/mw_pg23 .htm
496 Pollution Prevention, EPA Home Page
497 http://www.epa.gov/epahome/p2pgram.htm
498 Radioactive Waste Disposal
499 http://www.nrc.gov/NRC/radwaste.htm
500 20.9 References
501 20.9.1 Cited References
502 U.S. Department of Energy (DOE). Order O 435.1: Radioactive Waste Management. July 1,
503 1999. http://www.directives.doe.gov/pdfs/doe/doetext/neword/435/o4351 .html.
504 U. S. Department of Energy (DOE). M 43 5.1 -1. Radioactive Waste Management Manual. Office
505 of Environmental Management. July 9, 1999. http://www.directives.doe.gov/pdfs/doe/
506 doetext/neword/43 5/m43 51-1. html.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT 20-17 DO NOT CITE OR QUOTE
-------
Waste Management in a Radioanalytical Laboratory
507 U.S. Environmental Protection Agency (EPA). 1996. Profile and Management Options for EPA
508 Laboratory Generated Mixed Waste, EPA 402-R-96-015, August.
509 20.9.2 Other Sources
510 Lewandowski, Joseph J., Alan A. Moghissi. 1995. "Management of Mixed Waste at a Teaching,
511 Research, and Health Care Facility," Proceedings of the 3d Biennial Symposium of Mixed
512 Waste, Baltimore, MD, August.
513 Linens, Ilona, Robert C. Klein, Edward L. Gershey. 1991. "Management of Mixed Waste from
514 Biomedical Research," Health Physics, 61:3, pp. 421-426.
515 Lorenzen, William A. 1995. Operational Aspects of Harvard University's Waste Management
516 Program, pp. 415-420, August.
517 Methe, Brian M. 1993. "Managing Radioactively Contaminated Infectious Waste at a Large
518 Biomedical Facility," Health Physics, 64:2, pp. 187-191.
519 McCamey, R.B. 1995. "Building a Mixed-Waste Prevention Program at Comanche Peak,"
520 Radwaste Magazine, May, pp. 21-28.
521 National Research Council. 1995. Prudent Practices in the Laboratory; Handling and Disposal
522 of Chemicals, National Academy Press, Washington, DC.
523 Party, E. and E.L. Gershey. 1989. "Recommendations for Radioactive Waste Reduction in
524 Biomedical/Academic Institutions," Health Physics, 56:4, pp. 571-572.
525 Reinhardt, Peter A. (editor), Leonard K. Leigh, Peter C. Ashbrook. 1996. Pollution Prevention
526 and Waste Minimization in Laboratories, Boca Raton Press.
527 Ring, Joseph, William Lorenzen, Frank Osborne, Jacob Shapiro. 1995. Bio-Medical Radioactive
528 Waste Management, July 19.
529 Todisco, L.R. and L.R. Smith. 1995. "A Manufacturer's Perspective on Low-Level Mixed Waste
530 Treatment, Storage, and Disposal," E.I. DuPont and Company, Inc., NEN Products,
531 Proceedings of the 3d Biennial Symposium of Mixed Waste, Baltimore, MD, August.
532 U.S. Nuclear Regulatory Commission/U.S. Environmental Protection Agency. 1995. Low-Level
533 Mixed Waste Storage Guidance, Federal Register 60:40204-40211, August 7.
MARLAP JULY 2001
DO NOT CITE OR QUOTE 20-18 DRAFT FOR PUBLIC COMMENT
-------
GLOSSARY
The glossary will be prepared following public review
( ) Indicates the section in which the term is first used in the MARLAP document
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT i DO NOT CITE OR QUOTE
-------
APPENDIX A: DIRECTED PLANNING APPROACHES
2 A.I Directed Planning Approaches
3 There are a number of approaches being used for directed planning of environmental operations.
4 Some of these approaches were designed specifically for data collection activities; others are
5 applications of more general planning philosophies. Many variations to these approaches have
6 been made for specific applications. The following are some of the approaches being used:
7 • Data Quality Objectives (DQO);
8 • Observational Approach (OA);
9 • Streamlined Approach for Environmental Restoration (SAFER);
10 • Technical Project Planning (TPP);
11 • Expedited Site Characterization (ESC);
12 • Value Engineering;
13 • Systems Engineering;
14 • Total Quality Management (TQM); and
15 • Partnering.
16 Employing any of these approaches assures that sufficient planning is carried out to define a
17 problem adequately, determine its importance, and develop an approach to solutions prior to
18 spending resources.
19 This appendix discusses some elements that are common to direct planning processes
20 (Section A.2) and provides in Sections A.3 through A. 11 very brief descriptions of the planning
21 approaches listed above. References are listed at the end of the appendix on each of the
22 approaches to provide sources of more detailed information.
23 Several directed planning approaches have been implemented by the Federal sector for
24 environmental data collection activities. Project planners should be cognizant of agency
25 requirements for planning. MARLAP does not endorse any one planning approach. Users of this
26 manual are encouraged to consider all the available approaches and choose a directed planning
27 process that is appropriate to their project and agency.
28 A.2 Elements Common to Directed Planning Approaches
29 To achieve the outcomes desired from directed planning, all of these approaches address the
30 following essential elements:
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT A-1 DO NOT CITE OR QUOTE
-------
Directed Planning Approaches
31 1. Defining the problem or need. Identifying the problem(s) facing the stakeholder/customer
32 that requires attention, or the concern that requires streamlining.
33 2. Establishing the optimum result. Defining the decision, response, product, or result that
34 will address the problem or concern and satisfy the stakeholder/customer.
35 3. Defining the strategy and determining the quality of the solution. Laying out a decision
36 rule or framework, roadmap, or wiring diagram to get from the problem or concern to the
37 desired decision or product and defining the quality of the decision, response, product, or
38 result that will be acceptable to the stakeholder/customer by establishing specific,
39 quantitative, and qualitative performance measures (e.g., acceptable error in decisions,
40 defects in product, false positive responses).
41 4. Optimizing the design. Determining what is the optimum, cost-effective way to reach the
42 decision or create the product while satisfying the desired quality of the decision or
43 product.
44 To most problem solvers, these four elements stem from the basic tenets of the scientific method:
45 "Principles and procedures for the systematic pursuit of knowledge involving the recognition and
46 formulation of a problem, the collection of data through observation and experiment, and the
47 formulation and testing of hypotheses" (Webster's Dictionary).
48 Each approach requires that a team of customers, stakeholders, and decision makers defines the
49 problem or concern; a team of technical staffer line operators have the specific knowledge and
50 expertise to define and then provide the desired product; and both groups work together to
51 understand each other's needs and requirements and to agree on the product to be produced. The
52 approaches represent slightly different creative efforts in the problem-solving process. All are
53 intended to facilitate the achievement of optimum results at the lowest cost, generally using team
54 work and effective communication to succeed.
55 A.3 Data Quality Objectives Process
56 The Data Quality Objectives (DQO) process was created by the U. S. Environmental Protection
57 Agency's Quality Assurance Management Staff (QAMS) to promote effective communications
58 between decision makers, technical staff, and stakeholders on defining and planning the
59 remediation of environmental problems.
60 The DQO process consists of seven basic steps:
MARLAP JULY 2001
DO NOT CITE OR QUOTE A-2 DRAFT FOR PUBLIC COMMENT
-------
Directed Planning Approaches
61 1. State the problem
62 2. Identify the decision
63 3. Identify inputs to the decision
64 4. Define the study boundaries
65 5. Develop a decision rule
66 6. Specify limits on decision errors
67 7. Optimize the design
68 Applying the DQO steps requires effective communication between the parties who have the
69 problem and the parties who must provide the solution. Additional information about the DQO
70 Process is provided in Appendix B to this manual.
71 A.4 Observational Approach
72 The Observational Approach (OA) emphasizes determining what to do next by evaluating
73 existing information and iterating between collecting new data and taking further action. The
74 name "observational approach" is derived from observing parameters during implementation.
75 OA was developed by Karl Terzaghi (Peck, 1969) for geological applications. In mining
76 operations, there may be substantial uncertainty in the location of valuable geological formations.
77 Information on soil and mineral composition would help to identify such formations. Application
78 of O A utilizes the sampling information on soil and mineral composition to direct the digging
79 locations. OA should be encouraged in situations where uncertainty is large, the vision of what is
80 expected or required is poor, and the cost of obtaining more certainty is very high.
81 The philosophy of OA when applied to waste site remediation is that remedial action can be
82 initiated without fully characterizing the nature and extent of contamination. The approach
83 provides a logical decision framework through which planning, design, and implementation of
84 remedial actions can proceed with increased confidence. OA incorporates the concepts of data
85 sufficiency, identification of reasonable deviations, preparation of contingency plans, observation
86 of the systems for deviations, and implementation of the contingency plans. Determinations of
87 performance measures and the quality of new data are done as the steps are implemented.
88 The iterative steps of site characterization, developing and refining a site conceptual model, and
89 identifying uncertainties in the conceptual model are similar to traditional approaches. The
90 concept of addressing uncertainties as reasonable deviations is unique to OA and offers a
91 qualitative description of data sufficiency for proceeding with site remediation.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT A-3 DO NOT CITE OR QUOTE
-------
Directed Planning Approaches
92 A.5 Streamlined Approach for Environmental Restoration
93 The Streamlined Approach for Environmental Restoration (SAFER) is an integration of the DQO
94 process and OA developed by the U. S. Department of Energy (DOE). The planning and
95 assessment steps of SAFER are the DQO process. The implementation steps of SAFER are the
96 Observational Approach. The approach emphasizing team work between decision makers and
97 technical staff reduces uncertainty with new data collection and manages remaining uncertainty
98 with contingency plans. The labels in each SAFER step are slightly different from the DQO and
99 OA steps, but the basic logic is the same. The SAFER Planning steps are:
100 • Develop a conceptual model;
101 • Develop remedial objectives and general response actions;
102 • Identify priority problem(s);
103 • Identify reasonable deviations and possible contingencies;
104 • Pursue limited field studies to focus and expedite scoping;
105 • Develop the decision rule;
106 • Establish acceptable conditions and acceptable uncertainty for achieving objective; and
107 • Design the work plan.
108 A.6 Technical Project Planning
109 Technical Project Planning (TPP) (formerly Data Quality Design), developed by the U. S. Army
110 Corps of Engineers, is intended for developing data collection programs and defining data quality
ill objectives for hazardous, toxic, and radioactive waste sites (HTRW). This systematic process
112 (USAGE, 1998) entails a four-phase planning approach in which a planning team—comprised of
113 decision makers, data users, and data providers—identifies the data needed to support specific
114 project decisions and develops a data collection program to obtain those data. In Phase I, an
115 overall site strategy and a detailed project strategy are identified. The data user's data needs,
116 including the level of acceptable data quality, are defined in Phase II. Phase in entails activities
117 to develop sampling and analysis options for the data needed. During phase IV, the TPP team
118 finalizes a data collection program that best meets the decision makers' short- and long-term
119 needs within all project and site constraints. The technical personnel complete Phase IV by
120 preparing detailed project objectives and data quality objectives, finalizing the scope of work,
121 and preparing a detailed cost estimate for the data collection program. The TPP process uses a
122 multi-disciplinary team of decision makers, data users, and data implementors focused on site
123 closeout.
MARLAP JULY 2001
DO NOT CITE OR QUOTE A-4 DRAFT FOR PUBLIC COMMENT
-------
Directed Planning Approaches
124 A.7 Expedited Site Characterization
125 Expedited Site Characterization (ESC) was developed to support DOE's Office of Science and
126 Technology's Characterization, Monitoring, and Sensor Technology (CMST) program
127 (Burton, 1993). The ESC process has been developed by American Society for Testing and
128 Materials (ASTM) as a provisional standard for rapid field-based characterization of soil and
129 groundwater (ASTM, 1996). The process is also known as QUICKSITE and "expedited site
130 conversion." ESC is based on a core multi-disciplinary team of scientists participating throughout
131 the processes of planning, field implementation, data integration, and report writing. ESC
132 requires clearly defined objectives and data quality requirements that satisfy the needs of the ESC
133 client, the regulatory authority, and the stakeholders. The technical team uses real-time field
134 techniques, including sophisticated geophysical and environmental sampling methods and an on-
135 site analytical laboratory, to collect environmental information. Onsite computer support allows
136 the expert team to analyze data each day and decide where to focus data collection the next day.
137 Within a framework of an approved dynamic work plan, ESC relies on the judgment of the
138 technical team as the primary means for selecting the type and location of measurements and
139 samples throughout the ESC process. The technical team uses on-site data reduction, integration
140 and interpretation, and on-site decision making to optimize the field investigations.
141 Traditional site investigations generally are based on a phased engineering approach that collects
142 samples based on a pre-specified grid pattern and does not provide the framework for making
143 changes in direction in the field. A dynamic work plan (Robatt, 1997; Robatt et al., 1998)
144 relies—in part—on an adaptive sampling and analysis program. Rather than specify the sample
145 analyses to be performed, the number of samples to be collected and the location of each sample,
146 dynamic work plans specify the decision making logic that will be used in the field to determine
147 where the samples will be collected, when the sampling will stop, and what analyses will be
148 performed. Adaptive sampling and analysis programs change or adapt based on the analytical
149 results produced in the field (Robatt, 1998; Johnson, 1993a,b).
150 A.8 Value Engineering
151 Value methodology was developed by Lawrence D. Miles in the late 1940s. He used a function-
152 based process ("functional analysis") to produce goods with greater production and operational
153 efficiency. Value methodology has evolved and, depending on the specific application, is often
154 referred to as "value engineering," "value analysis," "value planning," or "value management."
155 In the mid-1960s value engineering was adopted by three Federal organizations: the Navy Bureau
156 of Shipyards and Docks, the U. S. Army Corp of Engineers, and the U. S. Bureau of Reclama-
157 tion. In the 1990s, Public Law 104-106 (1996) and OMB Circulars A-131 (1993) and A-ll
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT A-5 DO NOT CITE OR QUOTE
-------
Directed Planning Approaches
158 (1997) set out the requirements for the use of value engineering, as appropriate, to reduce
159 nonessential procurement and program costs.
160 Value Engineering is a systematic and organized decision-making process to eliminate, without
161 impairing essential functions, anything that increases acquisition, operation, or support costs. The
162 techniques used analyze the functions of the program, project, system, equipment, facilities,
163 services, or supplies to determine "best value," or the best relationship between worth and cost.
164 The method generates, examines, and refines creative alternatives that would produce a product
165 or a process that consistently performs the required basic function at the lowest life-cycle cost
166 and is consistent with required performance, reliability, quality, and safety.
167 A standard job plan is used to guide the process. The six phases of the value engineering job plan
168 are:
169 • Information;
170 • Speculation (or creative);
171 • Evaluation (or analysis);
172 • Evolution (or development);
173 • Presentation (or reporting); and
174 • Implementation (or execution).
175 Value engineering can be used alone or with other management tools, such as TQM and
176 Integrated Product and Process Development (IPPD).
177 A.9 Systems Engineering
178 Systems Engineering brings together a group of multi-disciplinary team members in a structured
179 analysis of project needs, system requirements and specifications, and a least-cost strategy for
180 obtaining the desired results. Systems engineering is a logical sequence of activities and
181 decisions that transforms an operational need into a preferred system configuration and a
182 description of system performance parameters. Problem and success criteria are defined through
183 requirements analysis, functional analysis, and systems analysis and control. Alternative
184 solutions, evaluation of alternatives, selection of the best life-cycle balanced solution, and the
185 description of the solution through the design package are accomplished through synthesis and
186 systems analysis and control.
187 The systems engineering process involves iterative application of a series of steps:
MARLAP JULY 2001
DO NOT CITE OR QUOTE A-6 DRAFT FOR PUBLIC COMMENT
-------
Directed Planning Approaches
188 • Mission analysis or requirements understanding;
189 • Functional analysis and allocation;
190 • Requirements analysis;
191 • Synthesis; and
192 • System analysis and control.
193 A. 10 Total Quality Management
194 Total Quality Management (TQM) is a customer-based management philosophy for continuously
195 improving the quality of products (or how work is performed) in order to meet customer
196 expectations of quality and to measure and produce results aligned with strategic objectives.
197 TQM grew out of two systems developed by Walter Shewhart of Bell Laboratories in the 1920s.
198 Statistical process control was used to measure variance in production systems and to monitor
199 consistency and diagnose problems in work processes. The "Plan-Do-Check-Act" cycle applied a
200 systematic approach to improving work processes. The work of Deming and others in Japan
201 following World War II expanded the quality philosophy beyond production and inspection to all
202 functions within an organization and defined quality as "fit for customer use."
203 TQM has been defined as "the application of quantitative methods and the knowledge of people
204 to assess and improve (a) materials and services supplied to the organizations, (b) all significant
205 processes within the organization, and (c) meeting the needs of the end-user, now and in the
206 future" (Houston and Dockstader, 1997). The goal of TQM is to enhance effectiveness of
207 providing services or products. This is achieved through an objective, disciplined approach to
208 making changes in processes that affect performance. Process improvement focuses on
209 preventing problems rather than fixing them after they occur. TQM involves everyone in an
210 organization in controlling and continuously improving how work is done.
211 A. 11 Partnering
212 Partnering is intended to bring together parties that ordinarily might have differing or competing
213 interests to create a synergistic effect on an outcome each views as desirable. Partnering is a team
214 building and relationship enhancing technique that seeks to identify and communicate the needs,
215 expectations, and strengths of the participants. Partnering combines the talents of the
216 participating organizations in order to develop actions that promote their common goals and
217 objectives. In the synergistic environment of partnering, creative solutions to problems can be
218 developed. Like TQM, partnering enfranchises all stakeholders (team members) in the decision
219 process and holds them accountable for the end results. Each team member (customer, manage-
220 ment, employee) agrees to share the risks and benefits associated with the enterprise. Like the
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT A-7 DO NOT CITE OR QUOTE
-------
Directed Planning Approaches
221 other approaches, partnering places a premium on open and clear communication among
222 stakeholders to define the problem and the solution, and to decide upon a course of action.
223 A. 12 References
224 A.12.1 Data Quality Objectives
225 Guidance:
226 American Society for Testing and Materials (ASTM). 1995. Standard Practice for Generation of
221 Environmental Data Related to Waste Management Activities: Development of Data Quality
228 Objectives, D5792-95.
229 U.S. Environmental Protection Agency (EPA). 2000. Guidance for the Data Quality Objective
230 Process (EPA QA/G-4). EPA/600/R-96/055, Washington, DC. available from www.epa.gov/
231 quality l/qa_docs.html.
232 U. S. Environmental Protection Agency (EPA). 1993. Data Quality Objectives Process for
233 Superfund. EPA/540/G-93/071 (Interim Final Guidance). Office of Emergency and Remedial
234 Response. OSWER Directive 9355.9-01. September.
235 Papers:
236 Blacker, S. M. 1993. "The Data Quality Objective Process—What It Is and Why It Was
237 Created." Proceedings of the Twentieth Annual National Energy and Environmental Quality
238 Division Conference of the American Society for Quality Control.
239 Blacker, S. and D. Goodman. 1994a. "Risk-Based Decision Making An Integrated Approach for
240 Efficient Site Cleanup." Environmental Science & Technology, 28(11): 466A-470A.
241 Blacker, S. and D. Goodman. 1994b. "Risk-Based Decision Making Case Study: Application at a
242 Superfund Cleanup." Environmental Science & Technology, 28(11): 471A-477A.
243 Blacker, S. M. and P. A. Harrington. 1994. "Use of Process Knowledge and Sampling and
244 Analysis in Characterizing FFC Act Waste - Applying the Data Quality Objective (DQO)
245 Process to Find Solutions." Proceedings of the Twenty First Annual National Energy and
246 Environmental Quality Division Conference, American Society for Quality Control.
MARLAP JULY 2001
DO NOT CITE OR QUOTE A-8 DRAFT FOR PUBLIC COMMENT
-------
Directed Planning Approaches
247 Blacker, S. M. and J. Maney. 1993. "The System DQO Planning Process." Environmental
248 Testing and Analysis. July/August.
249 Blacker, S. M., J. D. Goodman and J. M. Clark. 1994. "Applying DQOs to the Hanford Tank-
250 Waste Remediation." Environmental Testing and Analysis, 3(4): 38.
251 Blacker, S., D. Neptune, B. Fairless and R. Ryti. 1990. "Applying Total Quality Principles to
252 Superfund Planning." Proceedings of the 17th Annual National Energy Division Conference,
253 American Society For Quality Control.
254 Carter, M. and D. Bottrell. 1994. "Report on the Status of Implementing Site-Specific
255 Environmental Data Collection Project Planning at the Department of Energy's (DOE) Office
256 of Environmental Restoration and Waste Management (EM)." Proceedings of the Waste
257 Management '94 Conference. Vol 2. pp. 1379-1383.
258 Goodman, D. and S. Blacker. 1997. "Site Cleanup: An integrated Approach for Project
259 Optimization to Minimize Cost and Control Risk." In: The Encyclopedia of Environmental
260 Remediation. John Wiley & Sons, New York, NY.
261 Michael, D. I. 1992. "Planning Ahead to Get the Quality of RI Data Needed for Remedy
262 Selection: Applying the DQO Process to Superfund Remedial Investigations." Proceedings of
263 the Air and Waste Management Association 85th Annual Meeting.
264 Michael, D. I. and E. A. Brown. 1992. "Planning Tools that Enhance Remedial Decision
265 Making." Proceedings of the Nineteenth Annual Energy and Environmental Quality Division
266 Conference for the American Society for Quality Control.
267 Neptune, M. D. and S. M. Blacker. 1990. "Applying Total Quality Principles to Superfund
268 Planning: Part I: Upfront Planning in Superfund." Proceedings of the 17th Annual National
269 Energy Division Conference, American Society for Quality Control.
270 Neptune, D., E. P. Brantly, M. J. Messner and D. I. Michael. 1990. "Quantitative Decision-
271 Making in Superfund: A Data Quality Objectives Case Study." Hazardous Material Control,
272 3:18-27.
273 Ryti, R. T. and D. Neptune. 1991. "Planning Issues for Superfund Site Remediation." Hazardous
274 Materials Control, 4:47-53.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT A-9 DO NOT CITE OR QUOTE
-------
Directed Planning Approaches
275 A.12.2 Observational Approach
276 Papers:
111 Brown, S. M. 1990. "Application of the Observational Method to Groundwater Remediation."
278 Proceedings of HAZMAT'90, Atlantic City, NJ.
279 Ferguson, R. D., G. L. Valet, and F. J. Hood. 1992. Application of the Observational Approach,
280 Weldon Springs Case Study.
281 Mark, D. L. et al. 1989. "Application of the Observational Method to an Operable Unit
282 Feasibility Study - A Case Study." Proceedings of Superfund'89, Hazardous Material Control
283 Research Institute, Silver Springs, MD, pp. 436-442.
284 Myers, R. S. and Gianti, S. J. 1989. "The Observational Approach for Site Remediation at
285 Federal Facilities." Proceedings of Superfund'89, Washington, D.C.
286 Peck, R. B. 1969. "Ninth Rankine Lecture, Advantages and Limitations of the Observational
287 Method in Applied Soil Mechanics." Geotechnique, 19, No. 2, pp. 171-187.
288 Smyth, J. D. and R. D. Quinn. 1991. "The Observational Approach in Environmental
289 Restoration." Proceedings of the ASCE National Conference of Environmental Engineering,
290 Reno, NV.
291 Smyth, J. D., J. P. Amaya and M. S. Peffers. 1992. "DOE Developments: Observational
292 Approach Implementation at DOE Facilities." Federal Facilities Environmental Journal,
293 Autumn, pp. 345-355.
294 Smyth, J. D., J. P. Kolman, and M. S. Peffers. 1992. "Observational Approach Implementation
295 Guidance: Year-End Report." Pacific Northwest Laboratory Report PNL-7999.
296 A. 12.3 Streamlined Approach for Environmental Restoration (Safer)
297 Guidance:
298 U. S. Department of Energy (DOE). 1993. Remedial Investigation/Feasibility Study (RI/FS)
299 Process, Elements and Techniques Guidance, Module 7 Streamlined Approach for
300 Environmental Restoration, Office of Environmental Guidance, RCRA/CERCLA Division
MARLAP JULY 2001
DO NOT CITE OR QUOTE A-10 DRAFT FOR PUBLIC COMMENT
-------
Directed Planning Approaches
301 and Office of Program Support, Regulatory Compliance Division Report DOE/EH-
302 94007658.
303 Papers:
304 Bottrell, D. 1993. "DOE's Development and Application of Planning to Meet Environmental
305 Restoration and Waste Management Data Needs." Proceedings of the Twentieth Annual
306 National Energy & Environmental Quality Division Conference, American Society for
307 Quality Control.
308 Dailey, R., D. Lillian and D. Smith. 1992. "Streamlined Approach for Environmental Restoration
309 (SAFER): An Overview." Proceedings of the 1992 Waste Management and Environmental
310 Sciences Conference.
311 Gianti, S., R. Dailey, K. Hull and J. Smyth. 1993. "The Streamlined Approach For
312 Environmental Restoration." Proceedings of Waste Management '93. Vol 1. pp. 585-587.
313 Smyth, J. D. and J. P. Amaya. 1994. Streamlined Approach for Environmental Restoration
314 (SAFER): Development, Implementation and Lessons Learned. Pacific Northwest Laboratory
315 Report PNL-9421/UC-402, Richland, WA.
316 A.12.4 Technical Project Planning
317 Guidance:
318 U. S. Army Corps of Engineers (USAGE). 1995. Technical Project Planning Guidance for
319 Hazardous, Toxic and Radioactive Waste (HTRW) Data Quality Design. Engineer Manual
320 EM-200-1-2 (superceded by EM-200-1-2, 1998).
321 U. S. Army Corps of Engineers (USAGE). 1998. Technical Project Planning Process. Engineer
322 Manual EM-200-1-2.
323 A.12.5 Expedited Site Characterization
324 Guidance:
325 American Society for Testing and Materials (ASTM). 1996. Standard Provisional Guide for
326 Expedited Site Characterization of Hazardous Waste Contaminated Sites. D585-96.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT A-11 DO NOT CITE OR QUOTE
-------
Directed Planning Approaches
327 Papers:
328 Bottrell, D. 1993. "DOE's Development and Application of Planning Processes to Meet
329 Environmental Restoration and Waste Management Data Needs." Twentieth Annual National
330 Energy and Environmental Quality Division Conference.
331 Burton, J. C., et al. 1993. "Expedited Site Characterization: A Rapid Cost-Effective Process for
332 Preremedial Site Characterization." Proceeding of Superfund XIV, Vol. n, Hazardous
333 Materials Research and Control Institute, Greenbelt, MD, pp. 809-826.
334 Burton, J. C. 1994. "Expedited Site Characterization for Remedial Investigations at Federal
335 Facilities." Proceedings Federal Environmental Restoration in and Waste Minimization n
336 Conference, Vol. II, pp. 1407-1415.
337 Johnson, R. 1993a "Adaptive Sampling Program Support for Expedited Site Characterization."
338 ER'93 Environmental Remediation Conference Proceedings.
339 Johnson, R. 1993b. "Adaptive Sampling Program Support for the Unlined Chromic Acid Pit,
340 Chemical Waste Landfill, Sandia National Laboratory, Albuquerque, New Mexico." ANL-
341 EAD/TM-2.
342 Robatt, A. 1997. "A Guideline for Dynamic Work Plans and Field Analytics: The Keys to Cost
343 Effective Site Cleanup." Tufts University Center for Field Analytical Studies and Technology
344 and U.S. EPA, Region 1, Hazardous Waste Division.
345 Robatt, A. 1998. "A Dynamic Site Investigation: Adaptive Sampling and Analysis Program for
346 Operable Unit 1 at Hanscom Air Force Base, Bedford, Massachusetts." Tufts University
347 Center for Field Analytical Studies and Technology and U.S. EPA, Region 1, Office of Site
348 Remediation and Restoration, Boston, MA.
349 Robbat, A., S. Smarason, and Y. Gankin. 1998. "Dynamic Work Plans and Field Analytics, The
350 Key to Cost-Effective Hazardous Waste Site Investigations," Field Analytical Chemistry and
351 Technology 2(5):253-65.
352 Starke, T. P., C. Purdy, H. Belencan, D. Ferguson and J. C. Burton. 1995. "Expedited Site
353 Characterization at the Pantex Plant." Proceedings of the ER'95 Conference.
MARLAP JULY 2001
DO NOT CITE OR QUOTE A-12 DRAFT FOR PUBLIC COMMENT
-------
Directed Planning Approaches
354 A. 12.6 Value Engineering
355 Guidance:
356 The February 1996 Amendment to the Office of Federal Procurement Policy Act (41 U.S.C. 401
357 et. seq.) (Public Law 104-106, Sec 4306 amended this.)
358 Federal Acquisitions Regulations. FAR, Part 48, Value Engineering. September 19, 1992.
359 Federal Acquisitions Regulations. FAR, Part 52.248-1,-2,-3, Value Engineering Solicitation
360 Provisions and Contract Clauses. January 31, 1989.
361 National Defense Authorization Act for Fiscal Year 1996. PL 104-106, Law Requiring Value
362 Engineering in Executive Agencies. February 10, 1996.
363 Office of Management and Budget (OMB). 1993. OMB Circular A-131, Value Engineering.
364 Office of Management and Budget (OMB). 1997. OMB Circular A-11, Preparation and
365 Submission of Budget Estimates.
366 U. S. Army. Value Engineering. Army Regulation AR 5-4, Chapter 4 (Reference only).
367 U. S. Army Corps of Engineers (USAGE). Engineer Regulation. ER 5-1-11.
368 U. S. Department of Energy. 1997. Value Management. Good Practice Guide (GPG-FM-011).
369 U. S. Department of the Interior (Dol). 1995. Departmental Manual, Management Systems and
370 Procedures, Part 369, Value Engineering, Chapter 1, General Criteria and Policy. May 18,
371 1995.
372 Books:
373 Fallen, C. 1990. Value Analysis. The Miles Value Foundation, 2nd Edition.
374 Kauffman, J. J. 1985. Value Engineering for the Practitioner. North Carolina State University,
375 Raleigh, NC.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT A-13 DO NOT CITE OR QUOTE
-------
Directed Planning Approaches
376 Miles, L. D. 1989. Techniques of Value Analysis and Engineering. McGraw-Hill Book
377 Company, New York, NY.
378 Mudge, A. E. 1989. Value Engineering, A Systematic Approach. J. Pohl Associates.
379 Parker, D. 199x. Value Engineering Theory. The Miles Value Foundation.
380 Papers:
381 Al-yousefi, A. 1996. "Total Value Management (TVM): A VE-TQM Integration." Proceedings
382 of the 1996 SAVE Conference, Society of American Value Engineers.
383 Blumstein, G. 1996. "FAST Diagramming: A Technique to Facilitate Design Alternatives."
384 Proceedings of the 1996 SAVE Conference, Society of American Value Engineers.
385 Maynor, D. 1996. Value Engineering for Radiation Hazards Remediation at the Fernald OU1,
386 Ohio. U. S. DOE, Ohio Field Office.
387 Morrel, C. 1996. Value Engineering for Radiation Hazards Remediation at Fernald OU4, Ohio.
388 U.S. DOE Reclamation Technical Service Center.
389 Wixson, J. R. 1987. "Improving Product Development with Value Analysis/Value Engineering:
390 A Total Management Tool." Proceedings of the Society of American Value Engineers, Vol.
391 22, pp.51-66.
392 A. 12.7 Systems Engineering
393 Guidance:
394 Electronic Industries Alliance (EIA). 1994. Systems Engineering. Standard EIA/IS-632.
395 Electronic Industries Alliance (EIA). 1997. Upgrade IS-632, Process for Engineering a System.
396 EIA/SP-3537 Part 1: Process Characteristics and EIA/SP-4028 Part 2: Implementation
397 Guidance.
398 International Electrical and Electronics Engineers (IEEE). 1994. Standard for Application and
399 Management of the Systems Engineering Process. P1220.
MARLAP JULY 2001
DO NOT CITE OR QUOTE A-14 DRAFT FOR PUBLIC COMMENT
-------
Directed Planning Approaches
400 U. S. Department of Defense (DoD). 1992. Systems Engineering. MIL-STD-499B.
401 U. S. Department of Energy (DOE). 1996. Project Execution and Engineering Management
402 Planning. Good Practice Guide GPG-FM-010.
403 Books:
404 Boardman, J. 1990. Systems Engineering: An Introduction. Prentice Hall, New York, NY.
405 Chestnut, H. 1967. System Engineering Methods. John Wiley & Sons, New York, NY.
406 Churchman, C. W. 1968. The Systems Approach. Dell Publishing Co., Inc., New York, NY.
407 Eisner, H. 1998. Computer-Aided Systems Engineering (CASE). Prentice-Hall, Englewood Cliffs,
408 NJ.
409 Goode, H. H. 1957. Systems Engineering: An Introduction to the Design of Large-Scale Systems.
410 McGraw-Hill, New York, NY.
411 Machol, R. E. 1965. Systems Engineering Handbook. McGraw-Hill, New York, NY.
412 Smith, D. B. 1974. Systems Engineering and Management. Addison-Wesley Publ. Co., Reading,
413 MA.
414 Wymore, A. W. 1976. Systems Engineering Methodology for Interdisciplinary Teams. John
415 Wiley & Sons, New York, NY.
416 Papers:
417 Bensoussan, A. 1982. "Analysis and Optimization of Systems." Proceedings of the Fifth
418 International Conference on Analysis and Optimization of Systems, Versailles, France.
419 December 14-17, 1982.
420 David, H.T. and S. Yoo. 1993. "Where Next? Adaptive Measurement Site Selection for Area
421 Remediation." In: Environmental Statistics, Assessment and Forecasting (Richard Cathern,
422 Ed.). Lewis Publishers, MI.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT A-15 DO NOT CITE OR QUOTE
-------
Directed Planning Approaches
423 Ljunggren M. and J Sundberg. 199x. "A Systems Engineering Approach to National Solid Waste
424 Management — Case Study, Sweden." In: Proceedings 12th Int. Conference on Solid Waste
425 Management. November 17-20, 1996.
426 Pacific Northwest Laboratory. 1995. A Systems Engineering Analysis to Examine the Economic
427 Impact for Treatment of Tritiated Water in the Hanford K-Basin. Report No. PNL-S A-24970.
428 Richland, WA.
429 A. 12.8 Total Quality Management
430 Guidance:
431 U. S. Department of the Army. 1992. The Leadership for Total Army Quality Concept Plan.
432 U. S. Department of Energy (DOE). 1993. Total Quality Management Implementation
433 Guidelines. DOE/HR-0066.
434 U. S. Office of Personnel Management (OPM), Federal Quality Institute. 1990. Federal Total
435 Quality Management Handbook, How to Get Started, Booklet 1: Implementing Total Quality
436 Management, U. S. Government Printing Office.
437 Books:
438 Berk, J. and S. Berk. 1993. Total Quality Management: Implementing Continuous Improvement.
439 Sterling Publishing Co. Inc., New York, NY.
440 Carr, D. K. and I. D. Liftman. 1993. Excellence in Government. Coopers and Lybrand, Arlington,
441 VA.
442 Dobyns, L. and C. Crawford-Mason. 1994. Thinking about Quality: Progress, Wisdom and the
443 Deming Philosophy. Times Books, New York, NY.
444 Harrington, H. J. 1991. Business Process Improvement. McGraw-Hill, New York, NY.
445 Koehler, J. W. and J. M. Pankowski. 1996. Quality Government. Designing, Developing and
446 Implementing TQM. St. Lucie Press, Delray Beach, FL.
MARLAP JULY 2001
DO NOT CITE OR QUOTE A-16 DRAFT FOR PUBLIC COMMENT
-------
Directed Planning Approaches
447 Rao, A., et al. 1996. Total Quality Management: A Cross Functional Perspective. John Wiley &
448 Sons, New York, NY.
449 Walton, M. 1990. Deming Management at Work. Putnam., New York, NY.
450 Papers:
451 Blacker, S. 1990. "Applying Total Quality Concepts to Environmental Data Operations."
452 Proceedings of the Eight International Conference of the International Society for Quality
453 Control.
454 Breisch, R.E. 1996. "Are You Listening." Quality Progress, pp. 59-62.
455 Houston, A. and Dockstader, S. L. 1997. Total Quality Leadership: A Primer. Department of the
456 Navy, Total Quality Leadership Office Publication Number 97-02.
457 Kidder, P. J. and B. Ryan. 1996. "How the Deming Philosophy Transformed the Department of
458 the Navy." National Productivity Review 15(3).
459 A.12.9 Partnering
460 Guidance:
461 U. S. Department of the Army. 1993. Engineering and Design Quality Management, Appendix B
462 Partnering. ER-1110-1-12.
463 Books:
464 Hrebniak, L. 1994. We Force in Management: How to Build and Sustain Cooperation. Free
465 Press, New York, NY.
466 Maurer, R. 1992. Caught in the Middle: A Leadership Guide for Partnership in the Workplace.
467 Productivity Press, Portland, OR.
468 Poirier, C. C. 1994. Business Partnering for Continuous Improvement: How to Forge Enduring
469 Alliances Among Employees, Suppliers, and Customers. Berrett-Koehler, New York, NY.
470 Papers:
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT A-17 DO NOT CITE OR QUOTE
-------
Directed Planning Approaches
471 Brown, T. L. 1993. "Is there Power in Partnering?" Industry Week, 242(9): 13.
472 Covey, S. R. 1993. "Win-Win Partnerships." Executive Excellence, 10(11): 6-7.
473 Chem-Nuclear Systems, Inc. (CNSI). 1996. Community Partnering Plan: Pennsylvania Low-
474 Level Radioactive Waste Disposal Facility. S80-PL-021, Revision 0. Commonwealth of
475 Pennsylvania, Department of Environmental Protection, Bureau of Radiation.
476 Mosley, D. and C. C. Moore. 1994. "TQM and Partnering: An Assessment of Two Major Change
477 Strategies." PMNETwork, 18(9): 22-26.
478 Sanders, S. R. and M. M. Moore. 1992. "Perceptions on Partnering in the Public Sector." Project
479 Management Journal, 23(4): 13-19.
480 Simmons, J. 1989. "Partnering Pulls Everything Together." Journal for Quality & Participation.,
481 12:12-16.
482 U. S. Army Corps of Engineers (USAGE). 1996. "U.S. Corps of Engineers Adopts Partnering."
483 National Academy of Public Administration Foundation, Washington, DC.
MARLAP JULY 2001
DO NOT CITE OR QUOTE A-18 DRAFT FOR PUBLIC COMMENT
-------
i APPENDIX B: THE DATA QUALITY OBJECTIVES
2 PROCESS
3 Bl.O Introduction
4 MARLAP's objective in this appendix is to provide information about the basic framework of
5 the DQO process (ASTM 5792; EPA, 2000; NRC, 1998; MARSSIM, 1997). The DQO planning
6 process empowers both data users and data suppliers to take control and resolve issues in a
7 stepwise fashion. It brings together at the right time all key players from the data user and data
8 supplier constituencies and enables each participant to play a constructive role in clearly
9 defining:
10 • The problem that requires resolution;
11 • What type, quantity, and quality of data the decision maker needs to resolve that problem;
12 • Why the decision maker needs that type and quality of data;
13 • How much risk of making a wrong decision is acceptable; and
14 • How the decision maker will use the data to make a defensible decision.
15 The DQO Process provides a logic for setting well-defined, achievable objectives and developing
16 a cost-effective, technically sound sampling and analysis design. It balances the data user's
17 tolerance for uncertainty with the available resources for obtaining data. The number of visible
18 and successful applications of the DQO process has proven its value to the environmental
19 community. The DQO process is adaptable depending on the complexity of the project and the
20 input from the decision makers. Some users have combined DQO planning with remedy
21 selection for restoration projects (e.g., DOE's SAFER—see Appendix A.5). Other users have
22 integrated the project scoping meetings with the DQO Process. Much of the information that is
23 developed during the DQO process is useful for the development of the project plan documents
24 (Chapter 4) and the implementation of the data validation process (Chapter 8) and the data
25 quality assessment (DQA) process (Chapter 9).
26 Since its inception, the term "data quality objectives" has been adopted by many organizations,
27 and the definition has been adapted and modified (ee box on next page). Throughout this
28 document, MARLAP uses EPA's (2000) definition of DQOs: "Qualitative and quantitative
29 statements derived from the DQO process that clarify study objectives, define the appropriate
30 type of data, and specify the tolerable levels of potential decision errors that will be used as the
31 basis for establishing the quality and quantity of data needed to support decisions."
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT B-1 DO NOT CITE OR QUOTE
-------
The Data Quality Objectives Process
32
33
34
35
36
37
38
39
40
41
42
43
44
Definitions of Data Quality Objectives
(1) Statements on the level of uncertainty that a decision maker is willing to accept in
the results derived from environmental data (ASTM 5283; EPA, 1986).
(2) Qualitative and quantitative statements derived from the DQO process that clarify
study objectives, define the appropriate type of data, and specify the tolerable levels
of potential decision errors that will be used as the basis for establishing the quality
and quantity of data needed to support decisions (EPA, 2000).
(3) Qualitative and quantitative statements derived from the DQO process describing
the decision rules and the uncertainties of the decision(s) within the context of the
problem(s) (ASTM D5792).
(4) The qualitative and quantitative statements that specify the quality of the data
required to support decisions for any process requiring radiochemical analysis
(radioassay) (ANSI 42.23).
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
B2.0 Overview of the DQO Process
The DQO process (Figure Bl) consists of seven steps (EPA, 2000). In general, the first four steps
of the DQO Process require the project planning team to define the problem and qualitatively
determine required data quality. Once these steps have been addressed adequately, the last three
steps of the process establish quantitative performance measures for the decision and the data.
The last step of the process involves
developing the data collection design based on
the DQOs, which is dependent on a clear
understanding of the first six steps.
Although the DQO process is described as a
sequence of steps, it is inherently iterative. The
output from each step influences the choices
that will be made in subsequent steps. For
instance, a decision rule cannot be created
without first knowing the problem and desired
decision. Similarly, optimization of the
sampling and analysis design generally cannot
occur unless it is clear what is being optimized
—the results of the preceding steps. Often the
outputs of one step will trigger the need to
Stepl: State die Problem
I
Step 2: Identify die Decision
1
Step 3: Identify Inputs to die Decision
t
Step 4: Define die Study Boundaries
1
Step 5: Develop a Decision Rule
I
Step ft Specify Limits on Decision Errors
Step 7: Optimize die Design for Obtaining Data
Figure Bl—Seven steps of the DQO process.
MARLAP
DO NOT CITE OR QUOTE
B-2
JULY 2001
DRAFT FOR PUBLIC COMMENT
-------
The Data Quality Objectives Process
65 rethink or address issues that were not evaluated thoroughly in prior steps. These iterations lead
66 to a more focused sampling and analysis design for resolving the defined problem. The first six
67 steps should be completed before the sampling and analysis design is developed, and every step
68 should be completed before data collection begins. The DQO process is considered complete
69 with the approval of an optimal design for sampling and analysis to support a decision or when
70 available historical data are sufficient to support a decision.
71 In practice, project planning teams often do a cursory job on the first four steps, wanting to get
72 into technical design issues immediately. Without carefully defining the problem and the desired
73 result, the project planning team may develop a design that is technically sound but answers the
74 wrong question, or answers the questions only after the collection of significant quantities of
75 unnecessary data. Time spent on the first four steps is time well spent. Extra effort must be given
76 to assure that Steps 1 to 4 are adequately addressed.
77 When applying the DQO process, or any planning approach, it is important to document the
78 outputs of each step to assure that all participants understand and approve the interim products,
79 and that they have a clear record of their progress. It is sometimes useful to circulate an approval
80 copy with signature page to ensure agreement of the stakeholders.
si B3.0 The Seven Steps of the DQO Process
82 Each step of the DQO process will be discussed in the following sections. Not all items will be
83 applicable to every project. The project planning team should apply the concepts that are
84 appropriate to the problem.
85 B3.1 DQO Process Step 1: State the Problem
86 The first step is to define the problem clearly. The members of the project planning team present
87 their concerns, identify regulatory issues and threshold levels, and review the site history. The
88 project planning team should develop a concise description of the problem. Some elements to
89 include in the description might be the study objectives, regulatory context, groups who have an
90 interest in the study, funding and other resources available, previous study results, and any
91 obvious sampling design constraints. The more facts, perceptions and concerns of the key
92 stakeholders—including important social, economic, or political issues—that are identified
93 during this step, the better the chances are that the issues driving the decisions and actions will be
94 identified.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT B-3 DO NOT CITE OR QUOTE
-------
The Data Quality Objectives Process
95 The primary decision maker should be identified. The resources and relevant deadlines to address
96 the problem are also defined at this time. If possible, a "site conceptual model" should be
97 developed. This will help structure and package the diverse facts into an understandable picture
98 of what the various issues are and how those issues can be focused into a specific problem.
99 The expected outputs of Step 1 are:
100 • A conceptual model that packages all the existing information into an understandable picture
101 of the problem;
102 • A list of the project planning team members and identification of the decision maker;
103 • A concise description of the problem; and
104 • A summary of available resources and relevant deadlines for the study.
105 B3.2 DQO Process Step 2: Identify the Decision
106 During Step 2 of the DQO Process, the project planning team defines what decision must be
107 made or what question the project will attempt to resolve. The decision (or question) could be
108 simple, like whether a particular discharge is or is not in compliance, or the decision could be
109 complex, such as determining if observed adverse health is being caused by a non-point source
110 discharge. Linking the problem and the decision focuses the project planning team on seeking
111 only that information essential for decision making, saving valuable resources (time and money).
112 The result may be a comprehensive decision for a straightforward problem, or a sequence of
113 decisions for a complex problem. For complex problems with multiple concerns, these concerns
114 should be prioritized in order of importance. Often a complex concern is associated with a series
115 of decisions that need to be made. Once these decisions have been identified, they should be
116 sequenced in a logical order so the answer to one decision provides input in answering the next
117 decision. It may be helpful to develop a logic flow diagram (decision framework), arraying each
118 element of the issue in its proper sequence along with its associated decision that requires an
119 answer.
120 The term "action level" is used in this document to denote the numerical value that will cause the
121 decision maker to choose one of the alternative actions. The action level may be a derived
122 concentration guideline level, background level, release criteria, regulatory decision limit, etc.
123 The action level is often associated with the type of media, analyte and concentration limit. Some
124 action levels, such as the release criteria for license termination, are expressed in terms of dose or
MARLAP JULY 2001
DO NOT CITE OR QUOTE B-4 DRAFT FOR PUBLIC COMMENT
-------
The Data Quality Objectives Process
125 risk. The release criterion typically is based on the total effective dose equivalent (TEDE), the
126 committed effective dose equivalent (CEDE), risk of cancer incidence (morbidity) or risk of
127 cancer death (mortality) and generally can not be measured directly. A radionuclide-specific
128 predicted concentration or surface area concentration of specific nuclides that can result in a dose
129 (TEDE or CEDE) or specific risk equal to the release criterion is called the "derived concentra-
130 tion guideline level" (DCGL). A direct comparison can be made between the project's analytical
131 measurements and the DCGL (MARS SIM, 1997).
132 The project planning team should define the possible actions that may be taken to solve the
133 problem. Consideration should be given to the option of taking no action. A decision statement
134 can then be developed by combining the decisions and the alternative actions. The decision rule
135 and the related hypothesis test will be more fully developed in the DQO process at Steps 5 and 6.
136 By defining the problem and its associated decision clearly, the project planning team has also
137 begun to define the inputs and boundaries (DQO process Steps 3 and 4). At the end of Step 2, the
138 project planning team has:
139 • Identified the principal decisions or questions;
140 • Defined alternative actions that could be taken to solve the problem based on possible
141 answers to the principal decisions and questions;
142 • Combined the principal decisions and questions and the alternative actions into decision
143 statements that expresses a choice among alternative actions; and
144 • Organized multiple decisions.
145 B3.3 DQO Process Step 3: Identify Inputs to the Decision
146 During Step 3, the project planning team makes a formal list of the specific information required
147 for decision making. The project planning team should determine what information is needed and
148 how it can be acquired. The project planning team should specify if new measurements are
149 required for the listed data requirements. The data required are based on outcomes of discussion
150 during the previous two steps. The project planning team should define the basis for setting the
151 action level. Depending on the level of detail of the discussion during the previous steps, then
152 efforts associated with Step 3 may be primarily to capture that information. If the first two steps
153 have not defined the inputs with enough specificity, then those inputs should be defined here.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT B-5 DO NOT CITE OR QUOTE
-------
The Data Quality Objectives Process
154 However, before going further, the output should be reviewed to assure that the problem, the
155 decision steps and the input are compatible in complete agreement.
156 An important activity during Step 3 is to determine if the existing data or information, when
157 compared with the desired information, has significant gaps. If no gaps exist, then the existing
158 data or information may be sufficient to resolve the problem and make the decision. (Although
159 there may be no gaps in the data, the data may not have enough statistical power to resolve the
160 action level. See Step 6 for more discussion.) In order to optimize the use of resources, the
161 project planning team should maximize the use of historical information. If new data are
162 required, then this step establishes what new data (inputs) are needed. The specific environmental
163 variable or characteristic to be measured should be identified. The DQO Process clearly links
164 sampling and analysis efforts to an action and a decision. This linkage allows the project
165 planning team to determine when enough data have been collected.
166 If the project planning team determines that collection of additional data is needed, the analytical
167 laboratory acquisition strategy options should be considered at this stage. Identifying suitable
168 contracting options should be based on the scope, schedule, and budget of the project, and the
169 capability and availability of laboratory resources during the life of the project, and other
170 technical considerations of the project. If an ongoing contract with a laboratory is in place, it is
171 advisable to involve them with the radioanalytical specialists as early as possible.
172 The project planning team should ensure that there are analytical protocols available to provide
173 acceptable measurements. If analytical methods do not exist, the project planning team will need
174 to consider the resources needed to develop a new method, reconsider the approach for providing
175 input data, or perhaps reformulate the decision statement.
176 The expected outputs of Step 3 are:
177 • A list of information needed for decision making;
178 • Determination of whether data exists and are sufficient to resolve the problem;
179 • Determination of what new data, if any, are required;
180 • Defined the characteristics that define the population and domain of interest;
181 • Defined the basis for the action level;
182 • Confirmation that appropriate analytical protocols exist to provide the necessary data; and
183 • A review of the planning output to assure the problem, decision and inputs are fully linked.
MARLAP JULY 2001
DO NOT CITE OR QUOTE B-6 DRAFT FOR PUBLIC COMMENT
-------
The Data Quality Objectives Process
184 B3.4 DQO Process Step 4: Define the Study Boundaries
185 In Step 4, the project planning team should define clearly the geographic area within which the
186 decisions will apply. The project planning team specifies the spatial and temporal boundaries
187 covered by the decision statement. The spatial boundaries define the physical aspects to be
188 studied in terms of geographic area, media, and any appropriate subpopulations (e.g., an entire
189 plant, entire river basin, one discharge, metropolitan air, emissions from a power plant). When
190 appropriate, divide the population into strata that have relatively homogeneous characteristics.
191 The temporal boundaries describe the time frame the study data will represent (e.g., possible
192 exposure to local residents over a 30-year period) and when samples should be taken (e.g.,
193 instantaneous samples, hourly samples, annual average based on monthly samples, samples after
194 rain events). Changing conditions that could impact the success of sampling and analysis and
195 interpretation need to be considered. These factors include weather, temperature, humidity, or
196 amount of sunlight and wind.
197 The scale of decision is also defined during this step. The scale of decision selected should be the
198 smallest, most appropriate subset of the population for which decisions will be made based on
199 the spatial or temporal boundaries. During Step 4, the project planning team also should identify
200 practical constraints on sampling and analysis that could interfere with full implementation of the
201 data collection design. These include time, personnel, equipment, and seasonal or meteorological
202 conditions when sampling is not possible or may bias the data.
203 In practice, the study boundaries are discussed when the decision makers agree on the problem
204 and its associated decision. For instance, a land area that may be contaminated or a collection of
205 waste containers would be identified as part of the problem and decision definition in Steps 1 and
206 2. The boundaries also would be considered when determining inputs to the decision in Step 3. If
207 the study boundaries had not been addressed before Step 4 or if new issues were raised during
208 Step 4, then Steps 1, 2, and 3 should be revisited to determine how Step 4 results are now
209 influencing the three previous steps.
210 The outputs of Step 4 are:
211 • A detailed description of the spatial and temporal boundaries of the problem; and
212 • Any practical constraints that may interfere with the sampling and analysis activities.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT B-7 DO NOT CITE OR QUOTE
-------
The Data Quality Objectives Process
213 B3.5 Outputs of DQO Process Steps 1 to 4 Lead Into Steps 5 to 7
214 At this stage in the DQO process, the project planning team has defined with a substantial degree
215 of detail the problem, its associated decision, and the inputs and boundaries for addressing that
216 problem. The project planning team knows whether it needs new data to fill specific gaps and
217 what that data should be. The remaining three steps are highly technical and lead to the selection
218 of the sampling and analysis design. Even when new data is not required (i.e., a data collection
219 design is not needed), the project planning team should continue with Steps 5 and 6 of the DQO
220 Process. By establishing the formal decision rule and the quantitative estimates of tolerable
221 decision error rates, the project planning team is assured that consensus has been reached on the
222 actions to be taken and information to establish criteria for DQA process.
223 It is important to emphasize that every effort must be made to assure that Steps 1 to 4 are
224 adequately addressed. If the necessary time is taken in addressing carefully the first four steps
225 and assuring consensus among the project planning team, then the three remaining steps are less
226 difficult.
227 B3.6 DQO Process Step 5: Develop a Decision Rule
228 In Step 5, the project planning team determines the appropriate statistical parameter that
229 characterizes the population, specifies the action level, and integrates previous DQO process
230 outputs into a single "if..., then ..." statement (called a "decision rule") that describes a logical
231 basis for choosing among alternative actions. (The statistical parameters are discussed in more
232 detail in Chapter 19, Measurement Statistics.)
233 The four main elements to the decision rule are:
234 1. THE PARAMETER OF INTEREST. A descriptive measure (e.g., mean, median, or proportion) that
235 specifies the characteristic or attribute that the decision maker would like to know and that
236 the data will estimate. The characteristics that define the population and domain of interest
237 was established in Step 3.
238 2. THE SCALE OF DECISION MAKING. The smallest, most appropriate subset for which decisions
239 will be made. The scale of decision making was previously defined in Step 4.
240 3. THE ACTION LEVEL. A threshold value of the parameter of interest that provides the criterion
241 for choosing among alternatives. Action levels may be based on regulatory standards or they
MARLAP JULY 2001
DO NOT CITE OR QUOTE B-8 DRAFT FOR PUBLIC COMMENT
-------
The Data Quality Objectives Process
242 may be derived from site- and analyte-specific criteria such as dose or risk analysis. The basis
243 for the action level was determined in Step 3.
244 4. THE ALTERNATIVE ACTIONS. The actions the decision maker would take, depending on the
245 "true value" of the parameter of interest. The alternative actions were determined in Step 2.
246 The decision rule is a logical, sequential set of steps to be taken to resolve the problem. For
247 example, "If one or more conditions exits then take action 1, otherwise take action 2."
248 The outputs of Step 5 are:
249 • The action level;
250 • The statistical parameter of interest; and
251 • An "if..., then ..." statement that defines the conditions that would cause the decision maker
252 to choose among alternative courses of action.
253 B3.7 DQO Process Step 6: Specify the Limits on Decision Errors
254 In Step 6 of the DQO process, the project planning team assesses the potential consequences of
255 making a wrong decision and establishes a tolerable level for making a decision error. The
256 project planning team defines the types of decision errors (Type I and n) and the tolerable limits
257 on the decision error rates. In general, a Type I error is deciding against the default assumption
258 (the null hypothesis) when it is actually true; a Type II error is not deciding against the null
259 hypothesis when it is actually false (see Attachment Bl and Appendix C for detailed
260 discussions). The limits on the decision errors will be used to establish measurement
261 performance criteria for the data collection design.
262 Traditionally, the principles of statistical hypothesis testing (see Chapter 19) have been used to
263 determine tolerable levels of decision error rates. Other approaches applying decision theory have
264 been applied (Bottrell, et al., 1996a,b). Based on an understanding of the possible consequences
265 of making a wrong decision in taking alternative actions, the project planning team chooses the
266 null hypotheses and judges what decision error rates are tolerable for making a Type I or Type n
267 decision error.
268 The project planning team also specifies a range of possible values where the consequences of
269 decision errors are relatively minor (the gray region). Specifying a gray region is necessary
270 because variability in the population and imprecision in the measurement system combine to
271 produce variability in the data such that the decision may be "too close to call" when the true
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT B-9 DO NOT CITE OR QUOTE
-------
The Data Quality Objectives Process
272 value is very near the action level. The gray region establishes the minimum distance from the
273 action level where it is most important that the project planning team control Type II errors. (For
274 additional information on the gray region, hypothesis testing, and decision errors, see EPA
275 (2000), NRC (1998), and Chapter 19, Measurement Statistics.)
276 The tolerable decision error rates are used to establish performance goals for the data collection
277 design. Overall variability in the result can be attributed to several sources, including sample
278 location, collection, and handling; laboratory handling and analysis; and data handling and
279 analysis. In many environmental cases, sampling is a much larger source of uncertainty than
280 laboratory analyses. The goal is to develop a sampling and analysis design that reduces the
281 chance of making a wrong decision. The greater certainty demanded by the decision makers, the
282 more comprehensive and expensive the data collection process is likely to be. In this step, the
283 project planning team has to come to an agreement on how to determine acceptable analytical
284 uncertainty and how good the overall data results are required to be. The team has to reach a
285 consensus on the trade-off between the cost of more information and the increased certainty in
286 the resulting decision.
287 Often the project planning team does not feel comfortable with the concepts and terminology of
288 hypothesis testing (Type I and Type II errors, gray zone, critical region, tolerable decision error
289 rates). As a result the project planning team may have difficulty (or want to skip) this step of the
290 directed planning process. If these steps are skipped or insufficiently addressed, it is more likely
291 that the data will not be of the quality needed for the project. Attachment Bl is provided to give
292 some additional guidance on these concepts. MARLAP recommends that for each radionuclide
293 of concern an action level, gray region and limits on decision error rates be established during a
294 directed planning process.
295 Figure B2 summarizes the outputs of the decisions made by the project planning team in a
296 Decision Performance Goal Diagram (EPA, 2000). The horizontal axis represents the (unknown)
297 true value of the parameter being estimated. The vertical axis represents the decision maker's
298 desired probability of concluding that the parameter exceeds an action limit. The "gray region"
299 (bounded on one side by the action level) defines an area where the consequences of decision
300 error are relatively minor (in other words, it defines how big a divergence from the action level
301 we wish to distinguish). The gray region is related to the desired precision of the measurements.
302 The height of the indicated straight lines to the right and left of the gray region depict the
303 decision maker's tolerance for Type I and Type II errors.
MARLAP JULY 2001
DO NOT CITE OR QUOTE B-10 DRAFT FOR PUBLIC COMMENT
-------
The Data Quality Objectives Process
1
! 0.8
! 0.8
| ,7
; M
I M
l»
! °-2
I 0-1
jD.05
L 0
Tolerable
Type II
Decision
Error Rates
Type I Decision
Error Rates
Performance Curve
Gray Region
(Relatively Large
Decision ErrorRates are
Considered Tolerable.)
100 12* 1<0 175
Action Level
True Value of the Parameter (Mean Concentration)
j.ss
0.8
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
l|
*3
II
!'
&
i
i!
1
0.8
0.8
0.7
0.6
0.5
M
0.3
0.2
j 0.1
\°°t
Tolerable
Type I
Decision
Error Rates
Tolerable
Type 11 Decision
Error Rates
Performance
Curve
Gray Region ~
(Relatively Large -
Decision ErrorRates are
Considered Tolerable.)
100 125 150 175 200
"-Action Level
True Value of the Parameter (Mean Concentration)
Figure B2(a)—Decision performance goal diagram null
hypothesis: the parameter exceeds the action level.
Figure B2(b)—Decision performance goal diagram null
hypothesis: the parameter is less than the action level.
304 For purposes of this example, the default assumption (null hypothesis) was established as the
305 measured concentration exceeded the action level (Figure B2a). The Type I error (5 percent at
306 true concentration between 100 and 150; Ipercent at >150 units) making a decision NOT to take
307 action to solve an environmental problem (e.g., remediate) when that action was in fact required
308 (e.g., analyte concentrations are really above an action level). The Type n error (5 percent at true
309 concentrations <25 units; 10 percent between 25 and 75 units) is understood as taking an action
310 when in fact that action is not required (e.g., analyte concentrations are really below the action
311 level).
312 In Figure B2(b), the default assumption (null hypothesis) was established as the measured
313 concentration is less than the action level. The Type I error (5 percent at true concentrations <25
314 units; 10 percent between 25 and 100 units) is understood as taking an action when in fact that
315 action is NOT required (e.g., analyte concentrations are really below the action level). The Type
316 II error (10 percent at true concentration between 100 and 150; 5 percent at >150 units) is
317 understood as making a decision not to take action to solve an environmental problem (e.g.,
318 remediate) when that action was in fact required (e.g., analyte concentrations are really above an
319 action level).
320 The output of Step 6 is:
321 • The project planning team's quantitative measure of tolerable decision error rates based
322 on consideration of proj ect resources.
JULY 2001
DRAFT FOR PUBLIC COMMENT
B-ll
MARLAP
DO NOT CITE OR QUOTE
-------
The Data Quality Objectives Process
323 B3.8 DQO Process Step 7: Optimize the Design for Obtaining Data
324 By the start of Step 7, the project planning team has established their priority of concerns, the
325 definition of the problem, the decision or outcome to address the posed problem, the inputs and
326 boundaries, and the tolerable decision error rates. They have also agreed on decision rules that
327 incorporate all this information into a logic statement about what action to take in response to the
328 decision. During Step 7, the hard decisions are made between the planning team's desire to have
329 measurements with greater certainty and the reality of the associated resource needs (time, cost,
330 etc.) for obtaining that certainty.
331 During Step 7, the project planning team optimize the sampling and analytical design and
332 established the measurement quality objectives (MQOs) so the resulting data will meet all the
333 established constraints in the most resource-effective manner. The goal is to determine the most
334 efficient design (combination of sample type, sample number and analytical procedures) to meet
335 all the constraints established in the previous steps. Once the technical specialists and the rest of
336 the project planning team come to agreement about the sampling and analysis design, the
337 operational details and theoretical assumptions of the selected design should be documented.
338 If a proposed design cannot be developed to meet the limits on decision error rates within budget
339 or other constraints, then the project planning team will have to consider relaxing the error
340 tolerance, adjusting the width of the gray region, redefining the scale of decision, or committing
341 more funding. There is always a trade off between quality, cost and time. The project planning
342 team will need to develop a consensus on how to balance resources and data quality. If the
343 proposed design requires analysis using analytical protocols not readily available, the project
344 planning team must consider the resources (time and cost) required to develop and validate a
345 method, generate method detection limits relevant to media of concern, and develop appropriate
346 QA/QC procedures and criteria (Chapter 6, Selection and Application of an Analytical Method).
347 If the project entails a preliminary investigation of a site or material for which little is known, the
348 planners may choose to employ MQOs and requirements that typically are achieved by the
349 selected sampling and analytical procedures. At this early point in the project, the lack of detailed
350 knowledge of the site or material may postpone the need for the extra cost of more expensive
351 sampling and analytical procedures and large numbers of samples, until more site or material
352 knowledge is acquired. The less-demanding MQOs, however, should be adequate to further
353 define the site or material. For situations when the measured values are distant from an action
354 level the MQO-compliant data could also be sufficient to support the project decision.
MARLAP JULY 2001
DO NOT CITE OR QUOTE B-12 DRAFT FOR PUBLIC COMMENT
-------
The Data Quality Objectives Process
355 The planning of data collection activities is typically undertaken to determine if a characteristic
356 of an area or item does or does not exist above an action level. Since the area of interest (popula-
357 tion) is usually too large to be submitted to analyses, in its entirety, these data collection activities
358 generally include sampling. If sampling is done correctly, the field sample or set of field samples
359 will represent the characteristics of interest and, if analyzed properly, the information gleaned
360 from the samples can be used to make decisions about the larger area. However, if errors occur
361 during implementation of the project, the samples and associated data may not accurately reflect
362 the material from which the samples were collected and incorrect decisions could be made.
363 The planning team attempts to anticipate, quantify, and minimize the uncertainty in decisions
364 resulting from imprecision, bias, and blunders—or in other words, attempts to manage uncer-
365 tainty by managing its sources. The effort expended in managing uncertainty is project dependent
366 and depends upon what constitutes an acceptable level of decision uncertainty and the proximity
367 of the data to a decision point. For example, Figure B3(a) presents a situation where the data
368 have significant variability. Yet the variability of the data does not materially add to the
369 uncertainty of the decision since the measurements are so far removed from the action level.
370 More resources could be expended to control the variability. However, the additional expenditure
371 would be unnecessary, since they would not alter the decision or measurably increase confidence
372 in the decision.
373 In contrast, Figure B3(b) depicts data with
374 relatively little variability, yet this level of
375 variability is significant since the measured
376 data are adjacent to the action level, which
377 results in increased uncertainty in the
378 decision. Depending upon the consequences
379 of an incorrect decision, it may be advisable
380 to expend more resources with the intention
381 of increasing confidence in the decision.
382 The output of Step 7 is:
383 • The most resource-effective design for
384 sampling and analysis that will obtain
385 the specific amount and quality of data
386 needed to resolve the problem within
387 the defined constraints; and
JULY 2001
DRAFT FOR PUBLIC COMMENT
(a)
Action
Level
I Mean
Concentration
(b)
j Action
1 Level
Mean
Concentration
Figure B3 — How Proximity to the action level determines
what is an acceptable level of uncertainty.
MARLAP
B-13 DO NOT CITE OR QUOTE
-------
The Data Quality Objectives Process
388 • Detailed plans and criteria for data assessment.
389 B3.9 References
390 American National Standards Institute (ANSI) N42.23. American National Standard
391 Measurement and Associated Instrument Quality Assurance for Radioassay Laboratories.
392 1996.
393 American Society for Testing and Materials (ASTM) D5283. Standard Practice for Generation
394 of Environmental Data Related to Waste Management Activities: Quality Assurance and
3 95 Quality Control Planning and Implementation. 1992.
396 American Society for Testing and Materials (ASTM) D5792. Standard Practice for Generation
397 of Environmental Data Related to Waste Management Activities: Development of Data
398 Quality Objectives, 1995.
399 American Society for Testing and Materials (ASTM) D6051. Standard Guide for Composite
400 Sampling and Field Subsampling for Environmental Waste Management A ctivities. 1996.
401 Bottrell, D., S. Blacker and D. Goodman. 1996a. "Application of Decision Theory Methods to
402 the Data Quality Objectives Process." In, Proceedings of the Computing in Environmental
403 Resource Management Conference, Air and Waste Management Association.
404 Bottrell, D., N. Wentworth, S. Blacker and D. Goodman. 1996b. "Improvements to Specifying
405 Limits on Decision Errors in the Data Quality Objectives Process." In, Proceedings of the
406 Computing in Environmental Resource Management Conference, Air and Waste
407 Management Association.
408 MARSSEVI. 1997. Multi-Agency Radiation Survey and Site Investigation Manual. NUREG-
409 1575, EPA 402-R-97-016.
410 U.S. Environmental Protection Agency (EPA). 2000. Guidance for the Data Quality Objective
411 Process (EPA QA/G-4). EPA/600/R-96/055, Washington, DC. available from www.epa.gov/
412 qualityl/qa_docs.html.
413 U.S. Environmental Protection Agency (EPA). 1986. Development of Data Quality Objectives,
414 Description of Stages I and II. Washington, DC.
MARLAP JULY 2001
DO NOT CITE OR QUOTE B-14 DRAFT FOR PUBLIC COMMENT
-------
The Data Quality Objectives Process
415 U.S. Nuclear Regulatory Commission (NRC). 1998. A Nonparametric Statistical Methodology
416 for the Design and Analysis of Final Status Decommissioning Surveys. NUREG-1505, Rev.
417 1.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT B-15 DO NOT CITE OR QUOTE
-------
418
419
ATTACHMENT B-l DECISION ERROR RATES
AND THE GRAY REGION
420
B-l.l Introduction
421 This attachment is provided to present some additional discussion on decision error rates and the
422 gray region. The project planning team will need to specify a range of possible values where the
423 consequences of decision errors are relatively minor—the "gray region." Specifying a gray region
424 is necessary because variability in the population and imprecision in the measurement system
425 combine to produce variability in the data such that the decision may be "too close to call" when
426 the true value is very near the action level. The gray region establishes the minimum distance
427 from the action level, where it is most important that the project planning team control Type II
428 errors.
429 B-l.2 The Region of Interest
430 The first step in constructing the
431 gray region is setting the range of
432 concentrations that is a region of
433 interest (a range of possible values).
434 Usually there is an action level (such
435 as the derived concentration guide-
436 line level, a regulatory limit) that
437 should not be exceeded. If the
438 project planning team wants a
439 method to measure sample concen-
440 trations around this level, they would
441 not select one that worked at concen-
442 trations at 10 to 100 times the action
443 level, nor would they select one that
444 worked from zero to half the action level. They would want a method that worked well around
445 the action level—perhaps from 0.1 to 10 times the action level, or from one-half to two times the
446 action level. For the purpose of the example in this attachment, the action level is 1.0 and the
447 project planning team selected a region of interest that is zero to twice the action level (0-2), as
448 shown on the x-axis in Figure B-l.l.
>
y
&
Actic
*
f
n Ee^
el
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1
Concentration
8 2
FIGURE B-l.l
DRAFT MARLAP
DO NOT CITE OR QUOTE
B-16
JUNE 2001
NOT FOR DISTRIBUTION
-------
Decision Error Rates and the Gray Region
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
B-1.3 Measurement Uncertainty at the Action Level
The action level marks the concentration level that the project planning team must be able to
distinguish. The project planning team wants to be able to tell if the measured concentration is
above or below the action level. Does this mean that the project planning team needs to be able
to distinguish 0.9999 times the action level from 1.0001 times the action level? Sometimes, but
not usually. This is fortunate, because current measurement techniques are probably not good
enough to distinguish that small a difference in concentrations.
How close to the action level can
the project planning team plan to
measure? For this example, we will
assume that the standard uncertainty
(1 sigma, a) of the measured
concentration is 10 percent of the
action level. With that kind of
measurement "precision," can the
project planning team tell the
difference between a sample with
0.9 times the action level from one
right at the action level? Not
always. Figure B-1.2 shows the
distribution of the concentration that
is measured (assuming a normal distribution). This means that about 16 percent of the time, the
measured concentration (in the shaded area) will appear to be 0.9 times the action level or less,
even though the true concentration is exactly equal to the action level.
Concentra tion
FIGURE B-1.2
Similarly, about 16 percent of the
time, the measured concentration
will appear to be at or above the
action level (as shown in the shaded
area in Figure B-1.3), even though
the true concentration is only 0.9
times the action level.
The problem is, when there is only
the measurement result to go by, the
project planning team cannot tell the
JUNE 2001
NOT FOR DISTRIBUTION
FIGURE B-1.3
B-17
DRAFT MARLAP
DO NOT CITE OR QUOTE
-------
Decision Error Rates and the Gray Region
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
difference with confidence. If the measured concentration is 0.9, it is more likely that the true
concentration is 0.9 than it is 1.0, but there remains a chance that it is really 1.0.
B-1.4 The Null Hypothesis
If the measured concentration is 0.95,
it is equally likely that the true
concentration is 0.9 as it is 1.0 (see
Figure B-1.4). How does the project
planning team decide what is the true
concentration? The project planning
team starts by asking:
"Which mistake is worse: (1) saying
the true concentration is 0.9 when it
is 1.0 or more? or (2) saying the true
concentration is 1.0 when it is 0.9 or
less?"
0.95
\ \
0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8
Conce ntra tion
FIGURE B-1.4
498 What does the project planning team mean by "worse"? The project planning team really does
499 not want to make a mistake that is likely to remain undiscovered or will be difficult or expensive
500 to correct.
501
Case 1: Assume The True Concentration is Over 1.0
502 If a true concentration of 1.0 or more is over a regulatory limit, the project planning team will not
503 want to make mistake (1) above. If the project planning team decides the true concentration is
504 less than 1.0, the project planning team is not likely to look at the sample again. That would
505 mean that the mistake would probably not be discovered until much later, if at all. On the other
506 hand, if the project planning team decides that the true concentration is over 1.0 when it really is
507 not, the project planning team will discover the mistake while they are trying to figure out how to
508 "correct" the high reading. So the project planning team will make a rule: Assume the true
509 concentration is over 1.0 unless they are really sure it is under. This is the default assumption, the
510 "null hypothesis."
511 How sure does the project planning team need to be? For this example, we will assume that the
512 project planning team would like to be 95 percent sure. To be 95 percent sure, they would have to
DRAFT MARLAP
DO NOT CITE OR QUOTE
B-18
JUNE 2001
NOT FOR DISTRIBUTION
-------
Decision Error Rates and the Gray Region
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
stay with their assumption that the
true concentration is over 1.0 unless
the measured concentration is 0.84
or less (Figure B-1.5). The project
planning team knows that this will
only happen about 5 percent of the
time when the true concentration is
really 1.0. That is, the measurement
has to be less than 0.84 to be 95
percent sure the true concentration
is less than 1.0.
FIGURE B-1.5
But what if the true concentration is
0.9 or less—mistake (2) above?
Under the new rule (default assumption or null hypothesis), how often will the project planning
team say that the true concentration is over 1.0 when it is really only 0.84? As seen in Figure B-
1.6, there is only a 50-50 chance of making the right decision when the true concentration really
is 0.84. That is the price of being sure they are not over the action level.
530
53 1
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
How low does the true concentration
have to be in order to have a pretty
good chance of deciding that the
true concentrati on i s b el ow the
limit? To be 95 percent sure, the
true concentration needs to be twice
as far below the action level as the
decision point, namely at about 0.68.
That is, the project planning team
will need a concentration of 0.68 or
less to be 95 percent sure that they
will be able to decide the true
concentration is less than 1.0 (see
the unshaded portion in Figure B-
1.7). In other words, it is only when the true concentration is 0.68 or less that the project planning
team can be pretty sure that they will decide the true concentration is less than 1 .0. (Note how
similar this looks to an MDC in reverse.)
0.2
0.4
0.6
0.8 1 1.2
Concentra tion
1.6
FIGURE B-1.6
JUNE 2001
NOT FOR DISTRIBUTION
B-19
DRAFT MARLAP
DO NOT CITE OR QUOTE
-------
Decision Error Rates and the Gray Region
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
0.2
0.4
0.6
1.4
1.6
Case 2: Assume The True
Concentration is 0.9
As stated previously, the mistake
that is most serious determines the
null hypothesis. Suppose that the
project planning team determined
that it is worse to decide that the true
concentration is over 1.0 when it is
0.9 (than it is to decide it is 0.9
when it is 1.0). Then, the default
assumption (the null hypothesis)
would be that the true concentration
is 0.9, unless the measured
concentration is large enough to convince the planning team otherwise. Only when the measured
concentration reaches 1.06 does the planning team decide the true concentration is over 1.0
(Figure B-1.8). The team will have to have a true concentration of 1.22 or more to be 95 percent
sure that they will be able to decide the true concentration is over 1.0.
B-1.5 The Critical Region
0.8 1 1.2
Concentration
FIGURE B-1.7
0.4
0.6
0.8 1 1.2
Concontrstion
1.4
1.6
1.8
The mistake that is "worse" defines
the null hypothesis and also defines
a "Type I" error. The probability of a
Type I error happening is called the
"Type I error rate," and is denoted
by alpha (a). Under the original null
hypothesis (Case 1: Assume the true
concentration is over 1.0), a Type I
error would be deciding that the
concentration was less than 1.0
when it really was not. In general, a
Type I error is deciding against the null hypothesis when it is actually true. (A Type I error is also
called a "false positive." This can be confusing when the null hypothesis appears to be a
"positive" statement. Therefore, MARLAP uses the neutral terminology.)
The "less serious" mistake is called a Type n error, and the probability of it happening is the
"Type n error rate," denoted by beta (p). Under the original null hypothesis that the concentration
FIGURE B-1.8
DRAFT MARLAP
DO NOT CITE OR QUOTE
B-20
JUNE 2001
NOT FOR DISTRIBUTION
-------
Decision Error Rates and the Gray Region
581 was 1.0 or more, a Type II error would be deciding that the concentration was more than 1.0
582 when it really was not. In general, a Type II error is not deciding against the null hypothesis when
583 it is actually false.
584 In both Case 1 and Case 2, the probability of both Type I errors and Type n errors were set to 5
585 percent. The probabilities were calculated at multiples of the standard deviation, assuming a
586 normal distribution. This will not always be the case. However, the probability of a Type I error
587 is always calculated as the probability that the project planning team will decide to reject the null
588 hypothesis when it is actually true. This is simple enough, as long as there is a clear boundary for
589 the parameter of interest.
590 The parameter of interest in both Case 1 and Case 2 was the true concentration. The true
591 concentration had a limit of 1.0. Therefore, all the project planning team had to do was calculate
592 the probability that they would get a measured concentration that would cause them to decide
593 that the true concentration was less than 1.0, even though it was equal to 1.0. In the example, the
594 project planning team actually started with the probability (5 percent) and worked out the critical
595 value. The "critical value" (or decision point) is the measured value that divides the measurement
596 results into two different sets: (1) those values that will cause us to reject the null hypothesis and
597 (2) those values that will cause us to leave the null hypothesis as the default. Set (1) is called the
598 "critical region."
599 The Type I and Type n error rates, a and p, often are both set at 5 percent. This is only by
600 tradition. They do not have to be equal. Neither error rate needs to be set at 5 percent. The way
601 the project planning team should set the value is by examining the consequences of making a
602 Type I or a Type II error. What consequences will happen as a result of making each type of
603 error? This is a little different than the criterion that was used to define the null hypothesis. It
604 may be that in some circumstances, a Type II error is riskier than a Type I error. In that case,
605 consider making a bigger than p
606 B-1.6 The Gray Region
607 In the previous sections (B-l. 1 to B-l.4) the project planning team:
608 • Set the region of interest for the measured concentrations between zero and about twice the
609 action level;
610 • Assumed that the true concentration exceeds 1.0, unless they measure "significantly" below
611 that, the default assumption (null hypothesis);
JUNE 2001 DRAFT MARLAP
NOT FOR DISTRIBUTION B-21 DO NOT CITE OR QUOTE
-------
Decision Error Rates and the Gray Region
612 • Defined "significantly below" to mean a concentration that would be observed less than 5
613 percent of the time, when the true concentration is actually 1.0. To describe their uncertainty,
614 the project planning team used the normal distribution, with a relative standard deviation of
615 10 percent at the action level, as a model;
616 • Developed an operational decision rule: If the measured concentration is less than 0.84, then
617 decide the true concentration is less than 1.0. Otherwise, decide there is not enough reason to
618 change the default assumption (null hypothesis); and
619 • Found using this operational decision rule that they were pretty sure (95 percent) of deciding
620 that the true concentration is less than 1.0 only when the true concentration is actually 0.68 or
621 less.
622 If the true concentration is between 0.68 and 1.0, all the project planning team really can say is
623 that the probability of deciding that the true concentration is less than 1.0 will be between 5
624 percent (when the true concentration is 1.0) and 95 percent (when the true concentration is 0.68).
625 Conversely, when the true concentration is in this range, the probability of deciding that the true
626 concentration is not less than 1.0 (i.e., the probability of a Type n error) will be between 5
627 percent (when the true concentration is 0.68) and 95 percent (when the true concentration is just
628 under 1.0). This range of concentrations is called the "gray region."
629 When the null hypothesis is that the true concentration exceeds the action level (1.0), the gray
630 region is bounded from above by the action level. This is where a is set. It is bounded from
631 below at the concentration where p is set. There is some flexibility in setting the lower boundary
632 of the gray region (LBGR). If the project planning team specifies a concentration, they can
633 calculate the probability p. If they specify p, they can calculate the value of the true concentration
634 that will be correctly detected as being below 1.0 with probability 1-p.
635 In our example, the project planning team found that they needed the true concentration to be
636 0.68 or less to be at least 95 percent sure that they will correctly decide (by observing a measured
637 value of 0.84 or less) that the true concentration is less than 1.0. If the project planning team
638 doesn't like that, the project planning team can find that a true concentration of 0.71 will be
639 correctly detected 90 percent of the time (also by observing a measured value of 0.84 or less).
640 The critical value, or decision point, is determined by a, not p.
641 If the project planning team decides to raise the LBGR (i.e., narrow the gray region) the Type n
642 error rate at the LBGR goes up. If they lower the LBGR (i.e., widen the gray region) the Type n
DRAFT MARLAP JUNE 2001
DO NOT CITE OR QUOTE B-22 NOT FOR DISTRIBUTION
-------
Decision Error Rates and the Gray Region
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
0.92
error rate at the LBGR goes down. Nothing substantive is really happening. The project planning
team is merely specifying the ability to detect that the null hypothesis is false.
If the project planning team wants to make a substantive change, they need to change the
probability that an error is made. That is, they need to change the uncertainty (standard deviation)
of the measurements. Suppose the relative standard deviation of the measurements at the action
level is 5 percent instead of 10 percent. Then the value of the true concentration that will be
correctly detected to be below the action level (by observing a measured value of 0.92 or less) 95
percent of the time, is 0.84. Cutting
the standard deviation of the
measurement in half has cut the
(absolute) width of the gray region
in half, but left the width of the gray
region in standard deviations
unchanged. Previously, with a = 10
percent, the width of the gray region
was 1.0 - 0.68 = 0.32 = 3.2 (0.10) =
3.2a. As Figure B-1.9 illustrates,
with a = 5 percent, the width of the
gray region is 1.0-0.84 = 0.16 = 3.2
(0.05) = 3.2a.
o.e 1 1.2
Conce ntration
FIGURE B-1.9
What is important is the width of the gray region in standard deviations; not the width of the gray
region in concentration. In order to achieve the same specified Type n error rate at the LBGR, the
action level and the LBGR must be separated by the same number of standard deviations. The
width of the gray region (action level minus LBGR) will be denoted by delta (A), the "shift." A/a
is how many standard deviations wide the gray region is. A/a is called the "relative shift."
If the gray region is less than one standard deviation wide, the Type n error rate may be high at
the LBGR. The only way to improve the situation would be to decrease the standard deviation
(i.e., increase the relative shift, A/a). This can be done by employing a more precise measurement
method or by averaging several measurements. When the width of the gray region is larger than
about three standard deviations (i.e., A/a exceeds 3), it is overkill. It may be possible to use a
simpler, less expensive measurement method or take fewer samples.
JUNE 2001
NOT FOR DISTRIBUTION
B-23
DRAFT MARLAP
DO NOT CITE OR QUOTE
-------
i APPENDIX C
2 MEASUREMENT QUALITY OBJECTIVES
3 FOR METHOD UNCERTAINTY AND
4 DETECTION AND QUANTIFICATION CAPABILITY
5 C.I Introduction
6 This appendix expands on issues related to measurement quality objectives (MQOs) for several
7 method performance characteristics which are introduced in Chapter 3, Key Analytical Planning
8 Issues and Developing Analytical Protocol Specifications. Specifically, this appendix provides
9 the rationale and guidance for establishing project-specific MQOs for the following method per-
10 formance characteristics: method uncertainty, detection capability and quantification capability.
11 In addition, it provides guidance in the development of these MQOs for use in the method selec-
12 tion process and guidance in the evaluation of laboratory data based on the MQOs. Section C.2 is
13 a brief overview of statistical hypothesis testing as it is commonly used in a directed planning
14 process, such as the Data Quality Objectives (DQO) Process (EPA 2000). More information on
15 this subject is provided in Chapter 2, Directed Planning Process and Appendix B, The Data
16 Quality Objectives Process. Section C.3 derives MARLAP's recommended criteria for establish-
17 ing project-specific MQOs for method uncertainty, detection capability, and quantification capa-
18 bility. These criteria for method selection will meet the requirements of a statistically based
19 decision-making process. Section C.4 derives MARLAP's recommended criteria for evaluation
20 of the results of quality control analyses by project managers and data reviewers (see also Chap-
21 ter 8, Radiochemical Data Verification and Validation).
22 It is assumed that the reader is familiar with the concepts of measurement uncertainty, detection
23 capability, and quantification capability, and with terms such as "standard uncertainty," "mini-
24 mum detectable concentration," and "minimum quantifiable concentration," which are intro-
25 duced in Chapter 1, Introduction to MARLAP, and discussed in more detail in Chapter 19,
26 Measurement Statistics. MARLAP also uses the term "method uncertainty" to refer to the pre-
27 dieted uncertainty of the result that would be measured if the method were applied to a hypo-
28 thetical laboratory sample with a specified analyte concentration. The method uncertainty is a
29 characteristic of the analytical method and the measurement process.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT C-1 DO NOT CITE OR QUOTE
-------
MQOs For Method Uncertainty and Detection and Quantification Capability
so C.2 Hypothesis Testing
31 Within the framework of a directed planning process, one considers an action level, denoted here
32 by AL, which is the contaminant concentration in either a population (e.g., a survey unit) or an
33 individual item (e.g., a laboratory sample) that should not be exceeded. Statistical hypothesis
34 testing is used to decide whether the actual contaminant concentration Xis greater than AL. For
35 more information on this topic, see EPA QA/G-4, MARSSIM, NUREG-1505 (EPA 2000,
36 MARSSIM 2000, NRC 1998), or Appendix B of this manual.
37 In hypothesis testing, one formulates two hypotheses about the value of X, and evaluates the
38 measurement data to choose which hypothesis to accept and which to reject.1 The two hypotheses
39 are called the null hypothesis H0 and the alternative hypothesis Hj. They are mutually exclusive
40 and together describe all possible values of Asunder consideration. So, in any given situation, one
41 and only one of the hypotheses must be true. The null hypothesis is presumed true unless the data
42 provide evidence to the contrary. Thus the choice of the null hypothesis determines the burden of
43 proof in the test.
44 Most often, if the action level is not zero, one assumes it has been exceeded unless the measure-
45 ment results provide evidence to the contrary. In this case, the null hypothesis is H0: X> AL and
46 the alternative hypothesis is H^ X< AL. If one instead chooses to assume the action level has not
47 been exceeded unless there is evidence to the contrary, then the null hypothesis is FL; X< AL
48 and the alternative hypothesis is H^ X> AL. The latter approach is the only reasonable one if
49 AL = 0, because it is virtually impossible to obtain statistical evidence that an analyte concentra-
50 tion is exactly zero.
51 In any hypothesis test, there are two possible types of decision errors. A Type /error occurs if the
52 null hypothesis is rejected when it is, in fact, true. A Type //error occurs if the null hypothesis is
53 not rejected when it is false.2 Since there is always measurement uncertainty, one cannot elimi-
54 nate the possibility of decision errors. So instead, one specifies the maximum Type I decision
55 error rate a that is allowable when the contaminant concentration is at or above the action
1 In hypothesis testing, to "accept" the null hypothesis only means not to reject it, and for this reason many
statisticians avoid the word "accept" in this context. A decision not to reject the null hypothesis does not imply the
null hypothesis has been shown to be true.
2 The terms "false positive" and "false negative" are synonyms for "Type I error" and "Type II error,"
respectively. However, MARLAP deliberately avoids these terms here, because they may be confusing when the
null hypothesis is an apparently "positive" statement, such as X> AL.
MARLAP JULY 2001
DO NOT CITE OR QUOTE C-2 DRAFT FOR PUBLIC COMMENT
-------
MQOs For Method Uncertainty and Detection and Quantification Capability
56 level AL. This maximum usually occurs when the concentration is exactly equal to AL. The most
57 commonly used value of a is 0.05, or 5%. One also chooses another concentration DL (the "dis-
58 crimination limit") that one wishes to be able to distinguish reliably from the action level. One
59 specifies the maximum Type II decision error rate p that is allowable when the contaminant con-
60 centration equals DL, or, equivalently, the "power" 1 - p of the statistical test at X= DL. The
61 gray region is then defined as the interval between the two concentrations AL and DL.
62 The gray region is a set of concentrations close to the action level, where one is willing to tol-
63 erate a Type n decision error rate that is higher than p. For concentrations above the upper bound
64 of the gray region or below the lower bound, the decision error rate is no greater than the speci-
65 fied value (either a or p as appropriate). Ideally, the gray region should be narrow, but in practice,
66 its width is determined by balancing the costs involved, including the cost of measurements and
67 the estimated cost of a Type n error, possibly using prior information about the project and the
68 parameter being measured.
69 If H0 is X> AL (presumed contaminated), then the upper bound of the gray region is AL and the
70 lower bound is DL. If H0 is x < AL (presumed uncontaminated), then the lower bound of the gray
71 region is AL and the upper bound is DL. Since no assumption is made here about which form of
72 the null hypothesis is being used, the lower and upper bounds of the gray region will be denoted
73 by LBGR and UBGR, respectively, and not by AL and DL. The width of the gray region
74 (UBGR - LBGR) is denoted by A and called the shift or the required minimum detectable
75 difference in concentration (EPA 2000, MARSSEVI 2000, NRC 1998). See Appendix B, The
76 Data Quality Objectives Process, for graphical illustrations of these concepts.
77 Chapter 3 of MARLAP recommends that for each radionuclide of concern, an action level, gray
78 region, and limits on decision error rates be established during a directed planning process.
79 Section C.3 presents guidance on the development of MQOs for the selection and development
80 of analytical protocols. Two possible scenarios are considered. In the first scenario, the parameter
81 of interest is the mean analyte concentration for a sampled population. The question to be
82 answered is whether the population mean is above or below the action level. In the second
83 scenario a decision is to be made about individual items or specimens, and not about population
84 parameters. This is the typical scenario in bioassay, for example. Some projects may involve both
85 scenarios. For example, project planners may want to know whether the mean analyte concentra-
86 tion in a survey unit is above an action level, but they may also be concerned about individual
87 samples with high analyte concentrations.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT C-3 DO NOT CITE OR QUOTE
-------
MQOs For Method Uncertainty and Detection and Quantification Capability
88 C.3 Development of MQOs for Analytical Protocol Selection
89 This section derives MARLAP's recommendations for establishing MQOs for the analytical
90 protocol selection and development process. Guidance is provided for establishing project-
91 specific MQOs for method uncertainty, detection capability, and quantification capability. Once
92 selected, these MQOs are used in the initial, ongoing, and final evaluations of the protocols.
93 MARLAP considers two scenarios and develops MQOs for each.
94 SCENARIO I: A Decision Is to Be Made about the Mean of a Sampled Population
95 In this scenario the total variance of the data a2 is the sum of two components
96 a2 = ^ + 0$
97 where a^is the average analytical method variance (M= "method") and a^is the variance of the
98 sampled population. The sampling standard deviation a5 may be affected by the spatial and tem-
99 poral distribution of the analyte, the extent of the survey unit, the physical sample sizes, and the
100 sample collection procedures. The analytical standard deviation aMis affected by laboratory
101 sample preparation, subsampling, and analysis procedures. The value of aMmay be estimated by
102 the combined standard uncertainty of a measured value for a sample whose concentration equals
103 the hypothesized population mean concentration (see Chapter 19, Measurement Statistics).
104 The ratio A / a, called the "relative shift," determines the number of samples required to achieve
105 the desired decision error rates a and p. The target value for this ratio should be between 1 and 3,
106 as explained in MARSSIM and NUREG-1505 (MARSSIM 2000, NRC 1998). Ideally, to keep
107 the required number of samples low, one prefers that A / a ~ 3. The cost in number of samples
108 rises rapidly as the ratio A / a falls below 1, but there is little benefit from increasing the ratio
109 much above 3.
110 Generally, it is easier to control aMthan a5. If a5 is known (approximately), a target value for aM
111 can be determined. For example, if a5 < A / 3, then a value of aM no greater than JA2 / 9 - a^
112 ensures that a < A / 3, as desired. If a5 > A / 3, the requirement that the total a be less than A / 3
113 cannot be met regardless of aM. In the latter case, it is sufficient to make aM negligible in com-
114 parison to a5.
115 Often one needs a method for choosing aMin the absence of specific information about a5. In this
116 situation, MARLAP recommends the requirement aM < A / 10 by default. The recommendation is
117 justified below.
MARLAP JULY 2001
DO NOT CITE OR QUOTE C-4 DRAFT FOR PUBLIC COMMENT
-------
MQOs For Method Uncertainty and Detection and Quantification Capability
118 Since it is desirable to have a < A / 3, this condition is adopted as a primary requirement.
119 Assume for the moment that a5 is large. Then aM should be made negligible by comparison.
120 Generally, aM is considered negligible if it is no greater than about a5 / 3. When this condition is
121 met, further reduction of aMhas little effect on a and therefore is usually not cost-effective. So,
122 the inequality aM < a5/ 3 is adopted as a second requirement.
123 Algebraic manipulation of the equation a2 = a^+ a^ and the required inequality aM < a5/ 3 gives
124 a
•/To
125 The inequalities a < A / 3 and aM < a / JlQ together imply the requirement
126 A
127 or approximately
128
129 The required upper bound for the standard deviation aM will be denoted by <5MR. MARLAP
130 recommends
131 A
GM?~ 10
132 by default as a requirement in Scenario I when a5 is unknown. This upper bound was derived
133 from the assumption that a5 was large, but it also ensures that the primary requirement a < A / 3
134 will be met if a5is small. When the analytical standard deviation aMis less than GMK, the primary
135 requirement will be met unless the sampling variance a| is so large that a^ is negligible by com-
136 parison, in which case little benefit can be obtained from further reduction of aM.
137 The recommended value of <5MR is based on the assumption that any known bias in the measure-
138 ment process has been corrected and that any remaining bias is much smaller than the shift, A,
139 when a concentration near the gray region is measured.
140 Achieving an analytical standard deviation aMless than the recommended limit, A / 10, may be
141 difficult in some situations, particularly when the shift, A, is only a fraction of UBGR. When the
142 recommended requirement for aMis too costly to meet, project planners may allow a^to be
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT C-5 DO NOT CITE OR QUOTE
-------
MQOs For Method Uncertainty and Detection and Quantification Capability
143 larger, especially if a5 is believed to be small or if it is not costly to analyze the additional
144 samples required because of the larger overall data variance (a^+ a|). In this case, project
145 planners may choose om to be as large as A / 3 or any calculated value that allows the data
146 quality objectives to be met at an acceptable cost.
147 The true standard deviation, aM, is a theoretical quantity and is never known exactly, but the lab-
148 oratory may estimate its value using the methods described in Chapter 19, and Section 19.6.13 in
149 particular. The laboratory's estimate of aM will be denoted here by uMand called the "method
150 uncertainty." The method uncertainty, when estimated by uncertainty propagation, is the
151 predicted value of the combined standard uncertainty ("one-sigma" uncertainty) of the analytical
152 result for a laboratory sample whose concentration equals UBGR. Note that the term "method
153 uncertainty" and the symbol UM actually apply not only to the method but to the entire
154 measurement process.
155 In theory, the value om is intended to be an upper bound for the true standard deviation of the
156 measurement process, aM, which is unknown. In practice, <5MR is actually used as an upper bound
157 for the method uncertainty, t/M, which may be calculated. Therefore, the value of om will be
158 called the "required method uncertainty" and denoted by um. As noted in Chapter 3, MARLAP
159 recommends that project planners specify an MQO for the method uncertainty, expressed in
160 terms of UMR> for each analyte and matrix.
161 The MQO for method uncertainty is expressed above in terms of the required standard deviation
162 of the measurement process for a laboratory sample whose analyte concentration is at or above
163 the upper bound of the gray region, UBGR. In principle the same MQO may be expressed as a
164 requirement that the minimum quantifiable concentration (MQC) be less than or equal to UBGR.
165 Chapter 19 defines the MQC as the analyte concentration at which the relative standard deviation
166 of the measured value (i.e., the relative method uncertainty) is 1 / kQ, where kQ is some specified
167 positive value. The value of kQ in this case should be specified as kQ = UBGR / um. In fact, if the
168 lower bound of the gray region is zero, then one obtains kQ = 10, which is the value most com-
169 monly used to define the MQC in other contexts. In practice the requirement for method uncer-
170 tainty should only be expressed in terms of the MQC when kQ = 10, since to define the MQC
171 with any other value of kQ may lead to confusion.
MARLAP JULY 2001
DO NOT CITE OR QUOTE C-6 DRAFT FOR PUBLIC COMMENT
-------
MQOs For Method Uncertainty and Detection and Quantification Capability
172
173
174
175
176
177
EXAMPLE: Suppose the action level is 1 Bq/kg and the lower bound of the gray region is 0.6
Bq/kg. If decisions are to be made about survey units based on samples, then the required
method uncertainty at 1 Bq/kg is
MR
MR
= °-04
If this uncertainty cannot be achieved, then an uncertainty as large as A / 3 =0.13 Bq/kg may
be allowed if a5 is small or if more samples are taken per survey unit.
178 A common practice in the past has been to select an analytical method based on the minimum
179 detectable concentration (MDC), which is defined in Chapter 19, Measurement Statistics. For
180 example, the Multi-Agency Radiation Survey and Site Investigation Manual (MARSSEVI 2000)
181 says:
182 During survey design, it is generally considered good practice to select a measure-
183 ment system with an MDC between 10-50% of the DCGL [action level].
184 Such guidance implicitly recognizes that for cases when the decision to be made concerns the
185 mean of a population that is represented by multiple laboratory samples, criteria based on the
186 MDC may not be sufficient and a somewhat more stringent requirement is needed. It is inter-
187 esting to note that the requirement that the MDC (about 3 times aM) be 10-50% of the action
188 level is tantamount to requiring that aMbe 0.03 to 0.17 times the action level — i.e. the relative
189 standard deviation should be approximately 10% at the action level. Thus, the requirement is
190 more naturally expressed in terms of the MQC.
191 SCENARIO II: Decisions Are to Be Made about Individual Items
192 In this scenario, the total variance of the data equals the analytical variance, a^. Consequently the
193 data distribution in most instances should be approximately normal. The decision in this case
194 may be made by comparing the measured concentration, x, plus or minus a multiple of its com-
195 bined standard uncertainty to the action level, AL. The combined standard uncertainty, uc(x), is
196 assumed to be an estimate of the true standard deviation of the measurement process as applied
197 to the item being measured; so, the multiplier of uc(x) equals zl_ a, the (1 - a)-quantile of the stan-
198 dard normal distribution (see Appendix G, Statistical Tables).
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT C-7 DO NOT CITE OR QUOTE
-------
MQOs For Method Uncertainty and Detection and Quantification Capability
199 Alternatively, if AL is zero, so that any detectable amount of analyte is of concern, the decision
200 may involve comparing xto the critical value of the concentration, xc, as defined in Chapter 19,
201 Measurement Statistics.
202 Case II-l: Suppose the null hypothesis is x > AL, so that the action level, AL, equals the upper
203 bound of the gray region, UBGR. Given the analytical variance a^, only a measured result that is
204 less than about UBGR - zl_fl<5M will be judged to be clearly less than the action level. Then the
205 desired power of the test 1 - p is achieved at the lower bound of the gray region only if LBGR <
206 UBGR - zl_fl<5M- Z[_paM. Algebraic manipulation transforms this requirement to
UBGR- LBGR A
Zl-a+Zl-B Zl-a+Zl
207 Case II-2: Suppose the null hypothesis is x < AL, so that the action level, AL, equals the lower
208 bound of the gray region, LBGR. Then only a measured result that is greater than about LBGR +
209 z1_0!aM will be judged to be clearly greater than the action level. Then the desired power of the
210 test 1 - p is achieved at the upper bound of the gray region only if UBGR > LBGR + zl_aGM +
211 ZI-^OM- Algebraic manipulation transforms this requirement to
UBGR- LBGR A
7+7 7+7
1-a 1-p 1-a 1-p
212 So, in either case, we have the requirement:
A
7 +7
Z- Z
213 Therefore, MARLAP recommends the use of
MR
7 +7
1-a 1-p
214 as an MQO for method uncertainty when decisions are to be made about individual items (i.e.,
215 laboratory samples) and not about population parameters.
MARLAP JULY 2001
DO NOT CITE OR QUOTE C-8 DRAFT FOR PUBLIC COMMENT
-------
MQOs For Method Uncertainty and Detection and Quantification Capability
216 If both a and p are at least 0.05, one may use the value UMR = 0.3A.
217 If LBGR = 0, then A = UBGR and om = A / (zl_a + Zj_p) implies
UBGR
218 This requirement is essentially equivalent to requiring that the MDC not exceed UBGR. Thus,
219 when LBGR = 0, the MQO may be expressed in terms of the detection capability of the analytical
220 method.
221 Note that when AL = LBGR = 0, the MQO for detection capability may be derived directly in
222 terms of the MDC, since the MDC is defined as the analyte concentration at which the proba-
223 bility of detection is 1 - p when the detection criterion is such that the probability of false detec-
224 tion in a sample with zero analyte concentration is at most a.
225
226
227
228
EXAMPLE: Suppose the action level is 1 Bq/L, the lower bound of the gray region is 0.5
Bq/L, a = 0.05, and p = 0.10. If decisions are to be made about individual items, then the
required method uncertainty at 1 Bq/L is
A 1-0.5 0.5 -
^0.90 I- 645 + 1.282
229 C.4 The Role of the MQO for Method Uncertainty in Data Evaluation
230 This section provides guidance and equations for determining warning and control limits for QC
231 sample results based on the project-specific MQO for method uncertainty. In the MARLAP
232 Process as described in Chapter 1, these warning and control limits are used in the ongoing eval-
233 uation of protocol performance (see Chapter 7, Evaluating Protocols and Laboratories) and in
234 the evaluation of the laboratory data (see Chapter 8, Radiochemical Data Verification and
235 Validation).
236 C.4.1 Uncertainty Requirements at Various Concentrations
237 When project planners follow MARLAP's recommendations for establishing MQOs for method
238 uncertainty for method selection and development, the maximum allowable standard deviation,
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT C-9 DO NOT CITE OR QUOTE
-------
MQOs For Method Uncertainty and Detection and Quantification Capability
239 aMft at the upper bound of the gray region (UBGR) is specified. During subsequent data evalua-
240 tion, the standard deviation at any concentration less than UBGR should be at most aMff, and the
241 relative standard deviation at any concentration greater than UBGR should be at most
242 <5MR I UBGR, which will be denoted here by q>MK. Note that, since the true standard deviation can
243 never be known exactly, in practice the requirement is expressed in terms of the required method
244 uncertainty, UMK, to which the combined standard uncertainty of each result may be compared.
245
246
247
248
249
250
251
252
253
254
EXAMPLE: Consider the preceding example, in which AL = UBGR = 1 Bq/L, LBGR =
0.5 Bq/L, and UMR= 0.17 Bq/L. In this case the combined standard uncertainty for any meas-
ured result ^should be at most 0.17 Bq/L if x< 1 Bq/L, and the relative combined standard
uncertainty should be at most 0.17 / 1, or 17%, if x> 1 Bq/L.
In Scenario I, where decisions are made about the mean of a population based on multiple physi-
cal samples (e.g., from a survey unit), if the default value <5MR = A /10 is assumed for the required
method uncertainty, then the required bound for the analytical standard deviation as a function of
concentration is as shown in Figure C. 1 below. The figure shows that the bound, aReq, is constant
at all concentrations, x, below UBGR, and aReq increases with ^rwhen ^ris above UBGR. So,
aReq
, when x < UBGR and aReq = x •
/ UBGR when x> UBGR.
5
T3
w
C3
a>
0 LBGR UBGR
True Concentration (x)
FIGURE C.I — Required Analytical Standard Deviation (
-------
MQOs For Method Uncertainty and Detection and Quantification Capability
255 These requirements can be relaxed somewhat for samples with very high analyte concentrations
256 as long as the project's requirements for decision uncertainty are met. However, MARLAP does
257 not provide specific guidance to address this issue for Scenario I.
258 In Scenario II, where decisions are made about individual physical samples, it is possible to
259 widen the required bounds for the standard deviation at any concentration outside the gray
260 region. For example, suppose the upper bound of the gray region (UBGR) is at the action level
261 (AL), the lower bound (LBGR) is set at some concentration below UBGR, and the decision error
262 probabilities a and p are specified. Then the project planners require the probability of a Type I
263 error not to exceed a when the true concentration is at or above UBGR, and they require the
264 probability of a Type II error not to exceed p when the true concentration is at or below LBGR.
265 The decision rule is based on the combined standard uncertainty of the measurement result: any
266 sample whose measured concentration, X, exceeds AL minus z1_a times the combined standard
267 uncertainty, uc(x), is assumed to exceed the action level. So, assuming uc(x) is an adequate esti-
268 mate of the analytical standard deviation, the planners' objectives are met if
UBGR - x . f T orM,
, if x < LBGR
Zl-a+Zl-P
jr-LBGR .,. TTD™,
, if x > UBGR
Zl-a+Zl-p
Zl-a+Zl-P
-, if LBGR 1.0Bq/L
1.0 Bq/L
So, ifx = 0, the requirement is uc(x) < 1 / 2.927 = 0.34 Bq/L, and, ifx= 2, the requirement is
uc(x) < (2 - 0.5) / 2.927 = 0.51 Bq/L, which is approximately 26% in relative terms.
JULY 2001
DRAFT FOR PUBLIC COMMENT
C-ll
MARLAP
DO NOT CITE OR QUOTE
-------
MQOs For Method Uncertainty and Detection and Quantification Capability
275 C.4.2 Acceptance Criteria for Quality Control Samples
276 The next issue to be addressed is how to set warning and control limits for quality control (QC)
277 sample results. These limits will be used by project data assessors to determine whether the lab-
278 oratory appears to be meeting MQOs. Presumably the lab has stricter internal QC requirements
279 (see Chapter 18, Laboratory Quality Control).
280 The development of acceptance criteria for QC samples will be illustrated with an example.
281 Assume the upper bound of the gray region (UBGR) is 5 Bq/kg (soil) and the lower bound of the
282 gray region (LBGR) is 1.5 Bq/kg. The width of the gray region is A = 5 - 1.5 = 3.5 Bq/kg.
283 Project planners, following MARLAP's guidance, choose the required method uncertainty at 5
284 Bq/kg (UBGR) to be
^V = |p 0-35 Bq/kg
285 or 7%. So, the maximum standard uncertainty at analyte concentrations less than 5 Bq/kg should
286 be um = 0.35 Bq/kg, and the maximum relative standard uncertainty at concentrations greater
287 than 5 Bq/kg should be (pm = 0.07, or 7%.
288 Although it is possible to relax these uncertainty criteria for samples with very high analyte con-
289 centrations, MARLAP recommends that the original criteria be used to develop acceptance limits
290 for the results of QC sample analyses.
291 C.4.2. 1 Laboratory Control Samples
292 It is assumed that the concentration of a laboratory control sample (LCS) is high enough that the
293 relative uncertainty limit (pMK = 0.07 is appropriate. The percent deviation for the LCS analysis is
294 defined as
SSR " SA
%D=
SA
295 where
296 SSR is the measured result (spiked sample result) and
297 SA is the spike activity (or concentration) added.
298 It is assumed that the uncertainty of SA is negligible; so, the maximum allowable relative stan-
299 dard deviation of %Z?is the same as that of the measured result itself, or q>MK x 100%. Then the 2-
MARLAP JULY 2001
DO NOT CITE OR QUOTE C- 1 2 DRAFT FOR PUBLIC COMMENT
-------
MQOs For Method Uncertainty and Detection and Quantification Capability
300
301
302
303
304
305
306
307
sigma warning limits for %D are ± 2(pMff x 100% and the 3-sigma control limits are
± 3(pMff x 100%. (In situations where (pMK is very small, the uncertainty of SA should not be
ignored.)
The requirements for LCSs are summarized below.
Laboratory Control Samples
Statistic:
%D=
SSR
SA
Warning limits: ±2^x100%
Control limits: ± 3qm x 100%
308
309
310
311
312
313
314
315
316
317
EXAMPLE
(UBGR = 5 Bq/kg, um = 0.35 Bq/kg, qm = 0.07.)
Suppose an LCS is prepared with a concentration of SA = 10 Bq/kg and the result of the
analysis is 11.61 Bq/kg with a combined standard uncertainty of 0.75 Bq/kg. Then
10
100% = 16.1%
The warning limits in this case are
and the control limits are
100% = ± 14%
x 100% = ±21%
So, the calculated value of %D is above the upper warning limit but below the control limit.
JULY 2001
DRAFT FOR PUBLIC COMMENT
C-13
MARLAP
DO NOT CITE OR QUOTE
-------
MQOs For Method Uncertainty and Detection and Quantification Capability _
318 C.4.2.2 Duplicate Analyses
319 Acceptance criteria for duplicate analysis results depend on the sample concentration, which is
320 estimated by the average x of the two measured results xl and x2.
x=
321 When x < UBGR, the warning limit for the absolute difference | xl - x2 \ is
2t/ v/2-2.83 t/
AffiV MR
322 and the control limit is
3 um UBGR, the acceptance criteria may be expressed in terms of the relative percent
325 difference (RPD), which is defined as
\x - x I
RPD = '_ 2 x 100%
x
326 The warning limit for RPD is
2m t/2 x 100% ~ 2.83 m x 100%
1 sVlK * ' sVlK
327 and the control limit is
3cp ,_, x 100%
1 sVlK * ' sVlK
328 The requirements for duplicate analyses are summarized below.
MARLAP JULY 2001
DO NOT CITE OR QUOTE C-14 DRAFT FOR PUBLIC COMMENT
-------
MQOs For Method Uncertainty and Detection and Quantification Capability
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
Duplicate Analyses
If* 5 Bq/kg, the acceptance criteria are expressed in terms of RPD.
RPD = 1?-^—Hi! x 100% = 37.84%
11.1
JULY 2001
DRAFT FOR PUBLIC COMMENT
C-15
MARLAP
DO NOT CITE OR QUOTE
-------
MQOs For Method Uncertainty and Detection and Quantification Capability
348
349
350
351
The warning and control limits for RPD are
Warning limit = 2.83
Control limit = 4.24
0.07 x 100% = 19.81%
0.07 x 100% = 29.68%
In this case, the value of RPD is above the control limit. (Also note that the relative standard
uncertainties are larger than the 7% required for concentrations above 5 Bq/kg.)
352
C.4.2.3 Method Blanks
353 Case 1. If an aliquant of blank material is analyzed, or if a nominal aliquant size is used in the
354 data reduction, the measured blank result is an activity concentration. The target value is zero,
355 but the measured value may be either positive or negative. So, the 2-sigma warning limits are
356 ± 2uMR and the 3-sigma control limits are ± 3 UMK.
357 Case 2. If no blank material is involved (only reagents, tracers, etc., are used), the measured
358 result may be a total activity, not a concentration. In this case the method uncertainty limit UMK
359 should be multiplied by the nominal or typical aliquant size, Ms. Then the 2-sigma warning limits
360 are ± 2 uMRMs and the 3-sigma control limits are ± 3 uMRMs.
361 The requirements for method blanks are summarized below.
362
363
364
365
366
367
368
369
370
Method Blanks
Concentration:
Statistic:
Warning limits:
Control limits:
Total Activity:
Statistic:
Warning limits:
Control limits:
Measured concentration
±2u.
-MR
'MR
Measured total activity
± 2 umMs
± 3 uMRMs
MARLAP
DO NOT CITE OR QUOTE
C-16
JULY 2001
DRAFT FOR PUBLIC COMMENT
-------
MQOs For Method Uncertainty and Detection and Quantification Capability
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
EXAMPLE
(UBGR = 5 Bq/kg, um= 0.35 Bq/kg, 9^= 0.07)
Suppose a method blank is analyzed and the result of the measurement is
x= 0.00020 Bq with combined standard uncertainty uc(x) = 0.00010 Bq
Assuming the nominal aliquant mass is 1.0 g, or Ms= 0.001 kg, the result is evaluated by
comparing xto the warning and control limits:
± 2 uMRMs = ± 0.00070 Bq
± 3 uMRMs = ± 0.00105 Bq
In this case xis within the warning limits.
C.4.2.4 Matrix Spikes
The acceptance criteria for matrix spikes are more complicated than those described above for
laboratory control samples because of pre-existing activity in the unspiked sample, which must
be measured and subtracted from the activity measured after spiking. The percent deviation for a
matrix spike is defined as
%D =
SSR SR SA
SA
100%
where
SSR is the spiked sample result
SR is the unspiked sample result
SA is the spike concentration added (total activity divided by aliquant mass).
However, warning and control limits for %D depend on the measured values; so, %Z?is not a
good statistic to use for matrix spikes. Instead we define a "Zscore"
Z =
SSR - SR - SA
max(SR, UBGR)2
JULY 2001
DRAFT FOR PUBLIC COMMENT
C-17
MARLAP
DO NOT CITE OR QUOTE
-------
MQOs For Method Uncertainty and Detection and Quantification Capability
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
where "max(^r, y)" denotes the maximum of xand y. Then warning and control limits for Zare set
at ± 2 and ± 3, respectively. (It is assumed again that the uncertainty of SA is negligible.)
The requirements for matrix spikes are summarized below.
Matrix Spikes
Statistic:
Z =
SSR - SR - SA
max(SR, UBGR)2
Warning limits: ± 2
Control limits: ± 3
EXAMPLE
(UBGR = 5 Bq/kg, um = 0.35 Bq/kg, qm = 0.07)
Suppose a matrix spike is analyzed. The result of the original (unspiked) analysis is
SR = 3.5 with combined standard uncertainty t/c(SR) = 0.29
the spike concentration added is
SA = 10.1 with combined standard uncertainty t/c(SA) = 0.31
and the result of the analysis of the spiked sample is
SSR =11.2 with combined standard uncertainty t/c(SSR) = 0.55
Since SR is less than UBGR (5), max(SR, UBGR) = UBGR = 5. So,
SSR-SR-SA 11.2 -3.5 - 10.1 _ _.
cp^SSR2 + UBGR2 0.07^/1 1.22 + 52
So, Zis less than the lower warning limit (-2) but slightly greater than the lower control limit
(-3).
MARLAP
DO NOT CITE OR QUOTE
C-18
JULY 2001
DRAFT FOR PUBLIC COMMENT
-------
MQOs For Method Uncertainty and Detection and Quantification Capability
412 C.5 References
413 Environmental Protection Agency (EPA). 2000. Guidance for the Data Quality Objectives
414 (DQO) Process. EPA QA/G-4. EPA/600/R-96/055, EPA, Quality Staff, Washington, DC.
415 MARSSEVL 2000. Multi-agency Radiation Survey and Site Investigation Manual (MARSSIM)
416 Rev. 1. NUREG-1575, Nuclear Regulatory Commission, Washington, DC. EPA 402-R-97-
417 016, Environmental Protection Agency, Washington, DC.
418 Nuclear Regulatory Commission (NRC). 1998. A Nonparametric Statistical Methodology for the
419 Design and Analysis of Final Status Decommissioning Surveys. NUREG-1505. NRC,
420 Washington, DC.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT C-19 DO NOT CITE OR QUOTE
-------
APPENDIX D CONTENT OF PROJECT PLAN DOCUMENTS
2 Dl.O Introduction
3 Project plan documents were discussed in Chapter 4, Project Plan Documents. This appendix
4 will discuss appropriate content of plan documents. The content of project plan documents,
5 regardless of the document title or format, will include similar information, including the project
6 description and objectives, identification of those involved in the project activities and their
7 responsibilities and authorities, enumeration of the quality control (QC) procedures to be
8 followed, reference to specific standard operating procedures (SOPs) that will be followed for all
9 aspects of the projects, and Health and Safety protocols.
10 The discussion of project plan document content in this appendix will rely on EPA's guidance on
11 elements for a QA project plan (QAPP). MARLAP selected EPA's QAPP as a model for content
12 of a project plan document since it is closely associated with the data quality objective (DQO)
13 planning process and because other plan documents lack widely accepted guidance regarding
14 content. MARLAP hopes that presentation of a project plan document in one of the most
15 commonly used plan formats will facilitate plan writing by those less familiar with the task,
16 provide a framework for reviewing plan documents, and aid in tracking projects.
17 The discussion of plan content in sections D2 to D5 follows the outline developed by EPA
18 requirements (EPA, 1998b) and guidance (EPA, 1998a) for QAPPs for environmental data
19 operations. The QAPP elements are presented in four major sections (Table Dl) that are referred
20 to as "groups":
21 • Proj ect Management;
22 • Measurement/Data Acquisition;
23 • Assessment/Oversight; and
24 • Data Validation and Usability.
25 There are many formats that can be used to present the project plan elements. MARLAP does not
26 recommend any particular plan format over another. The project planning team should focus on
27 the appropriate content of plan documents needed to address the necessary quality assurance
28 (QA), QC, and other technical activities that must be implemented to ensure that the results of
29 the work performed will satisfy the stated performance criteria. Table D2 provides a crosswalk
30 between the table of contents of two example project plan documents—a QAPP and a work
31 plan—and EPA's (1998a) project plan document elements.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT D-1 DO NOT CITE OR QUOTE
-------
Content of Project Plan Documents
TABLE Dl—QAPP Groups and Elementsab
GROUP
A Project Management
B Measurement/Data
Acquisition
C Assessment/Oversight
D Data Validation and
Usability
ID
Al
A2
A3
A4
A5
A6
A7
A8
A9
Bl
B2
B3
B4
B5
B6
B7
B8
B9
BIO
Cl
C2
Dl
D2
D3
ELEMENT
Title and Approval Sheet
Table of Contents
Distribution List
Project/Task Organization
Problem Definition/Background
Project/Task Description
Quality Objectives and Criteria for
Measurement Data
Special Training Requirements/Certifications
Documentation and Record
Sampling Process Design
Sample Methods Requirements
Sample Handling and Custody Requirements
Analytical Methods Requirements
QC Requirements
Instrument/Equipment Testing, Inspection and
Maintenance Requirements
Instrument Calibrations and Frequency
Inspection/ Acceptance Requirements for
Supplies and Consumables
Data Acquisition Requirements (Non-direct
Measurements)
Data Management
Assessments and Response Actions
Reports to Management
Verification and Validation Requirements
Verification and Validation Methods
Reconciliation with Data Quality Objectives
APPENDIX
SECTION
D2.1
D2.2
D2.3
D2.4
D2.5
D2.6
D2.7
D2.8
D2.9
D3.1
D3.2
D3.3
D3.4
D3.5
D3.6
D3.7
D3.8
D3.9
D3.10
D4.1
D4.2
D5.1
D5.2
D5.3
MARLAP
CHAPTER
NA
NA
NA
2
2
2
2,3
7
7, 17
NA
NA
11
6
18
15
18
NA
2
17
7
9
8
8
9
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
(a) Based on EPA, 1998a.
(b) MARLAP recommends a graded approach to project plan documents. All elements may not be applicable,
especially for a small project. See Chapter 4, Section 4.3, "A Graded Approach to Project Plan Documents"
and Section 4.5.3, "Plan Content for Small Projects."
This appendix also will discuss how the project plan document is linked to the outputs of the
project planning process. Directed project planning is discussed in Chapter 2, Project Planning
Process. The discussion of project plan documents in this appendix will use the DQO process
DRAFT MARLAP
DO NOT CITE OR QUOTE
D-2
JUNE 2001
NOT FOR DISTRIBUTION
-------
Content of Project Plan Documents
67 (EPA, 1994) as a model for directed planning (see Appendix B, The Data Quality Objectives
68 Process). References will be made in this appendix to the steps of the DQO process, where
69 appropriate, to illustrate the linkage between the direct planning process and plan documents.
70 It should be noted that although the project plan documents will address both sampling and
71 analysis, MARLAP does not provide guidance on sampling design issues or sample collection.
72 Discussion in D3.1, "Sample Process Design," and D3.2, "Sample Methods Requirements," are
73 provided for completeness and consistency.
74
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
D2.0 Group A: Project Management
75 This group consists of nine elements that address project management issues such as organiza-
76 tion of the plan itself, management systems, and a description of project goals, participants and
77 activities. These elements ensure that the project goals are clearly stated, the approach to be used
78 is understood, and the project planning decisions are documented.
TABLE D2—Comparison of Project Plan Contents
I. Example QAPPa using EPA Guidance11 and EPA QAPP Elements0
QA PROJECT PLAN FOR RADIOLOGICAL
MONITORING TABLE OF CONTENTS
Title Page
Approval Sheet
Distribution List
1.0 Table of Contents
2.0 Project Description
2.1 Site History
2.2 Project Objectives and Requirements
2.3 DQOs
3.0 Project Organization and Responsibility
4.0 QA Objectives for Measurement Data (Precision,
Accuracy, Representativeness, Comparability,
Completeness)
5.0 Sampling Procedures, including QC [Cited Field
Sampling and Analysis Plan]
6.0 Sample Custody
6.1 Sample
6.2 Sample Identification
6.3 COC Procedures
7.0 Calibration Procedures and Frequency (Field and
Laboratory)
EPA G-5 QA PROJECT PLAN ELEMENTS
Al Title and Approval Sheet
A3 Distribution List
A2 Table of Contents
A5 Problem Definition/Background
A6 Project/Task Description
A4 Project/Task Organization
A7 Quality Objectives and Criteria for Measurement
Data
B 1 Sampling Process Designs
B2 Sampling Methods Requirements
B3 Sample Handling and Custody Requirements
B7 Instrument Calibration and Frequency
JULY 2001
DRAFT FOR PUBLIC COMMENT
D-3
MARLAP
DO NOT CITE OR QUOTE
-------
Content of Project Plan Documents
QA PROJECT PLAN FOR RADIOLOGICAL
MONITORING TABLE OF CONTENTS
8.0 Analytical Procedures
8.1 Background
8.2 Specific Analytical Procedures
8.3 Test Methods
8.4 Control of Testing
8.5 Limits of Detection
9.0 Data Reduction, Validation and Reporting and
Record
10.0 Internal QC Checks
11.0 Performance and Systems Audits
11.1 Systems Audits
11.2 Surveillance
11.3 Performance Audits
11.4 Resolution of Discrepancies
11.5 Review of Contractor Procedures
12.0 Preventive Maintenance
13.0 Specific Routine Procedures to Assess Data
Precision, Accuracy, Completeness
14.0 Corrective Action
15.0 QA Report to Management
16.0 References
EPA G-5 QA PROJECT PLAN ELEMENTS
B4 Analytical Methods Requirements
B6 Instrument/Equipment Testing, Inspection, and
Maintenance Requirements
BIO Data Management
Dl Data review, Validation, and Verification
Requirements
A9 Documentation and Records
B5 Quality Control Requirements
Cl Assessment and Response Actions
B6 Instrument/Equipment Testing, Inspection, and
Maintenance Requirements
D3 Reconciliation with DQOs
C2 Response to Management
A8 Special Training Requirements/Certification
B8 Inspection/ Acceptance Requirements for Supplies
and Consumables
B9 Data Acquisition Requirement for Non-direct
Measurements
D2 Verification and Validation Methods
II. Example Work Pland and EPA QA/G-5 QAPP Elements0
Work Plan Table of Contents
Cover Letter
Title Page (including Document Number, Prepared
by/Prepared for Identification)
Approvals
Table of Contents
EPA QAPP Elements
A3 Distribution List
Al Title and Approval Sheet
Al Title and Approval Sheet
A2 Table of Contents
1 Introduction/Background
Site and Regulatory Background
A5 Problem Definition/Background
DRAFT MARLAP
DO NOT CITE OR QUOTE
D-4
JUNE 2001
NOT FOR DISTRIBUTION
-------
Content of Project Plan Documents
Work Plan Table of Contents
Project Scope and Purpose
Project Organization and Management
Data Quality Objectives and Approach
Environmental Setting
Sampling Site Selection, Locations and
Identification
EPA QAPP Elements
A6 Project/Task Description
A4 Project/Task Organization
A7 Quality Objectives and Criteria for Measurement Data
A5 Problem Definition/Background
B 1 Sampling Process Design
2 Sampling and Analysis Plan
Objective
QA Objectives for Field Measurements, Laboratory
Measurements (including Calibration Procedures
and Frequency)
Sample Collection Procedures
Sample Identification, Handling and Transport
Sample Analysis
Sample Tracking and Records
Data Reduction, Validation and Reporting
Internal QC Checks
B 1 Sampling Process Design
A7 Quality Objectives and Criteria for Measurement Data
B7 Instrument Calibrations and Frequency
B2 Sample Methods Requirements
B3 Sample Handling and Custody Requirements
B4 Analytical Methods Requirements
BIO Data Management
Dl Data Review, Verification, and Validation
Requirements
D2 Verification and Validation Methods
B5 QC Requirements
3 QA Project Plan
QA Training and Awareness
Performance and Systems Audits
Preventive Maintenance
Quality Improvement
QA Reports to Management
Purchase Items and Service Control
4 Data and Records Management Plan
Objectives
Data Management
Document Control
Records Management System
Administrative Records
5 Data Interpretation Plan
Approach for Data Evaluation
Data Interpretation and Comparisons
6 Risk Analysis Plan
7 Hoaltli ctnrt ftstfotv Plan
Cl Assessments and Response Actions
B6 Instrument/Equipment Testing, Inspection, and
Maintenance Requirements
B6 Instrument/Equipment Testing, Inspection, and
Maintenance Requirements
C2 Reports to Management
B8 Inspection/ Acceptance Requirements for Supplies and
Consumables
A9 Documentation and Record
BIO Data Management
D3 Reconciliation with DQOs
JULY 2001
DRAFT FOR PUBLIC COMMENT
D-5
MARLAP
DO NOT CITE OR QUOTE
-------
Content of Project Plan Documents
Work Plan Table of Contents
EPA QAPP Elements
B9
A8
Data Acquisition Requirements (Non-direct
Measurements)
Special Training Requirements/Certifications
168
169
170 (a) Plan elements adapted from DOE, 1997.
171 (b) EPA, 1980.
172 (c) EPA, 1998a
173 (d) Plan elements adapted from DOE, 1996.
174 D2.1 Project Management (Al): Title and Approval Sheet
175 The project title sheet should:
176 • Clearly identify the project in an unambiguous manner;
177 • Include references to organizational identifiers such as project numbers (when appropriate);
178 • Clearly label and distinguish between draft and approved versions;
179 • Include the date of issuance of drafts or final approved version;
180 • Include revision or version numbers;
181 • Indicate if the document represents only a portion of the QAPP (e.g., Volume 1 of 4
182 Volumes);
183 • Include names of the organization(s) preparing the plan document and, if different, for whom
184 the plan was prepared; and
185 • Identify clearly on the title page if the document is a controlled copy and subjected to no-
186 copying requirements. If so, indicate the document control number.
187 QAPPs should be reviewed on an established schedule. QAPPs should be kept current and
188 revised when necessary. Documented approval, as an amendment to the QAPP, should be
189 obtained for modifications to the QAPP.
190 The approval sheet documents that the QAPP has been reviewed and approved prior to
191 implementation. The approval sheet should consist of the name, title, organization, signature and
192 signature date for:
DRAFT MARLAP JUNE 2001
DO NOT CITE OR QUOTE D-6 NOT FOR DISTRIBUTION
-------
Content of Project Plan Documents
193 • The proj ect manager or other person with overall responsibility for the proj ect;
194 • The QA manager or other person with overall responsibility for the quality of the project
195 outputs;
196 • The project managers or QA managers for all organizations (e.g., sampling organization,
197 laboratories, data validators) implementing project activities; and
198 • The representative of any oversight or regulatory organization.
199 The project manager or other person with overall responsibility for the project should require an
200 approved QA program, management plan, or quality manual that supports all technical
201 operations, including data collection and assessment activities.
202 D2.2 Project Management (A2): Table of Contents
203 The table of contents should:
204 • List all sections and subsections of the document, references, glossaries, acronyms and
205 abbreviations, appendices (including sections and subsections) and the associated page
206 numbers;
207 • List all attachments and the associated page numbers;
208 • List all tables and associated page numbers;
209 • List all figures and diagrams and associated page numbers; and
210 • List titles of other volumes, if the QAPP consists of more than one volume.
211 A document control format is useful in maintaining reference to the latest version of the planned
212 document, especially when only portions of a document have been copied and are being used to
213 implement or discuss project activities.
214 D2.3 Project Management (A3): Distribution List
215 The distribution list should identify all individuals, along with their titles and organizations, who
216 will receive copies and revisions of the approved QAPP and subsequent revisions. Listed
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT D-7 DO NOT CITE OR QUOTE
-------
Content of Project Plan Documents
217 individuals should include, at a minimum, all managers and QA personnel responsible for the
218 implementation and quality of the data collection activities. The project planning team or the core
219 group (see Chapter 2, Section 2.4) should be included on the document distribution list.
220 D2.4 Project Management (A4): Project/Task Organization
221 This QAPP element should:
222 • Identify the individuals and/or organizations participating in the project, as well as contact
223 information (address, telephone number, fax number, e-mail). The stakeholders, data users,
224 decision makers, and technical planning team members, and the person or organization that
225 will be responsible for project implementation, are identified during the directed planning
226 process (Appendix B, The DQO Process, Steps 1 and 7).
227 • Discuss the roles and responsibilities of the individuals and/or organizations that participate
228 in the data collection, including the roles and responsibilities of the data users, decision
229 makers, and QA manager.
230 • Include an organizational chart clearly showing the relationship, lines of authority and
231 communication, and mechanisms for information exchange among all project participants.
232 Complex projects may require more than one organizational chart to properly describe the
233 relationships among participants. At times, to clearly detail an organizations responsibilities and
234 communications, a general inter-organizational chart with primary contacts, responsibilities, and
235 communications may need to be accompanied by secondary charts that describe intra-
236 organizational contacts, responsibilities, and lines of communication.
237 One of the keys to successful projects is communication. The QAPP should identify the point of
238 contact for resolving field and laboratory problems. The QAPP may also summarize the points of
239 contact for dissemination of data to managers, users and the public.
240 D2.5 Project Management (A5): Problem Definition/Background
241 The "Problem Definition/Background" element (A5) and the subsequent elements "Project/Task
242 Description" (A6) and "Quality Objectives and Criteria" (A7) constitute the project description.
243 Separating the project description into three elements focuses and encourages the plan authors to
244 address all key issues (identification of problem to be solved, description of site history,
245 description of tasks and the quality objectives and data-acceptance criteria), some of which can
DRAFT MARLAP JUNE 2001
DO NOT CITE OR QUOTE D-8 NOT FOR DISTRIBUTION
-------
Content of Project Plan Documents
246 be overlooked if a larger, less-focused section is written. Table D3 provides bulleted components
247 for these three elements. This section and sections D2.6 and D2.7 provide a more detailed
248 discussion of these elements.
249
250
251
252
253
254
255
256
257
258
259
260
261
262
TABLE D3—Content of the Three Elements that Constitute the Project Description
Problem Project/Task Objectives
Definition/Background Description and Criteria
(A5) (A6) (A7)
263
264
265
266
267
268
269
Serves as an Introduction
Identifies the "problem
to be solved" or the
"question to be
answered"
Identifies the regulatory,
legal or "informational
needs" drivers
Presents the historical
perspective
Describes measurements
Identifies regulatory
standards and action levels
Identifies special
personnel, procedural and
equipment requirements
Summarizes assessment
tools
Details schedule and
milestones
Identifies record and report
requirements
Quality Objectives
Problem definition/Site
history
Data inputs
Population boundaries
Tolerable decision error
rates
Criteria for Measurement
Data
Measurement quality
objectives (MQOs; such as
the measurement uncer-
tainty at some concentra-
tion; the detection capa-
bility; the quantification
capability; the range; the
specificity; and the
ruggedness of the method)
The Problem Definition/Background element provides a discussion of the problem and pertinent
background so that the implementation team can understand the context of the project. This
section does not discuss the details of project activities, which are described in a subsequent
project management element. Much of the information needed for this element was collected and
discussed during Step 1 of the DQO process (Appendix B3.1). The decision statement was
developed during Step 2 of the DQO process.
The "Problem Definition/Background" element should:
JULY 2001
DRAFT FOR PUBLIC COMMENT
D-9
MARLAP
DO NOT CITE OR QUOTE
-------
Content of Project Plan Documents
270 • Serve as an introduction to the project;
271 • Identify the "problem to be solved" or the "question to be answered" upon successful
272 completion of the project—the decision rule (Appendix B3.6);
273 • Discuss the assumptions, limitations, and scope of the project;
274 • Identify the regulatory, legal, or "informational needs" drivers that are the underlying reasons
275 for the project;
276 • Describe the context of the project so that it can be put into a historical perspective. This
277 section may include a description and maps of a facility or site, its location, its use, site
278 topography, geology and hydrogeology, past data collection activities, historical data
279 including analytes and concentrations, past and present regulatory status, past releases,
280 seriousness and potential risk of any release, site maps, and utilities; and
281 • If the data collection activity is in support of a technology evaluation, include a discussion of
282 the purpose of the demonstrations, how the technology works, operating conditions, required
283 utilities, effluents and waste by-products and residues, past and expected efficiencies and
284 multi-media mass-balances by analyte and matrix.
285 D2.6 Project Management (A6): Project/Task Description
286 This element of the QAPP provides a discussion of the project and underlying tasks for the
287 implementation teams. It should provide a description of the work to be performed to resolve the
288 problem or answer the question, including the following information:
289 • A description of the measurements and the associated QA/QC procedures that are to be made
290 during the course of the project. DQO Step 3 describes existing and needed data inputs, while
291 Step 7 yields the optimized sampling and analytical designs as well as quality criteria.
292 - Identification of the analytes of interest.
293 - A summary (preferably a table) of samples type (e.g., grab, spatial or temporal
294 composite), number of samples, analyte or analyte class (e.g., "Tc, transuranic, gamma
295 emitters) and analytical protocol specifications or method.
296 • A discussion of applicable regulatory standards or action levels to which measurements will
297 be compared. Identify any applicable regulatory standard (e.g., gross alpha drinking water
298 maximum contamination limit), or applicable or relevant and appropriate requirements
DRAFT MARLAP JUNE 2001
DO NOT CITE OR QUOTE D-10 NOT FOR DISTRIBUTION
-------
Content of Project Plan Documents
299 (ARARs) that will be used as a metric or action level during decision-making. The DQO Step
300 6 details action levels and tolerable decision errors that will be the basis for decisions.
301 • Identify any special requirements required to implement project tasks.
302 - Identify any special training (e.g., hazardous waste site health and safety training (29 CFR
303 1910.120), radiation safety training).
304 - Identify any special protective clothing and sampling equipment.
305 - Identify any boundary conditions (e.g., only sample after a rainfall of more than 1 inch).
306 - Specify any special document format, chain-of-custody, or archival procedures.
307 - Identify any special sample handling (e.g., freezing of tissue samples), instrumentation, or
308 non-routine analytical protocols that are required to achieve specified performance
309 criteria (e.g., very low detection limits) (see also Chapter 3, Critical Analytical Planning
310 Issues and Developing Analytical Protocol Specifications).
311 • Summarize the assessment tools that will be employed to determine whether measurement
312 data complied with performance criteria and are suitable to support decision-making. Include
313 a schedule of the assessment events. Assessment tools include performance evaluations,
314 program technical reviews, surveillance, technical and systems audits, and verification and
315 validation. Briefly outline:
316 - A first tier of reviews (e.g., when field or lab personnel check each other's notes or
317 calculations).
318 - Reviews of the work, notes and calculations of subordinates by the supervisor (e.g.,
319 review and sign all notebook entries).
320 - The percentage of data subject to review by internal QA staff.
321 - Data verification and validation to be performed by an independent party and the
322 guidelines or plan to be used.
323 - Assessment of proj ect activities to be conducted by personnel independent of proj ect
324 activities (e.g., performance evaluation samples, surveillance, audits).
325 - Assessment of how results of the proj ect will be reconciled with the proj ect DQOs ("data
326 quality assessment").
327 • Supply a schedule that includes start and completion dates for tasks and a list of completion
328 dates for important milestones. Dates can be calendric, or as number of days following
329 approval of the QAPP, or number of days following commencement of field operations.
330 DQO Steps 1 and 4 identify deadlines and other constraints that can impact scheduling.
331 • Identify the records and reports that will be required. This should be a brief but complete
332 listing of necessary reports and records (e.g., field and lab notebooks, sample logbooks,
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT D-11 DO NOT CITE OR QUOTE
-------
Content of Project Plan Documents
333 spectra, sample tracking records, laboratory information system print-outs, QA reports,
334 corrective action reports).
335 • Identify whether the original documents are required or if photocopies are sufficient. More
336 detailed information will be presented in "Documentation and Records" (A9) and "Data
337 Management" (BIO).
338 D2.7 Project Management (A7): Quality Objectives and Criteria for Measurement Data
339 This element addresses two closely related but different issues, quality objectives for the project
340 and criteria used to evaluate the quality of measurement data. The element summarizes outputs
341 from all steps of the DQO process. A fundamental principle underlying plan documents is that
342 requirements for the data quality must be specified by the project planning team and documented.
343 By clearly stating the intended use of the data and specifying qualitative and quantitative criteria
344 for system performance, a critical link between the needs of the project planning team and the
345 performance requirements to be placed on the laboratory data is established. (See Chapter 3 for a
346 discussion of MQOs.)
347 D2.7.1 Project's Quality Objectives
348 The project's quality objectives or data quality objectives (DQOs) are qualitative and quantitative
349 statements that:
350 • Clarify the intended use of the data (e.g., data will be used to determine if lagoon sediment
351 contains 232Th at concentrations greater than or equal to the action level);
352 • Define the type and quantity of data per matrix needed to support the decision (e.g., 232Th
353 concentrations in 300 composite sediments samples each composite consisting of 10 samples
354 randomly collected from a 100 m2 sampling grid adjacent to the point of discharge);
355 • Identify the conditions under which the data should be collected (e.g., sediment samples
356 collected from the top 6 cm of sediment within a 100 m radius of the point of discharge into
357 lagoon #1, following de-watering of the lagoon and prior to sediment removal); and
358 • Specify tolerable limits on the probability of making a decision error due to uncertainty in the
359 data and any associated action levels (e.g., 95 percent confidence that the true concentration
360 is actually below the action level).
DRAFT MARLAP JUNE 2001
DO NOT CITE OR QUOTE D-12 NOT FOR DISTRIBUTION
-------
Content of Project Plan Documents
361 Authors of project plan documents are often encouraged to condense the DQO outputs in a
362 summary statement. This approach can have value as long as critical information is not lost in the
363 summary process and the original information is cited and available for all project participants.
364 The following is an example of a DQO summary statement:
365 "The purpose of this project is to determine, to within a lateral distance of 10 m, the extent of
366 232Th in soil along a pipeline at concentrations at or above 1,145 mBq/g, with a false positive
367 rate less than or equal to 5 percent; and to define within 1 m the vertical extent of measured
368 232Th concentrations greater than 7,400 mBq/g."
369 D2.7.2 Specifying Measurement Quality Objectives
370 Measurement quality objectives (MQOs) or measurements performance criteria are essential to
371 the success of a project since they establish the necessary quality of the data. The quality of data
372 can vary as a result of the occurrence and magnitude of three different types of errors (Taylor,
373 1990).
374 • BLUNDERS—mistakes that occur on occasion and produce erroneous results (e.g., mis-
375 labeling or transcription errors);
376 • SYSTEMATIC ERRORS—mistakes that are always the same sign and magnitude and produce
377 bias (i.e., they are constant no matter how many measurements are made); and
378 • RANDOM ERRORS—mistakes that vary in sign and magnitude and are unpredictable on an
379 individual basis (i.e., random differences between repetitive readings) but will average out if
380 enough measurements are taken.
381 The frequent occurrence of these types of errors is the reason why data quality is subject to
382 question, why there is uncertainty when using data to make decisions and why measurement
383 performance criteria are necessary.
384 During the DQO process, project DQOs are used to establish the MQOs. An MQO is a statement
385 of a performance objective or requirement for a particular method performance characteristic.
386 Examples of method performance characteristics include the measurement uncertainty at some
387 concentration; the detection capability; the quantification capability; the range; the specificity;
388 and the ruggedness of the method. MQOs for the project should be identified and described
389 within this element of the QAPP. MARLAP provides guidance for developing MQOs for select
390 method performance characteristics in Chapter 3 and Appendix C.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT D-13 DO NOT CITE OR QUOTE
-------
Content of Project Plan Documents
391 D2.7.3 Relation between the Project DQOs, MQOs, and QC Requirements
392 The ultimate goal of all data collection operations is the collection of appropriately accurate data.
393 Appropriately accurate data are data for which errors caused by imprecision and bias are
394 controlled such that it is suitable for use in the context outlined by the DQOs (i.e., the overall
395 error is less than that specified in the acceptable decision error). During the optimization of
396 design in the planning process, DQO-specified decision error rates are translated into MQOs with
397 the intention of monitoring, detecting, quantifying and controlling imprecision and analytical
398 bias. During optimization, precautions are also incorporated into the design with the intention of
399 preventing blunders and types of non-measurable bias not susceptible to measurement by QC
400 samples.
401 The MQOs provide acceptance or rejection criteria for the quality control samples whose types
402 and frequency are discussed in the Quality Control Requirements element (B5) (Appendix C).
403 QC samples and the project's associated MQOs are key—but not the sole mechanisms—for
404 monitoring the achievement of DQOs.
405 In summary, translating acceptable decision error rates into a design that will produce data of
406 appropriate precision and bias is often a complex undertaking. The team must consider the
407 synergistic and antagonistic interactions of the different options for managing errors and
408 uncertainty. Accurate data require not only control of imprecision, but must also control the
409 various forms of bias.
410 D2.8 Project Management (A8): Special Training Requirements/Certification
411 All project personnel should be qualified and experienced in their assigned task(s). The purpose
412 of this element is to add additional information regarding special training requirements and how
413 they will be managed during implementation of the project. This element should:
414 • Identify and describe any mandated or specialized training or certifications that are required;
415 • Indicate if training records or certificates are included in the QAPP as attachments;
416 • Explain how training will be implemented and certifications obtained; and
417 • Identify how training documentation and certification records will be maintained.
418 D2.9 Project Management (A9): Documentation and Record
419 This element of the QAPP will identify which records are critical to the project, from data
420 generation in the field to final use. It should include what information needs to be contained in
DRAFT MARLAP JUNE 2001
DO NOT CITE OR QUOTE D-14 NOT FOR DISTRIBUTION
-------
Content of Project Plan Documents
421 these records and reports, the formats of the records and reports, and a brief description of
422 document control procedures. The following are suggested records and content:
423 • SAMPLE COLLECTION RECORDS should include sampling procedures, the names of the persons
424 conducting the activity, sample number, sample collection points, maps and diagrams,
425 equipment/protocol used, climatic conditions, and unusual observations. Bound field
426 notebooks, pre-printed forms, or computerized notebooks can serve as the recording media.
427 Bound field notebooks are generally used to record raw data and make references to
428 prescribed procedures, changes in planned activities and implementation of corrective
429 actions. Preferably, notebooks will contain pre-numbered pages with date and signature lines
430 and entries will be made in ink. Field QC issues such as field, trip, and equipment rinsate
431 blanks, co-located samples, field-spiked samples, and sample preservation should be
432 documented. Telephone logbooks and air bill records should be maintained.
433 • SAMPLE TRACKING RECORDS document the progression of samples as they travel from the
434 original sampling location to the laboratory and finally to their disposal or archival. These
435 records should contain sample identification, the project name, signatures of the sample
436 collector, the laboratory custodian and other custodians, and the date and time of receipt. The
437 records should document any sample anomalies. If chain-of-custody (COC) is required for
438 the project, the procedures and requirements should be outlined (Chapter 11, Sample Receipt,
439 Inspection and Tracking).
440 • ANALYTICAL QC issues that should be documented include standard traceability, and
441 frequency and results of QC samples, such as, method and instrument blanks, spiked
442 samples, replicates, calibration check standards and detection limit studies.
443 • ANALYTICAL RECORDS should include standard operating procedures for sample receipt,
444 preparation, analysis and report generation. Data report formats and the level of supporting
445 information is determined by data use and data assessment needs.
446 • PROJECT ASSESSMENT RECORDS should include audit check lists and reports, performance
447 evaluation (PE) sample results, data verification and validation reports, corrective action
448 reports. The project may want to maintain copies of the laboratory proposal package, pre-
449 award documentation, initial precision and bias test of the analytical protocol and any
450 corrective action reports.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT D-15 DO NOT CITE OR QUOTE
-------
Content of Project Plan Documents
451 The QAPP should indicate who is responsible for creating, tracking, and maintaining these
452 records and when records can be discarded, as well as any special requirements for computer,
453 microfiche, and paper records.
454 D3.0 Group B: Measurement/Data Acquisition
455 The Measurement/Data Acquisition group consists of 10 elements that address the actual data
456 collection activities related to sampling, sample handling, sample analysis and the generation of
457 data reports. Although these issues may have been previously considered by project management
458 elements, the project management section of the QAPP dealt with the overall perspective. The
459 measurement/data section contains the details covering design and implementation to ensure that
460 appropriate protocols are employed and documented. This section also addresses quality control
461 activities that will be performed during each phase of data collection from sampling to data
462 reporting.
463 D3.1 Measurement/Data Acquisition (Bl): Sampling Process Design
464 This element of the QAPP describes the finalized sampling design that will be used to collect
465 samples in support of project objectives. The design should describe the matrices to be sampled,
466 where the samples will be taken, the number of samples to be taken, and the sampling frequency.
467 A map of the sampling locations should be included to provide unequivocal sample location
468 determination and documentation.
469 If a separate sampling and analysis plan or a field sampling and analysis plan has been
470 developed, it can be included by citation or as an appendix. This element will not address the
471 details of standard operating procedures for sample collection, which will be covered in
472 subsequent elements. This element will describe the sampling design and the underlying logic, so
473 that implementation teams can understand the rationale behind and better implement the
474 sampling effort. Understanding the rationale for the decisions will help if plans have to be
475 modified due to conditions in the field. DQO Step 7 establishes the rationale for and the details
476 of the sampling design.
477 This element should restate the outputs of the planning process and any other considerations and
478 assumptions that impacted the design of the sampling plan, such as:
479 • The number of samples, including QC samples, sample locations and schedule, and rationale
480 for the number and location of samples;
DRAFT MARLAP JUNE 2001
DO NOT CITE OR QUOTE D-16 NOT FOR DISTRIBUTION
-------
Content of Project Plan Documents
481 • A brief discussion of how the sampling design will facilitate the achievement of project
482 objectives;
483 • A discussion of the population boundaries (temporal and spatial) and any accessibility
484 limitations;
485 • A description of how the sampling design accommodates potential problems caused by the
486 physical properties of the material being sampled (e.g., large particle size), the characteristic
487 of concern (e.g., potential losses due to the volatility of tritium) or heterogeneity;
488 • A discussion of the overarching approach to sampling design (e.g., worse case or best case
489 sampling versus average value) and assumptions made in selecting this approach (e.g., an
490 assumption that the darkened soil adjacent to the leaking tank would present a worse case
491 estimate of soil contamination);
492 • A listing of guidance and references that were relied upon when designing the sampling plan;
493 • Identification of the characteristics of interest (e.g.,"Tc activity), associated statistical
494 parameters (e.g., mean, standard deviations, 99th percentile), and acceptable false error rates
495 (e-g-, false negative rate of less than 5%);
496 • Identification of relevant action level and how data will be compared to the action level
497 (Appendix B3.2);
498 • A discussion of the anticipated range of the characteristic of interest and assumed temporal
499 and spatial variations (heterogeneity), anticipated variance, anticipated sources and
500 magnitude of error (e.g., heterogeneity of material being sampled, sampling imprecision,
501 analytical imprecision), anticipated mean values and distribution of measurements and the
502 basis (e.g., historical data, similar processes or sites) for any associated assumptions;
503 • If any level of bias is assumed, what is the assumed magnitude and the basis of the
504 assumption (e.g., historical data, typical analytical bias for matrix type);
505 • It is usually assumed that the magnitude of measurements made at individual sampling
506 locations are independent of each other (e.g., no correlation of concentration with location).
507 Geostatistical approaches may be more appropriate if measurements are significantly
508 correlated with locations (e.g., serial-correlation, auto-correlation) since serial-correlation can
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT D-17 DO NOT CITE OR QUOTE
-------
Content of Project Plan Documents
509 bias estimates of variance and invalidate traditional probabilistic techniques such as
510 hypothesis testing; and
511 • A discussion of the rationale for choosing non-routine sampling protocols and why these non-
512 routine protocols are expected to produce acceptable precision and bias.
513 D3.2 Measurement/Data Acquisition (B2): Sampling Methods Requirements
514 This element of the QAPP describes the detailed sampling procedures that will be employed
515 during the project. The preliminary details of sampling methods to be employed were established
516 during Step 7 of the DQO process. The selected sampling procedures should be appropriate to (1)
517 ensure that a representative sample is collected, (2) avoid the introduction of contamination
518 during collection, and (3) properly preserve the sample to meet project objectives. Written SOPs
519 should be included as attachments to the QAPP. This element and the appendices or other
520 documents that it references should in total contain all the project specific details needed to
521 successfully implement the sampling effort as planned. If documents to be cited in the QAPP are
522 not readily available to all project participants, they must be incorporated as appendices. All
523 sampling personnel should sign that they have read the sampling procedures and the health and
524 safety procedures.
525 Correct sampling procedures and equipment used in conjunction with a correct sampling design
526 should result in a collection of samples that in total will represent the population of interest. A
527 detailed discussion of sampling procedures, equipment and design are beyond the scope of
528 MARLAP. In general, the selected procedures must be designed to ensure that the equipment is
529 used properly and that the collected samples represent the individual sampling unit from which
530 samples are collected. The sampling equipment should be chemically and physically compatible
531 with the analyte of concern as well as the sample matrix. The sampling design should facilitate
532 access to individual sampling units, result in an appropriate mass/volume of sample such that it
533 meets or exceeds minimum analytical sample sizes, accommodates short-range heterogeneity
534 (i.e., does not preclude large particle sizes or lose small particles) and reduce or prevent loss of
535 volatile components, if appropriate.
536 This element of the QAPP should:
537 • Identify the sampling methods to be used for each matrix, including the method number if a
538 standardized method. If methods are to be implemented differently than specified by the
539 standard method or if the standard method offers alternatives for implementation, the
540 differences and alternatives should be specified;
DRAFT MARLAP JUNE 2001
DO NOT CITE OR QUOTE D-18 NOT FOR DISTRIBUTION
-------
Content of Project Plan Documents
541 • Identify the performance requirements of the sampling method. If the sampling method of
542 choice is unlikely to be able to achieve the level of performance demanded by the project
543 DQO, the project planning team should be notified;
544 • Identify the required field QC samples (e.g., trip blank, co-located duplicate);
545 • Identify any sample equipment preparation (e.g., sharpening of cutting edges, degreasing and
546 cleaning) or site preparation (e.g., removal of overburden, establishing dust-free work space
547 for filtering) for each method;
548 • Identify and preferably generate a list of equipment and supplies needed. For example, the
549 sampling devices, decontamination equipment, sampling containers, consumables (e.g., paper
550 towels), chain-of-custody seals and forms, shipping materials (e.g., bubble-pack, tape), safety
551 equipment and paper work (e.g., pens, field books);
552 • Identify and detail logistical procedures for deployment, sample shipment and demobili-
553 zation. If a mobile lab will be used, explain its role and the procedures for sample flow to the
554 mobile lab and data flow to the data-user;
555 • Identify, preferably in a tabular form, sample container types, sizes, preservatives, and
556 holding times;
557 • Identify procedures that address and correct problems encountered in the field (variances and
558 nonconformance to the established sampling procedures);
559 • Identify for each sampling method, decontamination procedures and the procedures for
560 disposing of contaminated equipment and used-decontamination chemicals and waters;
561 • Identify the disposal procedures for waste residuals generated during the sampling process
562 (e.g., purged well waters, drilling dregs) for each method; and
563 • Identify oversight procedures (e.g., audits, supervisor review) that ensure that sampling
564 procedures are implemented properly. The person responsible for implementing corrective
565 actions should be identified.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT D-19 DO NOT CITE OR QUOTE
-------
Content of Project Plan Documents
566 D3.3 Measurement/Data Acquisition (B3): Sample Handling and Custody Requirements
567 This element of the QAPP details how sample integrity will be maintained and how the sample
568 history and its custody will be documented ensuring that (1) samples are collected, transferred,
569 stored, and analyzed by authorized personnel, (2) the physical, chemical and legal integrity of
570 samples is maintained, and (3) an accurate written record of the history of custody is maintained.
571 DQO Step 1 describes the regulatory situation which can be used to identify the appropriate level
572 of sample tracking. The QAPP should state whether COC is required. Sample handling, tracking
573 and COC requirements are discussed in detail in Chapter 11, Sample Receipt and Tracking.
574 In the QAPP, the following elements should be documented:
575 • INTEGRITY OF SAMPLE CONTAINERS: Describe records to be maintained on the integrity of
576 sample container and shipping container seals upon receipt. Describe records to be
577 maintained if specially prepared or pre-cleaned containers are required.
578 • SECURITY: If wells are being sampled, whether the wellheads were locked or unlocked should
579 be noted. Security of remote sampling sites or automatic samplers not maintained in locked
580 cages should be discussed.
581 • SAMPLE IDENTIFICATION: The assignment of sample numbers and the labeling of sample
582 containers is explained. If samples are to be assigned coded sample identifications (IDs) to
583 preclude the possibility of bias during analysis, the sample code is one of the few items that
584 will not be included in the QAPP, since the lab will receive a copy. The code and sample ID
585 assignment process will have to be described in a separate document, which is made available
586 to the field team and the data validators. An example of a sample label should be included in
587 the QAPP.
588 • TRACKING OR CUSTODY IN THE FIELD: Procedures for sample tracking or custody while in the
589 field and during sample shipment should be described. When COC is required, a copy of the
590 COC form and directions for completion should be included. A list of all materials needed
591 for tracking or custody procedures should be provided (e.g., bound notebooks, shipping
592 containers, shipping labels, tape, custody seals, COC forms).
593 • SAMPLE PRESERVATION: Sample preservation procedures, if desired, should be clearly
594 described. Preservation of radiological samples is discussed in Chapter 10, Requirements
595 When Collecting, Preserving, and Shipping Samples That Require Analytical Measurement.
DRAFT MARLAP JUNE 2001
DO NOT CITE OR QUOTE D-20 NOT FOR DISTRIBUTION
-------
Content of Project Plan Documents
596 • TRACKING OR CUSTODY IN THE LABORATORY: A decision must be made as to whether the
597 laboratory in general is considered a secure area such that further security is not required once
598 the sample is officially received by the laboratory or whether internal tracking or custody
599 procedures will be required as the samples are handled by different personnel within the lab.
600 The laboratory's sample receipt SOP, laboratory security procedures, and—if needed—
601 internal tracking or custody procedures should be described.
602 • SPECIAL REQUIREMENT: Any special requirements, such as shipping of flammable or toxic
603 samples, or requirements for verification of sample preservation upon sample receipt by the
604 laboratory should be clearly described.
605 • ARCHIVAL: Document the rationale for the request to archive samples, extracts, and
606 digestates. Describe how samples, extracts, and digestates will be archived. Identify how long
607 samples, extracts, digestates, reports, and supporting documentation must be maintained.
608 D3.4 Measurement/Data Acquisition (B4): Analytical Methods Requirements
609 This element of the QAPP should identify the Analytical Protocol Specifications (APSs)
610 including the MQOs that were employed by the laboratory to select the analytical protocols. (See
611 Chapter 3 for guidance on developing APs.) This element integrates decisions from three DQO
612 steps: Step 3 which identified the analyte of interest and needed inputs to the decision, Step 6
613 which identifies the allowable uncertainty, and Step 7 which identifies the optimized analytical
614 design. Input from all three steps drive the choice of analytical protocols. The discussion of the
615 selected analytical protocols should address: subsampling, sample preparation, sample clean-up,
616 radiochemical separations, the measurement system, confirmatory analyses and pertinent data
617 calculation and reporting issues. A tabular summary of the analytical protocol by matrix type can
618 facilitate reference for both the plan document development team and the laboratory analytical
619 team.
620 This element of the QAPP should clearly describe the expected sample matrices (e.g.,
621 groundwater with no sediments, soils with no rocks larger than 2 cm in diameter) and what
622 should be done or who should be contacted if sample matrices are different than expected.
623 Subsampling is a key link in the analytical process which is often overlooked during planning
624 leaving important decisions to laboratory staff, this element should specify appropriate
625 subsampling procedures.
626 This QAPP element should:
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT D-21 DO NOT CITE OR QUOTE
-------
Content of Project Plan Documents
627 • Identify the laboratories supplying analytical support. If more than one laboratory will be
628 used, detail the analyses supplied by each laboratory;
629 • Identify analyses to be performed in the field using portable equipment or by a mobile lab;
630 • Identify the sample preparation techniques. Non-routine preparatory protocols, such as novel
631 radiochemical separations, should be described in detail and documented in an SOP including
632 pertinent literature citations and the results of validations studies and other performance data,
633 when they exist;
634 • Identify the analytical protocols to be used. The protocol documentation should describe all
635 necessary steps including the necessary reagents, apparatus and equipment, standards
636 preparation, calibration, sample introduction, data calculation, quality control, interferences,
637 and waste disposal;
638 • If the selected analytical protocols have not been demonstrated for the intended application,
639 the QAPP should include information about the intended procedure, how it will be validated,
640 and what criteria must be met before it is accepted for the project's application (Chapter 6,
641 Selection and Application of an Analytical Protocol);
642 • If potential analytical protocols were not identified during the project planning process and
643 existing analytical protocols can not meet the MQOs, an analytical protocol will have to be
644 developed and validated (Chapter 6, Section 6.5, "Method Validation"). If this issue was not
645 identified by the project planning team, the project planning team must be contacted because
646 the original project objectives and the associated MQOs may have to be revisited and
647 changed (Appendix B);
648 • If both high concentration and low concentration samples are expected, discuss how the two
649 sample types will be identified and handled in a manner that will prevent cross-contamination
650 or other analytical problems;
651 • Discuss reporting requirements (e.g., suitable data acquisition and print-outs or electronic
652 data archival that will capture all necessary information), the proper units (dry weight versus
653 wet weight), the method to be employed to report the final result and its uncertainty, and
654 reporting package format requirements; and
655 • Identify oversight procedures (e.g., QC samples, audits, supervisor review) for ensuring that
656 analytical procedures are implemented properly and procedures for correcting problems
DRAFT MARLAP JUNE 2001
DO NOT CITE OR QUOTE D-22 NOT FOR DISTRIBUTION
-------
Content of Project Plan Documents
657 encountered in the laboratory. The person responsible for implementing corrective actions in
658 the lab should be identified.
659 The project plan document should be a dynamic document, used and updated over the life of the
660 project as information becomes available or changes. For example, under a performance based
661 approach, the analytical protocols requirements in the project plan documents should initially
662 reflect the Analytical Protocol Specifications established by the project planning team and issued
663 in the statement of work (or task order). When the analytical laboratory has been selected
664 (Appendix E, Contracting Analytical Services) the project plan document should be updated to
665 reflect the identification of the selected laboratory and the analytical protocols, that is, the actual
666 analytical protocols to be used should be included by citation or inclusion of the SOPs as
667 appendices.
668 D3.5 Measurement/Data Acquisition (B5): Quality Control Requirements
669 This element of the QAPP should include enough detail that the use and evaluation of QC
670 sample results and corrective actions will be performed as planned and support project activities.
671 The QC acceptance limits and the required corrective actions for non-conformances should be
672 described. DQO Step 7 identified the optimized analytical design and the desired MQOs which
673 will help determine the QC acceptance criteria. Refer to Chapter 19.8.1 for information on
674 control charts and Chapter 18, Quality Assurance and Quality Control, for a detailed discussion
675 of radioassay QC and quality indicators. A discussion of QC requirements in the QAPP should
676 include the following information:
677 • A list of all QC sample types by matrix;
678 • The frequency of QC sample collection or analysis, preferably a tabular listing;
679 • A list of QC sample acceptance criteria or warning limits and control limits;
680 • Procedures for documenting QC sample results;
681 • Equations and calculations used to evaluate QC sample results and to determine measurement
682 performance acceptability;
683 • Actions to be taken if QC samples fail to meet the acceptance criteria; and
684 • Identification of the appropriate responsible person to whom QC reports should be sent.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT D-23 DO NOT CITE OR QUOTE
-------
Content of Project Plan Documents
685 Acceptance criteria for QC samples should be based on the project MQOs, in particular the MQO
686 for measurement uncertainty at some concentration. Appendix C provides guidance on
687 developing acceptance criteria for QC samples based on the project's MQO for the method's
688 measurement uncertainty at some concentration, typically the action level.
689 D3.6 Measurement/Data Acquisition (B6): Instrument/Equipment Testing, Inspection,
690 and Maintenance Requirements
691 The QAPP should include a discussion of testing, inspection and maintenance requirements that
692 will be followed to ensure that equipment and instrumentation will be in working order during
693 implementation of project activities. An instrument or testing equipment will be deemed to be in
694 working order if it is maintained according to protocol and it has been inspected and tested and
695 meets acceptance criteria.
696 This element of the QAPP should:
697 • Discuss the maintenance policy for all essential instrumentation and equipment, what it
698 involves, its frequency, whether it is performed by internal staff or if it is a contracted service,
699 and whether an inventory of spare parts is maintained;
700 • Describe the inspection protocols for instrumentation and equipment. This ranges from the
701 routine inspections (i.e, gases, nebulizers, syringes and tubing) prior to instrument or
702 equipment use and more detailed inspections employed while troubleshooting an instrument
703 or equipment problem. Mandatory inspection hold points, beyond which work may not
704 proceed, should be identified; and
705 • Address the frequency and details of equipment and instrument testing. This may involve the
706 weighing of volumes to test automatic diluters or pipets, the use of a standard weight prior to
707 weighing sample aliquots to the use of standards to test sophisticated instrumentation. If
708 standards (e.g., National Institute of Standards and Technology [NIST] standard reference
709 material [SRM]) are used during testing, the type, source and uncertainty of standard should
710 be identified.
711 There is not always a clear distinction between the testing component of this element and the
712 previous element addressing the use of QC samples to determine whether an instrument is within
713 control. In any case, it is important to describe in either of these elements of the QAPP, all
DRAFT MARLAP JUNE 2001
DO NOT CITE OR QUOTE D-24 NOT FOR DISTRIBUTION
-------
Content of Project Plan Documents
714 procedures that are deemed important to determining whether an instrument/equipment is in
715 working order and within control.
716 D3.7 Measurement/Data Acquisition (B7): Instrument Calibration and Frequency
717 This element of the QAPP details the calibration procedures including standards, frequencies,
718 evaluation, corrective action measures and documentation. Summary tables may be used to
719 complement the more detailed discussions in the text. The following issues should be addressed
720 in this element:
721 • Identify all tools, gauges, sampling devices, instruments, and test equipment that require
722 calibration to maintain acceptable performance;
723 • Describe the calibration procedures in enough detail in this element or by citation to readily
724 available references so that the calibration can be performed as intended;
725 • Identify reference equipment (e.g., NIST thermometers) and standards, their sources, and how
726 they are traceable to national standards. Where national standards are not available, describe
727 the procedures used to document the acceptability of the calibration standard used;
728 • Identify the frequency of calibration and any conditions (e.g., failed continuing calibration
729 standard, power failure) that may be cause for unscheduled calibration;
730 • Identify the procedure and the acceptance criteria (i.e., in control) to be used to evaluate the
731 calibration data;
732 • Identify the corrective actions to be taken if the calibration is not in control. When calibration
733 is out of control, describe the evaluations to be made to determine the validity and
734 acceptability of measurements performed since the last calibration; and
735 • Identify how calibration data will be documented, archived and traceable to the correct
736 instrument/equipment.
737 See Chapter 16, Instrument Calibration and Test Source Preparation, for a discussion of
738 radiochemical instrument calibration.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT D-25 DO NOT CITE OR QUOTE
-------
Content of Project Plan Documents
739 D3.8 Measurement/Data Acquisition (B8): Inspection/Acceptance Requirements for
740 Supplies and Consumables
741 This element of the QAPP deals with inspecting and accepting all supplies and consumables that
742 may directly or indirectly affect the quality of the data. For some projects, this information may
743 be provided by citation to a chemical safety and hygiene plan. The contents of this element
744 should contain enough supportive information that the project and the data will be sufficient to
745 undergo solicited and unsolicited reviews. The following detail should be included in this
746 element, so the inspection process can be accurately implemented:
747 • Identify and document all supplies and consumables (e.g., acids, solvents, preservatives,
748 containers, reagents, standards) that have the potential of directly or indirectly impacting the
749 quality of the data collection activity;
750 • Identify the significant criteria that should be used when choosing supplies and consumables
751 (e-g-, grade, purity, activity, concentration, certification);
752 • Describe the inspection and acceptance procedures that will be used for supplies or
753 consumables, including who is responsible for inspection, the timing of inspections and the
754 acceptance and rejection criteria. This description should be complete enough to allow
755 replication of the inspection process. Standards for receiving radiological packages are
756 provided in 10 CFR 20 Section 20.1906 "Procedures for Receiving and Opening Packages"
757 or an Agreement State equivalent;
758 • Describe the procedures for checking the accuracy of newly purchased standards, other than
759 SRMs, by comparison to other standards purchased from other sources;
760 • Identify any special handling and storage (e.g., refrigerated, in the dark, separate from high
761 concentration standards, lead shielding) conditions that must be maintained;
762 • Describe the method of labeling, dating and tracking supplies and consumables and the
763 disposal method for when their useful life has expired; and
764 • Describe the procedures and indicate by job function who is responsible for documenting the
765 inspection process and the status of inventories.
DRAFT MARLAP JUNE 2001
DO NOT CITE OR QUOTE D-26 NOT FOR DISTRIBUTION
-------
Content of Project Plan Documents
766 D3.9 Measurement/Data Acquisition (B9): Data Acquisition Requirements for Non-Direct
767 Measurement Data
768 This element of the QAPP addresses the use of existing data. Non-direct measurement data is
769 defined as existing data that is independent of the data generated by the current project's
770 sampling and analytical activities. Non-direct data may be of the same type (e.g., mBq/g of 232Th
771 in soil) that will complement the data being collected during the project. Other non-direct data
772 may be of a different type such as weather information from the National Weather Service, or
773 geological and hydrogeological data from the U.S. Geological Survey.
774
775 To achieve project objectives it is important that the data obtained from non-direct sources be
776 subjected to scrutiny prior to acceptance and use. Use of existing data is discussed during Step 1
777 and 3 of the DQO process. If existing data of the same type is to be used to achieve project
778 objectives, it has to be evaluated in terms of its ability to comply with MQOs established in DQO
779 Step 7. The limitations on the use of non-direct measurements should be established by the
780 project planning team.
781 This element should:
782 • Identify the type and source of all non-direct data that will be needed to achieve the project
783 objectives;
784 • State whether the same quality criteria and QC sample criteria will be applied to the non-
785 direct measurement data. If the same criteria cannot be applied, then identify criteria that will
786 be acceptable for the non-direct data but at the same time won't bias or significantly add to
787 the uncertainty of decisions for the project;
788 • Identify whether the data will support qualitative decisions (e.g., rain occurred on the third
789 day of sampling) or if the data will be used quantitatively (e.g., used to calculate a mean
790 concentration that will be compared to an action level);
791 • Identify whether enough information exists to evaluate the quality of the non-direct data (e.g.,
792 spike and collocated sample data, minimum detectable concentrations, reported measurement
793 uncertainties); and
794 • If the non-direct data are to be combined with project-collected data, identify the criteria that
795 will be used to determine if the non-direct data are comparable (e.g., sampled the same
796 population, same protocol).
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT D-27 DO NOT CITE OR QUOTE
-------
Content of Project Plan Documents
797 D3.10 Measurement/Data Acquisition (BIO): Data Management
798 This element of the QAPP should present an overview of the data management process from the
799 receipt of raw data to data storage. The overview should address all interim steps, such as, data
800 transformations, transmittals, calculations, verifications, validations and data quality assess-
801 ments. The procedures should address how internal checks for errors are made. Laboratories
802 should follow accepted data management practices (EPA, 1995). Applicable SOPs should be
803 included as attachments to the QAPP. (See Chapter 17, Data Generation, Reduction and
804 Reporting for a discussion of radiochemical data generation and reduction.)
805 The discussion of data management should address the following issues:
806 • DAT A RECORDING: The process of the initial data recording steps (e.g., field notebooks,
807 instrument printouts, electronic data storage of alpha and gamma spectra) should be
808 described. Examples of unique forms or procedures should be described. Describe the
809 procedures to be used to record final results (e.g., negative counts) and the uncertainty.
810 • CONVERSIONS AND TRANSFORMATIONS: All data conversions (e.g., dry weight to wet weight),
811 transformations (conversion to logs to facilitate data analysis) and calculation of statistical
812 parameters (e.g., uncertainties) should be described, including equations and the rationale for
813 the conversions, transformations and calculations. Computer manipulation of data should be
814 specified (e.g., software package, macros).
815 • DATA TRANSMITTALS: Data transmittals occur when data are sent to another location or
816 person or when it is converted to another format (incorporated into a spreadsheet) or media
817 (hardcopy reports keyed into a computer database). All transmittals and associated QA/QC
818 steps taken to minimize transcription errors should be described in enough detail to ensure
819 their proper implementation.
820 • DATA REDUCTIONS: Identify and explain the reasons for data reductions. Data reduction is the
821 process of changing the number of data items by arithmetic or statistical calculations,
822 standard curves, or concentration factors. A laboratory information management system may
823 use a dilution factor or concentration factor to change raw data. These changes often are
824 irreversible and in the process the original data are lost.
DRAFT MARLAP JUNE 2001
DO NOT CITE OR QUOTE D-28 NOT FOR DISTRIBUTION
-------
Content of Project Plan Documents
825 • DATA VERIFICATION, VALIDATION AND ASSESSMENTS : Since these assessment issues are
826 discussed in a subsequent element of the QAPP (D2), only an overview should be provided
827 identify the timing and frequency of these assessments.
828 • DATA TRACKING, STORAGE AND RETRIEVAL: Describe the system for tracking and compiling
829 data as samples are being analyzed, how data are stored, and the mechanism for retrieving
830 data (e.g., from archived back-up tapes or disks).
831 • SECURITY: Describe procedures for data and computer security.
832 D4.0 Group C: Assessment/Oversight
833 The elements of this group are intended to assess progress during the project, facilitate corrective
834 actions in a timely manner (Section D4.1), and provide reports to management (Section D4.2). It
835 should be stressed that early detection of problems and weaknesses—before project commence-
836 ment or soon thereafter—and initiation of corrective actions are important for a project's success.
837 The focus of the elements in this group is the implementation of the project as defined in the
838 QAPP. This group is different from the subsequent group, data validation and usability, which
839 will assesses project data after the data collection activity is complete.
840 D4.1 Assessment/Oversight (Cl): Assessment and Response Actions
841 The QAPP authors have a range of assessment choices that can be employed to evaluate on-going
842 project activities, which include surveillance, peer review, systems reviews, technical systems
843 audits (of field and laboratory operations), and performance evaluations. A detailed discussion of
844 laboratory evaluation is presented in Chapter 7, Evaluating Radiological Laboratories. It is
845 important to schedule assessments in a timely manner. An assessment has less value if its
846 findings become available after completion of the activity. The goal is to uncover problems and
847 weaknesses before project commencement or soon thereafter and initiate corrective actions so the
848 project is a success.
849 This element of the QAPP should:
850 • Identify all assessments by type, frequency and schedule;
851 • Identify the personnel who will implement the assessments;
852 • Identify the criteria, documents, and plans upon which assessments will base their review;
853 • Describe the format of assessment reports;
854 • Identify the time frame for providing the corrective action plan; and
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT D-29 DO NOT CITE OR QUOTE
-------
Content of Project Plan Documents
855 • Identify who is responsible for approving corrective actions and ensuring that they are
856 implemented.
857 D4.2 Assessment/Oversight (C2): Reports To Management
858 Reports to management are a mechanism for focusing management's attention on project quality
859 and on proj ect issues that may require the management's level of authority. To be effective
860 reports to management and management's review and response must be timely. The benefit of
861 these status reports is the opportunity to alert management of data quality problems, propose
862 viable solutions and procure additional resources.
863 At the end of the project, a final project report which includes the documentation of the DQA
864 findings should be prepared (Chapter 9, Data Quality Assessment). It may also be beneficial for
865 future planning efforts for the project planning team to provide a summary of the "lesson
866 learned" during the project, such as key issues not addressed during planning and discovered in
867 implementation or assessment, specialist expertise needed on the planning team, experience with
868 implementing performance-based analytical protocol selection.
869 This element of the QAPP should address the following issues:
870 • Identify the various proj ect reports that will be sent to management;
871 • Identify non-project reports that may discuss issues pertinent to the project (e.g., backlog
872 reports);
873 • Identify QA reports that provide documentary evidence of quality (e.g., results of independent
874 performance testing, routine QC monitoring of system performance);
875 • Identify the content of "reports to management" (e.g., project status, deviations from the
876 QAPP and approved amendments, results of assessments, problems, suggested corrective
877 actions, status on past corrective actions);
878 • Identify the frequency and schedule for reports to management;
879 • Identify the organization or personnel who are responsible for authoring reports; and
880 • Identify the management personnel who will receive and act upon the assessment reports.
DRAFT MARLAP JUNE 2001
DO NOT CITE OR QUOTE D-3 0 NOT FOR DISTRIBUTION
-------
Content of Project Plan Documents
881 D5.0 Group D: Data Validation and Usability
882 This group of elements ensures that individual data elements conform to the project specific
883 criteria. This section of the QAPP discusses data verification, data validation and data quality
884 assessment (DQA), three processes employed to accept, reject or qualify data in an objective and
885 consistent manner. Although there is good agreement as to the range of issues that the three
886 elements, in total, should address, within the environmental community there are significant
887 differences as to how verification, validation and DQA are defined. The discussion of this group
888 of elements will use the definitions which are defined Chapter 8, Radiochemical Data
889 Verification and Validation.
890 D5.1 Data Validation and Usability (Dl): Verification and Validation Requirements
891 This element of the QAPP addresses requirements for both data verification and data validation.
892 The purpose of this element is to clearly state the criteria for deciding the degree to which each
893 data item and the data set as a whole has met the quality specifications described in the
894 "Measurement/Data Acquisition" section of the QAPP. The strength of the conclusions that can
895 be drawn from the data is directly related to compliance with and deviations from the sampling
896 and analytical design. The requirements can be presented in tabular or narrative form.
897 Verification procedures and criteria should be established prior to the data evaluation.
898 Requirements for data verification include the following criteria:
899 • Criteria for determining if specified protocols were employed (e.g., compliance with essential
900 procedural steps);
901 • Criteria for determining if methods were in control (e.g., QC acceptance criteria);
902 • Criteria for determining if a data report is complete (e.g., list of critical components that
903 constitute the report);
904 • Criteria for determining if the analysis was performed according to the QAPP and the SOW;
905 • Criteria and codes used to qualify data; and
906 • Criteria for summarizing and reporting the results of verification.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT D-31 DO NOT CITE OR QUOTE
-------
Content of Project Plan Documents
907 A discussion of verification can be found in Chapter 8, Radiochemical Data Verification and
908 Validation.
909 Data validation should be performed by an organization independent of the group that generated
910 the data to provide an unbiased evaluation. Validation procedures and criteria should be
911 established prior to the data evaluation. Requirements for data validation include the following:
912 • An approved list of well-defined MQOs and the action level(s) relevant to the project DQOs;
913 • Criteria for assigning qualifiers based on the approved list of MQOs;
914 • Criteria for identifying situations when the data validator's best professional judgement can
915 be employed and when a strict protocol must be followed; and
916 • Criteria for summarizing and reporting the results of validation.
917 A discussion of verification can be found in Chapter 8, Radiochemical Data Verification and
918 Validation.
919 D5.2 Data Validation and Usability (D2): Verification and Validation Methods
920 D5.2.1 Data Verification
921 Data verification or compliance with the SOW is concerned with: complete, consistent,
922 compliant and comparable data. Since the data verification report documents whether laboratory
923 conditions and operations were compliant with the SOW, the report is often used to determine
924 payment for laboratory services. Chapter 5, Obtaining Laboratory Services, discusses the need to
925 prepare a SOW for all radioanalytical laboratory work regardless of whether the work is
926 contracted out or performed in-house.
927 This element of the QAPP should address the following issues to ensure that data verification
928 will focus on the correct issues:
929 • Identify the documents (e.g., other QAPP sections, SOW, contracts, standard methods) that
930 describe the deliverables and criteria that will be used to evaluate compliance;
DRAFT MARLAP JUNE 2001
DO NOT CITE OR QUOTE D-32 NOT FOR DISTRIBUTION
-------
Content of Project Plan Documents
931 • Identify the performance indicators that will be evaluated (e.g., yield, matrix spikes,
932 replicates). See Chapter 18, Laboratory Quality Control, for a discussion of radiochemistry
933 performance indicators;
934 • Identify the criteria that will be used to determine "in-control" and "not-in-control"
935 conditions;
936 • Identify who will perform data verification;
937 • Describe the contents of the verification report (e.g., a summary of the verification process as
938 applied; required project activities not performed or not on schedule or not according to
939 required frequency; procedures that were performed but did not meet acceptance criteria;
940 affected samples; exceptions); and
941 • Identify who will receive verification reports and the mechanism for its archival.
942 D5.2.2 Data Validation
943 Chapter 8, Radiochemical Data Verification and Validation, discusses radiochemical data
944 validation in detail. MARLAP recommends that a data validation plan document be included as
945 an appendix to the QAPP. The data validation report will serve as the major input to the process
946 that evaluates the reliability of measurement data.
947 This element of the QAPP should address the following issues:
948 • Describe the deliverables, measurement performance criteria and acceptance criteria that will
949 be used to evaluate data validity;
950 • Identify who will perform data validation;
951 • Describe the contents of the validation report (e.g., a summary of the validation process as
952 applied; summary of exceptional circumstances; list of validated samples, summary of
953 validated results); and
954 • Identify who will receive validation reports and the mechanism for its archival.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT D-3 3 DO NOT CITE OR QUOTE
-------
Content of Project Plan Documents
955 D5.3 Data Validation and Usability (D3): Reconciliation with Data Quality Objectives
956 This element of the QAPP describes how project data will be evaluated to determine its usability
957 in decision-making. This evaluation is referred to as the "data quality assessment." DQA is the
958 process that scientifically and statistically evaluates project-wide knowledge in terms of the
959 project objectives to assess the usability of data. DQA should be ongoing and integrated into the
960 project data collection activities. On project diagrams and data life cycles, it is often shown as the
961 last phase of the data collection activity. However, like any assessment process, DQA should be
962 considered throughout the data collection activity to ensure usable data. EPA guidance (EPA,
963 1996) provides a detailed discussion of that part of the DQA process that addresses statistical
964 manipulation of the data. In addition to statistical considerations, the DQA process integrates and
965 considers information from the validation report, assessment reports, the field, the conceptual
966 model and historical data to arrive at its conclusions regarding data usability. DQA is discussed
967 in Chapter 9, Data Quality Assessment.
968 The DQA considers the impact of a myriad of data collection activities in addition to measure-
969 ment activities. This element of the QAPP should direct those performing the DQA to:
970 • Review the QAPP and DQOs;
971 • Review the validation report;
972 • Review reports to management;
973 • Review identified field, sampling, sample handling, analytical and data management
974 problems associated with project activities;
975 • Review all corrective actions; and
976 • Review all assessment reports and findings (e.g., surveillances, audits, performance
977 evaluations, peer reviews, management and technical system reviews).
978 In addition to the above, this element of the QAPP should address the following issues:
979 • Identify who will perform the DQA;
980 • Identify what issues will be addressed by the DQA;
981 • Identify any statistical tests that will be used to evaluate the data (e.g., tests for normality);
982 • Describe how MQOs will be used to determine the usability of measurement data (i.e., did
983 the measurement uncertainty in the data significantly affect confidence in the decision?);
984 • Describe how the representativeness of the data will be evaluated (e.g., review the sampling
985 strategy, the suitability of sampling devices, subsampling procedures, assessment findings);
986 • Describe how the potential impact of non-measurable factors will be considered;
987 • Identify what will be included in the DQA report; and
DRAFT MARLAP JUNE 2001
DO NOT CITE OR QUOTE D-34 NOT FOR DISTRIBUTION
-------
Content of Project Plan Documents
988 • Identify who will receive the report and the mechanism for its archival.
989 D6.0 References
990 American National Standards Institute and the American Society for Quality Control (ANSI/
991 ASQC). 1994. Specifications and Guidelines for Quality Systems for Environmental Data
992 Collection and Environmental Technology Programs, National Standard E-4.
993 Taylor, J. K. 1990. Quality Assurance of Chemical Measurements. Lewis, Chelsea, Michigan.
994 U. S. Department of Energy (DOE). 1996. Project Plan for the Background Soils Project for the
995 Paducah Gaseous Diffusion Plant, Paducah, Kentucky. Report DOE/OR/07-1414&D2. May.
996 U.S. Department of Energy (DOE). 1997. Quality Assurance Project Plan for Radiological
997 Monitoring at the U.S. DOE Paducah Gaseous Diffusion Plant, Paducah, Kentucky.
998 February.
999 U.S. Environmental Protection Agency (EPA). 1980. Interim Guidelines and Specifications for
1000 Preparing Quality Assurance Project Plans., QAMS-005/80. Office of Monitoring Systems
1001 and Quality Assurance, Washington, DC.
1002 U.S. Environmental Protection Agency (EPA). 1994. Guidance for the Data Quality Objective
1003 Process ( EPA QA/G-4). EPA/600/R-96/055, EPA, Washington, DC.
1004 U.S. Environmental Protection Agency (EPA). 1995. Good Automated Laboratory Practices.
1005 Report 2185, EPA, Washington, DC.
1006 U.S. Environmental Protection Agency (EPA). 1996. Guidance for Data Quality Assessment:
1007 Practical Methods for Data Analysis. EPA QA/G-9, EPA/600/R-96/084, EPA, Washington,
1008 DC.
1009 U.S. Environmental Protection Agency (EPA). 1998a. EPA Guidance for Quality Assurance
1010 Project Plans (EPA QA/G-5). EPA/600/R-98/018, EPA, Washington, DC.
1011 U.S. Environmental Protection Agency (EPA). 1998b. EPA Requirements for Quality Assurance
1012 Project Plans for Environmental Data Operations. EPA QA/R-5, External Review Draft
1013 Final, EPA, Washington, DC.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT D-3 5 DO NOT CITE OR QUOTE
-------
i APPENDIX E: CONTRACTING LABORATORY
2 SERVICES
3 E.I Introduction
4 This appendix provides general guidance on Federal contracting and contracting terminology as
5 used for negotiated procurements. Federal Agencies, and laboratories doing business with them,
6 must follow applicable provisions of the Federal Acquisition Regulations (FAR) and Agency-
7 specific supplements. The examples provided in this appendix are based primarily on procedures
8 followed by the U.S. Geological Survey (USGS).
9 This appendix addresses selecting a laboratory to establish services that supplement an Agency's
10 in-house activities through the contracting of additional outside support. This appendix offers a
11 number of principles that may be used when selecting a service provider, establishing a
12 contractual agreement, and later working with a contract laboratory. These principles may also be
13 applied to contractors that are located outside of the United States. In such cases, legal counsel
14 will need to review and advise an Agency concerning pertinent issues related to international
15 contracts.
16 This appendix also covers laboratory audits that are part of a final selection process and other
17 activities that take place until the contract is concluded. Chapter 5 supports this appendix with a
18 general description on how to obtain laboratory services. Chapter 7 complements this appendix
19 by considering information related to laboratory evaluations that are conducted throughout the
20 term of a project—whether or not this work is specifically covered by a contract.
21 Obtaining support for laboratory analyses is already a practice that is familiar to a number of
22 Federal and State Agencies. The following discussion will apply:
23 • Agency - a Federal or State government office or department, (or potentially any other public
24 or private institution) that offers a solicitation or other mechanism to obtain outside services;
25 • Proposer - a contracting firm or commercial facility that submits a proposal related to
26 providing services; and
27 • Contractor - a firm that is awarded the contract and is engaged in providing analytical
28 services.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT E-1 DO NOT CITE OR QUOTE
-------
Contracting Laboratory Services
29 Furthermore, the size and complexity of some agency projects will clearly exceed the extent of
30 the information presented here. In its present form, this appendix serves to touch on many of the
31 issues and considerations that are common to all projects, be they large or small.
32 MARLAP draws attention to another dimension of the overall contracting process by considering
33 how the Data Quality Objectives (DQOs) and Measurement Quality Objectives (MQOs) are
34 incorporated into every stage of a project—as described earlier in greater detail (Chapters 2 and
35 3). In this regard, an Agency's Project Managers and staff are given an opportunity to consider
36 options with some foresight and to examine the larger picture, which concerns planning short- or
37 long-term projects that utilize a contractor's services. As services are acquired, and later as work
38 is performed, the specific concepts and goals outlined by the DQOs and MQOs will be revisited.
39 This becomes an iterative process that offers the possibility to further define objectives as work is
40 conducted. Whenever the DQOs or MQOs are changed, the contract should be modified to
41 reflect the new specifications. Employing the MQOs and tracking the contractor's progress
42 provides a means by which Project Managers and contract-laboratory technical staff can return
43 and review the project at any point during the contract period. This allows for repeated
44 evaluations to further optimize a project's goals and, if anticipated in the contract's language,
45 perhaps even provides for the option to revise or redirect the way performance-based work is
46 conducted.
47 The Office of Federal Procurement Policy (OFPP, 1997) has developed a Performance-Based
48 Service Contracting review checklist to be used as a guide in developing a performance-based
49 solicitation. The checklist contains minimum required elements that should be present for a
50 contract to be considered performance-based. Performance-Based Service Contracting focuses on
51 three elements: a performance work statement; a quality assurance project plan (QAPP); and
52 appropriate incentives, if applicable. The performance work statement defines the requirements
53 in terms of the objective and measurable outputs. The performance work statement should
54 answer five basic questions: what, when, where, how many, and how well. The work statement
55 should structure and clearly define the requirements, performance standards, acceptable quality
56 levels, methods of surveillance, incentives if applicable and evaluation criteria. A market survey
57 should be conducted so that the marketplace and other stakeholders are provided the opportunity
58 to comment on draft performance requirements and standards, the proposed QA project plan, and
59 performance incentives, if applicable.
60 A number of benefits arise from establishing a formal working relationship between an Agency
61 and a contractor. For example:
MARLAP JULY 2001
DO NOT CITE OR QUOTE E-2 DRAFT FOR PUBLIC COMMENT
-------
Contracting Laboratory Services
62 • A contract is a legal document that clearly defines activities and expectations for the benefit
63 of both parties engaged in the contractual relationship.
64 • The process of drafting language to cover legal considerations may well include contributions
65 from legal staff. Legal guidance may be obtained as needed at any time during the planning
66 stages or later when a contract is in place. However, the core of a contractor's proposal, and
67 eventually the contract itself, provide the foundation of technical work that is required to
68 complete a project or attain an ongoing program goal. In this regard, aside from legal issues
69 that are an integral part of every contract, this appendix's principal focus is on the
70 laboratory process or technical work-related content of the contract.
71 • The statement of work (SOW) first appears as part of the Agency's request for proposal
72 (RFP) and later is essentially incorporated into the proposal by the proposer when responding
73 to the RFP. When work is underway, the SOW becomes a working document that both the
74 Agency and contractor refer to during the life of the contract.
75 • Legal challenges concerning project results (i.e., laboratory data) may arise during the
76 contract period. The language in a contract should offer sufficient detail to provide the means
77 to circumvent potential or anticipated problems. For example, attention to deliveries of
78 samples to the laboratory on weekends and holidays or data reporting requirements that are
79 designed to support the proper presentation of data in a legal proceeding are important
80 aspects of many Federal- and State-funded contracts.
81 Overall, this appendix incorporates a sequence that includes both a planning and a selection
82 process. Figure E-l illustrates a series of general steps from planning before a contract is even in
83 place to the ultimate termination of the contract. An Agency first determines a need as part of
84 planning, and along the way advertises this need to solicit proposals from outside service
85 providers who operate analytical laboratory facilities. Planning future work, advertising for, and
86 later selecting services from proposals submitted to an Agency takes time—perhaps six or more
87 months pass before a laboratory is selected, a contract is in place, and analytical work begins.
88 The total working duration of a contract, for example, might cover services for a brief time
89 (weeks or months) and in other cases, many contracts may run for a preset one-year period or for
90 a more extended period of three to five years with optional renewal periods during that time.
91 The MARLAP user will find that planning employs a thought process much like that used to
92 prepare an RFP. In general, one starts with questions that define a project's needs. Further, by
93 developing Analytical Protocol Specifications (APSs) which include specific MQOs, one enters
94 an iterative process such that—at various times—data quality is checked in relation to work
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT E-3 DO NOT CITE OR QUOTE
-------
Contracting Laboratory Services
Planning
i
r
Procurement
Process
i
r
Request for
Proposal
i
r
|
I
I
Contractor Proposal |
i
r
Review
Evaluate
Score
Audit
i
r
Award
i
r
Contractor Servici
i
-^
r
1
|
BS |
•
•
Contract Completion |
« Includes SOW
Section E.2
^ Reflects SOW
Section E.3
Section E.4
Section E.5
Section E.6
Section E.7
Periodic Evaluations
Work According to SOW
Section E.8
FIGURE E.I — General Sequence Initiating and Later
Conducting Work with a Contract Laboratory
MARLAP
DO NOT CITE OR QUOTE
E-4
JULY 2001
DRAFT FOR PUBLIC COMMENT
-------
Contracting Laboratory Services
95 performed both in-house and by the outside service provider. Overall, planning results in the
96 development of a project plan document (e.g., QAPP). During planning, a Project Manager and
97 the Agency staff can consider both routine and special analytical services that may be required to
98 provide data of definable quality. The SOW serves to integrate all technical and quality aspects
99 of the project, and to define how specific quality-assurance and quality-control activities are
100 implemented during the time course of a contract. Also, at an early stage in planning, the Agency
101 may choose to assemble a team to serve as the Technical Evaluation Committee (TEC; Section
102 E.5.1). The main role of the TEC is in selecting the contract laboratory by reviewing proposals
103 and by auditing laboratory facilities. The TEC is discussed later in this appendix, however, the
104 key issue here concerns the benefit to establishing this committee early on, even to the point of
105 including TEC members in the initial planning activities. The result is a better informed
106 evaluation committee and a team of individuals that can help make adjustments when the
107 directed planning process warrants an iterative evaluation of the way work is performed under
108 the contract. Overall, planning initiates the process that characterizes the nature of the contracting
109 process to follow.
no E.2 Procurement of Services
111 Recognizing that the procurement process differs from Agency to Agency, the following
112 guidance provides a general overview to highlight considerations that may already be part of—or
113 be incorporated into—the current practice. First, the request for specific analytical services can
114 be viewed as a key product of both the Agency's mission and the directed planning process. As
115 Agency staff ask questions, list key considerations to address during the work, and in turn define
116 objectives, they also eliminate unnecessary options to help focus on the most suitable contracting
117 options that satisfy the APSs. Thereafter, the scope of the work, schedule, manpower constraints,
118 availability of in-house engineering resources, and other technical considerations all enter into
119 estimating and defining a need for project support. This approach refines the objectives and
120 establishes needs that may be advertised in a solicitation for outside services. The resulting work
121 or project plan should clearly articulate what is typically known but not limited to the following:
122 • Site conditions;
123 • Analytes of interest;
124 • Matrices of concern;
125 • How samples are to be collected and handled;
126 • Custody requirements;
127 • Data needs and APSs, including the MQOs;
128 • Stipulated analytical methods, if required
129 • Applicable regulations; and
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT E-5 DO NOT CITE OR QUOTE
-------
Contracting Laboratory Services
130 • Data reporting.
131 All of this defines the scope of work, such that the Agency can initiate a formal request for
132 proposals or arrange for an analysis request as part of a less formal procurement.
133 E.2.1 Request for Approval of Proposed Procurement Action
134 If required within an Agency, a request is processed using forms and related paperwork to
135 document information typically including, but not limited to, the following:
136 • Identification of product or service to be procured;
137 • Title of program or project;
138 • Description of product or service;
139 • Relationship of product or service to overall program or project;
140 • Funding year, projected contract life, amounts, etc.;
141 • Name and phone number of Project Officer(s);
142 • Signature of Project Officer and date
143 • Name and phone number of Contracting Officer; and
144 • Signature of Contracting Officer and date.
145 An Agency may also be required to collect or track information for an RFP with regard to:
146 • New procurements: type of contract, grant, agreement, proposal, etc. Continuing
147 procurements: pre-negotiated options, modifications, justification for non-competitive
148 procurement, etc.
149 • Source information: small business or other set aside, minority business, women-owned
150 business, etc.
151 In addition to the information listed above, Agency-specific forms used to initiate a procurement
152 request may also provide a place to indicate Agency approval with names, signature lines, and
153 date spaces for completion by officials in the office responsible for procurement and contracts.
154 An Agency administrator or director above the level of the office of procurement may also sign
155 this form indicating Agency approval.
156
157
158
E.2.2 Types of Procurement Mechanisms
Table E.I lists many of the procurement options available to the Project Manager. Each option
offers a solution to a specific need. For example, a purchase order is typically appropriate for
MARLAP JULY 2001
DO NOT CITE OR QUOTE E-6 DRAFT FOR PUBLIC COMMENT
-------
Contracting Laboratory Services
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
tasks with a somewhat limited scope and thus is perhaps most useful when samples are to be
processed on a one-time basis. In some cases where only one or a limited number of vendors can
fulfill the needs of the project, e.g., low-level tritium analysis by helium ingrowth within a
specified time period, a sole source solicitation is commonly used.
TABLE E.I— Examples of Procurement Options to Obtain Materials or Services.
Procurement Mechanism
Purchase order
Sole source solicitation
Request for Quotation (RFQ)
Request for Proposal (RFP)
Modification to an existing contract
or delivery order
Basic Ordering Agreement (BOA)
Example of Specific Use or Application
In-house process handled through purchasing staff; limited to small needs
without a formal request or used in conjunction with a solicitation
(competitive process) and a limited amount of funding; commonly used to
purchase equipment and supplies, but may be used for processing samples.
In specific instances, a single or a limited number of service providers are
able to offer specific services.
Formal, main process for establishing contracts — generally addresses a
major, long-term need for contractor support; this is a competitive process
based mainly on cost.
Formal, main process for establishing contracts — generally addresses a
major, long-term need for contractor support; this is a competitive process
based mainly on technical capability.
This approach meets a need that is consistent with the type of contract that
is in place, e.g., Agency amends contract to add a method for sample
processing that is similar to work already covered.
Work is arranged with a pre-approved laboratory as described in
Section E.2.2.
The process leading to a formal contract provides a more comprehensive view of nearly every
aspect of the work that an Agency expects from a contractor. The formal process includes three
types of procurement: Request for Quotation (RFQ), Request for Proposal (RFP), and the Basic
Ordering Agreement (BOA). The RFQ solicits bidders to provide a quotation for laboratory
services that have been detailed in the solicitation. The specifications may include the technical,
administrative, and contractual requirements for a project. For the RFQ, the contract typically is
awarded to the lowest bidder that can fulfill the contract specifications without regard to the
quality of the service. What appears to be a good price may not entail the use of the best or most
appropriate method or technology. There may be significant advantages in seeking to acquire
high-technology services as a primary focus in advance of, or along with, concerns pertaining to
price.
For an RFP, there is considerably more work for the Agency and the laboratory. The laboratory
must submit a formal proposal addressing all key elements of the solicitation that include how,
JULY 2001
DRAFT FOR PUBLIC COMMENT
E-7
MARLAP
DO NOT CITE OR QUOTE
-------
Contracting Laboratory Services
185 why, what, when ,where and by whom the services are to be performed. The TEC or Contracting
186 Officer must review all proposals, rank them according to a scoring system and finally assess the
187 cost effectiveness of the proposals before making the final award.
188 The BOA provides a process that serves to pre-approved service providers. This includes a
189 preliminary advertisement for a particular type of work, such as radioanalytical services. The
190 Agency then selects and approves a number of candidates that respond to the advertisement.
191 With this approach, the Agency assembles a potential list of approved laboratories that are
192 contacted as needed to support specific needs. The Agency may choose to simply write a task
193 order (defining a specific scope of work) with a specific pre-approved laboratory, or the Agency
194 may initiate a competitive bidding process for the task order between several or all members on
195 the list of pre-approved laboratories. Once chosen, the laboratory may be guided by a combined
196 Statement of Work or Task Order that is issued by the Agency.
197 Mechanisms that permit an Agency to obtain analyses for a limited number of samples—without
198 an established contractual relationship with a specific contractor—may simply be necessitated by
199 the small number of samples, time constraints where specific analyses are not part of an existing
200 contract, limitations related to funding, or other consideration. The formal business and legal
201 requirements of a long-term relationship warrant a stronger contractual foundation for work
202 conducted in a timely fashion, on larger numbers of samples, and over specified periods of time.
203 The contracts described above, with the exception of a BOA, are considered "requirement"
204 contracts and requires the group initiating the solicitation to use only the contracted laboratory,
205 without exception, for the contract period to perform the sample analyses.
206 E.3 Request for Proposals—The Solicitation
207 To appreciate the full extent of a competitive process leading to a formal working relationship—
208 between an Agency and a contractor—the primary example used hereafter is the solicitation and
209 selection process that starts with the issuance of a RFP, as shown in Figure E-l.
210 Federal announcements of certain RFPs can be found in the Commerce Business Daily (CBD).
211 The CBD primarily provides a synopsis or brief description of the type of work the Agency is
212 interested in purchasing. States and local governments also solicit proposals and announce the
213 availability of work in USABID (a compilation of solicitations from hundreds of city, county,
214 and state agencies). Internet sites that offer access to the CBD (http://cbdnet.access.gpo.gov/) and
215 USABID listings can be located through electronic searches using Web Browser software. Once
216 a site is located, the information can be viewed through public access or commercial Internet-
217 based services. In other cases, a State or Federal Agency may maintain a mailing list with names
MARLAP JULY 2001
DO NOT CITE OR QUOTE E-8 DRAFT FOR PUBLIC COMMENT
-------
Contracting Laboratory Services
218 and addresses for potentially interested parties. This might include contractors that previously
219 supported the Agency or others who have volunteered information for the mailing list.
220 Once the RFP, State advertisement, or other form of solicitation is publicized, interested parties
221 can contact the appropriate Agency to obtain all the specific information relevant to completing a
222 candidate laboratory's contract proposal. For the present discussion, this information is contained
223 in the text of the RFP document. The RFP may be accompanied by a cover letter stating an
224 invitation to applicants and general information related to the content of a proposal and specific
225 indication for the types of sections or sub-sections the proposal will contain. For example, a
226 proposal divided into three sections technical proposal, representations and certifications, and
227 price proposal allows the Agency to separate pricing from technical information. In this way, the
228 Agency considers each candidate first on technical merits before the price of services enters the
229 selection process.
230 The Agency's RFP is designed to provide a complete description of the proposed work. For
231 example, a RFP should inform all candidate laboratories (i.e., proposers) of the estimated number
232 of samples that are anticipated for processing under the contract. The description of work in the
233 RFP as described in the SOW serves to indicate the types of radionuclide analyses required for
234 the stated sample types and the number of samples to undergo similar or different processing
235 protocols. The estimate also has a bearing on cost and other specific project details as described
236 in the SOW. Additional information provided with the RFP serves to instruct the proposer
237 regarding other technical requirements (APSs), the required number of copies of each section of
238 the proposal, proposal deadline, address where proposals are to be sent, and other general
239 concerns or specifications relevant to the solicitation.
240 The cover letter may indicate how each proposer will be notified if its proposal is dropped from
241 the competitive range of candidates during the selection process. The letter may also include
242 precautionary notes concerning whom to contact or not contact at the Agency regarding the
243 potential contract during the competitive process. Finally, if particular sources are encouraged to
244 apply (e.g., minority or small business), this information will be mentioned in the Agency's
245 invitation to apply.
246 E.3.1 Market Research
247 The Office of Federal Procurement Policy (OFPP, 1997) recommends that the marketplace and
248 other stakeholders be provided the opportunity to comment on draft performance requirements
249 and standards. This practice allows for feedback from those people working in the technical
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT E-9 DO NOT CITE OR QUOTE
-------
Contracting Laboratory Services
250 community so that their comments may be incorporated into the final RFP and the potential
251 offerers can develop intelligent proposals.
252 E.3.2 Length of Contract
253 The time and resources involved in writing and awarding a major contract generally make it
254 impractical and cost ineffective to award contracts for less than one or more years. While
255 contracts running for shorter terms are sometimes established, single or multiple year terms are
256 commonly used to provide the necessary services for some Federal or State programs.
257 Monitoring programs are likely to go long periods of time with renewals or RFPs that continue
258 the work into the future. Elsewhere, relatively large projects conducting radiation survey and site
259 investigations may require a contract process that, for the most part, estimates the time services
260 will be needed to finish work through to the completion of a final status survey. In this case, the
261 contract may specify any length of time, but also include the option to renew the contract for a
262 period of time to bring the project to a close. The relationship between the length of a contract
263 and the type of project can be part of the structured planning process that seeks to anticipate
264 every facet of a proj ect from start to finish.
265 Multi-year contracts are typically initiated with an award for the first year followed by an
266 additional number of one-year options. In this way, a five-year contract is awarded for 1 year
267 with four one-year option periods to complete the contract's full term. Problems that arise during
268 any year may result in an Agency review of the MQOs or an examination of the current working
269 relationship that may result in the Agency's decision to not extend the contract into the next
270 option year.
271 E.3.3 Subcontracts
272 For continuity or for quality assurance (QA), the contract may require one laboratory to handle
273 the entire analytical work load. However, subcontracting work with the support of an additional
274 laboratory facility may arise if the project plan calls for a large number of samples requiring
275 quick turnaround times and specific methodologies that are not part of the primary laboratory's
276 support services. A proposer may choose to list a number of subcontractors in the proposal. The
277 listing may or may not include other laboratories with whom the proposer has an existing or prior
278 working relationship. The choice of subcontracting firms may be limited during the proposal
279 process. There may be many qualified service providers to meet specific project needs. However,
280 once work is under way, using a limited number of laboratories that qualify for this secondary
281 role helps maintain greater control of quality and thus the consistency of data coming from more
MARLAP JULY 2001
DO NOT CITE OR QUOTE E-10 DRAFT FOR PUBLIC COMMENT
-------
Contracting Laboratory Services
282 than a single laboratory alone. Furthermore, the contractor may prefer working with a specific
283 subcontractor, but this arrangement is subject to Agency approval.
284 The use of multiple service providers adds complexity to the Agency's tasks of auditing,
285 evaluating, and tracking services. The contractor and their subcontractor(s) are held to the same
286 terms and conditions of the contract. The prime contractor is held responsible for the
287 performance of its subcontract laboratories. In some instances, certain legal considerations
288 related to chain of custody, data quality and reporting, or other concern may limit an Agency's
289 options and thus restrict the number of laboratories that are part of any one contract.
290 E.4 Proposal Requirements
291 The Agency's RFP will state requirements that each proposer is to cover in its proposal. The
292 proposal document itself becomes first the obj ect of evaluation and is a reflection of how the
293 contract and the SOW are structured. Whether one works with a formal contract or a simpler
294 analysis request, the Agency and contractor need to agree to all factors concerning the specific
295 analytical work. Where written agreements are established, the language should be specific to
296 avoid disputes. Clear communication and complete documentation are critical to a project's
297 success. For example, the Agency's staff asks questions of itself during the planning process to
298 create and later advertise a clearly stated need in the RFP. The contractor then composes a
299 proposal that documents relevant details concerning their laboratory's administrative and
300 technical personnel, training programs, instrumentation, previous project experience, etc.
301 Overall, the proposer should make an effort to address every element presented in the RFP. The
302 proposer should be as clear and complete as possible to ensure a fair and proper evaluation
303 during the Agency's selection process.
304 The planning process will reveal numerous factors related to technical requirements necessary to
305 tailor a contract to specific project needs. The following sections may be reviewed by Agency
306 staff (radiochemist or TEC) during planning to determine if additional needs are required beyond
307 those listed in this manual. Agency personnel should consider carefully the need to include every
308 necessary detail to make a concise RFP. The proposer can read the same sections to anticipate the
309 types of issues that are likely to appear in an RFP and that may be addressed in a proposal.
310 E.4.1 RFP and Contract Information
311 There are two basic areas an Agency can consider when assembling information to include in an
312 RFP. The proposer is expected to respond with information for each area in its proposal. The first
313 area includes a listing of General Laboratory Requirements and Activities. The second area,
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT E-11 DO NOT CITE OR QUOTE
-------
Contracting Laboratory Services
314 Technical Components to Laboratory Functions., complements the first, but typically includes
315 more detailed information.
316 1) General Laboratory Requirements
317 • Personnel;
318 • Facilities;
319 • Meeting Contract Data Quality Requirements;
320 • Schedule;
321 • Quality Manual;
322 • Data Deliverables Including Electronic Format;
323 • Licenses and Certifications; and
324 • Experience: Previous and Current Contracts; Quality of Performance.
325 2) Technical Components to Laboratory Functions
326 • Standard Operating Procedures;
327 • Instrumentation
328 • Training
329 • Performance Evaluation Programs; and
330 • Quality System.
331 The laboratory requirements and technical components indicated above are addressed in this
332 appendix. Beyond this, there are additional elements that may be required to appear with detailed
333 descriptions in an RFP and later in a formal proposal. One significant portion of the RFP, and a
334 key element appearing later in the contract itself, is the SOW. This is the third area a proposer is
335 to address, and information in a SOW may vary depending on the nature of the work.
336 The Agency will provide specifications in the RFP regarding the work the contractor will
337 perform. This initiates an interaction between a proposer and the Agency and further leads to two
338 distinct areas of contractor-Agency activity. The first concerns development and submitting of
339 proposals stating how the laboratory work will be conducted to meet specific Agency needs. The
340 second concerns Agency evaluations of the laboratory's work according to contract specifications
341 (Section E.5) and the SOW. Once the contract is awarded, a contractor is bound to perform the
342 work as proposed.
343 Specific sections of each contract cover exactly what is expected of the contractor and its
344 analytical facilities to fulfill the terms and conditions of the contract. The SOW describes the
345 required tasks and deliverables, and presents technical details regarding how tasks are to be
346 executed. A well written SOW provides technical information and guidance that directs the
MARLAP JULY 2001
DO NOT CITE OR QUOTE E-12 DRAFT FOR PUBLIC COMMENT
-------
Contracting Laboratory Services
347 contractor to a practice that is technically qualified, meets all relevant regulatory requirements,
348 and appropriately coordinates all work activities. A sample checklist for key information that
349 may be in a SOW is presented in Table E.2. Note that not all topics in the list are appropriate for
350 each project, and in some cases, only a subset is required. The list may also be considered in
351 relation to less formal working relationships (e.g., purchase order), as well as tasks covered in
352 formal contracts.
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
TABLE E.2 — SOW Checklists for the Agency and Proposer
SAMPLE HISTORY
General background on the problem
Site conditions
Regulatory background
Sample origin
Analytes and interferences (chemical forms and estimated concentration range)
Safety issues
Data use
Regulatory compliance
Litigation
ANALYSIS RELATED
Number of samples
Matrix
Container type and volume
Receiving and storage requirements
Special handling considerations
Custody requirements
Preservation requirements, if any
Analytes of interest (specific isotopes or nuclide)
Measurement Quality Objectives
Proposed method (if appropriate) and method validation documentation
Regulatory reporting time requirement (if applicable)
Analysis time requirements (time issues related to half-lives)
QC requirements (frequency, type, and acceptance criteria)
Waste disposal issues during processing
Licenses and accreditation
OVERSIGHT
Quality manual
Required Performance Evaluation Program participation
Criteria for (blind) QC
Site visit/data assessment
Audit (if any)
REPORTING REQUIREMENTS
Report results as gross, isotopic....
Reporting units
Reporting basis (dry weight,....)
How to report measurement uncertainties
JULY 2001
DRAFT FOR PUBLIC COMMENT
E-13
MARLAP
DO NOT CITE OR QUOTE
-------
Contracting Laboratory Services
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
Reporting Minimum Detectable Concentration and Minimum Quantifiable Concentration
Report contents desired and information for electronic data transfer
Turn-around time requirements
Electronic deliverables
Data report format and outline
NOTIFICATION
Exceeding predetermined Maximum Concentration Levels - when applicable
Batch QC failures or other issues
Failure to meet analysis or turnaround times
Violations related to radioactive material license
Change of primary staff associated with contract work
SCHEDULE
Expected date of delivery
Method of delivery of samples
Determine schedule (on batch basis)
Method to report and resolve anomalies and nonconformance in data to the client
Return of samples and disposition of waste
CONTACT
Name, address, phone number of responsible parties
410
E.4.2 Personnel
411 The education, working knowledge, and experience of the individuals that supervise operations,
412 conduct analyses, operate laboratory instruments, process data, and create the deliverables is of
413 key importance to the operation of a laboratory. The Agency is essentially asking: Who is
414 sufficiently qualified to meet the proposed project's needs? (The answer to this question may
415 come from an Agency's guidance or other specific requirements generated by the structured
416 planning process.) The laboratory staff that will perform the analyses should be employed,
417 trained, and qualified prior to the award of the contract.
418 In response to the RFP, the proposer should include a listing of staff members capable of
419 managing, receiving, logging, preparing, and processing samples; providing reports in the format
420 specified by the project; preparing data packages with documentation to support the results;
421 maintaining the chain of custody; and other key work activities. The laboratory should list the
422 administrative personnel and appoint a technical person to be a point of contact for the proposed
423 work. This person should fully understand the project's requirements and be reasonably available
424 to respond to every project need. A proposal should include the educational background and a
425 brief resume for all key personnel. The level of training for each technician should be included.
MARLAP
DO NOT CITE OR QUOTE
E-14
JULY 2001
DRAFT FOR PUBLIC COMMENT
-------
Contracting Laboratory Services
426 Tables E.3 and E.4 are examples that briefly summarize the suggested minimum experience,
427 education, and training for the listed positions. Note, some Agency-specific requirements may
428 exceed the suggested qualifications and this issue should be explored further during the planning
429 process. The goal here is to provide basic guidance with examples that the MARLAP user can
430 employ as a starting point during planning. Once specific requirements are established, this
431 information will appear in the RFP.
432 Table E.3 provides a listing for the types of laboratory technical supervisory personnel that are
433 likely to manage every aspect of a laboratory's work. Each position title is given a brief
434 description of responsibilities, along with the minimum level of education and experience. Table
435 E.4 presents descriptions for staff members that may be considered optional personnel or, in
436 some cases, represent necessary support that is provided by personnel with other position titles.
437 Table E.5 indicates the minimum education and experience for laboratory technical staff
438 members. In some cases, specific training may add to or be substituted for the listed education or
439 experience requirement. Training may come in a number of forms, such as instrument-specific
440 classes offered by a manufacturer, to operational or safety programs given by outside trainers or
441 the laboratory's own staff.
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
TABLE E.3 — Laboratory Technical Supervisory Personnel Listed by Position Title and
Examples for Suggested Minimum Qualifications.
All personnel are responsible to perform their work to meet all terms and conditions of the contract.
Technical Supervisory Personnel
Position Title
and Responsibilities
Education
Experience
Radiochemical Laboratory
Supervisor, Director, or Manager.
Responsible for all technical
efforts of the radiochemical
laboratory.
Minimum of Bachelor's degree in
any scientific/engineering discip-
line, with training in radiochemis-
try, radiation detection instrumen-
tation, statistics, and QA.
Minimum of three years of radioanalyti-
cal laboratory experience, including at
least one year in a supervisory position.
Training in laboratory safety, including
radiation safety.
Quality Assurance Officer
Responsible for overseeing the
quality assurance aspects of the
data and reporting directly to
upper management.
Minimum of Bachelor's degree in
any scientific/engineering discip-
line, with training in physics,
chemistry, and statistics.
Minimum of three years of laboratory
experience, including at least one year of
applied experience with QA principles
and practices in an analytical laboratory
or commensurate training in QA
principles.
JULY 2001
DRAFT FOR PUBLIC COMMENT
E-15
MARLAP
DO NOT CITE OR QUOTE
-------
Contracting Laboratory Services
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
TABLE E.4 — Laboratory Technical Personnel Listed by Position Title and Examples for
Suggested Minimum Qualifications and Examples of Optional Staff Members
Optional Technical Personnel
Position Title
and Responsibilities
Education
Experience
Systems Manager
Responsible for the management and
quality control of all computing systems;
generating, updating, and quality control
for deliverables.
Minimum of Bachelor's degree
with intermediate courses in
programming, information
management, database manage-
ment systems, or systems
requirements analysis.
Minimum of three years
experience in data or systems
management of programming,
including one year experience
with the software being utilized
for data management and
generation of deliverables.
Programmer Analyst
Responsible for the installation, opera-
tion, and maintenance of software and
programs, generating, updating, and
quality of controlling analytical databases
and automated deliverables.
Minimum of Bachelor's degree
with intermediate courses in
programming, information
management, information systems,
or systems requirements analysis.
Minimum of two years
experience in systems or
applications programming,
including one year experience
with the software being utilized
for data management and
generation of deliverables.
474
475
476
477
478
479
480
TABLE E.5 — Laboratory Technical Staff Listed by Position Title and Examples for
Suggested Minimum Qualifications
All personnel are responsible to perform their work to meet all terms and conditions of the contract.
Technical Staff
Position Title
Gamma
Spectrometrist
Alpha
Spectrometrist
Education
• Minimum of Bachelor's degree in chemistry or
any physical scientific/engineering discipline.
• Training courses in gamma spectrometry.
Minimum of Bachelor's degree in chemistry or
any physical scientific/engineering discipline.
• Training courses in alpha spectrometry.
Experience
• Minimum two years experience in
spectrometric data interpretation.
• Formal training or one year experience with
spectral analysis software used to analyze
data.
Formal training or one year experience with
spectral analysis software used to analyze
data.
481
482
MARLAP
DO NOT CITE OR QUOTE
E-16
JULY 2001
DRAFT FOR PUBLIC COMMENT
-------
Contracting Laboratory Services
Position Title
Radiochemist
Counting Room
Technician
Laboratory
Technician
Education
Minimum of Bachelor's degree in chemistry or
any physical scientific/engineering discipline. In
lieu of the educational requirement, two years of
additional, equivalent radioanalytical experience
may be substituted.
Minimum of Bachelor's degree in chemistry or
any scientific/engineering discipline.
Minimum of high school diploma and a college
level course in general chemistry or
equivalent — or college degree in another
scientific discipline (e.g., biology, geology, etc.)
Experience
Minimum of two years experience with
chemistry laboratory procedures, with at least
one year of radiochemistry in conjunction
with the educational qualifications, including
(for example): 1) Operation and maintenance
of radioactivity counting equipment; 2)
Alpha/gamma spectrometric data interpreta-
tion; 3) Radiochemistry analytical procedures;
and 4) Sample preparation for radioactivity
analysis.
Minimum of one year experience in a
radioanalytical laboratory.
Minimum of one year experience in a
radioanalytical laboratory.
483
484
485
486
487
488
E.4.3 Instrumentation
489 A proposer's laboratory must have in place and in good working order the types and required
490 number of instruments necessary to perform the work advertised by the Agency. Specific factors
491 are noted in the RFP, such as: an estimate for the number of samples, length of the contract, and
492 expected turnaround times which influence the types of equipment needed to support the
493 contract.
494 Analytical work can be viewed as a function of current technology. Changes may occur from
495 time to time, especially in relation to scientific advancements in equipment, software, etc.
496 Instrumentation represents the mechanical interface between prepared samples and the data
497 generated in the laboratory. The capacity to process larger and larger numbers of samples while
498 sustaining the desired level of analytical sensitivity and accuracy is ultimately a function of the
499 laboratory's equipment, and the knowledge and experience of the individuals who operate and
500 maintain the instruments. Additional support for the laboratory's on-line activities or the state of
501 readiness to maintain a constant or an elevated peak work load comes in the form of back-up
502 instruments that are available at all times. Information concerning service contracts that provide
503 repairs or replacement when equipment fails to perform is important to meeting contract
504 obligations. Demonstrating that this support will be in place for the duration of the contract is a
505 key element for the proposer to clearly describe in a proposal.
JULY 2001
DRAFT FOR PUBLIC COMMENT
E-17
MARLAP
DO NOT CITE OR QUOTE
-------
Contracting Laboratory Services
506 E.4.3.1 Type, Number, and Age of Laboratory Instruments
507 A description of the types of instruments at a laboratory is an important component of the
508 proposal. The number of each type of instrument available for the proposed work should be
509 indicated in the proposal. This includes various counters, detectors, or other systems used for
510 radioanalytical work. A complete description for each instrument might include the age or
511 acquisition date. This information may be accompanied by a brief description indicating the level
512 of service an instrument provides at its present location.
513 E.4.3.2 Service Contract
514 The types and numbers of service contracts may vary depending on the service provider. Newly
515 purchased instruments will be covered by a manufacturer's warranty. Other equipment used
516 beyond the initial warranty period may either be supported by extensions to the manufacturer's
517 warranties or by other commercial services that cover individual instrument or many instruments
518 under a site-wide service contract. Whatever type of support is in place, the contractor will need
519 to state how having or not having such service contracts affects the laboratory's ability to meet
520 the terms of the contract and the potential impact related to the SOW.
521 E.4.4 Narrative to Approach
522 A proposal can "speak" to the Agency's evaluation team by providing a logical and clearly
523 written narrative of how the proposer will attend to every detail listed in the RFP. This approach
524 conveys key information in a readable format to relate a proposer's understanding, experience,
525 and working knowledge of the anticipated work. In this way, the text also illustrates how various
526 components of the proposal work together to contribute to a unified view of the laboratory
527 functions given the proposed work load as described in the RFP and as detailed in the SOW. The
528 next four sections provide examples of proposal topics for which the proposer may apply a
529 narrative format to address how the laboratory is qualified to do the proposed work.
530 E.4.4.1 Analytical Methods or Protocols
531 The proposer should list all proposed methods they plan to use. The proposal should also furnish
532 all required method validation documentation to gain approval for use. When addressing use of
533 methods, the proposer can describe how a method exhibits the best performance and also offer
534 specific solutions to meet the Agency's needs.
MARLAP JULY 2001
DO NOT CITE OR QUOTE E-18 DRAFT FOR PUBLIC COMMENT
-------
Contracting Laboratory Services
535 E.4.4.2 Meeting Contract Measurement Quality Objectives
536 The Agency's planning process started with a review of questions and issues concerned with
537 generating specific project APSs/MQOs. Stating how a proposer intends to meet the APSs/
538 MQOs data quality requirements adds an important section to the proposal. This allows the
539 competing laboratories to demonstrate that they understand the requirements of the contract and
540 their individual approaches to fulfilling these requirements. Further evidence in support of the
541 proposer's preparations to meet or exceed the Agency's data quality needs is generally covered in
542 a contract laboratory's Quality Manual (Section E.4.5).
543 E.4.4.3 Data Package
544 The proposer responds to the RFP by stating how data will be processed under the contract. A
545 narrative describing the use of personnel, equipment, and facilities illustrates every step in
546 obtaining, recording, storing, formatting, documenting and reporting sample information and
547 analytical results. The specific information related to all these activities and the required
548 information as specified by the SOW is gathered into a data package. For example, a standard
549 data package includes a case narrative, the results (in the format specified by the Agency), a
550 contractor data review checklist, any non-conformance memos resulting from the work, Agency
551 and contractor-internal chains of custody, sample and quality control (QC) sample data (this
552 includes a results listing, calculation file, data file list, and the counting data) and continuing
553 calibration data, and standard and tracer source-trace information, when applicable. At the
554 inception of a project, initial calibration data are provided for detectors used for the work. If a
555 detector is re-calibrated, or a new detector is placed in service, initial calibration data are
556 provided whenever those changes apply to the analyses in question.
557 Specific data from the data package may be further formatted in reports, including electronic
558 formats, as the required deliverables which the contractor will send to the Agency. The delivery
559 of this information is also specified according to a set schedule.
560 E.4.4.4 Schedule
561 The RFP will provide information that allows the proposer to design a schedule that is tailored to
562 the Agency's need. For example, samples that are part of routine monitoring will arrive at the
563 laboratory and the appropriate schedule reflects a cycle of activity from sample preparation to
564 delivering a data package to the Agency. This type of schedule is repeatedly applied to each set
565 of samples. Other projects, surveys, or studies may follow a time line of events from start to
566 completion, with distinct sets of samples and unique needs that arise at specific points in time.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT E-19 DO NOT CITE OR QUOTE
-------
Contracting Laboratory Services
567 The proposer will initially outline a schedule that may utilize some cycling of activities at various
568 stages of the work, but overall the nature of the work may change from stage to stage. The
569 schedule in this case will reflect how the contractor expects to meet certain unique milestones on
570 specific calendar dates.
571 Some projects will have certain requirements to process samples according to a graded
572 processing schedule. The SOW should provide the requirements for the radiological holding time
573 and sample processing turnaround time. Radiological holding time refers to the time required to
574 process the sample—the time differential from the sample receipt date to the final sample matrix
575 counting date. The sample processing turnaround time normally means the time differential from
576 the receipt of the sample at the laboratory (receipt date) to the reporting of the analytical results
577 to the Agency (analytical report date). As such, the turnaround time includes the radiological
578 holding time, the time to generate the analytical results, and the time to report the results to the
579 Agency.
580 Typically, three general time-related categories are stated: routine, expedited, and rush. Routine
581 processing is normally a 30-day turnaround time, whereas expedited processing may have a
582 turnaround time greater than five days but less than 30 days. Rush sample processing may have a
583 radiological holding time of less than five days. For short-lived nuclides, the RFP should state the
584 required radiological holding time, wherein the quantification of the analyte in the sample must
585 be complete within a certain time period. The reporting of such results may be the standard 30-
586 day turnaround time requirement. The Agency should be reasonable and technically correct in
587 developing the required radiological holding and turnaround times.
588 The RFP should specify a schedule of liquidated or compensatory damages that should be
589 imposed when the laboratory is non-compliant relative to technical requirements, radiological
590 holding times, or turnaround times.
591 E.4.4.5 Sample Storage and Disposal
592 The RFP should specify the length of time the contractor must store samples after results are
593 reported. In addition, it should state who is economically and physically responsible for the
594 disposal of the samples. The laboratory should describe how the samples will be stored for the
595 specified length of time and how it plans to dispose of the samples in accordance with local,
596 State and Federal regulations.
MARLAP JULY 2001
DO NOT CITE OR QUOTE E-20 DRAFT FOR PUBLIC COMMENT
-------
Contracting Laboratory Services
597 E.4.5 Quality Manual
598 Only those radiochemistry laboratories that adhere to well-defined quality assurance procedures
599 —pertaining to data validation, internal and external laboratory analytical checks, instrument
600 precision and accuracy, personnel training, and setting routine laboratory guidelines—can insure
601 the highest quality of scientifically valid and defensible data. In routine practice, a laboratory
602 prepares a written description of its quality manual that addresses, at a minimum, the following
603 items:
604 • Organization and Management
605 • Quality System Establishment, Audits, Essential Quality Controls and Evaluation and Data
606 Verification;
607 • Personnel (Qualifications and Resumes);
608 • Physical Facilities - Accommodations and Environment;
609 • Equipment and Reference Materials;
610 • Measurement Traceability and Calibration;
611 • Test Methods and Standard Operating Procedures (Methods);
612 • Sample Handling, Sample Acceptance Policy and Sample Receipt;
613 • Records;
614 • Subcontracting Analytical Samples;
615 • Outside Support Services and Supplies; and
616 • Complaints.
617 The quality manual may be a separately prepared document that may incorporate or reference
618 already available and approved laboratory standard operating procedures (SOPs). This manual
619 provides sufficient detail to demonstrate that the contractor's measurements and data are
620 appropriate to meet the MQOs and satisfy the terms and conditions of the contract. The manual
621 should clearly state the objective of the SOP, how the SOP will be executed, and which
622 performance standards will be used to evaluate the data. Work-related requirements based on
623 quality assurance are also an integral part of the SOW.
624 When a proposal is submitted for review, the contracting laboratory generally sends along a
625 current copy of its quality manual. Additional details pertaining to the content of a quality
626 manual can be found in NELAC (2000), ASQC (1995), EPA (1993, 1994, 1997a), ISO/IEC
627 (17025), and MARS SIM (2000).
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT E-21 DO NOT CITE OR QUOTE
-------
Contracting Laboratory Services
628 E.4.6 Licenses and Accreditations
629 All laboratories must have appropriate licenses from the U.S. Nuclear Regulatory Commission
630 (NRC) or other jurisdictions (Agreement State, host nation, etc.) to receive, possess, use, transfer,
631 or dispose of radioactive materials (i.e., those licensable as indicated in 10 CFR 30.70, Schedule
632 A—Exempt concentrations). A license number and current copy of a laboratory's licenses are
633 typically requested with paperwork that one submits to obtain radionuclide materials—for
634 example, when ordering and arranging to use laboratory standards. Overall, a laboratory's license
635 permits work with certain radionuclides and limits to the quantity of each radionuclide at the
636 laboratory. A proposer's license should allow for new work with the types and anticipated
637 amounts of radionuclides as specified in an RFP. Part of the licensing requirement ensures that
638 the laboratory maintains a functioning radiation safety program and properly trains its personnel
639 in the use and disposal of radioactive materials. For more complete information on license
640 requirements, refer to either the NRC, the appropriate State office, or 10 CFR 30.
641 The laboratory may need to be certified for radioassays by the State in which the lab resides.
642 The RFP should request a copy of the current standing certification(s) to be submitted with the
643 proposal. If the Agency expects a laboratory to process samples from numerous States across the
644 United States, then additional certifications for other States may or will be required. To request
645 that a proposer arrange for certification in multiple States prior to submitting a proposal may be
646 viewed as placing an unfair burden on a candidate laboratory who as yet to learn if it will be
647 awarded a contract. Additional fees, for each State certification, potentially add to a proposer's
648 cost to simply present a proposal. In such cases, an Agency may indicate that additional
649 certification(s)—above that already held for the laboratory's State of residence—may be required
650 once the contract is awarded and just prior to initiating the work.
651 E.4.7 Experience
652 The contractor, viewed as a single entity made of all its staff members, may have an extensive
653 work history as is exemplified through the number and types of projects and contracts that were
654 previously or are currently supported by its laboratory services. This experience is potentially an
655 important testimonial to the kind of work the contractor is presently able to handle with a high
656 degree of competence. The Agency's evaluation team will review this information relative to the
657 need(s) stated in the RFP. The more applicable the track record, the stronger a case the proposer
658 has when competing for the award.
MARLAP JULY 2001
DO NOT CITE OR QUOTE E-22 DRAFT FOR PUBLIC COMMENT
-------
Contracting Laboratory Services
659 E.4.7.1 Previous or Current Contracts
660 In direct relation to the preceding section, the proposer's staff should respond directly to the RFP
661 when asked to provide a list of contracts previously awarded and those they are presently
662 fulfilling. Of primary importance, the list should contain contracts that are similar to the one
663 under consideration (i.e., similar work load and technical requirements), with the following
664 information:
665 • Name of the company or Agency awarding the contract;
666 • Address;
667 • Phone number;
668 • Name of contact person; and
669 • Scope of contract.
670 E.4.7.2 Quality of Performance
671 The Agency's TEC (Section E.5.1) is likely to check a laboratory's results for its participation in
672 a proficiency program which is sponsored by one of several Federal agencies. For example, the
673 U.S. Department of Energy (DOE), and National Institute of Standards and Technology (NIST)
674 offer proficiency programs. Records for the laboratory's results may be reviewed to cover a
675 number of years. This review indicates quality and consistency in relation to the types of samples
676 that the Federal Agency sends to each laboratory. Thus, at designated times during each year, a
677 laboratory will receive, process, and later report findings for proficiency program samples. This
678 routine is also required for certification by an Agency, such as the U.S. Environmental Protection
679 Agency (EPA) for drinking water analysis. In this case, to obtain or maintain a certification, the
680 laboratory must pass (i.e., successfully analyze) on the basis of a specific number of the total
681 samples.
682 E.5 Proposal Evaluation and Scoring Procedures
683 The initial stages of the evaluation process separate technical considerations from cost. Cost will
684 enter the selection process later on. The Agency's TEC will consider all proposals and then make
685 a first cut (Table E.6 and Section E.5.3 below), whereby some proposals are eliminated based on
686 the screening process. This selection from among the candidates is based on predetermined
687 criteria that are related to the original MQOs and how a proposer's laboratory is technically able
688 to support the contract. A lab that is obviously unequipped to perform work according to the
689 SOW is certain to be dropped early in the selection process. In some cases, the stated ability to
690 meet the analysis request should be verified by the Agency, through pre-award audits and
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT E-23 DO NOT CITE OR QUOTE
-------
Contracting Laboratory Services
691 proficiency testing, as described below. Letters notifying unsuccessful bidders may be sent at this
692 time. For information concerning a proposer's response to this letter, see Section E.5.7.
693 E.5.1 Evaluation Committee
694 The Agency personnel initially involved in establishing a new contract and starting the selection
695 process include the Contract Officer (administrative, non-technical) and Contracting Officer's
696 Representative (technical staff person). Once all proposals are accepted by the Agency, a team of
697 technical staff members score the technical portion of the proposal. The team is lead by a
698 chairperson who oversees the activities of this TEC. It is recommended that all members of the
699 TEC have a technical background relevant to the subject matter of the contract.
700 One approach to evaluation includes sending copies of all proposals to each member of the
701 committee for individual scoring (Table E.6). The Agency, after an appropriate length of time,
702 may conduct a meeting or conference call to discuss the scores and reach a unified decision.
703 Using this approach, each proposal is given a numerical score and these are listed in descending
704 order. A "break-point" in the scores is chosen. All candidates above this point are accepted for a
705 continuation of the selection process. Those below the break point may be notified at this point in
706 time. Note that evaluations performed by some agencies may follow variations on this scoring
707 and decision process.
708 The TEC must have a complete technical understanding of the subject matter related to the
709 proposed work and the contract that is awarded at the end of the selection process. These
710 individuals are also responsible for responding to any challenge to the Agency's decision to
711 award the contract. Their answers to such challenges are based on technical merit in relation to
712 the proposed work (Section E.5.7).
713 E.5.2 Ground Rules — Questions
714 The Agency's solicitation should clearly state if and when questions from an individual proposer
715 will be allowed during the selection process. Information furnished in the Agency's response is
716 simultaneously sent to all competing laboratories.
717 E.5.3 Scoring/Evaluating Scheme
718 The Agency should prepare an RFP that includes information concerning scoring of proposals or
719 weights for areas of evaluation. This helps a proposer to understand the relative importance of
720 specific sections in a proposal and how a proposal will be scored. In this case, the method of
MARLAP JULY 2001
DO NOT CITE OR QUOTE E-24 DRAFT FOR PUBLIC COMMENT
-------
Contracting Laboratory Services
721 evaluation and the scoring of specific topic areas is outlined in the solicitation. If this information
722 is not listed in the solicitation and because evaluation formats differ Agency to Agency,
723 proposers may wish to contact the Agency for additional Agency-specific details concerning this
724 process.
725 An Agency may indicate the relative weight an evaluation area holds with regard to the proposed
726 work for two principle reasons. First, the request is focused to meet a need for a specific type of
727 work for a given study, project, or program. This initially allows a proposer to concentrate on
728 areas of greatest importance. Second, if the contractor submits a proposal that lacks sufficient
729 information to demonstrate support in a specific area, the Agency can then indicate how the
730 proposal does not fulfill the need as stated in the request.
731 Listed below is an example of some factors and weights that an Agency might establish before an
732 RFP is distributed:
733 Description Weight
734 Factor I ... .Technical Merit 25
735 Factor n . . .Proposer's Past Performance 25
736 Factor HI. . .Understanding of the Requirements 15
737 Factor IV . .Adequacy and Suitability of Laboratory Equipment and
Resources 15
738 Factor V . . .Academic Qualifications and Experience of Personnel . . 10
739 Factor VI . .Proposer's Related Experience 10
740 The format presented above assigns relative weights for each factor—with greater weight given
741 to more important elements of the proposal. Technical merit (Factor I) includes technical merit,
742 method validation and the ability to meet the MQOs, etc. Factor II includes how well the
743 proposer performed in previous projects or related studies. A proposer's understanding (Factor
744 IE) is demonstrated by the laboratory's programs, commitments as well as certifications, licenses,
745 etc., to ensure the requirements of the RFQ will be met. Adequacy and suitability (Factor IV) is
746 generally an indication that the laboratory is presently situated to accept samples and conduct the
747 work as proposed. Factor V focuses on topics covered previously in Section E.4.2 while the
748 proposer's experience (Factor VI) is considered in Section E.4.7.
749 An Agency may use a Technical Evaluation Sheet—in conjunction with the Proposal Evaluation
750 Plan as outlined in the next section (Table E.6)—to list the total weight for each factor and to
751 provide a space for the evaluator's assigned rating. The evaluation sheet also provides areas to
752 record the RFP number, identity of the proposer, and spaces for total score, remarks, and
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT E-25 DO NOT CITE OR QUOTE
-------
Contracting Laboratory Services
753 evaluator's signature. The scoring and evaluation scheme is based on additional, more detailed,
754 considerations which are briefly discussed in the next three sections (E.5.3.1 to E.5.3.3)
755 E.5.3.1 Review of Technical Proposal and Quality Manual
756 Each bidding-contractor laboratory will be asked to submit a technical proposal and a copy of its
757 Quality Manual. This document is intended to address all of the technical and general laboratory
758 requirements. The proposal and Quality Manual are reviewed by members of the TEC who are
759 both familiar with the proposed project and are clearly knowledgeable in the field of
760 radiochemistry.
761 Table E.6 is an example of a Proposal Evaluation Plan (based on information from the U.S.
762 Geological Survey). This type of evaluation can be applied to proposals as they are considered by
763 the TEC.
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
TABLE E.6 — Example of a Proposal Evaluation Plan
Proposal Evaluation
Objective: To ensure impartial, equitable, and comprehensive evaluation of proposals from contractors desiring
to accomplish the work as outlined in the Request for Proposals and to assure selection of the contractor whose
proposal, as submitted, offers optimum satisfaction of the government's objective with the best composite blend
of performance, schedules, and cost.
Basic Philosophy. To obtain the best possible technical effort which satisfies all the requirements of the
procurement at the lowest overall cost to the government.
Evaluation Procedures
1. Distribute proposals and evaluation instructions to Evaluation Committee.
2. Evaluation of proposals individually by each TEC member. Numerical values are recorded with a concise
narrative justification for each rating.
3. The entire committee by group discussion prepares a consensus score for each proposal. Unanimity is
attempted, but if not achieved, the Chairperson shall decide the score to be given.
4. A Contract Evaluation Sheet listing the individual score of each TEC member for each proposal and the
consensus score for the proposal is prepared by the Chairperson. The proposals are then ranked in
descending order.
5. The Chairperson next prepares an Evaluation Report which includes a Contract Evaluation Sheet, the rating
MARLAP
DO NOT CITE OR QUOTE
E-26
JULY 2001
DRAFT FOR PUBLIC COMMENT
-------
Contracting Laboratory Services
sheets of each evaluator, a narrative discussion of the strong and weak points of each proposal, and a list of
questions which must be clarified at negotiation. This summary shall be forwarded to the Contracting
Officer.
6. If required, technical clarification sessions are held with acceptable proposers.
7. Analysis and evaluation of the cost proposal will be made by the Contracting Officer for all proposals
deemed technically acceptable. The Chairperson of the TEC will perform a quantitative and qualitative
analysis on the cost proposals or those firms with whom cost negotiations will be conducted.
Evaluation Criteria
The criteria to be used in the evaluation of this proposal are selected before the RFP is issued. In accordance with
the established Agency policy, TEC members prepare an average or consensus score for each proposal on the
basis of these criteria and only on these criteria.
A guideline for your numerical rating and rating sheets with assigned weights for each criteria are outlined next
under Technical Evaluation Guidelines for Numerical Rating.
Technical Evaluation Guidelines for Numerical Rating
1. Each item of the evaluation criteria will be based on a rating of 0 to 10 points. Therefore, each evaluator will
score each item using the following guidelines:
a. Above normal: 9 to 10 points (a quote element which has a high probability of exceeding the expressed
RFP requirements).
b. Normal. 6 to 8 points (a quote element which, in all probability, will meet the minimum requirements
established in the RFP and Scope of Work).
c. Below normal. 3 to 5 points (a quote element which may fail to meet the stated minimum requirements,
but which is of such a nature that it has correction potential).
d. Unacceptable: 0 to 2 points (a quote element which cannot be expected to met the stated minimum
requirements and is of such a nature that drastic revision is necessary for correction).
2. Points will be awarded to each element based on the evaluation of the quote in terms of the questions asked.
3. The evaluator shall make no determination on his or her own as to the relative importance of various items
of the criteria. The evaluator must apply a 0 to 10 point concept to each item without regard to his or her
own opinion concerning one item being of greater significance than another. Each item is given a
predetermined weight factor in the Evaluation Plan when the RFP is issued and these weight factors must be
used in the evaluation.
JULY 2001
DRAFT FOR PUBLIC COMMENT
E-27
MARLAP
DO NOT CITE OR QUOTE
-------
Contracting Laboratory Services
812 E.5.3.2 Review of Laboratory Accreditation
813 A copy of the current accreditation(s) should be submitted with the proposal. The Agency should
814 confirm the laboratory's accreditation by contacting the Federal or State Agency that provided
815 the accreditation. In some cases, a public listing or code number is provided. Confirming that a
816 specific code number belongs to a given laboratory will require contacting the Agency that issued
817 the code.
818 E.5.3.3 Review of Experience
819 The laboratory should furnish references in relation to its past or present work (Section E.4.7.1).
820 To the extent possible, this should be done with regard to contracts or projects similar in
821 composition and size to the proposed project. One or more members of the TEC are responsible
822 for developing a list of pertinent questions and then contacting each reference listed by the
823 proposer. The answers obtained from each reference are recorded for use later in the evaluation
824 process. In some cases, the laboratory's previous performance for the same Agency should be
825 given special consideration.
826 E.5.4 Pre-Award Proficiency Samples
827 Some agencies may elect to send proficiency or performance testing (PT) samples to the
828 laboratories that meet a certain scoring criteria in order to demonstrate the laboratory's analytical
829 capability. The composition and number of samples should be determined by the nature of the
830 proposed project. The PT sample matrix should be composed of well-characterized materials. It
831 is recommended that site-specific PT matrix samples or method validation reference material
832 (MVRM; Chapter 6) be used when available. The matrix of which the PT sample is composed
833 must be well characterized and known to the Agency staff who supply the sample to the
834 candidate laboratory. For example, if an Agency is concerned with drinking water samples, then
835 the Agency's laboratory may use its own source of tap water as a base for making PT samples.
836 This water, with or without additives, may be supplied for this purpose.
837 Each competing lab should receive an identical set of PT samples. The RFP should specify who
838 will bear the cost of analyzing these samples, as well as the scoring scheme, (e.g., pass/fail) or a
839 sliding scale. Any lab failing to submit results should be automatically disqualified. The results
840 should be evaluated and each lab given a score. This allows the Agency to narrow the selection
841 further—after which only two or three candidate laboratories are considered.
MARLAP JULY 2001
DO NOT CITE OR QUOTE E-28 DRAFT FOR PUBLIC COMMENT
-------
Contracting Laboratory Services
842 At this point, two additional selection phases remain. A visit to each candidate's facilities comes
843 next (Section E.5.5) and thereafter, once all technical considerations are reviewed, the cost of the
844 contractor's service is examined last (Section E.5.6).
845 E.5.5 Pre-Award Audit
846 A pre-award audit, which may be an initial audit, is often performed to provide assurance that a
847 selected laboratory is capable of performing the required analyses in accordance with the SOW.
848 In other words, is the laboratory's representation (proposal) realistic when compared to the
849 actual facilities? To answer this question, auditors will be looking to see that a candidate
850 laboratory appears to have all the required elements to meet the proposed contract's needs. In
851 some cases, it may be appropriate to conduct both a pre-award audit, followed by an evaluation
852 after the work begins (see Section E.6.7 for information on ongoing laboratory evaluations).
853 The two or three labs with the highest combined scores (for technical proposals and proficiency
854 samples) may be given an on-site audit.
855 The pre-award audit is a key evaluating factor that is employed before the evaluation committee
856 makes a final selection. Many Federal agencies, including DOE, EPA, and USGS, have
857 developed forms for this purpose. Some of the key items to observe during an audit include:
858 • Sample Security - Will the integrity of samples be maintained for chain of custody? If
859 possible, examine the facility's current or past chain-of-custody practice.
860 • Methods - Are copies of SOP's available to every analyst? In some cases, one may check
861 equations used to identify and quantitate the radionuclides of interest. Additional concerns
862 include the potential for interferences, total propagated uncertainty, decision levels, and
863 minimum detectable concentrations.
864 • Method Validation Documentation - Verify the method validation documentation provided
865 in the response to the RFP. Have there been any QA/QC issues related to the methods? Are
866 the identified staff (provided in the RFP) qualified to perform the methods?
867 • Adherence to SOPs - This may include looking to see that sample preparation, chemical
868 analysis, and radiometric procedures are performed according to the appropriate SOP.
869 • Internal QC - Check the files and records.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT E-29 DO NOT CITE OR QUOTE
-------
Contracting Laboratory Services
870 • External QC/PT samples - Check files and records pertaining to third-party programs.
871 • Training - Check training logs. Examine analysts' credentials, qualifications, and proficiency
872 examination results.
873 • Instrumentation - Check logs. Are instruments well maintained, is there much down time, are
874 types and numbers listed in technical proposal correct? Look for QC chart documentation.
875 • Instrumentation - Calibration records. Do past and current calibration records indicate that
876 the laboratory's instruments are capable of providing data consistent with project needs?
877 Look at instrumentation characteristics, including resolution, detection efficiency, typical
878 detection limits, etc. Are NIST-traceable materials used for detector calibration and chemical
879 yield determinations?
880 • Personnel - Talk with and observe analysts. Verbal interaction with laboratory staff during an
881 audit helps auditors to locate the information and likewise provide evidence for the
882 knowledge and understanding of persons who conduct work in the candidate laboratory.
883 • Log-In - Is this area well-organized to reduce the possibility of sample mix-ups?
884 • Tracking-Is there a system of tracking samples through the lab?
885 Information about each laboratory may be gathered in various ways. One option available to the
886 Agency is to provide each candidate laboratory with a list of questions or an outline for
887 information that will be collected during the audit (Table E.7). The Agency's initial contact with
888 the laboratory can include a packet with information about the audit and questions that the
889 laboratory must address prior to the Agency's on-site visit. For example, from the checklist
890 presented in Table E.7, one can see the laboratory will be asked about equipment. In advance of
891 the audit, laboratory personnel can create a listing of all equipment or instruments that will be
892 used to support the contract. Table E.7 also indicates information to be recorded by the auditors
893 during the visit. The audit record includes the Agency's on-site observations, along with the
894 laboratory's prepared responses.
895
896
897
898
899
900
TABLE
E.7 — Sample Checklist for Information
Recorded During a Pre-Award Laboratory Audit
Laboratory:
Date:
Auditors:
1.
2.
MARLAP JULY 2001
DO NOT CITE OR QUOTE E-3 0 DRAFT FOR PUBLIC COMMENT
-------
Contracting Laboratory Services
A. Review packet that was sent to laboratory for completion:
1. Laboratory Supervisor
2. Laboratory Director
3. Current Staff
4. Is the laboratory responsible for all analyses? If not, what other laboratory(s) is (are) responsible?
5. Agency responsible for [drinking water] program in the State.
6. Does the laboratory perform analyses of environmental samples around nuclear power facilities, or
from hospitals, colleges, universities, or other radionuclide users?
7. Agency responsible for sample collections in item 6.
B. Laboratory Facilities:
1. Check all items in the laboratory packet.
2. Comments
3. Is there a Hot Laboratory or a designated area for samples from a nuclear power facility that would
represent a nuclear accident or incident? Is this documented in the SOP or QA Manual?
C. Laboratory Equipment and Supplies:
1. Check all items on the laboratory packet. Includes analytical balances, pH meters, etc.
2. Comments
3. Radiation counting instruments:
a. Thin window gas-flow proportional counters
b. Windowless gas-flow proportional counters
c. Liquid scintillation counter
d. Alpha scintillation counter
e. Radon gas-counting system
f. Alpha spectrometer
g. Gamma spectrometer systems:
1. Ge (HPGe) detectors
2. Nal detectors
3. Multichannel analyzer(s)
D. Analytical Methodology:
1. Check all items on the laboratory packet.
2. Comments
E. Sample Collection, Handling, and Preservation:
1. Check all items on the laboratory packet.
2. Comments
F. Quality Assurance Section:
1. Examine laboratory SOP
a. Comments
2. Examine laboratory's Quality Manual
a. Comments
JULY 2001
DRAFT FOR PUBLIC COMMENT
E-31
MARLAP
DO NOT CITE OR QUOTE
-------
Contracting Laboratory Services
3. Performance Evaluation Studies (Blind)
a. Comments and results
4. Maintenance records on counting instruments and analytical balances.
a. Comments and results
5. Calibration data
a. Gamma Spectrometer system
1. Calibration source
2. Sufficient energy range
3. Calibration frequency
4. Control charts
a. Full Peak Efficiency
b. Resolution
c. Background
b. Alpha/Beta counters
1. Calibration source
2. Calibration frequency
3. Control charts
a. Alpha
b. Beta
c. Background
c. Radon counters
1. Calibration source
2. Frequency of radon cell background checks
d. Liquid Scintillation Analyzer
1. Calibration sources
2. Calibration frequency
3. Control charts
a. H-3
b. C-14
c. Background
d. Quench
6. Absorption and Efficiency curves:
a. Alpha absorption curve
b. Beta absorption curve
c. Ra-226 efficiency determination
d. Ra-228 efficiency determination
e. Sr-89, Sr-90, and Y-90 efficiency determinations
f. Uranium efficiency determination
MARLAP
DO NOT CITE OR QUOTE
E-32
JULY 2001
DRAFT FOR PUBLIC COMMENT
-------
Contracting Laboratory Services
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
7. Laboratory QC Samples
a. Spikes
b. Replicates/duplicates
c. Blanks
d. Cross check samples
e. Frequency of analysis
f. Contingency actions if control samples are out of specification
g. Frequency of analysis
E. Records and Data Reporting
1. Typical data package
2. Electronic data deliverable format
3. Final data report
H. Software Verification and Validation
1. Instrumentation and Equipment Control and Calibrations
2. Analytical Procedure Calculations/Data Reduction
3. Record Keeping/Laboratory/Laboratory Information Management System/Sample Tracking
4. Quality Assurance Related — QC sample program/instrument QC
995 E.5.6 Comparison of Prices
996 To this point, the selection process focuses on technical issues related to conducting work under
997 the proposed contract. Keeping this separate from cost considerations simplifies the process and
998 helps to sustain reviewer objectivity. Once the scoring of labs is final, the price of analyses may
999 be reviewed and compared. Prices are now considered along with inspection results. This part of
1000 the process is best performed by technical personnel, including members of the TEC who work
1001 in either a laboratory or the field setting, and who possess the knowledge to recognize a price that
1002 is reasonable for a given type of analysis. Various scenarios may apply where prices differ:
1003 • Candidates are dropped generally if their proposed prices are extreme.
1004 • Laboratories that score well—aside from their prices that may still be on the high side—are
1005 given an opportunity to rebid with a best and final cost. This lets laboratories know they have
1006 entered the final stage of the selection process.
1007 A final ranking is based on the technical evaluation, including the proficiency examination and
1008 audit if conducted, and the best-and-final prices submitted by each laboratory.
1009 While there is no way to determine how evaluations may be conducted in the future, some extra
1010 consideration may be given to proposals that offer greater technical capabilities (i.e., those that
JULY 2001
DRAFT FOR PUBLIC COMMENT
E-33
MARLAP
DO NOT CITE OR QUOTE
-------
Contracting Laboratory Services
1011 house state-of-the-art or high-tech analytical services) as opposed to fulfilling the minimum
1012 requirements of the RFP.
1013 E.5.7 Debriefing of Unsuccessful Vendors
1014 At an appropriate time in the selection process, all unsuccessful bidders are sent a letter outlining
1015 the reasons that they were not awarded the contract. As noted previously, the RFP should be very
1016 explicit in illustrating what a proposal should contain and which areas carry more or less weight
1017 with regard to the Agency's evaluation. If so, the Agency is able to provide a written response to
1018 specifically identify areas of the proposal where the contractor lacks the appropriate services or is
1019 apparently unable to present a sufficiently strong case documenting an ability to do the work.
1020 Also, as stated previously, the proposer must present as clear a case as possible and write into the
1021 proposal all relevant information. A simple deletion of key information will put a capable
1022 proposer out of the running in spite of the experience, support, and services they are able to
1023 render an Agency.
1024 If a contractor wishes an individual debriefing, the Agency can arrange to have the TEC meet
1025 with the contractor's representatives. This meeting allows for an informal exchange to further
1026 explore issues to the satisfaction of the proposer. This exchange may offer the Agency an
1027 opportunity to restate and further clarify the expected minimum qualifications that are required of
1028 the proposer.
1029 A more formal approach contesting the Agency's decision follows after a protest is lodged by the
1030 contractor. In this case, the Agency's TEC and the contractor's representatives are accompanied
1031 by legal council for both sides.
1032 E.6 The Award
1033 The selection process ends when the Agency personnel designate which contractor will receive
1034 the award. Several steps follow in advance of formally presenting the award. This essentially
1035 includes in-house processing, a review by the Agency's legal department, and a final review by
1036 the contract staff. These activities verify that the entire selection process was followed properly
1037 and that the contract's paperwork is correct. The Agency's contracts office then signs the proper
1038 documents and the paperwork is sent to the contractor. The contract becomes effective as of the
1039 date when the government's contracting officer signs.
MARLAP JULY 2001
DO NOT CITE OR QUOTE E-3 4 DRAFT FOR PUBLIC COMMENT
-------
Contracting Laboratory Services
1040 E.7 For the Duration of the Contract
1041 After the award is made, the Agency enters into a working relationship with the contract
1042 laboratory and work begins. Over the period of the contract, the Agency will send samples,
1043 receive deliverables, and periodically check the laboratory's performance. The work according to
1044 the SOW and the activities associated with performance checks and laboratory evaluations are
1045 topics covered beginning with the next section. Furthermore, as data are delivered to the Agency,
1046 invoices will be sent by the contractor to the Agency. The Agency will process the invoices in
1047 steps: that receipt of data is initially confirmed, the results are appropriate (i.e., valid), and finally
1048 that the invoice is paid. This activity may occur routinely as invoices arrive—weekly, monthly, or
1049 at some other time interval throughout the course of a contract.
1050 Keep in mind that the structured planning process is iterative in nature and may come into play at
1051 any point during a contract period. For example, Federal or State laboratories engaging contract-
1052 support services may be involved in routine monitoring of numerous sampling sites. For sets of
1053 samples that are repeatedly taken from a common location over the course of years, only the
1054 discovery of unique results or change in performance-based methods may instigate an iteration
1055 and a review of the MQOs. For other types of projects, such as a location undergoing a
1056 MARSSEVI-site survey, the project plan may change as preliminary survey work enters a period
1057 of discovery—e.g., during a scoping or characterization survey (MARSSEVI, 2000). Even during
1058 a final status survey, discovery of some previously unknown source of radioactive contamination
1059 may force one to restate not only the problem, but to reconsider every step in the planning
1060 process. Modification of a contract may be necessary to address these circumstances.
1061 E.7.1 Managing a Contract
1062 Communication is key to the successful management and execution of the contract. Problems,
1063 schedule, delays, potential overruns, etc., can only be resolved quickly if communications
1064 between the laboratory and Agency are conducted promptly.
1065 A key element in managing a contract is the timely verification (assessment) of the data packages
1066 provided by the laboratory. Early identification of problems allows for corrective actions to
1067 improve laboratory performance and, if necessary, the cessation of laboratory analyses until
1068 solutions can be instituted to prevent the production of large amounts of data which are unusable.
1069 Note that some sample matrices and processing methods can be problematic for even the best
1070 laboratories. Thus the contract manager must be able to discern between failures due to
1071 legitimate reasons and poor laboratory performance.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT E-3 5 DO NOT CITE OR QUOTE
-------
Contracting Laboratory Services
1072 E.7.2 Responsibility of the Contractor
1073 First and foremost, the responsibility of the laboratory is to meet the performance criteria of the
1074 contract. If the SOW is appropriately written, this provides guidance necessary to ensure the data
1075 produced will meet the project planning goals and be of definable quality. Likewise, the
1076 laboratory must communicate anticipated or unforeseen problems as soon as possible. Again, this
1077 could easily occur with complex, unusual, or problematic sample matrices. Communication is
1078 vital to make sure that matrix interferences are recognized as early as possible, and that
1079 subsequent analyses are planned accordingly.
1080 The laboratory's managers must plan the analysis—that is, have supplies, facilities, staff, and
1081 instruments available as needed—and schedule the analysis to meet the Agency's due date. In the
1082 latter case, a brief buffer period might be included for unanticipated problems and delays, thus
1083 allowing the laboratory the opportunity to take appropriate corrective action on problems
1084 encountered during an analysis.
1085 E.7.3 Responsibility of the Agency
1086 During the period of the contract, the Agency is responsible for employing external quality
1087 assurance oversight. Thus the performance of the laboratory should be monitored continually to
1088 insure the Agency is receiving compliant results. Just because a laboratory produces acceptable
1089 results at the beginning of its performance on a contract does not necessarily mean that it will
1090 continue to do so throughout the entire contract period. For example, the quality of the data can
1091 degenerate at times when an unusually heavy workload is encountered by an environmental
1092 laboratory. One way to monitor this performance is to review the results of internal and external
1093 quality assurance programs. This may in part take the form of site visits (including onsite audits),
1094 inclusion of QC samples, evaluation of performance in Performance Evaluations or
1095 intercomparison programs, desk audits, and data assessments.
1096 E.7.4 Anomalies and Nonconformance
1097 The contractor must document and report all deviations from the method and unexpected
1098 observations that may be of significance to the data user. Such deviations should be documented
1099 in the narrative section of the data package produced by the contract laboratory. Each narrative
1100 should be monitored closely to assure that the laboratory is documenting departures from
1101 contract requirements or acceptable practice. The Agency's reviewer should assure that the
1102 reason(s) given for the departures are clearly explained and are credible. The repeated reporting
1103 of the same deviation may be an indication of internal laboratory problems.
MARLAP JULY 2001
DO NOT CITE OR QUOTE E-3 6 DRAFT FOR PUBLIC COMMENT
-------
Contracting Laboratory Services
1104 E.7.5 Laboratory Assessment
1105 As work under a contract progresses over time, there are two principle means to assess a
1106 laboratory's performance: by having the laboratory process quality control samples (Section
1107 E.7.5.1 and E.7.5.2), and by Agency personnel visiting the laboratory to conduct on-site
1108 evaluations (Section E.7.5.3).
1109 E.7.5.1 Performance and Quality Control Samples
1110 A laboratory's performance is checked in one of several ways, including the use of Agency QC
1111 samples, the laboratory's QC samples, laboratory participation in a performance evaluation
1112 program, Agency certification program, and through Agency audits, which may include an on-
1113 site visit.
1114 There are several approaches to determining that an analysis is accurate and that the data reflect a
1115 true result. One check on each analysis comes from the laboratory's own QC measures. The
1116 contractor will routinely run standards, prepared spiked samples, and blanks, along with the
1117 samples submitted by the Agency. Calibrations are also performed and a laboratory technician is
1118 expected to record information to document instrument performance.
1119 Another avenue for QC comes with measures taken by the Agency, including the incorporation
1120 of a number of double-blind samples, with each batch of samples sent to the contract laboratory.
1121 The preparation of double-blind samples for matrices other than water is difficult. A sample
1122 designated as a blind sample is one that the contractor knows is submitted by the Agency for QC
1123 purposes. A double-blind sample is presented to the laboratory as if it were just another sample
1124 with no indication that this is for QC purposes. In the former case, the samples may be labeled in
1125 such a manner that the laboratory recognizes these as QC samples. In the latter case, unless the
1126 Agency takes steps to use very similar containers and labeling as that for the field samples, the
1127 laboratory may recognize the double-blind samples for what they are. This in effect compromises
1128 the use of a double-blind sample. In each case, the Agency knows the level or amount of each
1129 radionuclide in the blind sample.
1130 When the analysis for a set of samples is complete and data are sent to the Agency, the Agency in
1131 turn checks the results for the QC samples and then performs data validation. In the case of
1132 characterization studies, one may continue to check results for QC samples, but data validation
1133 packages may not be required. If the double-blind results are not within reasonable limits, the
1134 Agency will need to examine how these specific data may indicate a problem. In the meantime,
1135 work on subsequent sample sets cannot go forward until the problem is resolved. Some or all
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT E-3 7 DO NOT CITE OR QUOTE
-------
Contracting Laboratory Services
1136 samples in the questionable batch may need to be reanalyzed depending on the findings for the
1137 QC samples. This is a case where storage of samples by the laboratory—e.g., from three to six
1138 months after analyses are performed—allows the Agency to back track and designate specific
1139 samples for further or repeated analyses. The one exception to going back and doing additional
1140 analyses arises for samples containing radionuclides with short half lives. This type of sample
1141 requires a more immediate assessment to allow for repeated analyses, if needed.
1142 Where data validation is required, the Agency will routinely look at results for the QC samples
1143 that are added to the sample sets collected in the field. An additional QC measure includes a
1144 routine examination—for example, on a monthly or quarterly basis—of the laboratory's results
1145 for their own internal QC samples. This includes laboratory samples prepared as spikes,
1146 duplicates, and blanks that are also run along with the Agency samples.
1147 The Agency can also schedule times to monitor a contractor laboratory's participation in a
1148 performance evaluation program—for example, those supported by the DOE, EPA, NIST, or
1149 NRC. Each laboratory, including the Agency's own facilities, are expected to participate in such
1150 programs. The Agency will also check to see if a laboratory's accreditation (if required) is current
1151 and this is something that should be maintained along with participation in a Federally sponsored
1152 performance evaluation program. In general, the States accredit laboratories within their
1153 jurisdiction.
1154 E.7.5.2 Laboratory Performance Evaluation Programs
1155 Participating in a collaborative interlaboratory testing program (such as the PT programs
1156 mentioned in E.5.4) is the best way for a laboratory to demonstrate or an Agency to evaluate a
1157 laboratory's measurement quality in comparison to other laboratories or to performance
1158 acceptance criteria. Furthermore, because MARLAP promotes consistency among radiochemistry
1159 laboratories, it is scientifically, programmatically, and economically advantageous to embrace the
1160 concept of a common basis for radioanalytical measurements—a measurement quality system
1161 that is ultimately linked to the national physical standards. ANSI N42.23, Measurement and
1162 Associated Instrument Quality Assurance for Radioassay Laboratories., defines a system in
1163 which the quality and traceability of service laboratory measurements to the national standards
1164 can be demonstrated through reference (and monitoring) laboratories. The service (in this case
1165 the contracted) laboratory shall analyze NIST traceable reference performance testing materials
1166 to examine the bias and precision of an analytical methodology or an analyst. Traceable reference
1167 material, a sample of known analyte concentration, is prepared from NIST Standard Reference
1168 Material or derived reference material supplied by a NIST traceable radioactive source
MARLAP JULY 2001
DO NOT CITE OR QUOTE E-3 8 DRAFT FOR PUBLIC COMMENT
-------
Contracting Laboratory Services
1169 manufacturer (ANSI N42.22). Demonstration of measurement performance and traceability shall
1170 be conducted at an appropriate frequency.
1171 E.7.5.3 Laboratory Evaluations Performed During the Contract Period
1172 An audit before awarding a contract emphasizes an examination of availability of instruments,
1173 facilities, and the potential to handle the anticipated volume of work. This also includes
1174 recognizing that the proper personnel are in place to support the contract. After the award, a
1175 laboratory evaluation will place additional weight on how instruments and personnel are
1176 functioning on a daily basis. Thus, logbooks, charts, or other documentation that are produced as
1177 the work progresses are now examined. This type of evaluation during the contract period uses an
1178 approach that differs from the pre-award audit (Section E.5.5). The format and documentation for
1179 an on-site audit may differ from Agency to Agency. An Agency may wish to examine the EPA
1180 forms (EPA, 1997b) and either adopt these or modify them to accommodate radionuclide work
1181 that includes sample matrices other than water or additional nuclides not presently listed.
1182 There are two types of evaluations or audits that can be performed during the life of a contract.
1183 The first involves Agency personnel that visit the contractor's facilities. The second approach
1184 includes activities conducted by Agency personnel without visiting the laboratory.
1185 In the former case, Agency personnel examine documentation at the laboratory, including each
1186 instrument's logbook which is used to record background values, or to ensure that QC charts are
1187 current. During this type of evaluation, the Agency and contractor personnel have an opportunity
1188 to communicate face-to-face, which is a benefit to both parties when clarification or additional
1189 detail is needed. For example, this audit's goal essentially is to check the capability of the
1190 laboratory to perform the ongoing work according to the contract work. In this case, an auditor
1191 may request to see one or more data packages, and then follow the information described in each
1192 package—including such items as sample tracking and documentation concerning sample
1193 preparation and analysis—to verify that the laboratory is now accomplishing the work as
1194 described by the SOW and in conformance with the Quality Manual.
1195 In the latter case, one conducts what might be called a desk audit, where Agency personnel
1196 review the contract and examine records or documentation that have come in as part of the
1197 project's deliverables. For the most part, the Agency should constantly be monitoring activities
1198 under the contract, and in this sense, a desk audit is a daily activity without a formal process
1199 being applied at any specific point in time. However, depending on the Agency's practice, if on-
1200 site visits are not made, then a desk audit becomes the only means to track activities under the
1201 contract. One approach to a desk audit is thus a periodic review—for example, every 6 or 12
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT E-3 9 DO NOT CITE OR QUOTE
-------
Contracting Laboratory Services
1202 months—of QC records to track the laboratory's performance over that period of time. This
1203 allows the Agency to determine if there are deviations, shifts, or other trends that appear over
1204 time.
1205 Each evaluation presents an additional opportunity to monitor various laboratory parameters,
1206 such as turnaround time. This is most important in cases when samples contain radionuclides
1207 having short half lives. During an on-site evaluation, the Agency is able to determine if
1208 additional emphasis is required to tighten the time frame between sample receipt and analysis.
1209 The personal interaction between Agency and laboratory permits a constructive dialog and
1210 facilitates an understanding of the possible means to increase or maintain the efficiency when
1211 processing and analyzing samples at the contractor's facility.
1212 E.8 Contract Completion
1213 There are several general areas of concern at the close of a contract that may be addressed
1214 differently depending on the Agency or nature of the project under a given contract. For example,
1215 Agency personnel who monitor contracts will review invoices to be certain that work is complete
1216 and that the corresponding results are considered acceptable. Once such monitoring activity
1217 provides the proper verification that the work is complete, then the Agency's financial office
1218 processes all related bills and makes final payment for the work.
1219 The laboratory should send in final deliverables, including routine submissions of raw data or
1220 records, as is the practice under the contract. Also, when applicable, Agency-owned equipment
1221 shared with the laboratory during the contract period will be returned. The disposition of samples
1222 still in storage at the contractor's facility and additional records or other raw data must be
1223 decided and specified. The Agency may wish to receive all or part of these items—otherwise,
1224 disposal of sample materials and documents held by the contractor must be arranged.
1225 In some cases, work under the contract may create conditions where more time is necessary to
1226 process samples that remain or to process additional work that arises during the latter part of the
1227 contract period. Depending on the Agency, funding, nature of the project, or other factor, the
1228 contract may be extended for a period of time, which may vary from weeks to months.
1229 Otherwise, once the contract comes to a close, the work ceases.
MARLAP JULY 2001
DO NOT CITE OR QUOTE E-40 DRAFT FOR PUBLIC COMMENT
-------
Contracting Laboratory Services
1230 E.9 References
1231 American National Standards Institute (ANSI). 1995. American National Standard—Traceability
1232 of Radioactive Sources to the National Institute of Standards and Technology (NIST) and
1233 Associated Instrument Quality Control. N42.22.
1234 American National Standards Institute (ANSI). 1996. American National Standard—
1235 Measurement and Associated Instrument Quality Assurance for Radioassay Laboratories.
1236 N42.23.
1237 American Society for Quality Control (ASQC). 1995. Specifications and Guidelines for Quality
1238 Systems for Environmental Data Collection and Environmental Technology Programs.
1239 ANSI/ASQC E4-1994, ASQC, Milwaukee, Wisconsin.
1240 U.S. Environmental Protection Agency (EPA). 1993. Quality Assurance for Superfund
1241 Environmental Data Collection Activities. Publication 9200.2-16FS, EPA, Office of Solid
1242 Waste and Emergency Response, Washington, DC.
1243 U.S. Environmental Protection Agency (EPA). 1994. EPA Requirements for Quality Assurance
1244 Project Plans for Environmental Data Operations. EPA QA/R-5, EPA, Draft Interim Final,
1245 Quality Assurance Management Staff, Washington, DC.
1246 U.S. Environmental Protection Agency (EPA). 1997a. EPA Guidance for Quality Assurance
1247 Project Plans. Final, EPA QA/G-5, EPA, Office of Research and Development, Washington,
1248 DC.
1249 U.S. Environmental Protection Agency (EPA). 1997b. Manual for the Certification of
1250 Laboratories Analyzing Drinking Water. EPA 815-B-97-001.
1251 International Organization for Standardization (ISO/IEC) 17025. General Requirements for the
1252 Competence of Testing and Calibration Laboratories, International Organization for
1253 Standardization, Geneva, Switzerland. 1999.
1254 MARSSEVI. 2000. Multi-Agency Radiation Survey and Site Investigation Manual, Revision 1.
1255 NUREG-1575 Rev 1, EPA 402-R-97-016 Revl, DOE/EH-0624 Revl. August. Available
1256 from http://www.epa.gov/radiation/marssim/filesfm.htm.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT E-41 DO NOT CITE OR QUOTE
-------
Contracting Laboratory Services
1257 NELAC 2000. National Environmental Laboratory Accreditation Conference, Quality Systems,
1258 July 2000.
1259 Office of Federal Procurement Policy (OFPP). 1997. Performance-Based Service Contracting
1260 (PBSC) Solicitation/Contract/Task Order Review Checklist. August 8.
1261 http://www.arnet.gov/Library/OFPP/PolicyDocs/pbscckls.html.
1262 U.S. Code of Federal Regulations, Title 10, Part 30, Rules of General Applicability to Domestic
1263 Licensing of Byproduct Material.
MARLAP JULY 2001
DO NOT CITE OR QUOTE E-42 DRAFT FOR PUBLIC COMMENT
-------
1264 APPENDIX E: CONTRACTING LABORATORY SERVICES E-l
1265 E.I Introduction E-l
1266 E.2 Procurement of Services E-5
1267 E.2.1 Request for Approval of Proposed Procurement Action E-6
1268 E.2.2 Types of Procurement Mechanisms E-6
1269 E.3 Request for Proposals—The Solicitation E-8
1270 E.3.1 Market Research E-9
1271 E.3.2 Length of Contract E-10
1272 E.3.3 Subcontracts E-10
1273 E.4 Proposal Requirements E-l 1
1274 E.4.1 RFP and Contract Information E-l 1
1275 E.4.2 Personnel E-14
1276 E.4.3 Instrumentation E-17
1277 E.4.3.1 Type, Number, and Age of Laboratory Instruments E-18
1278 E.4.3.2 Service Contract E-18
1279 E.4.4 Narrative to Approach E-18
1280 E.4.4.1 Analytical Methods or Protocols E-18
1281 E.4.4.2 Meeting Contract Measurement Quality Objectives E-19
1282 E.4.4.3 Data Package E-19
1283 E.4.4.4 Schedule E-19
1284 E.4.4.5 Sample Storage and Disposal E-20
1285 E.4.5 Quality Manual E-21
1286 E.4.6 Licenses and Accreditations E-22
1287 E.4.7 Experience E-22
1288 E.4.7.1 Previous or Current Contracts E-23
1289 E.4.7.2 Quality of Performance E-23
1290 E.5 Proposal Evaluation and Scoring Procedures E-23
1291 E.5.1 Evaluation Committee E-24
1292 E.5.2 Ground Rules — Questions E-24
1293 E.5.3 Scoring/Evaluation Scheme E-24
1294 E.5.3.1 Review of Technical Proposal and Quality Manual E-26
1295 E.5.3.2 Review of Laboratory Accreditation E-28
1296 E.5.3.3 Review of Experience E-28
1297 E.5.4 Pre-Award Proficiency Samples E-28
1298 E.5.5 Pre-Award Audit E-29
1299 E.5.6 Comparison of Prices E-33
1300 E.5.7 Debriefing of Unsuccessful Vendors E-34
1301 E.6 The Award E-34
1302 E.7 For the Duration of the Contract E-35
1303 E.7.1 Managing a Contract E-35
1304 E.7.2 Responsibility of the Contractor E-36
-------
1305 E.7.3 Responsibility of the Agency E-36
1306 E.7.4 Anomalies and Nonconformance E-36
1307 E.7.5 Laboratory Assessment E-37
1308 E.7.5.1 Performance and Quality Control Samples E-37
1309 E.7.5.2 Laboratory Performance Evaluation Programs E-38
1310 E.7.5.3 Laboratory Evaluations Performed During the Contract Period E-39
1311 E.8 Contract Completion E-40
1312 E.9 References E-41
1313 Table E.I— Examples of Procurement Options to Obtain Materials or Services E-7
1314 Table E.2 — SOW Checklists for the Agency and Proposer E-13
1315 Table E.3 — Laboratory Technical Supervisory Personnel Listed by Position Title and Examples
1316 for Suggested Minimum Qualifications E-15
1317 Table E.4 — Laboratory Technical Personnel Listed by Position Title and Examples for
1318 Suggested Minimum Qualifications and Examples of Optional Staff Members E-16
1319 Table E.5 — Laboratory Technical Staff Listed by Position Title and Examples for Suggested
1320 Minimum Qualifications E-16
1321 Table E.6 — Example of a Proposal Evaluation Plan E-26
1322 Figure E-l — General Sequence Initiating and Later Conducting Work with a Contract
1323 Laboratory E-4
-------
APPENDIX F LABORATORY SUBSAMPLING
2 F.I Introduction
3 In most cases a sample that arrives at the laboratory cannot be analyzed in its entirety. Usually
4 only a small subsample is taken for analysis, and the analyte concentration of the subsample is
5 assumed to be approximately equal to that of the sample itself. Obviously a subsample cannot be
6 perfectly representative of a heterogeneous sample. Improper subsampling may introduce a
7 significant bias into the analytical process. Even when done properly, subsampling increases the
8 variability of the measured result. There are simple methods for controlling the bias, but
9 estimating and controlling the random variability is less straightforward.
10 French geologist Pierre Gy has developed a theory of particulate sampling for applications in
11 mining exploration and development (Gy, 1992), and his work has been promoted in the United
12 States by Francis Pitard (Pitard, 1993). The basic concept of the theory is that the variability in
13 the analyte concentration of a laboratory sample depends on the mass of the sample and the
14 distribution of particle types and sizes in the material sampled. The parti culate sampling theory
15 developed by Gy is applicable to the sampling of soils and radioactive waste (EPA 1992a,
16 1992b). In this appendix, the theory is applied in qualitative and quantitative approaches to the
17 subsampling of parti culate solids in the radiation laboratory.
18 There are many examples of the use of Gy's theory in the mining industry (Assibey-Bonsu 1996;
19 Stephens and Chapman, 1993; Bilonick, 1990; Borgman et al., 1996), and a computer program
20 has been developed for its implementation (Minkkinen, 1989). The theory has recently been
21 adapted for use in environmental science. To date, most environmental applications have been in
22 laboratory and field sampling for hazardous chemicals in Superfund cleanups (Borgman et al.,
23 1994; Shefsky 1997), and there are several applications of the theory that involve mixed
24 radioactive and hazardous wastes (Tamura, 1976).
25 In principle, particulate sampling theory applies to materials of any type, since even gases and
26 liquids are composed of particles (molecules). However, sampling large numbers of randomly
27 distributed molecules in a fluid presents few statistical difficulties; so, the theory is more often
28 applied to particulate solids.
29 One of the most likely applications of Gy's theory in the radiation laboratory is the subsampling
30 of soils. Natural soils are complex mixtures of different particle types, shapes, densities, and
31 sizes. Soil particles range from fine clays at less than 4 jim diameter to coarse sand that ranges
32 over 2 mm in diameter, spanning about 4 orders of magnitude. Contaminants may be absorbed or
33 chemically combined into the soil matrix, adsorbed onto the surfaces of particles, or may occur in
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT F-1 DO NOT CITE OR QUOTE
-------
Laboratory Subsampling
34 discrete particles that are not bound to the soil matrix. Contaminant particles in soil can vary in
35 size from fine airborne deposits of less than 1 jim diameter to relatively large pellets. These
36 factors and others, including radionuclide half-lives, significantly affect the sampling problem.
37 F.2 Basic Concepts
38 This appendix applies Gy's sampling theory to subsampling. To avoid confusion, the terms "lot"
39 and "sample" will be used here instead of "sample" and "subsample," respectively. There may be
40 several subsampling stages at the laboratory, and all of the stages must be considered. At any
41 stage of sampling, the lot is the collection of particles from which a portion is to be taken, and
42 the sample is the portion taken to represent the lot.
43 In Gy's theory, the chemical or physical component whose proportion in a lot is of interest is
44 called the critical component. In the context of radiochemistry, the critical component may be a
45 radionuclide, but, if the chemical form of the radionuclide is known, it may be more useful to
46 consider the critical component to be a chemical compound. Certain applications of Gy's theory
47 require knowledge of the density, so the physical form of the compound may also be important.
48 In the limited context of this appendix, however, the critical component will be identified with
49 the analyte, which is usually a radionuclide.
50 The proportion of critical component by mass in a lot, sample, or particle is called the critical
51 content. In the context of radiochemistry, the critical content is directly related to the activity
52 concentration, or massic activity, of the analyte, but it is expressed as a dimensionless number
53 between 0 and 1. Many of the mathematical formulas used in Gy's sampling theory are equally
54 valid if the critical content is replaced everywhere by analyte concentration. All the formulas in
55 this appendix will be expressed in terms of analyte concentration, not critical content.
56 The sampling error of a sample S is defined, for our purposes, as the relative error in the analyte
57 concentration of the sample, or (zs - zz) / ZL, where zs is the analyte concentration of the sample
58 and ZL is the analyte concentration of the lot. If the sample is the entire lot, the sampling error is
59 zero by definition.
60 A lot may be heterogeneous with respect to many characteristics, including particle size, density,
61 and analyte concentration. Of these, analyte concentration is most important for the purposes of
62 this appendix. A lot may be considered perfectly homogeneous when all particles have the same
63 concentration of analyte.
MARLAP JULY 2001
DO NOT CITE OR QUOTE F-2 DRAFT FOR PUBLIC COMMENT
-------
Laboratory Subsampling
64 The term "heterogeneity" is commonly used with more than one meaning. Gy attempts to clarify
65 the concepts by distinguishing between two types of heterogeneity. The constitution hetero-
66 geneityof a lot is determined by variations among the particles without regard to their locations
67 in the lot. It is an intrinsic property of the lot itself, which cannot be changed without altering
68 individual particles. The distribution heterogeneity of a lot depends not only on the variations
69 among particles but also on their spatial distribution.1 Thus, the distribution heterogeneity may
70 change, for example, when the material is shaken or mixed. In Gy's theory, both constitution
71 heterogeneity and distribution heterogeneity are quantitative terms, which are defined
72 mathematically.
73 Heterogeneity is also sometimes described as either "random" or "nonrandom" (ASTM D5956).
74 Random heterogeneity is exhibited by well-mixed material, in which dissimilar particles are
75 randomly distributed. Nonrandom heterogeneity occurs when particles are not randomly
76 distributed, but instead are stratified. There is a natural tendency for a randomly heterogeneous
77 lot to become more stratified when shaken, bounced, or stirred. The same material may exhibit
78 both random and nonrandom heterogeneity at different times in its history.2
79 In MARLAP's terminology, the representativeness of a sample denotes the closeness of the
80 analyte concentration of the sample to the analyte concentration of the lot. A sample is
81 representative if its analyte concentration is close to the concentration of the lot, just as a
82 measured result is accurate if its value is close to the value of the measurand. Representativeness
83 may be affected by bias and imprecision in the sampling process, just as accuracy may be
84 affected by bias and imprecision in the measurement process.3
85 The concept of representativeness is related to the question of heterogeneity. If a lot is completely
86 homogeneous, then any sample is perfectly representative of the lot, regardless of the sampling
87 strategy, but as the degree of heterogeneity increases, it becomes more difficult to select a
88 representative sample.
:ASTM D5956 uses the terms "compositional heterogeneity" and "distributional heterogeneity."
2A state of random heterogeneity exists when the distribution heterogeneity is zero. A state of
nonrandom heterogeneity exists when the distribution heterogeneity is positive.
3The term "representativeness" is also like "accuracy" inasmuch as it is used with different
meanings by different people. The definition provided here is MARLAP's definition.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT F-3 DO NOT CITE OR QUOTE
-------
Laboratory Subsampling
89 F.3 Sources of Measurement Error
90 The total variance of the result of a measurement is the sum of the variances of a series of error
91 components, including errors produced in the field and in the laboratory. Errors in the laboratory
92 may be divided into those associated with sampling and those associated with sample preparation
93 and analysis.
94 Note that the practical significance of any error, including sampling error, depends on its
95 magnitude relative to the other errors. If a crude analytical procedure is used or if there is a
96 relatively large counting uncertainty, the sampling error may be relatively unimportant. In other
97 cases the sampling error may dominate. If the standard uncertainty from either source is less than
98 about one-third of the standard uncertainty from the other, the smaller uncertainty component
99 contributes little to the combined standard uncertainty.
100 This appendix focuses only on sampling errors, which include the following:
101 • Sampling bias;
102 • The fundamental error; and
103 • Grouping and segregation errors.
104 The following sections define the three types of sampling errors and present methods for
105 controlling or quantifying them. (See Chapter 19, Measurement Statistics, for a more general
106 discussion of laboratory measurement errors.)
107 F.3.1 Sampling Bias
108 Sampling bias is often related to distribution heterogeneity. When there is a correlation between
109 the physical properties of a particle and its location in the lot, care is required to avoid taking a
110 biased sample. For example, if the analyte is primarily concentrated at the bottom of the lot, the
ill analyte concentration of a sample taken from the top will be biased low. Situations like this may
112 occur frequently in environmental radiochemical analysis, since non-natural radioactive materials
113 often tend to be concentrated in the smallest particles, which tend to settle to the bottom of the
114 container.
115 Sampling bias can be controlled by the use of "correct" sampling procedures. A sampling
116 procedure is called "correct" if every particle in the lot has the same probability of being selected
117 for the sample. As a practical rule, a sample is guaranteed to be unbiased only if the sampling
118 procedure is correct.
MARLAP JULY 2001
DO NOT CITE OR QUOTE F-4 DRAFT FOR PUBLIC COMMENT
-------
Laboratory Subsampling
119
120
RULE 1: A sample is guaranteed to be unbiased only if every particle in the lot has the same
probability of selection.
121 The preceding rule is not being followed, for example, if particles on the bottom or in recesses of
122 the container are never selected.
123 Actually the rule stated above is only approximately true.4 It is invalid if the sample consists of
124 only a few particles, or if only a few particles in the lot contain most of the mass. Therefore, a
125 second practical rule of sampling is that the sample must be many times larger (by weight) than
126 the largest particle of the lot.
127
RULE 2: The sample must be many times larger than the largest particle of the lot.
128 Grouping of particles should also be minimized. If the particles form clumps, the effective
129 number of particles in the lot is actually the number of clumps.
130 F.3.2 Fundamental Error
131 When a sample is taken, the existence of constitution heterogeneity in a lot leads to an
132 unavoidable sampling error, called the fundamental error. Its variance, called the fundamental
133 variance, is a property of the lot and the size of the sample. It represents the smallest sampling
134 variance that can be achieved without altering individual particles or taking a larger sample. The
135 fundamental variance is not affected by homogenizing, or mixing, and exists even when the
136 sampling procedure is correct. It cannot be eliminated, but it can be reduced either by increasing
137 the size of the sample or by reducing the particle sizes before sampling (e.g., by grinding).
138
139
140
RULE 3: The fundamental variance may be reduced by:
• Taking a larger sample
• Reducing the particle sizes before sampling
141 This theoretical minimum sampling variance is only achieved in practice when the lot is in a state
142 of pure random heterogeneity (and the sampling is performed correctly). If there is nonrandom
4A sample is unbiased if E(ZS I Ms) = ZL, where Zs is the total analyte activity in the sample, Ms is
the sample mass, ZL is the analyte activity concentration of the lot, and EQ denotes expected value.
Equal selection probabilities guarantee only that E(Z^ I E(M^ = ZL.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT F-5 DO NOT CITE OR QUOTE
-------
Laboratory Subsampling
143 heterogeneity at the time of sampling, the total sampling variance will be larger than the
144 fundamental variance.
145 Either method for reducing the fundamental variance may be difficult or costly to implement in
146 some situations. When large objects or consolidated materials are contained in the lot, particle
147 size reduction for every lot may be unrealistically expensive. Not all materials are amenable to
148 particle size reduction (e.g., steel). If available, knowledge of the expected contamination types
149 and distributions may be used to reduce the need for particle size reduction. For example, it may
150 be known that large objects in the lot are relatively free of analyte. If so, then such objects might
151 be removed or analyzed separately using different methods, depending on the project objectives.
152 When particle size reduction is required and trace levels of contamination are expected in the lot,
153 complete decontamination of grinding or milling equipment is required to avoid the possibility of
154 cross-sample contamination. The equipment should be constructed of non-contaminating
155 materials that are compatible with the chemical components of the lot. Glass, ceramic and
156 stainless steel are typical materials. Particle size reducers, such as ball mills and ceramic plate
157 grinders, require dried samples and thorough decontamination. Mechanical splitters may be
158 difficult to decontaminate. A grinding blank may be analyzed to check for contamination of the
159 grinding equipment.
160 Contamination from airborne sources (e.g., stack releases or incinerator emissions), leaching
161 (e.g., stored mill tailings), or from weathering of contaminated surfaces tends to be dispersed and
162 deposited as many fine particles. In these cases, as long as the particles of the matrix are small
163 relative to the sample size (Rule 2), grinding the material is unlikely to make dramatic
164 differences in the fundamental variance, but the variance tends to be small because of the large
165 number of contaminant particles.
166 If the lot contains only a few contaminant particles, all of which are very small, the fundamental
167 variance may remain large even after extensive grinding. However, the analytical procedure may
168 be amenable to modifications that permit larger samples to be processed. For example,
169 dissolution of a large solid sample may be followed by subsampling of the solution to obtain the
170 amount needed for further analysis. Since liquid solutions tend to be more easily homogenized
171 than solids, subsampling from the solution contributes little to the total sampling error.
172 If neither reducing the particle size nor increasing the sample size is feasible, more innovative
173 analytical techniques may have to be considered.
MARLAP JULY 2001
DO NOT CITE OR QUOTE F-6 DRAFT FOR PUBLIC COMMENT
-------
Laboratory Subsampling
174 F.3.3 Grouping and Segregation Error
175 Since the analyte is often more closely associated with particles having certain characteristics
176 (e.g., small or dense), it may become concentrated in one portion of the lot or in clumps spread
177 throughout the lot. Such effects tend to increase distribution heterogeneity.
178 The existence of distribution heterogeneity leads to a sampling error called the grouping and
179 segregation error. The grouping and segregation variance is not as easily quantified as the
180 fundamental variance, but there are methods for reducing its magnitude.
181 Although the traditional approach to reducing the grouping and segregation error is mixing, or
182 homogenizing, the material, Gy and Pitard warn that homogenizing heterogeneous materials is
183 often difficult, especially if a large quantity is involved. Using improper methods, such as
184 stirring, may actually tend to increase segregation, and, even if a degree of homogeneity is
185 achieved, it is likely to be short-lived, because of the constant influence of gravity. Agitation of
186 particulate matter during transport and handling also tends to produce segregation of particles by
187 size, shape, and density. During these processes, the denser, smaller, and rounder particles tend to
188 settle to the bottom of the container, while less dense, larger, and flatter particles tend to rise to
189 the top.
190
191
192
193
RULE 4: The effects of homogenizing heterogeneous solid material tend to be short-lived
because of the constant influence of gravity. Denser, smaller, and rounder particles tend to
settle to the bottom of a container, while less dense, larger, and flatter particles tend to rise to
the top.
194 As an alternative to homogenizing, Gy and Pitard recommend sampling procedures to reduce not
195 the distribution heterogeneity itself, but its effects on the grouping and segregation error. Gy
196 classifies sampling procedures into two categories: (1) increment sampling, and (2) splitting.
197 Increment sampling involves extracting a number of small portions, called increments., from the
198 lot, which are combined to form the sample. Splitting involves dividing the lot into a large
199 number of approximately equal-sized portions and recombining these portions into a smaller
200 number of potential samples. One of the potential samples is then randomly chosen as the actual
201 sample.
202 A sample composed of many increments will generally be more representative than a sample
203 composed of a single increment. For example, if a 25 g sample is required, it is better to take five
204 5 g increments, selected from different locations in the sample, than to take a single 25 g
205 increment.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT F-7 DO NOT CITE OR QUOTE
-------
Laboratory Subsampling
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
RULE 5: A sample composed of many increments taken from different locations in the lot is
usually more representative than a sample composed of a single increment.
The variance reduction achievable by increment sampling depends on the distribution
heterogeneity of the lot. If the lot is in a state of pure random heterogeneity, increment sampling
provides no benefit. On the other hand, if the lot is highly stratified, the standard deviation of the
analyte concentration of a small composite sample formed from n independent increments may
be smaller by a factor of 1 / \fn than the standard deviation for a sample composed of a single
increment.5 Variance reductions intermediate between these two extremes are most likely in
practice.
Figures F.I and F.2 illustrate what Gy calls "increment delimitation error" and "increment
extraction error," respectively. One method for extracting increments is the one-dimensional
"Japanese slab-cake" method (Gy 1992, Pitard 1993). First, the material in the lot is spread out
into an elongated pile with roughly constant width and height. Then a scoop or spatula is used to
delimit and extract evenly spaced cross-sections from the pile. A flat-bottomed scoop should be
used for this purpose to avoid leaving particles at the bottom of the pile. Ideally it should also
have vertical sides, as shown in Figure F.3, although such scoops may not be commercially
available. If a spatula is used, its width must be much larger than the largest particles to be
sampled, since particles will tend to fall off the edges (see Figure F.2).
FIGURE F.I — Incorrect increment delimitation using a round scoop
5This statement assumes the stratification is such that a single large increment is likely to have no
more constitution heterogeneity than any of the n smaller increment.
MARLAP
DO NOT CITE OR QUOTE
F-8
JULY 2001
DRAFT FOR PUBLIC COMMENT
-------
Laboratory Subsampling
FIGURE F.2 — Incorrect increment extraction using a spatula
FIGURE F.3 — Correct increment delimitation using a rectangular scoop
224 Splitting may be performed correctly by mechanical splitters, such as riffle splitters and sectorial
225 splitters, or it may be performed manually by "fractional shoveling" (or "fractional scooping" in
226 the laboratory). Fractional shoveling involves removing small portions of equal size from the lot
227 and depositing them into two or more empty containers (or piles), cycling through the containers
228 in order, and repeating the process until all the material has been deposited. When this process is
229 complete, one container is chosen at random to be the sample.
230 The traditional "coning and quartering" method for splitting, although correct, is not recommen-
231 ded because it produces a subsample from too few increments. With this method, the material is
JULY 2001
DRAFT FOR PUBLIC COMMENT
F-9
MARLAP
DO NOT CITE OR QUOTE
-------
Laboratory Subsampling
232 mixed by forming it into a cone, adding a fraction of the sample at a time to the apex of the cone.
233 After the entire sample is mixed in this way, the cone is flattened into a circular layer. Next the
234 circular layer of material is divided into quarters and two opposite quarters are discarded. This
235 process may be repeated until a suitable sample size is obtained (Shugar and Dean, 1990).
236 Homogenization may also be achieved with some types of grinding equipment, such as a ring-
237 and-puckmill.
238 According to Gy, small quantities of solid material, up to a few kilograms, can be homogenized
239 effectively in the laboratory. He recommends the use of a jar-shaker for this purpose and states
240 that immediately after the lot is shaken, the sample may be taken directly from the jar using a
241 spatula (Gy, 1992). Although Pitard recognizes the possibility of homogenizing small lots in the
242 laboratory using a mechanical mixer that rotates and tumbles a closed container, he also states
243 that homogenizing heterogeneous materials is often "wishful thinking" and recommends the one-
244 dimensional Japanese slab-cake procedure instead (Pitard, 1993, §14.4.3).
245 F.4 Implementation of the Particulate Sampling Theory
246 DISCLAIMER: Gy 's theory is currently the best-known and most completely developed theory of
247 particulate sampling, but the problem is a difficult one, and the mathematical approaches
248 offered may not give satisfactory results for all purposes. Quantitative estimates of the
249 fundamental variance are often crude. Conservative assumptions are sometimes needed to
250 permit mathematical solutions of the equations, leading to upper bounds for the fundamental
25 1 variance which may be significantly overestimated. It appears that the theory has not been
252 applied previously to sampling for radiochemical analysis, and no data are available to
253 demonstrate the limits of its applicability. Until such data are available, MARLAP recommends
254 the theory only for crude estimation.
255 F.4.1 The Fundamental Variance
2
256 Gy's sampling theory leads to the following equation for the fundamental variance aFE (Gy 1992,
257 Pitard 1993):
258 Here
259 Ms is the mass of the sample (g)
MARLAP
DO NOT CITE OR QUOTE
F- 1 0
JULY 2001
DRAFT FOR PUBLIC COMMENT
-------
Laboratory Subsampling
260 ML is the mass of the lot (g)
261 N is the number of particles in the lot
262 zt is the analyte concentration of the/h particle
263 ZL is the analyte concentration of the lot
264 mi is the mass of the/h particle (g)
265 Equation F. 1 is usually of only theoretical interest because it involves quantities whose values
266 cannot be determined in practice; however, it is the most general formula for the fundamental
267 variance and serves as a starting point for the development of more useful approximation
268 formulas, which are derived using known or assumed properties of the lot.
269 F.4.2 Scenario 1 - Natural Radioactive Minerals
270 Gy has derived a practical formula for the fundamental variance based on the following
271 assumptions (Gy, 1992):
272 • The analyte concentration (actually the critical content) of a particle does not depend on its
273 size. More precisely, if the lot is divided into fractions according to particle size and density,
274 the analyte concentration of each fraction is a function of particle density but not size.
275 • The distribution of particle sizes is unrelated to density. That is, if the lot is divided into
276 fractions by density, each fraction has approximately the same distribution of particle
277 diameters.
278 The first of these assumptions is often violated when environmental samples are analyzed for
279 non-natural radionuclides, because in these cases, the analyte concentration of a particle tends to
280 be inversely related to its size The second assumption may also be violated when non-natural
281 materials are involved. However, when natural materials are analyzed for naturally occurring
282 radionuclides, both assumptions may be valid.
283 Under the two stated assumptions, the fundamental standard deviation a^ is related to the mass
284 of the lot ML, the mass of the sample Ms, and the maximum particle diameter dby the equation
(F.2)
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT F-11 DO NOT CITE OR QUOTE
-------
Laboratory Subsampling
285 where ^is a constant of proportionality.6 The "maximum" diameter (/is defined as the length of
286 the edge of a square mesh that retains no more than a specified fraction of oversize by mass.
287 Thus, it is notthe size of the largest particle in the lot. Gy has found it most convenient to let dbe
288 the size of a square mesh that retains only 5% oversize, and his definition will be assumed here.
289 According to Gy, this value of (/also tends to be the approximate size of the largest particles that
290 are easily identifiable by sight.
291 When Ms is much smaller than ML, which is often the case, the fundamental standard deviation is
292 given more simply by
\
(F.3)
293 This formula implies that, to reduce the fundamental standard deviation by half, one may either
294 increase the sample size M5by a factor of 4 or reduce the maximum particle size dby a factor of
295 0.52/3 = 0.63.7
296 F.4.3 Scenario 2 - Hot Particles
297 As noted, the assumptions of Scenario 1 are often violated when environmental media are
298 analyzed for non-natural radionuclides, because there is usually a correlation between particle
299 size and radionuclide concentration. However, another approximation formula (not due to Gy)
300 may be used if the analyte occurs only in a minuscule fraction of the particles (i.e., "hot
301 particles").
302 It is assumed that:
303 • The maximum analyte concentration of a particle zmax is known;
304 • Every particle in the lot has concentration 0 or zmax (approximately); and
305 • The high-activity particles make up a small fraction of the lot both by number and by mass.
6Gy (1992) and Pitard (1993) provide more information about the constant k. MARLAP presents
only a brief summary of Scenario 1 because of the difficulty of estimating k.
7Equation F.3 also may be understood to say that the fundamental standard deviation is inversely
proportional to the square root of the number of particles in the sample.
MARLAP JULY 2001
DO NOT CITE OR QUOTE F-12 DRAFT FOR PUBLIC COMMENT
-------
Laboratory Subsampling
306
Under these assumptions the fundamental standard deviation a^ is described by the equation8
( 1 - 1
\ Zmax8//4
M M
s
L
2z
307 where
308 Ms is the sample mass (g)
309 ML is the mass of the lot (g)
310 8// is the average density of a high-activity particle (g / cm3)
311 dH is the maximum diameter of a high-activity particle, defined as in Scenario 1
312 k is a constant of proportionality.
313 The proportionality constant k depends on the distribution of sizes of the high-activity particles
314 but is most likely to lie between 0.5 and I.9
315 When Ms is much smaller than ML, Equation F.4 reduces to
^
Zmax5//^//
2zLMs
5)
316 If all the high-activity particles have approximately the same mass and the sample mass is much
317 smaller than the mass of the lot, then Equation F.5 may be rewritten in the simple form
JFE
MSnL
(F.6)
8 A more complete formula is OFE = f — - — 1 """"" ZL f Zm° Zl S/^//^ + &GkGdG] , where 8G, 1G, and
LI M« MJ 2^« V ZL )\
dG describe the zero-activity particles. Equation F.4 is obtained when zmax is much greater than ZL,
which happens when the mass of high-activity material is very small.
9The constant k equals the square root of Gy's "size distribution factor" g. Gy recommends the
value ^=0.25 by default for most uncalibrated materials of interest in the mining industry, but no
assumption is made here that the same default value is appropriate for hot particles. If all the particles
have the same size, g=\.
JULY 2001
DRAFT FOR PUBLIC COMMENT
F- 1 3
MARLAP
DO NOT CITE OR QUOTE
-------
Laboratory Subsampling
318 where nL is the number of hot particles in the lot. Equation F.6 can also be derived from the fact
319 that the number of hot particles in a small sample can be modeled by a Poisson distribution,
320 whose mean and variance are equal (Chapter 19, Measurement Statistics). The fundamental
321 standard deviation equals the coefficient of variation of the Poisson distribution, which is large
322 when the mean is small.
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
EXAMPLE 1
A 1-kg lot of soil contains approximately 1 Bq/g of 240Pu occurring as hot particles of
relatively pure plutonium dioxide (240PuO2, density 8^= 11.4 g/cm3, specific activity
zmax = 7.44 x 109 Bq/g) with "maximum" diameter dH= 1CT3 cm (10 jim). Assume the
distribution of particle sizes is such that k ~ 0.5. What is the fundamental standard deviation
for a 1-gram sample?
According to Equation F.5,
aFE = 0.5
(7.44xl09)(ll.4)(l(T3J3
3.3
Thus, the fundamental standard deviation is about 330%, indicating that a 1 g sample probably
is inadequate.
If all the hot particles had the same size, then k would equal 1 and the fundamental standard
deviation would be about 650%.
When the presence of a small number of hot particles makes it impossible to reduce the
fundamental standard deviation to an acceptable value by ordinary means (grinding the material
or increasing the sample size), then more innovative methods may be required. For example, the
entire lot may be spread into a thin layer and an autoradiograph made to locate the hot particles.
Then, if necessary, a biased sample containing essentially all of the hot particles may be taken
and analyzed, and the measured result corrected for sample size to obtain the average analyte
concentration of the lot.
342
F.4.4 Scenario 3 - Particle Surface Contamination
343 A third approximation formula may be used if the contaminant occurs in tiny particles, or even
344 molecules, which adhere randomlyto the surfaces of larger host particles of the matrix and
MARLAP
DO NOT CITE OR QUOTE
F-14
JULY 2001
DRAFT FOR PUBLIC COMMENT
-------
Laboratory Subsampling
345 cannot be selected without their hosts. In this case, the total mass of the contaminant particles is
346 assumed to be negligible. If the contaminant particles are also extremely numerous, so that many
347 of them adhere to a typical host particle, then the analyte concentration of a particle tends to be
348 inversely proportional to its diameter. In this case the fundamental variance depends primarily on
349 the characteristics of the host particles.10
350 Under the stated assumptions, the fundamental standard deviation a^ for typical soils is given by
(F.7)
Ws MLi
351 where
352 Ms is the sample mass (g)
353 ML is the mass of the lot (g)
354 k is a constant of proportionality
355 5 is the average particle density (g/cm3)
356 d is the "maximum" particle diameter (cm), as defined for Scenario 1
357 The factor ^may vary from lot to lot but is always less than 1 and is usually less than 0.5.
358 When the sample mass is small, Equation F.7 reduces to
2MS
(F.8)
359 The fundamental standard deviation a^ calculated using Equation F.8 is never greater than
360 Jdd3 /2MS, which is the square root of the ratio of the "maximum" particle mass 5 d312 to the
361 mass of the sample Ms. So, as long as the sample is much heavier than the heaviest particle in
362 the lot, the fundamental variance in Scenario 3 tends to be small. As in Scenario 1, reducing the
363 fundamental standard by half requires either increasing the sample mass M5by a factor of 4 or
364 reducing the particle diameter by a factor of 0.63. However, note that grinding may cause the
10The formula for OFE given here describes the variability of the total surface area in a sample. A
more complete expression includes a term for the variability of the analyte concentration per unit area,
but this term is negligible if the number of contaminant particles is sufficiently numerous.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT F-15 DO NOT CITE OR QUOTE
-------
Laboratory Subsampling
365 assumptions underlying Equation F.8 to be violated if the contaminant is not redistributed onto
366 the newly created particle surfaces.
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
EXAMPLE 2
Suppose a 1-kg lot of soil contains 90Sr, which is expected to adhere randomly to the surfaces
of the particles. The maximum particle diameter dis found to be approximately 0.2 cm. If
nothing more is known about the distribution of particles sizes, what is the maximum
fundamental standard deviation for a 1-g sample?
Assuming the density of the soil particles is 8 = 2.675 g/cm3, Equation F.8 with k=\ gives the
solution
(2-675X0.2)3 =
r LJ \
Note that since ^is usually less than 0.5, the fundamental standard deviation is more likely to
be less than 5%.
F.5 Summary
Results derived from particulate sampling theory provide sampling protocols that help to control
sampling errors, including sampling bias, fundamental error, and grouping and segregation
errors. Some of the important conclusions are listed below.
• For most practical purposes, a sample is guaranteed to be unbiased only if all particles in the
lot have the same probability of selection.
• The sample mass should be many times greater than the heaviest particle in the lot, and
clumping of particles should be minimized.
• The fundamental variance, which is considered to be the minimum achievable sampling
variance, may be reduced by increasing the size of the sample or reducing the particle sizes
before sampling.
• Grouping and segregation of particles, which occur because of the particles' differing
physical characteristics and the influence of gravity, tend to increase the sampling variance.
MARLAP
DO NOT CITE OR QUOTE
F-16
JULY 2001
DRAFT FOR PUBLIC COMMENT
-------
Laboratory Subsampling
390 • Grouping and segregation errors can be reduced by increment sampling or by splitting. The
391 more increments, the better.
392 • Correct sampling requires proper tools and procedures.
393 • Small quantities of particulate material can be homogenized effectively in the laboratory
394 using mechanical mixers that rotate and tumble a closed container, but the effects of mixing
395 tend to be short-lived.
396 • Estimation of the fundamental variance requires either knowledge or assumptions about the
397 characteristics of the material being analyzed. Quantitative estimates may be crude.
398 F.6 References
399 American Society for Testing and Materials (ASTM). D5633. "Standard Practice for Sampling
400 with a Scoop." 1994.
401 American Society for Testing and Materials (ASTM). D5956. "Standard Guide for Sampling
402 Strategies for Heterogeneous Wastes." 1996.
403 Assibey-Bonsu, W. 1996. "Summary of present knowledge on the representative sampling of ore
404 in the mining industry." Journal of The South African Institute of Mining and Metallurgy
405 96(6): 289-293.
406 Bilonick, Richard A. 1990. "Gy's parti culate material sampling theory." ASTM Special
407 Technical Publication n 1097. p75-92.
408 Borgman, L.; Anderson-Sprecher, R.; GerowK.; and Flatman, G. 1994. "Cost-effective selection
409 of a sampling plan for spatially distributed hazardous waste."
410 Borgman, L. E.; Kern, J. W.; Anderson-Sprecher R.; Flatman, G. T. 1996. "The sampling theory
411 of Pierre Gy: Comparisons, implementation, and applications for environmental sampling."
412 Principles of Environmental Sampling. 2nd ed.
413 Gy, Pierre M. 1992. Sampling of Heterogeneous and Dynamic Material Systems: Theories of
414 Heterogeneity, Sampling, and Homogenizing. Elsevier, Amsterdam, The Netherlands.
JULY 2001 MARLAP
DRAFT FOR PUBLIC COMMENT F-17 DO NOT CITE OR QUOTE
-------
Laboratory Subsampling
415 Environmental Protection Agency (EPA). 1992a. Preparation of Soil Sampling Protocols:
416 Sampling Techniques and Strategies. Office of Research and Development. EPA/600/R-
417 92/128, Washington, DC.
418 Environmental Protection Agency (EPA). 1992b. CharacterizingHeterogenous Wastes. EPA
419 Office of Research and Development, EPA/600/R-92/033. U.S. Department of Energy,
420 Office of Technology Development.
421 Minkkinen, P. 1989. "A computer program for solving sampling problems." Chemometrics and
422 Intelligent Laboratory Systems (Vol. #): 189-194.
423 Pitard, Francis. 1993. Pierre Gy's Sampling Theory and Sampling Practice: Heterogeneity,
424 Sampling Correctness, and Statistical Process Control, 2nd ed., CRC Press, Boca Raton, FL.
425 Shefsky, S. 1997. "Sample handling strategies for accurate lead-in-soil measurements in the field
426 and laboratory." International Symposium of Field Screening Methods for Hazardous Wastes
427 and Toxic Chemicals, Las Vegas, NV.
428 Shugar and Dean. 1990. The Chemist's Ready Reference Handbook. McGraw-Hill, New York,
429 NY.
430 Stephens, A.J.; Chapman, G. J. 1993. Optimisation of sampling procedures at the Fimiston Open
431 Pit, Kalgoorie" Conference Series - Australasian Institute of Mining and Metallurgy n 5,
432 p!85-194.
433 Tamura, T. 1976. "Physical and Chemical Characteristics of Plutonium, in Existing
434 Contaminated Soils and Sediments." Proceedings of the Symposium on Transuranic Nuclides
435 in the Environment, IAEA Publication ST1/PUB/410, Vienna.
MARLAP JULY 2001
DO NOT CITE OR QUOTE F-18 DRAFT FOR PUBLIC COMMENT
-------
APPENDIX G STATISTICAL TABLES
TABLE G.I — Quantiles of the standard normal distribution
3
4
5
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
p
0.51
0.52
0.53
0.54
0.55
0.56
0.57
0.58
0.59
0.60
0.61
0.62
0.63
0.64
0.65
0.66
0.67
0.68
0.69
0.70
0.71
0.72
0.73
0.74
0.75
1-P
0.49
0.48
0.47
0.46
0.45
0.44
0.43
0.42
0.41
0.40
0.39
0.38
0.37
0.36
0.35
0.34
0.33
0.32
0.31
0.30
0.29
0.28
0.27
0.26
0.25
*J>
0.02507
0.05015
0.07527
0.1004
0.1257
0.1510
0.1764
0.2019
0.2275
0.2533
0.2793
0.3055
0.3319
0.3585
0.3853
0.4125
0.4399
0.4677
0.4959
0.5244
0.5534
0.5828
0.6128
0.6433
0.6745
P
0.76
0.77
0.78
0.79
0.80
0.81
0.82
0.83
0.84
0.85
0.86
0.87
0.88
0.89
0.90
0.91
0.92
0.93
0.94
0.95
0.96
0.97
0.98
0.99
1.00
1-p
0.24
0.23
0.22
0.21
0.20
0.19
0.18
0.17
0.16
0.15
0.14
0.13
0.12
0.11
0.10
0.09
0.08
0.07
0.06
0.05
0.04
0.03
0.02
0.01
0.00
*J>
0.7063
0.7388
0.7722
0.8064
0.8416
0.8779
0.9154
0.9542
0.9945
1.036
1.080
1.126
1.175
1.227
1.282
1.341
1.405
1.476
1.555
1.645
1.751
1.881
2.054
2.326
CO
Note:
= -z
(Continued on next page)
JULY 2001
DRAFT FOR PUBLIC COMMENT
G-l
MARLAP
DO NOT CITE OR QUOTE
-------
Appendix G
31
32
TABLE G.I (Continued) — Quantiles of the standard normal distribution
(Critical Values)
33
34
35
36
37
38
39
40
41
42
43
p
0.90
0.95
0.975
0.99
0.995
0.9975
0.999
0.9995
0.99975
0.9999
1-p
0.10
0.05
0.025
0.01
0.005
0.0025
0.001
0.0005
0.00025
0.0001
^
1.282
1.645
1.960
2.326
2.576
2.807
3.090
3.291
3.481
3.719
44
45
MARLAP
DO NOT CITE OR QUOTE
G-2
JULY 2001
DRAFT FOR PUBLIC COMMENT
-------
Appendix G
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
TABLE G.2 — Quantiles of Student's t distribution
Degrees
of
Freedom
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
p = 0.90
1-/7 = 0.10
tp= 3.078
1.886
1.638
1.533
1.476
1.440
1.415
1.397
1.383
1.372
1.363
1.356
1.350
1.345
1.341
1.337
1.333
1.330
1.328
1.325
1.323
1.321
1.319
1.318
1.316
1.315
1.314
1.313
1.311
1.310
0.95
0.05
6.314
2.920
2.353
2.132
2.015
1.943
1.895
1.860
1.833
1.812
1.796
1.782
1.771
1.761
1.753
1.746
1.740
1.734
1.729
1.725
1.721
1.717
1.714
1.711
1.708
1.706
1.703
1.701
1.699
1.697
0.975
0.025
12.706
4.303
3.182
2.776
2.571
2.447
2.365
2.306
2.262
2.228
2.201
2.179
2.160
2.145
2.131
2.120
2.110
2.101
2.093
2.086
2.080
2.074
2.069
2.064
2.060
2.056
2.052
2.048
2.045
2.042
0.98
0.02
15.895
4.849
3.482
2.999
2.757
2.612
2.517
2.449
2.398
2.359
2.328
2.303
2.282
2.264
2.249
2.235
2.224
2.214
2.205
2.197
2.189
2.183
2.177
2.172
2.167
2.162
2.158
2.154
2.150
2.147
0.99
0.01
31.821
6.965
4.541
3.747
3.365
3.143
2.998
2.896
2.821
2.764
2.718
2.681
2.650
2.624
2.602
2.583
2.567
2.552
2.539
2.528
2.518
2.508
2.500
2.492
2.485
2.479
2.473
2.467
2.462
2.457
0.995
0.005
63.657
9.925
5.841
4.604
4.032
3.707
3.499
3.355
3.250
3.169
3.106
3.055
3.012
2.977
2.947
2.921
2.898
2.878
2.861
2.845
2.831
2.819
2.807
2.797
2.787
2.779
2.771
2.763
2.756
2.750
0.9975
0.0025
127.321
14.089
7.453
5.598
4.773
4.317
4.029
3.833
3.690
3.581
3.497
3.428
3.372
3.326
3.286
3.252
3.222
3.197
3.174
3.153
3.135
3.119
3.104
3.091
3.078
3.067
3.057
3.047
3.038
3.030
JULY 2001
DRAFT FOR PUBLIC COMMENT
G-3
MARLAP
DO NOT CITE OR QUOTE
-------
Appendix G
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
TABLE G.2 (Continued) — Quantiles of Student's t distribution
Degrees
of
Freedom
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
60
70
80
90
100
200
300
400
500
CO
p = 0.90
1-/7 = 0.10
1.309
1.309
1.308
1.307
1.306
1.306
1.305
1.304
1.304
1.303
1.303
1.302
1.302
1.301
1.301
1.300
1.300
1.299
1.299
1.299
1.296
1.294
1.292
1.291
1.290
1.286
1.284
1.284
1.283
1.282
0.95
0.05
1.696
1.694
1.692
1.691
1.690
1.688
1.687
1.686
1.685
1.684
1.683
1.682
1.681
1.680
1.679
1.679
1.678
1.677
1.677
1.676
1.671
1.667
1.664
1.662
1.660
1.653
1.650
1.649
1.648
1.645
0.975
0.025
2.040
2.037
2.035
2.032
2.030
2.028
2.026
2.024
2.023
2.021
2.020
2.018
2.017
2.015
2.014
2.013
2.012
2.011
2.010
2.009
2.000
1.994
1.990
1.987
1.984
1.972
1.968
1.966
1.965
1.960
0.98
0.02
2.144
2.141
2.138
2.136
2.133
2.131
2.129
2.127
2.125
2.123
2.121
2.120
2.118
2.116
2.115
2.114
2.112
2.111
2.110
2.109
2.099
2.093
2.088
2.084
2.081
2.067
2.063
2.060
2.059
2.054
0.99
0.01
2.453
2.449
2.445
2.441
2.438
2.434
2.431
2.429
2.426
2.423
2.421
2.418
2.416
2.414
2.412
2.410
2.408
2.407
2.405
2.403
2.390
2.381
2.374
2.368
2.364
2.345
2.339
2.336
2.334
2.326
0.995
0.005
2.744
2.738
2.733
2.728
2.724
2.719
2.715
2.712
2.708
2.704
2.701
2.698
2.695
2.692
2.690
2.687
2.685
2.682
2.680
2.678
2.660
2.648
2.639
2.632
2.626
2.601
2.592
2.588
2.586
2.576
0.9975
0.0025
3.022
3.015
3.008
3.002
2.996
2.990
2.985
2.980
2.976
2.971
2.967
2.963
2.959
2.956
2.952
2.949
2.946
2.943
2.940
2.937
2.915
2.899
2.887
2.878
2.871
2.839
2.828
2.823
2.820
2.807
MARLAP
DO NOT CITE OR QUOTE
G-4
JULY 2001
DRAFT FOR PUBLIC COMMENT
-------
Appendix G
4*
u
(M
O
J5
"•M
=
S
I/I
1^
ON
o
o
IT)
ON
o
i
o\
o
o
!C
0\
o
o
o
0\
o
* i
S o\
•i o
A
2
a. o
s 2
(3 2
^ =
01
1 i
g
o
o
ir)
o
o
o
o
o
o
o
Tt
§
O
If)
(S
§
o
" S
8 §
S «•- -a
ri
^r oo o *O c^
00 ^H Tf ^O ^H
in i— i oo o *o
m m rs i— i in
i o m ^H ^H
m ,— ; ,—: ,— ;
^000
co
•xt *o ^o -^f -— '
i o i— i oo m
o
CN) (N CN) CN) CS
^ o in m
(N O\ ^H TJ- 00
^o o oo c^ o
m
oo m ^o oo o
^o o m ^o o
c^ i— i rs m in
oo >n I-H ^o I-H
in
in ^o o l> oo
O m c^ o *o
m rs oo in rs
^T m m ^o o
rs o i— i m ^o
00 Tt O vo i— i in o
^r m o oo o
0 l-H ^H C^ t^
o ^r oo ^H m
rs m ^ ^o o
in ^ m in (^
oo i— i in oo i— i
oo o ^ oo
•^t o *o ^ m
^H o rs
in o *o i— i i>
^t in in ^o *o
^o o oo c^ o
oo o rs l> rs m
^O C^ t— i -^f ^o
rs m in ^o o
m Tf in *o *o
o oo ^o vo m
CS OO "n CS C^
CN) CN) m -^ T^
\O -— i \O
-------
Appendix G
4*
w
(M
O
•a
4>
s
=
o
U
CQ
o m ^o c^ rs
in ^o O oo o
in in in in *o
c^ c^ oo ^o ^r
i— i "^ i> o m
in in in in "7")
m oo m O o
C^
O\ i— i
^o m o i> m
in ^o o i> oo
*o m m
^r o i> m o
m "^ ^ in ^o
-— i rs m "^ in
en ro ro ro ro
^H Tf ^O 00 O
•^f i> o m o
^r m o oo o\
*o *o *o *o *o
oo oo oo oo o
m oo ^H ^r o
^ o n
•^ c^ ^t o in
\o ^f m rs o
m ^o o oo c^
O 00 C^ O O
o i> in m rs
o m o r- ^r
^o o oo oo c^
^o o oo c^ o
ro ro ro ro T^
-— i rs r- i> i> i>
in -^t rs o\ o
o m vo oo ^H
oo c^ o i— i m
^o ^o o i> i>
m i— i ^o i— i ^o
C^ CN) ^ I> C^
^r ^o o oo c^
*o *o *o *o *o
^o oo o\ o i— i
m o o\ rs •;^
o ^ in m
m ^o ^o o oo
m o\
^r o o\ i> i> i> i>
o ^r oo i> r- i> i>
m o r- ^r o
oo o ^ m m
rs ^ in ^o o
*o *o *o *o *o
^ I> ^ ^j- o
^o o c^ o ^
oo o\ o in rs c^
in m *o (^ i^
^ i^
m Tf in in ^o
^o o oo c^ o
^r ^r ^r ^r m
^r ^ o ^0 c^
m oo ^H
c^ rs m m -— i
i— i -^f *O OO O
c^ o ^ ^ ^o m c^ m
CO O CO CO CO
O *O I> OO C^
rs — ' ^— '
m i— i in
oo o oo m oo
in ' — i O"N (^ to
oo ^ ^t in ^o
• — i m ^r ^r m
-— i *o c^
rs \o -— i \o o
O rs m m ^t
• — i rs m ^ in
oo ^ o -— i m
(N 00 O
-------
Appendix G
114
TABLE G.4 — Critical values for the nonrandomized exact test
NB
0
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
a = 0.01
1
6
9
11
13
14
16
18
19
21
23
24
26
27
28
30
31
33
34
35
37
38
40
41
42
44
45
46
48
49
50
51
2
4
5
6
7
8
9
10
11
12
13
14
14
15
16
17
17
18
19
20
20
21
22
23
23
24
25
25
26
27
27
28
3
3
4
5
5
6
7
8
8
9
9
10
10
11
12
12
13
13
14
14
15
15
16
16
17
17
18
18
19
19
20
20
4
2
3
4
5
5
6
6
7
7
8
8
8
9
9
10
10
11
11
11
12
12
13
13
13
14
14
15
15
15
16
16
5
2
3
3
4
4
5
5
6
6
7
7
7
8
8
8
9
9
9
10
10
10
11
11
11
12
12
12
13
13
13
13
a = 0.05
1
4
6
8
9
11
12
14
15
17
18
19
21
22
23
25
26
27
29
30
31
32
34
35
36
37
39
40
41
42
44
45
2
2
3
4
5
6
7
8
8
9
10
11
11
12
13
14
14
15
16
16
17
18
18
19
19
20
21
21
22
23
23
24
3
2
3
3
4
4
5
6
6
7
7
8
8
9
9
10
10
11
11
12
12
12
13
13
14
14
15
15
16
16
16
17
4
1
2
3
3
4
4
5
5
5
6
6
7
7
7
8
8
8
9
9
9
10
10
11
11
11
12
12
12
13
13
13
5
1
2
2
3
3
3
4
4
5
5
5
6
6
6
6
7
7
7
8
8
8
9
9
9
9
10
10
10
10
11
11
JULY 2001
DRAFT FOR PUBLIC COMMENT
G-7
MARLAP
DO NOT CITE OR QUOTE
-------
Appendix G
147
TABLE G.4 (Continued) — Critical values for the nonrandomized exact test
NB
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
a = 0.01
1
53
54
55
57
58
59
60
62
63
64
65
67
68
69
70
72
73
74
75
77
78
79
80
82
83
84
85
86
88
89
2
29
29
30
31
32
32
33
33
34
35
35
36
37
37
38
39
39
40
41
41
42
43
43
44
45
45
46
46
47
48
3
21
21
22
22
22
23
23
24
24
25
25
26
26
27
27
27
28
28
29
29
30
30
31
31
31
32
32
33
33
34
4
16
17
17
17
18
18
19
19
19
20
20
20
21
21
21
22
22
22
23
23
23
24
24
24
25
25
25
26
26
26
5
14
14
14
15
15
15
16
16
16
16
17
17
17
18
18
18
18
19
19
19
20
20
20
20
21
21
21
22
22
22
a = 0.05
1
46
47
48
50
51
52
53
54
56
57
58
59
60
61
63
64
65
66
67
68
70
71
72
73
74
75
77
78
79
80
2
25
25
26
26
27
28
28
29
30
30
31
31
32
33
33
34
34
35
36
36
37
37
38
39
39
40
40
41
42
42
3
17
18
18
19
19
19
20
20
21
21
22
22
22
23
23
24
24
24
25
25
26
26
26
27
27
28
28
29
29
29
4
14
14
14
15
15
15
16
16
16
17
17
17
17
18
18
18
19
19
19
20
20
20
21
21
21
22
22
22
23
23
5
11
12
12
12
12
13
13
13
13
14
14
14
14
15
15
15
16
16
16
16
17
17
17
17
18
18
18
18
19
19
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
MARLAP
DO NOT CITE OR QUOTE
G-8
JULY 2001
DRAFT FOR PUBLIC COMMENT
-------
Appendix G
179
TABLE G.4 (Continued) — Critical values for the nonrandomized exact test
NB
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
a = 0.01
1
90
91
92
94
95
96
97
98
100
101
102
103
104
106
107
108
109
110
112
113
114
115
116
118
119
120
121
122
123
125
2
48
49
50
50
51
51
52
53
53
54
55
55
56
56
57
58
58
59
59
60
61
61
62
63
63
64
64
65
66
66
3
34
34
35
35
36
36
37
37
37
38
38
39
39
40
40
40
41
41
42
42
43
43
43
44
44
45
45
45
46
46
4
27
27
27
28
28
28
29
29
29
30
30
30
31
31
31
32
32
32
33
33
33
34
34
34
35
35
35
36
36
36
5
22
23
23
23
23
24
24
24
25
25
25
25
26
26
26
26
27
27
27
27
28
28
28
28
29
29
29
30
30
30
a = 0.05
1
81
82
83
85
86
87
88
89
90
91
93
94
95
96
97
98
99
100
102
103
104
105
106
107
108
110
111
112
113
114
2
43
43
44
45
45
46
46
47
47
48
49
49
50
50
51
52
52
53
53
54
54
55
56
56
57
57
58
58
59
60
3
30
30
31
31
31
32
32
33
33
33
34
34
35
35
35
36
36
37
37
37
38
38
38
39
39
40
40
40
41
41
4
23
23
24
24
24
25
25
25
26
26
26
26
27
27
27
28
28
28
29
29
29
30
30
30
30
31
31
31
32
32
5
19
19
20
20
20
20
21
21
21
21
22
22
22
22
23
23
23
23
24
24
24
24
25
25
25
25
26
26
26
26
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
JULY 2001
DRAFT FOR PUBLIC COMMENT
G-9
MARLAP
DO NOT CITE OR QUOTE
-------
Appendix G
211
TABLE G.4 (Continued) — Critical values for the nonrandomized exact test
NB
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
a = 0.01
1
126
127
128
129
130
132
133
134
135
136
137
139
140
141
142
143
144
146
147
148
149
150
151
152
154
155
156
157
158
159
2
67
67
68
69
69
70
70
71
72
72
73
73
74
75
75
76
76
77
78
78
79
79
80
81
81
82
82
83
84
84
3
47
47
48
48
48
49
49
50
50
50
51
51
52
52
52
53
53
54
54
55
55
55
56
56
57
57
57
58
58
59
4
37
37
37
37
38
38
38
39
39
39
40
40
40
41
41
41
42
42
42
43
43
43
43
44
44
44
45
45
45
46
5
30
31
31
31
31
32
32
32
32
33
33
33
33
34
34
34
34
35
35
35
35
36
36
36
36
37
37
37
37
38
a = 0.05
1
115
116
117
118
120
121
122
123
124
125
126
127
129
130
131
132
133
134
135
136
137
139
140
141
142
143
144
145
146
147
2
60
61
61
62
62
63
64
64
65
65
66
66
67
68
68
69
69
70
70
71
72
72
73
73
74
74
75
76
76
77
3
42
42
42
43
43
44
44
44
45
45
46
46
46
47
47
47
48
48
49
49
49
50
50
51
51
51
52
52
52
53
4
32
33
33
33
33
34
34
34
35
35
35
35
36
36
36
37
37
37
38
38
38
38
39
39
39
40
40
40
40
41
5
26
27
27
27
27
28
28
28
28
29
29
29
29
30
30
30
30
31
31
31
31
32
32
32
32
32
33
33
33
33
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
MARLAP
DO NOT CITE OR QUOTE
G-10
JULY 2001
DRAFT FOR PUBLIC COMMENT
-------
TABLE G.5 — Critical values of Filliben's statistic
>
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
32
Cd
r
o
o
0
§
w
H
9
1—1
a
o
o
H
O
H
W
O
o >
h-J ^
W T3
n
O
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
Significance Level (a)
0.005 0.01 0.025 0.05
0.867 0.869 0.872 0.879
0.813 0.822 0.845 0.868
0.803 0.822 0.855 0.879
0.818 0.835 0.868 0.890
0.828 0.847 0.876 0.899
0.841 0.859 0.886 0.905
0.851 0.868 0.893 0.912
0.860 0.876 0.900 0.917
0.868 0.883 0.906 0.922
0.875 0.889 0.912 0.926
0.882 0.895 0.917 0.931
0.888 0.901 0.921 0.934
0.894 0.907 0.925 0.937
0.899 0.912 0.928 0.940
0.903 0.916 0.931 0.942
0.907 0.919 0.934 0.945
0.909 0.923 0.937 0.947
0.912 0.925 0.939 0.950
0.914 0.928 0.942 0.952
0.918 0.930 0.944 0.954
0.922 0.933 0.947 0.955
0.926 0.936 0.949 0.957
0.928 0.937 0.950 0.958
0.930 0.939 0.952 0.959
0.932 0.941 0.953 0.960
0.934 0.943 0.955 0.962
0.937 0.945 0.956 0.962
0.938 0.947 0.957 0.964
n
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
55
60
65
70
75
80
85
90
95
100
Significance Level (a)
0.005 0.01 0.025 0.05
0.939 0.948 0.958 0.965
0.939 0.949 0.959 0.966
0.940 0.950 0.960 0.967
0.941 0.951 0.960 0.967
0.943 0.952 0.961 0.968
0.945 0.953 0.962 0.968
0.947 0.955 0.962 0.969
0.948 0.956 0.964 0.970
0.949 0.957 0.965 0.971
0.949 0.958 0.966 0.972
0.950 0.958 0.967 0.973
0.951 0.959 0.967 0.973
0.953 0.959 0.967 0.973
0.954 0.960 0.968 0.974
0.955 0.961 0.969 0.974
0.956 0.962 0.969 0.974
0.956 0.963 0.970 0.975
0.957 0.963 0.970 0.975
0.957 0.964 0.971 0.977
0.959 0.965 0.972 0.978
0.962 0.967 0.974 0.980
0.965 0.970 0.976 0.981
0.967 0.972 0.977 0.982
0.969 0.974 0.978 0.983
0.971 0.975 0.979 0.984
0.973 0.976 0.980 0.985
0.974 0.977 0.981 0.985
0.976 0.978 0.982 0.985
0.977 0.979 0.983 0.986
0.979 0.981 0.984 0.987
-------
273
274
275
276
277
278
279
280
281
282
283
284
285
286
o S
o £
o
o
o
o
TABLE G.6 — Summary of probability distributions
Q
to
H
O
Cd
r
KH
o
o ^
O c
11
F(x) denotes the gamma function. F(l/2) = 0t, F(l) = 1, and F(x + 1) = x • Y(x) for x> 0.
* If p = I, the mode is N. Otherwise, if Np + p is an integer and p>0, both Np + p and Np + p - I are modes.
t If X is a positive integer, both X and X - 1 are modes.
Distribution
Binomial
Poisson
Rectangular
Trapezoidal
Normal
Log-Normal
Student's?
Exponential
Chi-Square
Parameters
N,p
»
a , a+
at - a
2
u, a
V.
v
>
V
Values
k 0123
XE [a ,at]
x E [a ,a ]
XE (-00,00)
«.*->
„,»,,
«[*-,
Probability Function
(?),•(.-,*-
Xe +
k\
I
at-a_
x - a
I
at - x
a + a
Y^ ~ flB
2 P
a + a +
x ^ a p
a + a
2n R2si' " 2 "^
1
aV2S
exp(-ln(jc/u
-fr-"2""2
g)2/2(lnag)2)
*(lnagV^
L((v+l)/2)f
L(v/2)^ I
2V(v + D/2
VJ
,,-^
xv/2 - lg -M
2V/2 L(v/2)
Mode
[Np+p\*
w
Not unique
Not unique
H
v--*""
0
0
J 0, v < 2
[ v-2, v>2
Mean
A/p
^
a +a+
2
a + a+
2
"
v'"-'-
0 (v>l)
1
I
V
Standard Deviation
^/NP(^—pj
f.
at - a
if
.-. [77?
2 \ 6
a
/ 20na )2 flna )2
u ye 8 - e g
\ V
\| v - 2
1
I
^
------- |