United States Office of Research and EPA/600/R-98/111
Environmental Protection Development August 1998
Agency Washington, D.C. 20460
vvEPA Environmental Technology
Verification Report
Immunoassay Kit
Strategic Diagnostics Inc.
RaPID Assay System for PCB
Analysis
4^^^^^^^^^^ d^^^^^^^^^^^^^t
-------
EPA/600/R-98/111
August 1998
Environmental Technology
Verification Report
Immunoassay Kit
Strategic Diagnostics Inc.
RaPID Assay System for PCB
Analysis
By
Amy B. Dindal
Charles K. Bayne, Ph.D.
Roger A. Jenkins, Ph.D.
Oak Ridge National Laboratory
Oak Ridge Tennessee 37831-6120
Stephen Billets, Ph.D.
Eric N. Koglin
U.S. Environmental Protection Agency
Environmental Sciences Division
National Exposure Research Laboratory
Las Vegas, Nevada 89193-3478
This demonstration was conducted in cooperation with
U.S. Department of Energy
David Bottrell, Project Officer
Cloverleaf Building, 19901 Germantown Road
Germantown, Maryland 20874
Superfund Innovative Technology
Evaluation Program
on\l
-------
Notice
The U.S. Environmental Protection Agency (EPA), through its Office of Research and
Development (ORD), and the U.S. Department of Energy's Environmental Management (EM)
Program, funded and managed, through Interagency Agreement No. DW89937854 with Oak Ridge
National Laboratory, the verification effort described herein. This report has been peer and
administratively reviewed and has been approved for publication as an EPA document. Mention of
trade names or commercial products does not constitute endorsement or recommendation for use of
a specific product.
-------
£>
I
«t
UNITED STATES ENVIRONMENTAL PROTECTION AGENCY
Office of Research and Development
Washington, D.C. 20460
ENVIRONMENTAL TECHNOLOGY VERIFICATION PROGRAM
VERIFICATION STATEMENT
TECHNOLOGY TYPE: POLYCHLORINATED BIPHENYL (PCB) FIELD ANALYTICAL
TECHNIQUES
APPLICATION: MEASUREMENT OF PCBs IN SOILS AND SOLVENT EXTRACTS
TECHNOLOGY NAME: RaPID ASSAY SYSTEM FOR PCB ANALYSIS
COMPANY: STRATEGIC DIAGNOSTICS INC.
ADDRESS: 111 PENCADER DRIVE
NEWARK, DE 19702-3322
PHONE: (302) 456-6789
The U.S. Environmental Protection Agency (EPA) has created a program to facilitate the deployment of innovative
technologies through performance verification and information dissemination. The goal of the Environmental Technology
Verification (ETV) Program is to further environmental protection by substantially accelerating the acceptance and use
of improved and more cost-effective technologies. The ETV Program is intended to assist and inform those involved in
the design, distribution, permitting, and purchase of environmental technologies. This document summarizes the results
of a demonstration of the Strategic Diagnostics Inc. (SDI) RaPID Assay System for polychlorinated biphenyl (PCB)
Analysis.
PROGRAM OPERATION
The EPA, in partnership with recognized testing organizations, objectively and systematically evaluates the performance
of innovative technologies. Together, with the full participation of the technology developer, they develop plans, conduct
tests, collect and analyze data, and report findings. The evaluations are conducted according to a rigorous demonstration
plan and established protocols for quality assurance. EPA's National Exposure Research Laboratory, which conducts
demonstrations of field characterization and monitoring technologies, with the support of the U.S. Department of
Energy's Environmental Management program, selected Oak Ridge National Laboratory (ORNL) as the testing
organization for the performance verification of PCB field analytical techniques.
DEMONSTRATION DESCRIPTION
In July 1997, the performance of six PCB field analytical techniques was determined under field conditions. Each
technology was independently evaluated by comparing field analysis results to those obtained using approved reference
methods. Performance evaluation (PE) samples were also used to assess independently the accuracy and comparability
of each technology.
The demonstration was designed to detect and measure PCBs in soil and solvent extracts. The demonstration was
conducted at ORNL in Oak Ridge, Tennessee, from July 22 through July 29. The study was conducted under two climatic
conditions. The first site was outdoors, with naturally fluctuating temperatures and relative humidity conditions. The
second site was inside a controlled environmental chamber, with generally cooler temperatures and lower relative
humidities. Multiple soil types, collected from sites in Ohio, Kentucky, and Tennessee, were analyzed in this study.
Solutions of PCBs were also analyzed to simulate extracted surface wipe samples. The results of the soil and extract
analyses conducted under field conditions by the technology were compared with results from analyses of homogeneous
EPA-VS-SCM-17 The accompanying notice is an integral part of this verification statement August 1998
iii
-------
replicate samples conducted by conventional EPA SW-846 methodology in an approved reference laboratory. Details of
the demonstration, including a data summary and discussion of results, may be found in the report entitled Environmental
Technology Verification Report: Immunoassay Kit, Strategic Diagnostics Inc., RaPID Assay System for PCB Analysis,
EPA/600/R-98/111.
TECHNOLOGY DESCRIPTION
The RaPID Assay System applies the principles of enzyme-linked immunosorbent assay to the determination of PCBs.
The sample to be tested is added, along with an enzyme conjugate, to a disposable test tube, followed by paramagnetic
particles coated with PCB-specific antibodies. Both the analyte PCB (which may be in the sample) and the labeled PCB
(the enzyme conjugate) compete for the antibody binding sites on the paramagnetic particles. At the end of an incubation
period, a magnetic field is applied to hold the paramagnetic particles (which contain the analyte PCB and labeled PCB
bound to the antibodies in proportion to their original concentration) in the tube and allow the unbound reagents to be
decanted. After decanting, the particles are washed with washing solution. The presence of PCBs is detected by adding
the enzyme substrate (hydrogen peroxide) and the chromogen (3,3',5,5'-tetramethylbenzidine). The enzyme-labeled PCB
conjugate bound to the PCB-specific antibody catalyzes the conversion of the enzyme substrate/chromogen mixture to
a colored product. After an incubation period, the reaction is stopped and stabilized by the addition of acid. Because the
labeled PCB (enzyme conjugate) is in competition with the analyte PCBs (in the sample) for the antibody sites, the color
development is inversely proportional to the concentration of PCB in the sample (e.g. the darker the color, the less analyte
PCB is present in the sample).
VERIFICATION OF PERFORMANCE
The following performance characteristics of the RaPID Assay System were observed:
Detection limits: EPA defines the method detection limit (MDL) as the minimum concentration of a substance that can
be measured and reported with 99% confidence that the analyte concentration is greater than zero. The MDL was
calculated to be 1.5 ppm based on the performance evaluation sample analyses. This was slightly higher than SDFs
specified MDL of 0.5 ppm.
Throughput: Throughput was 10 to 11 samples/hour. This rate included sample preparation and analysis.
Ease of Use: Three operators analyzed samples during the demonstration, but the technology can be run by a single
trained operator. Minimal training (2 to 4 hours) is required to operate the RaPID Assay System, provided the user has
a fundamental understanding of basic chemical and field analytical techniques.
Completeness: The RaPID Assay System generated results for all 232 PCB samples for a completeness of 100%.
Blank results: No PCBs were detected in either the soil or extract blanks above the RaPID Assay's MDLs; therefore, the
percentage of false positive results was 0%. Two false negative results (1%) were reported for the nonblank soil samples.
Precision: The RaPID Assay System exhibited a significant "site effect" in terms of precision. The overall precision,
based on average relative standard deviations (RSDs), was 25% (under outdoor conditions) and 12% (under chamber
conditions) for soil samples. The outdoor precision was comparable to the reference laboratory's precision (21% RSD),
while the chamber precision was better. The RaPID Assay's precision was comparable to the reference laboratory's (12%
and 14%, respectively) for extract samples.
Accuracy: Accuracy was assessed using PE soil and extract samples. The data showed that the RaPID Assay System
exhibited both positive and negative bias depending on the Aroclor type present in the sample. Because the bias was evenly
distributed (positive and negative), this was not reflected in the overall accuracy (which was based on average percent
recoveries) of 103% for the PE soil samples. Extract measurements were relatively unbiased, with an overall average
EPA-VS-SCM-17 The accompanying notice is an integral part of this verification statement August 1998
iv
-------
percent recovery of 101 %. Evaluation of the data generated at each site indicated that there were no significant differences
between the two data sets based on environmental conditions.
Comparability: This demonstration showed that the RaPID Assay System generated data that exhibited a linear
correlation to the reference laboratory data. The coefficient of determination (R2), which is a measure of the degree of
correlation between the reference laboratory and the RaPID Assay data, was 0.754 when all soil samples (0 to 700 ppm)
were considered. Forthe concentration range from 0 to 125 ppm, the R2 value was 0.716. Approximately 36% of the soil
sample results had percent difference values within the range of ±25%. For extract samples, the data were highly
correlated with the reference laboratory, R2 of 0.977.
Regulatory decision-making: One objective of this demonstration was to assess the technology's ability to perform at
regulatory decision-making levels for PCBs, specifically 50 ppm for soils and 100 (jg/100cm2 for surface wipes. For PE
and environmental soil samples in the range of 40 to 60 ppm, the precision was 21% RSD with a mean accuracy of 126%
recovery. For extract samples representing surface wipe sample concentrations of 100 (jg/100cm2 and 1000 (jg/100cm2
(assuming a 100 cm2 wipe sample), measurements were precise (12% RSD) and accurate (101% recovery).
Data quality levels: The overall performance of the RaPID Assay System was characterized as slightly biased and precise,
under a given set of environmental conditions. Although there was a significant "site effect" in terms of the precision, it
should be noted that the RaPID's worst-case precision (25% RSD) was comparable to the best case precision (21% RSD,
excluding suspect values) for the reference laboratory.
The results of the demonstration show that the SDI RaPID Assay System for PCB analysis can provide useful, cost-
effective data for environmental problem-solving and decision-making. Undoubtedly, it will be employed in a variety of
applications, ranging from serving as a complement to data generated in a fixed analytical laboratory to generating data
that will stand alone in the decision-making process. As with any technology selection, the user must determine if this
technology is appropriate for the application and the project data quality objectives. For more information on this and other
verified technologies, visit the ETV web site at http://www. epa.gov/etv.
Gary J. Foley, Ph.D.
Director
National Exposure Research Laboratory
Office of Research and Development
NOTICE: EPA verifications are based on an evaluation of technology performance under specific, predetermined criteria and the
appropriate quality assurance procedures. EPA makes no expressed or implied warranties as to the performance of the technology
and does not certify that a technology will always, under circumstances other than those tested, operate at the levels verified. The end
user is solely responsible for complying with any and all applicable federal, state, and local requirements.
EPA-VS-SCM-17 The accompanying notice is an integral part of this verification statement August 1998
V
-------
VI
-------
Foreword
The U.S. Environmental Protection Agency (EPA) is charged by Congress with protecting the nation's
natural resources. The National Exposure Research Laboratory (NERL) is EPA's center for the
investigation of technical and management approaches for identifying and quantifying risks to human health
and the environment. NERL's research goals are to (1) develop and evaluate technologies for the
characterization and monitoring of air, soil, and water; (2) support regulatory and policy decisions; and (3)
provide the science support needed to ensure effective implementation of environmental regulations and
strategies.
EPA created the Environmental Technology Verification (ETV) Program to facilitate the deployment of
innovative technologies through performance verification and information dissemination. The goal of the
ETV Program is to further environmental protection by substantially accelerating the acceptance and use of
improved and cost-effective technologies. The ETV Program is intended to assist and inform those involved
in the design, distribution, permitting, and purchase of environmental technologies. This program is
administered by NERL's Environmental Sciences Division in Las Vegas, Nevada.
The U.S. Department of Energy's (DOE) Environmental Management (EM) Program has entered into
active partnership with EPA, providing cooperative technical management and funding support. DOE EM
realizes that its goals for rapid and cost-effective cleanup hinges on the deployment of innovative
environmental characterization and monitoring technologies. To this end, DOE EM shares the goals and
objectives of ETV.
Candidate technologies for these programs originate from the private sector and must be commercially
ready. Through the ETV Program, developers are given the opportunity to conduct rigorous demonstrations
of their technologies under realistic field conditions. By completing the evaluation and distributing the
results, EPA establishes a baseline for acceptance and use of these technologies.
Gary J. Foley, Ph.D.
Director
National Exposure Research Laboratory
Office of Research and Development
vn
-------
Vlll
-------
Abstract
In July 1997, the U.S. Environmental Protection Agency (EPA) conducted a demonstration of
polychlorinated biphenyl (PCB) field analytical techniques. The purpose of this demonstration was to
evaluate field analytical technologies capable of detecting and quantifying PCBs in soils and solvent
extracts. The fundamental objectives of this demonstration were (1) to obtain technology performance
information using environmental and quality control samples, (2) to determine how comparable the
developer field analytical results were with conventional reference laboratory results, and (3) to report on
the logistical operation of the technology. The demonstration design was subjected to extensive review and
comment by EPA's National Exposure Research Laboratory Environmental Sciences Division in Las
Vegas, Nevada; Oak Ridge National Laboratory (ORNL); EPA Regional Offices; the U.S. Department of
Energy; and the technology developers.
The demonstration study was conducted at ORNL under two sets of environmental conditions. The first site
was outdoors, with naturally variable temperature and relative humidity conditions typical of eastern
Tennessee in the summer. A second site was located inside a controlled environmental chamber having
lower, and relatively stable, temperature and relative humidity conditions. The test samples analyzed during
this demonstration were performance evaluation soil, environmental soil, and extract samples. Actual
environmental soil samples, collected from sites in Ohio, Kentucky, and Tennessee, were analyzed, and
ranged in concentration from 0.1 to 700 parts per million (ppm). Extract samples were used to simulate
surface wipe samples, and were evaluated at concentrations ranging from 0 to 100 (jg/mL. The reference
laboratory method used to evaluate the comparability of data was EPA SW-846 Method 8081.
The field analytical technologies tested in this demonstration were the L2000 PCB/Chloride Analyzer
(Dexsil Corporation), the PCB Immunoassay Kit (Hach Company), the 4100 Vapor Detector (Electronic
Sensor Technology), and three immunoassay kits: D TECH, EnviroGard, and RaPID Assay System
(Strategic Diagnostics Inc.). The purpose of an Environmental Technology Verification Report (ETVR) is
to document the demonstration activities, present demonstration data, and verify the performance of the
technology. This ETVR presents information regarding the performance of SDI's RaPID Assay System.
Separate ETVRs have been published for the other technologies demonstrated.
The RaPID Assay System is a field portable instrument that applies the principles of enzyme-linked
immunosorbent assay to the determination of PCBs. The RaPID Assay System uses a PCB-labeled enzyme
conjugate and paramagnetic particles coated with PCB-specific antibodies, where the analyte PCB (which
may be in the sample) and the labeled PCB compete for the antibody binding sites and bind in proportion to
their original concentration. The presence of PCBs is detected by a colored reaction, where the color
development is inversely proportional to the concentration of PCBs in the sample (e.g., the darker the color,
ix
-------
the less PCBs present in the sample). The RaPID Assay System provides no information on Aroclor
identification.
The RaPID Assay System's quantitative results were based on initial calibrations. The method detection
limit (MDL) is often defined as the minimum concentration of a substance that can be measured and
reported with 99% confidence that the analyte concentration is greater than zero. The calculated field-based
MDL (1.5 ppm) was slightly higher than SDFs specified MDL of 0.5 ppm. In general, the RaPID Assay
System's results were slightly biased. Because the bias was evenly distributed (both positive and negative
bias depending on the Aroclor type present in the sample), this was not reflected in the overall accuracy of
103% recovery for the performance evaluation soil samples. Extract measurements were relatively
unbiased, with an overall average percent recovery of 101%. The overall precision exhibited a significant
"site effect," where the average relative standard deviation (RSD) was 25% (under outdoor conditions) and
12% (under chamber conditions) for soil samples. The outdoor precision was comparable to the reference
laboratory's precision (21% RSD), while the chamber precision was better. The precision for extract
samples was comparable with the reference laboratory. Comparability, based on coefficients of
determination (R2), was 0.754 for all soil samples (0 to 700 ppm), where an R2 of 1.0 denotes perfect
correlation. Approximately 36% of the soil sample results had percent difference values within the range of
±25%. The data for the extract samples were highly correlated with the reference laboratory.
The demonstration found that the RaPID Assay System was light, easily transportable, and rugged,
requiring about one hour for initial setup and preparation for sample analysis. Once operational, the sample
throughput of the RaPID Assay System was 10 to 11 samples/h. Three operators analyzed samples during
the demonstration, but the technology can be run by a single operator. Minimal training (2 to 4 h) is required
to operate the RaPID Assay System, provided the user has a fundamental understanding of basic chemical
and field analytical techniques. No site effects (i.e., differences in performance due to environmental
conditions) were observed in terms of the accuracy of the measurements; however, the significant (but
comparable to the best case precision of the reference laboratory) site effect for precision should be
considered when using this technology. Overall, the performance of the RaPID Assay System was
characterized as slightly biased and precise, under a given set of environmental conditions.
-------
Table of Contents
Notice ii
Verification Statement iii
Foreword vii
Abstract ix
List of Figures xv
List of Tables xvii
List of Abbreviations and Acronyms xix
Acknowledgments xxi
Section 1 Introduction 1
Technology Verification Process 2
Needs Identification and Technology Selection 2
Demonstration Planning and Implementation 3
Report Preparation 3
Information Distribution 3
Demonstration Purpose 4
Section 2 Technology Description 5
Objective 5
Principle 5
Sample Preparation 6
Description of Kit Contents 6
Materials Not Provided in Kit 6
Procedure 7
Sample Weighing 7
Extraction 8
Filtration 8
Dilution 8
Assay Procedure 8
xi
-------
Reagents 8
Materials Required but Not Provided in Kit 9
Procedural Notes and Precautions 9
Method 10
Results 11
Manual Calculations 11
RPA-I RaPID Analyzer 11
Limitations 12
Section 3 Site Description and Demonstration Design 15
Objective 15
Demonstration Site Description 15
Site Name and Location 15
Site History 15
Site Characteristics 15
Experimental Design 16
Environmental Conditions during Demonstration 18
Sample Descriptions 18
Performance Evaluation Materials 18
Environmental Soil Samples 20
Extract Samples 20
Sampling Plan 20
Sample Collection 20
Sample Preparation, Labeling, and Distribution 20
Predemonstration Study 22
Predemonstration Sample Preparation 22
Predemonstration Results 23
Deviations from the Demonstration Plan 23
Section 4 Reference Laboratory Analytical Results and Evaluation 25
Objective and Approach 25
Reference Laboratory Selection 25
Reference Laboratory Method 26
Calibration 26
Sample Quantification 26
Sample Receipt, Handling, and Holding Times 27
Quality Control Results 27
Objective 27
Continuing Calibration Verification Standard Results 27
Instrument and Method Blank Results 28
Surrogate Spike Results 28
Laboratory Control Sample Results 28
Matrix Spike Results 29
Conclusions of the Quality Control Results 29
xii
-------
Xlll
-------
Data Review and Validation 29
Objective 29
Corrected Results 30
Suspect Results 30
Data Assessment 31
Objective 31
Precision 31
Performance Evaluation Samples 31
Environmental Soil Samples 32
Extract Samples 34
Accuracy 34
Performance Evaluation Soil Samples 34
Extract Samples 35
Representativeness 36
Completeness 36
Comparability 37
Summary of Observations 37
Section 5 Technology Performance and Evaluation 39
Objective and Approach 39
Data Assessment 39
Precision 39
Performance Evaluation Samples 39
Environmental Soil Samples 40
Extract Samples 42
Precision Summary 42
Accuracy 43
Performance Evaluation Soil Samples 43
Extract Samples 44
Accuracy Summary 44
False positive/False Negative Results 45
Representativeness 46
Completeness 46
Comparability 46
Summary of PARCC Observations 49
Regulatory Decision-Making Applicability 51
Additional Performance Factors 51
Detection Limits 51
Sample Throughput 52
Cost Assessment 52
RaPID Assay System Costs 52
Reference Laboratory Costs 54
Cost Assessment Summary 54
General Observations 55
xiv
-------
Performance Summary 55
Section 6 Technology Update and Representative Applications 57
Objective 57
Technology Update 57
Reconfiguration of Soil Extraction (Sample Preparation) Products 57
Instrument Consolidation 57
Representative Applications 58
Data Quality Objective Example 58
Section 7 References 59
Appendix A Description of Environmental Soil Samples 61
Appendix B Characterization of Environmental Soil Samples 65
Appendix C Temperature and Relative Humidity Conditions 69
Appendix D PCB Technology Demonstration Sample Data 75
Appendix E Data Quality Objective Example 87
Disclaimer 89
Background and Problem Statement 89
Data Quality Objective Goals 89
Use of Performance Information to Implement the Decision Rule 90
RaPID Assay Accuracy 91
Determining the Number of Samples 92
Determining the Action Level 94
Alternative FP Error Rate 95
xv
-------
List of Figures
2-1. Soil sample collection device 7
3-1. Schematic map of ORNL, indicating the demonstration area 17
5-1. Standard deviation vs average PCB concentration for the RaPID Assay measurements with average
values less than 100 ppm 41
5-2. Paired PCB measurements for RaPID Assay and reference laboratory for (a) all soil samples (RaPID
Assay results as ">200 ppm" not included) and (b) soil samples where the reference laboratory
results were less than or equal to 125 ppm. Lines denote perfect correlation 46
5-3. Range of percent difference values for the comparison of the RaPID Assay's soil sample results with
the reference laboratory results 48
5-4. Paired extract PCB measurements for the RaPID Assay and reference laboratory 48
5-5. Range of percent difference values for the comparison of the RaPID Assay's extract sample results
with the reference laboratory results 49
C-l. Summary of temperature conditions for outdoor site 71
C-2. Summary of relative humidity conditions for outdoor site 72
C-3. Summary of temperature conditions for chamber site 72
C-4. Summary of relative humidity conditions for chamber site 73
E-l. A linear model for predicting SDI's RaPID Assay results from the certified PE values with 95%
confidence intervals 91
E-2. A linear model for predicting SDI's RaPID Assay results from the reference laboratory results for
environmental soils up to 60 ppm 91
E-3. Decision performance curves for RSD =15% (solid line) and RSD = 25% (dashed lined) for the PCB
drum example 96
xvi
-------
-------
List of Tables
2-1. Tube labels 11
2-2. Recommended parameter settings for the RPA-I RaPID analyzer 12
2-3. Cross-reactivity of the RaPID Assay System for PCB analysis 13
3-1. Summary of experimental design by sample type 18
3-2. Summary of the RaPID Assay's predemonstration results 23
4-1. Suspect measurements within the reference laboratory data 30
4-2. Precision of the reference laboratory for PE soil samples 32
4-3. Precision of the reference laboratory for environmental soil samples 33
4-4. Precision of the reference laboratory for extract samples 34
4-5. Accuracy of the reference laboratory for PE soil samples 35
4-6. Accuracy of the reference laboratory for extract samples 36
4-7. Summary of the reference laboratory performance 38
5-1. Precision of the RaPID Assay System for PE soil samples 40
5-2. Precision of the RaPID Assay System for environmental soil samples 41
5-3. Precision of the RaPID Assay System for extract samples 42
5-4. Overall precision of the RaPID Assay System for all sample types 43
5-5. Accuracy of the RaPID Assay System for PE soil samples 44
5-6. Accuracy of the RaPID Assay System for extract samples 45
5-7. Overall accuracy of the RaPID Assay System for all sample types 45
5-8. Comparison of the reference laboratory's suspect data to the RaPID Assay System data 50
5-9. Summary of PARCC observations for the RaPID Assay System 50
5-10. Performance of the RaPID Assay System for soil samples between 40 and 60 ppm 51
5-11. Estimated analytical costs for PCB soil samples 53
5-12. Performance summary for the RaPID Assay System 56
A-l. Summary of soil sample descriptions 63
B-l. Summary of environmental soil characterization 67
C-l. Average temperature and relative humidity conditions during testing periods 71
D-l. RaPID Assay System PCB technology demonstration soil sample data 78
D-2. RaPID Assay System PCB technology demonstration extract sample data 84
D-3. Corrected reference laboratory data 85
E-l. Predicting SDI's RaPID Assay Kit results from graphs of performance data 92
xvm
-------
-------
List of Abbreviations and Acronyms
AL action level
ANOVA analysis of variance
ASTM American Society for Testing and Materials
CCV continuing calibration verification standard
CI confidence interval
CSCT Consortium for Site Characterization Technology
DCB decachlorobiphenyl
DOE U.S. Department of Energy
DQO data quality objective
EM Environmental Management (DOE)
EPA U.S. Environmental Protection Agency
ERA Environmental Resource Associates
ETTP East Tennessee Technology Park
ETV Environmental Technology Verification (Program)
ETVR Environmental Technology Verification Report
EvTEC Environmental Technology Evaluation Center
fn false negative result
FN false negative decision error rate
fp false positive result
FP false positive decision error rate
ID identifier
LCS laboratory control sample
LMES Lockheed Martin Energy Systems
MDL method detection limit
MS matrix spike
MSD matrix spike duplicate
n number of samples
xx
-------
NERL National Exposure Research Laboratory (EPA)
NCEPI National Center for Environmental Publications and Information
NRC U.S. Nuclear Regulatory Commission
ORNL Oak Ridge National Laboratory
ORO Oak Ridge Operations (DOE)
PARCC precision, accuracy, representativeness, completeness, comparability
PCB polychlorinated biphenyl
PDF portable document format
PE performance evaluation
ppb parts per billion
ppm parts per million; equivalent units: mg/kg for soils and (jg/mL for extracts
Pr probability
QA quality assurance
QC quality control
R2 coefficient of determination
RDL reporting detection limit
RH relative humidity
RFD request for disposal
RPD relative percent difference
RSD relative standard deviation (percent)
RT regulatory threshold
SD standard deviation
SDI Strategic Diagnostics Inc.
SITE Superfund Innovative Technology Evaluation
SMO Sample Management Office
SOP standard operating procedure
SSM synthetic soil matrix
TCMX tetrachloro-m-xylene
TSCA Toxic Substances Control Act
Z!_P the (1 - p)th percentile for the standard normal distribution
%D percent difference
xxi
-------
Acknowledgments
The authors wish to acknowledge the support of all those who helped plan and conduct the demonstration,
analyze the data, and prepare this report. In particular we recognize the technical expertise of Mitchell
Erickson (Environmental Measurements Laboratory), Viorica Lopez-Avila (Midwest Research Institute),
and Robert F. O'Brien (Pacific Northwest National Laboratory), who were peer reviewers of this report; for
internal peer review, Stacy Barshick (ORNL); for technical support during the demonstration, Todd Skeen
and Ralph Ilgner (ORNL); for site safety and health support, Kim Thomas, Marilyn Hanner, and Fred
Smith (ORNL); for administrative support, Betty Maestas and Linda Plemmons (ORNL); for sample
collection support, Wade Hollinger, Charlotte Schaefer, and Arlin Yeager (LMES) and Mike Rudacille and
W. T. Wright (EET Corporation); for preliminary soil characterization support, Frank Gardner, John
Zutman, and Bob Schlosser (ORNL, Grand Junction, Colorado); for sample management support, Angie
McGee, Suzanne Johnson, and Mary Lane Moore (LMES); for providing performance evaluation samples,
Michael Wilson (EPA's Office of Solid Waste and Emergency Response's Analytical Operations and Data
Quality Center); and for technical guidance and project management of the demonstration, David Garden,
Marty Atkins, and Regina Chung (DOE's Oak Ridge Operations Office), David Bottrell (DOE,
Headquarters), Deana Crumbling (EPA's Technology Innovation Office), and Dr. Stephen Billets, Gary
Robertson, and Eric Koglin (EPA's National Exposure Research Laboratory, Las Vegas, Nevada). The
authors also acknowledge the participation of Strategic Diagnostics Inc., in particular, Craig Kostyshyn,
Tim Lawruk, Chris Jones, and Penny Kosinski, who performed the analyses during the demonstration.
For more information on the PCB Field Analytical Technology Demonstration, contact:
Eric Koglin
Project Technical Leader
Environmental Protection Agency
Characterization and Research Division
National Exposure Research Laboratory
P.O. Box 93478
Las Vegas, Nevada 89193-3478
(702) 798-2432
For more information on the RaPID Assay System, contact:
Tim Lawruk
Strategic Diagnostics Inc.
Ill Pencader Drive
Newark, Delaware 19702-3322
(302)456-6789
xxii
-------
Section 1
Introduction
The performance evaluation of innovative and alternative environmental technologies is an integral part of
the U.S. Environmental Protection Agency's (EPA's) mission. Early efforts focused on evaluating
technologies that supported the implementation of the Clean Air and Clean Water Acts. In 1987, the Agency
began to evaluate the cost and performance of remediation and monitoring technologies under the Superfund
Innovative Technology Evaluation (SITE) program. This was in response to the mandate in the Superfund
Amendments and Reauthorization Act of 1986. In 1990, the U.S. Technology Policy was announced. This
policy placed a renewed emphasis on "making the best use of technology in achieving the national goals of
improved quality of life for all Americans, continued economic growth, and national security." In the spirit
of the Technology Policy, the Agency began to direct a portion of its resources toward the promotion,
recognition, acceptance, and use of U.S.-developed innovative environmental technologies both domestically
and abroad.
The Environmental Technology Verification (ETV) Program was created by the Agency to facilitate the
deployment of innovative technologies through performance verification and information dissemination. The
goal of the ETV Program is to further environmental protection by substantially accelerating the acceptance
and use of improved and cost-effective technologies. The ETV Program is intended to assist and inform
those involved in the design, distribution, permitting, and purchase of environmental technologies. The ETV
Program capitalizes upon and applies the lessons that were learned in the implementation of the SITE
Program to the verification of twelve categories of environmental technology: Drinking Water Systems,
Pollution Prevention/Waste Treatment, Pollution Prevention/ Innovative Coatings and Coatings Equipment,
Indoor Air Products, Air Pollution Control, Advanced Monitoring Systems, EvTEC (an independent,
private-sector approach), Wet Weather Flow Technologies, Pollution Prevention/Metal Finishing, Source
Water Protection Technologies, Site Characterization and Monitoring Technology [also referred to as the
Consortium for Site Characterization Technology (CSCT)], and Climate Change Technologies. The
performance verification contained in this report was based on the data collected during a demonstration of
polychlorinated biphenyl (PCB) field analytical technologies. The demonstration was administered by
CSCT.
For each pilot, EPA utilizes the expertise of partner "verification organizations" to design efficient
procedures for conducting performance tests of environmental technologies. To date, EPA has partnered
with federal laboratories and state, university, and private sector entities. Verification organizations oversee
and report verification activities based on testing and quality assurance protocols developed with input from
all major stakeholder/customer groups associated with the technology area.
In July 1997, CSCT, in cooperation with the U.S. Department of Energy's (DOE's) Environmental
Management (EM) Program, conducted a demonstration to verify the performance of six field analytical
technologies for PCBs: the L2000 PCB/Chloride Analyzer (Dexsil Corporation), the PCB Immunoassay Kit
1
-------
(Hach Company), the 4100 Vapor Detector (Electronic Sensor Technology), and three immunoassay kits
from Strategic Diagnostics Inc. (SDI): D TECH, EnviroGard, and RaPID Assay System. This
environmental technology verification report (ETVR) presents the results of the demonstration study for one
PCB field analytical technology, SDFs RaPID Assay System. Separate ETVRs have been published for the
other five technologies.
Technology Verification Process
The technology verification process is intended to serve as a template for conducting technology
demonstrations that will generate high-quality data that EPA can use to verify technology performance.
Four key steps are inherent in the process:
Needs identification and technology selection
• Demonstration planning and implementation
Report preparation
• Information distribution
Needs Identification and Technology Selection
The first aspect of the technology verification process is to determine technology needs of EPA and the
regulated community. EPA, DOE, the U.S. Department of Defense, industry, and state agencies are asked to
identify technology needs and interest in a technology. Once a technology need is established, a search is
conducted to identify suitable technologies that will address this need. The technology search and
identification process consists of reviewing responses to Commerce Business Daily announcements,
searches of industry and trade publications, attendance at related conferences, and leads from technology
developers. Characterization and monitoring technologies are evaluated against the following criteria:
• meets user needs;
• may be used in the field or in a mobile laboratory;
• is applicable to a variety of environmentally impacted sites;
• has high potential for resolving problems for which current methods are unsatisfactory;
• is cost competitive with current methods;
• performs better than current methods in areas such as data quality, sample preparation, or
analytical turnaround time;
-------
• uses techniques that are easier and safer than current methods; and
• is a commercially available, field-ready technology.
Demonstration Planning and Implementation
After a technology has been selected, EPA, the verification organization, and the developer agree to the
responsibilities for conducting the demonstration and evaluating the technology. The following tasks are
undertaken at this time:
• identifying demonstration sites that will provide the appropriate physical or chemical
environment, including contaminated media;
• identifying and defining the roles of demonstration participants, observers, and reviewers;
• determining logistical and support requirements (for example, field equipment, power and
water sources, mobile laboratory, communications network);
• arranging analytical and sampling support; and
• preparing and implementing a demonstration plan that addresses the experimental design,
sampling design, quality assurance/quality control (QA/QC), health and safety
considerations, scheduling of field and laboratory operations, data analysis procedures, and
reporting requirements.
Report Preparation
Innovative technologies are evaluated independently and, when possible, against conventional technologies.
The field technologies are operated by the developers in the presence of independent technology observers.
The technology observers are provided by EPA or a third-party group. Demonstration data are used to
evaluate the capabilities, limitations, and field applications of each technology. Following the
demonstration, all raw and reduced data used to evaluate each technology are compiled into a technology
evaluation report, which is mandated by EPA as a record of the demonstration. A data summary and
detailed evaluation of each technology are published in an ETVR.
Information Distribution
The goal of the information distribution strategy is to ensure that ETVRs are readily available to interested
parties through traditional data distribution pathways, such as printed documents. Documents are also
available on the World Wide Web through the ETV Web site (http://www.epa.gov/etv) and through a Web
site supported by the EPA Office of Solid Waste and Emergency Response's Technology Innovation Office
(http://CL U-in. com).
-------
Demonstration Purpose
The purpose of this demonstration was to obtain performance information for PCB field analytical
technologies, to compare the results with conventional fixed-laboratory results, and to provide supplemental
information (e.g., cost, sample throughput, and training requirements) regarding the operation of the
technology. The demonstration was conducted under two climatic conditions. One set of activities was
conducted outdoors, with naturally fluctuating temperatures and relative humidity conditions. A second set
was conducted in a controlled environmental facility, with lower, relatively stable temperatures and relative
humidities. Multiple soil types, collected from sites in Ohio, Kentucky, and Tennessee, were used in this
study. PCB soil concentrations ranged from approximately 0.1 to 700 parts per million (ppm). Developers
also analyzed 24 solutions of known PCB concentration that were used to simulate extracted wipe samples.
The extract samples ranged in concentration from 0 to 100 (jg/mL.
-------
Section 2
Technology Description
Objective
The objective of this section is to describe the technology being demonstrated, including the operating
principles underlying the technology and the overall approach to its use. The information provided here is
excerpted from that provided by the developer. Performance characteristics described in this section are
specified by the developer, which may or may not be substantiated by the data presented in Section 5.
Principle
The RaPID Assay System for PCB analysis (formerly the Ohmicron RaPID Assay System) applies the
principles of enzyme-linked immunosorbent assay (ELISA) to the determination of PCB [1]. In such an
assay, an enzyme has been chemically linked to a PCB molecule or PCB analog to create a labeled PCB
reagent. The labeled PCB reagent (called a conjugate) is mixed with an extract of native sample containing
the PCB contaminant. A portion of the mixture is applied to a surface to which an antibody specific for
PCBs has been affixed. The native PCB and PCB-enzyme conjugate compete for a limited number of
antibody sites. After a period of time, the solution is washed away, and what remains is either PCB-antibody
complexes or enzyme-PCB-antibody complexes attached to the test surface. The proportion of the two
complexes on the test surface is determined by the amount of native PCB in the original sample. The enzyme
present on the test surface is used to catalyze a color change reaction in a solution added to the test surface.
Because the amount of enzyme present is inversely proportional to the concentration of native PCB
contaminant, the amount of color development is inversely proportional to the concentration of PCB
contaminant.
In the case of the RaPID Assay System, the sample to be tested is added, along with an enzyme conjugate, to
a disposable test tube, followed by paramagnetic particles with antibodies specific to PCBs attached. Both
the PCBs (which may be in the sample) and the enzyme labeled PCB analog (the enzyme conjugate)
compete for antibody binding sites on the magnetic particles. At the end of an incubation period, a magnetic
field is applied to hold the paramagnetic particles (with PCBs and labeled PCB analog bound to the
antibodies on the particles in proportion to their original concentration) in the tube and allow the unbound
reagents to be decanted. After decanting, the particles are washed with washing solution.
The presence of PCBs is detected by adding the enzyme substrate (hydrogen peroxide) and the chromogen
(3,3',5,5'-tetramethylbenzidine). The enzyme-labeled PCB analog that is bound to the PCB antibody
catalyzes the conversion of the substrate/chromogen mixture to a colored product. After an incubation
period, the reaction is stopped and stabilized by the addition of acid. Since the labeled PCBs (conjugate)
were in competition with the unlabeled PCBs (sample) for the antibody sites, the color developed is inversely
proportional to the concentration of PCBs in the sample. The color developed is quantified with a small,
handheld photometer. Thus, the RaPID Assay System consists of three primary components, the kit used for
-------
sample preparation (the RaPID Prep Soil Collection Kit), the assay kit itself, and the RPA-I RaPID
Analyzer, which is the small photometer.
Sample Preparation
The RaPID Prep Soil Collection Kit is designed for collection, extraction, and subsequent filtration of soil
samples before analysis of environmental contaminants. The reagents contained in the RaPID Prep PCB
Sample Extraction Kit have been optimized for fast, efficient removal of PCB from soil and convenient
preparation of the sample for immunoassay at levels of interest to the investigator. The system allows for
reliable, convenient, and cost-effective determinations at the field testing or remediation site. This procedure
was validated using 10 g soil samples. Soil sampling should be conducted in a uniform and consistent
manner according to a plan appropriate for the site and the objectives of the study.
Description of Kit Contents
The following items are contained in the RaPID Assay Sample Preparation Kit:
• soil collection device with detachable plunger and screw cap
• filter caps
• extract collection vials
• chain-of-custody container labels
• portable Styrofoam tube holder
• PCB extraction solution (methanol with dispersion agent)
• PCB extract diluent (buffered saline solution containing preservatives and stabilizers
without detectable PCBs)
25-(iL precision pipet
Pipet tips
The components of the kit should be stored at 2 to 30 °C, and reagents should not be used after the
expiration date.
Materials Not Provided in Kit
In addition to materials provided, the following items will be necessary for the preparation of a soil sample:
• stopwatch
• permanent marking pen
-------
• protective gloves
• digital balance
Procedure
Sample Weighing
The soil collector (shown in Figure 2-1), with its plunger fully depressed (pushed to the top), will
reproducibly collect a volume of soil. The weight of this volume will vary depending on soil type. Sand,
clay, and loam soils collected with the device in this way will weigh approximately 12 g. The same volume
of organic soil weighs much less but can be reproducibly collected. If a site is undergoing a preliminary
screen for high levels of PCBs, the volume collection method could be used with an estimated average
sample weight appropriate for that site.
Weighing the soil before extraction is recommended for soils with high organic content and for combinations
of soil types. The following method was used during the demonstration: Remove the screw cap and plunger
rod from an empty collection tube. Position the plunger at the bottom of the collection tube. Attach the red
base piece provided, and place the tube in an upright position on the balance and tare weight. Weigh 10 ±0.1
g of soil into the tube. Record the soil weight.
Extraction
Position the soil collection tube containing a soil sample upright in the Styrofoam rack, and add one vial (20
mL) of the PCB extraction solution. Screw the cap (without filter) on tightly, and make sure that the luer
cap is secured. Shake vigorously and continuously for at least 60 s. Additional shaking may be required to
break up large or dry soil aggregates. Position the collection tube upright in the rack, and allow the mixture
1
BOTH
e=
1M -
Soil Collector
i
f *
™™™^^
V
TOP
4
TOP _ Screw Cap
Luer Cap
Plunger Rod Plunger
Luer Cone
Figure 2-1. Soil sample collection device.
-------
to sit at least 5 min. Longer extraction times may be desirable for some situations. If batch processing is
desired, up to 21 soil samples with added extraction solution can be loaded into the rack inside the kit box
base; the box lid is put in place, and the box is shaken vigorously for at least 60 s.
Filtration
Remove the screw cap and attach the filter cap. Hand tighten until resistance is felt. Attach the plunger rod
to the plunger of the soil collector. Invert the soil collector so that the luer cone is positioned over a
collection vial. Keep the collector inverted for a few seconds to wet the filter and to allow the filtrate to drip
through the filter into the luer cone. Apply slight pressure to the plunger handle. The filtrate will begin to
flow more quickly as gentle pressure is applied continuously. Fill the vial with approximately 10 to 20 drops
(0.5 to 1 mL). Cap the vial. This amount of filtrate is sufficient to perform multiple replicate analyses with
the RaPID Assay Kit. The vial will hold up to 5 mL of filtrate if additional extract volume is desired. The
filtrate is stable when stored in the extract collection vial for one week at room temperature (15 to 30°C).
Dilution
Using the pipet provided, transfer 25 (A of the extract directly into a vial of PCB extract diluent (25 mL).
Mix by inverting the vial several times.
NOTE: This is the step where the solvent extract analysis begins.
Assay Procedure
Reagents
The following are the reagents provided with the RaPID Assay System for PCB analysis:
PCB antibody coupled paramagnetic particles—PCB antibody (rabbit anti-PCB)
covalently bound to paramagnetic particles, which are suspended in buffered saline
containing preservative and stabilizers.
PCB enzyme conjugate—horseradish peroxidase-labeled PCB analog diluted in buffered
saline containing preservative and stabilizers.
• PCB standards—three PCB standard solutions standards in buffered saline containing
preservative and stabilizers. The concentrations of the standards are 0.25, 1.0, and 5.0 ppb
as Aroclor 1254. Each vial contains 2.0 mL.
• Control—a PCB solution in buffered saline containing preservative and stabilizers. The
concentration is approximately 3 ppb as Aroclor 1254. A 2.0-mL volume is supplied in
one vial.
• Diluent/zero standard—buffered saline containing preservative and stabilizers without any
detectable PCBs.
Color solution—solution of hydrogen peroxide and 3, 3', 5, 5'-tetramethylbenzidine in an
organic base.
-------
• Stopping solution—solution of sulfuric acid (0.5%).
• Washing solution—preserved deionized water.
• Test tubes—polystyrene tubes (36).
Store all reagents at 2 to 8 °C. Do not freeze. Reagents may be used until the expiration date on the box. The
test tubes require no special storage conditions and may be stored separately from the reagents. All reagents
must be allowed to come to room temperature, and the antibody-coupled paramagnetic particles should be
mixed thoroughly before use.
Materials Required but Not Provided in Kit
In addition to the reagents provided, the following items are essential for performance of the test:
Pipets—precision pipets capable of delivering 200, 250, and 500 ^L and a 1.0-mL
repeating pipet.
• Vortex mixer—Thermolyne Maxi Mix, Scientific Industries Vortex Genie, or equivalent.
• Magnetic separation rack.
• RPA-I RaPID analyzer—or equivalent photometer capable of readings at 450 nm.
Procedural Notes and Precautions
As with all immunoassay methods, a consistent technique is the key to optimal
performance. To obtain the greatest precision, be sure to treat each tube in an identical
manner.
Add reagents directly to the bottom of the tube while avoiding contact between the reagents
and the pipet tip. This will help ensure consistent quantities of reagent in the test mixture.
• Avoid cross-contaminations and carryover of reagents by using clean pipets for each
sample addition and by avoiding contact between reagent droplets on the tubes and pipet
tips.
• Avoid foam formation during vortexing.
• The magnetic separation rack consists of two parts: an upper rack that will securely hold
the test tubes and a lower separator that contains the magnets used to attract the antibody-
coupled paramagnetic particles. During incubations the upper rack is removed from the
lower separator so that the paramagnetic particles remain suspended during the incubation.
For separation steps, the rack and the separator are combined to pull the paramagnetic
particles to the sides of the tubes.
-------
• To obtain optimum assay precision, it is important to perform the separation steps carefully
and consistently. Decant the rack by slowly inverting it away from the operator using a
smooth turning action so that the liquid flows consistently along only one side of the test
tube. While the rack is still inverted, place it on an absorbent pad and allow it to drain.
Lifting the rack and replacing it gently onto the pad several times will ensure complete
removal of the liquid from the rim of the tube (this technique is demonstrated on a training
video available from SDI).
• Mix the antibody-coupled paramagnetic particles immediately before pipetting.
• Standard and control vials should remain capped when not in use to prevent evaporation.
• Do not use any reagents beyond their stated shelf life.
• Avoid contact of the stopping solution (sulfuric acid) with skin and mucous membranes. If
this reagent comes in contact with skin, wash with water.
• A control solution with approximately 3 ppb of PCBs (as Aroclor 1254) is provided with
the RaPID Assay PCB Kit. It is recommended that this solution be included in every run
and be treated in the same manner as unknown samples. Acceptable limits should be
established by each laboratory.
Method
Label test tubes for standards, control, and samples according to Table 2-1. Add 200 uL of the appropriate
standard, control, or sample to each tube. Add 250 uL of PCB enzyme conjugate to each tube. Mix the PCB
antibody-coupled paramagnetic particles thoroughly, and add 500 uL to each tube. Vortex for 1 to 2 s,
minimizing foaming. Incubate for 15 min at room temperature. Separate in the magnetic separation rack for
2 min. Decant and gently blot all tubes briefly in a consistent manner. Add 1 mL of washing solution to each
tube, and vortex tubes for 1 to 2 s. Return tubes and allow to remain in the magnetic separation unit for 2
min. Decant and gently blot all tubes briefly in a consistent manner. Repeat steps 10 and 11 one additional
time. Remove the rack from the separator, and add 500 uL of color solution to each tube. Vortex for 1 to 2
s, minimizing foaming. Incubate for 20 min at room temperature. Add 500 uL of stopping solution to each
tube. Add 1 mL of washing solution to a clean test tube to use as a blank. Within 15 min after adding the
stopping solution, read the results using the RPA-I RaPID analyzer set at 450 nm.
Table 2-1. Tube labels
Tube Label
1,2
Contents of Tube
Diluent/Zero Standard, 0 ppb
10
-------
3,4
5,6
7,8
9
10
11
12
Standard 1, 0.25 ppb
Standard 2, 1.0 ppb
Standards, 5.0 ppb
Control
Sample 1
Sample 2
Sample 3
Results
Manual Calculations
1.
Calculate the mean absorbance value for each of the standards.
2. Calculate the %B/Bo for each standard by dividing the mean absorbance value for the standard by
the mean absorbance value for the diluent/zero standard.
3. Construct a standard curve by plotting the %B/Bo for each standard on vertical logit (Y) axis vs the
corresponding PCB concentration on horizontal logarithmic (X) axis on the graph paper provided.
4. %B/Bo for controls and samples will then yield results in parts per billion (ppb) of PCBs by
interpolation using the standard curve.
RPA-I RaPID Analyzer
Using the RPA-I RaPID analyzer, calibration curves can be automatically calculated and stored. Refer to
the RPA-I operating manual for detailed instructions. To obtain results from the RaPID Assay on the RPA-
I, the recommended parameter settings are shown in Table 2-2. The following is a summary of the
performance characteristics of the RPA-I RaPID:
Recovery: PCB recoveries will vary depending on soil type, retention mechanism, solvent and extraction
apparatus used, length of extraction period, amount of agitation, and levels of potentially interfering
substances in the soil.
11
-------
Table 2-2. Recommended parameter settings for the RPA-I RaPID analyzer
Parameter
Data Reduct
Xformation
Read Mode
Wavelength
Units
# Reagent Blk
# of Cals
# of Reps
Calibration Concentrations
Range
Correlation
Replicate %CV
Recommended Setting
Linear Regression
Ln/LogitB
Absorbance
450 nm
ppb
0
4
2
0.00 ppb
0.25 ppb
1.00 ppb
5. 00 ppb
0.10 -5. 00 ppb
> 0.990
< 10%
Sensitivity: The RaPID Assay System for PCB analysis has an estimated minimum detectable
concentration, based on a 90% B/Bo of 100 ppt. Refer to appropriate application notes or procedures for
sensitivity in specific sample matrices.
Specificity: The cross-reactivity of the RaPID Assay System for PCB analysis for various Aroclors, as
shown in Table 2-3, can be expressed as the least detectable dose, which is estimated at 90% B/Bo, or as the
dose required to produce a 50% B/Bo response. The following compounds demonstrated no reactivity in the
RaPID Assay System for PCB analysis at concentrations up to 10,000 ppb: biphenyl, 2,5-dichlorophenol,
2,3,5-trichlorophenol, and di-n-octyl-phthalate.
Limitations
The RaPID Assay System for PCB analysis will detect PCBs to different degrees. Refer to specificity table
for data on various Aroclors and congeners. The PCB RaPID Assay Kit provides screening results. As with
any analytical technique (gas chromatography, high-pressure liquid chromatography, etc.), positive results
requiring some action should be confirmed by an alternate method.
12
-------
Table 2-3. Cross-reactivity of the RaPID Assay System for PCB analysis
Compound
Aroclor 1254
Aroclor 1260
Aroclor 1248
Aroclor 1242
Aroclor 1262
Aroclor 1232
Aroclor 1268
Aroclor 1016
Aroclor 1221
Least Detectable Dose (ppb) a
[90% B/Bo]
0.10
0.10
0.11
0.17
0.18
0.42
0.46
0.47
6.77
Dose Required To Produce 50%
B/Bo Response (ppb)
1.80
1.15
2.11
4.40
2.37
9.38
10.9
12.8
81.3
a Concentration in the extract or calibration standard, not in the soil.
The total time required for pipetting the magnetic particles should be kept to 2 min or less; therefore, the
total number of tubes that can be assayed in a run should be adjusted accordingly.
13
-------
Section 3
Site Description and Demonstration Design
Objective
This section describes the demonstration site, the experimental design for the verification test, and the
sampling plan (sample types analyzed and the collection and preparation strategies). Included in this section
are the results from the predemonstration study and a description of the deviations made from the original
demonstration design.
Demonstration Site Description
Site Name and Location
The demonstration of PCB field analytical technologies was conducted at Oak Ridge National Laboratory
(ORNL) in Oak Ridge, Tennessee. PCB-contaminated soils from three DOE sites (Oak Ridge; Paducah,
Kentucky; and Piketon, Ohio) were used in this demonstration. The soil samples used in this study were
brought to the demonstration testing location for evaluation of the field analytical technologies.
Site History
Oak Ridge is located in the Tennessee River Valley, 25 miles northwest of Knoxville. Three DOE facilities
are located in Oak Ridge: ORNL, the Oak Ridge Y-12 Plant, and East Tennessee Technology Park (ETTP).
Chemical processing and warhead component production have occurred at the Y-12 Plant, and ETTP is a
former gaseous diffusion uranium enrichment plant. At both facilities, industrial processing associated with
nuclear weapons production has resulted in the production of millions of kilograms of PCB-contaminated
soils. Two other DOE facilities—the Paducah plant in Paducah, Kentucky, and the Portsmouth plant in
Piketon, Ohio—are also gaseous diffusion facilities with a history of PCB contamination. During the
remediation of the PCB-contaminated areas at the three DOE sites, soils were excavated from the ground
where the PCB contamination occurred, packaged in containers ranging in size from 55-gal to 110-gal
drums, and stored as PCB waste. Samples from these repositories—referred to as "Oak Ridge,"
"Portsmouth," and "Paducah" samples in this report—were used in this demonstration.
In Oak Ridge, excavation activities occurred between 1991 and 1995. The Oak Ridge samples were
comprised of PCB-contaminated soils from both Y-12 and ETTP. Five different sources of PCB
contamination resulted in soil excavations from various dikes, drainage ditches, and catch basins. Some of
the soils are EPA-listed hazardous waste due to the presence of other contaminants (e.g., diesel fuels).
A population of over 5000 drums containing PCB-contaminated soils was generated from 1986 to 1987
during the remediation of the East Drainage Ditch at the Portsmouth Gaseous Diffusion Plant. The ditch
was reported to have three primary sources of potential contamination: (1) treated effluent from a
radioactive liquid treatment facility, (2) runoff from a biodegradation plot where waste oil and sludge were
disposed of, and (3) storm sewer discharges. In addition, waste oil was reportedly used for weed control in
15
-------
the ditch. Aside from PCB contamination, no other major hazardous contaminants were detected in these
soils. Therefore, no EPA hazardous waste codes are assigned to this waste.
Twenty-nine drums of PCB-contaminated soils from the Paducah plant were generated as part of a spill
cleanup activity at an organic waste storage area (C-746-R). The waste is considered a listed hazardous
waste for spent solvents (EPA hazardous waste code F001) because it is known to contain trichloroethylene.
Other volatile organic compounds, such as xylene, dichlorobenzene, and cresol, were also detected in the
preliminary analyses of some of the Paducah samples.
Site Characteristics
PCB-contaminated environmental soil samples from Oak Ridge, Portsmouth, and Paducah were collected
from waste containers at storage repositories at ETTP and Paducah. Many of the soils contained interfering
compounds such as oils, fuels, and other chlorinated compounds (e.g., trichloroethylene). Specific sample
descriptions of the environmental soils used in this demonstration are given in Appendix A. In addition, each
sample was characterized in terms of its soil composition, pH, and total organic carbon content. Those
results are summarized in Appendix B.
Field demonstration activities occurred at two sites at ORNL: a natural outdoor environment (the outdoor
site) and inside a controlled environmental atmosphere chamber (the chamber site). Figure 3-1 shows a
schematic map of a section of ORNL indicating the demonstration area where the outdoor field activities
occurred. Generally, the average summer temperature in eastern Tennessee is 75.6°F, with July and August
temperatures averaging 79.1 °F and 76.8 °F, respectively. Average temperatures during the testing periods
ranged from 79 to 85 °F, as shown in Appendix C. Studies were also conducted inside a controlled
environmental atmosphere chamber, hereafter referred to as the "chamber," located in Building 5507 at
ORNL. Demonstration studies inside the chamber were used to evaluate performance under environmental
conditions that were markedly different from the ambient outdoor conditions at the time of the test. Average
temperatures in the chamber during the testing periods ranged from 55 to 70 °F. The controlled experimental
atmosphere facility consists of a room-size walk-in chamber 10 ft wide and 12 ft long with air processing
equipment to control temperature and humidity. The chamber is equipped with an environmental control
system, including reverse osmosis water purification that supplies the chamber humidity control system.
Fiigh efficiency particulate air and activated charcoal filters are installed for recirculation and building
exhaust filtration.
Experimental Design
The analytical challenge with PCB analysis is to quantify a complex mixture that may or may not resemble
the original commercial product (i.e., Aroclor) due to environmental aging, and to report the result as a
single number [2]. The primary objective of the verification test was to compare the performance of the field
technology to laboratory-based measurements. Often, verification tests involve a direct one-to-one
comparison of results from field-acquired samples. However, because sample heterogeneity can preclude
replicate field or laboratory comparison, accuracy and precision data must often be derived from the
analysis of QC and performance evaluation (PE) samples. In this study, replicates of all three sample types
(QC, PE, and environmental soil) were analyzed. The ability to use
16
-------
Figure 3-1. Schematic map of ORNL, indicating the demonstration area.
environmental soils in the verification test was made possible because the samples, collected from drums
containing PCB-contaminated soils, could be thoroughly homogenized and characterized prior to the
demonstration. This facet of the design, allowing additional precision data to be obtained on actual field-
acquired samples, provided an added performance factor in the verification test.
Another objective of this demonstration was to evaluate the field technology's capability to support
regulatory compliance decisions. For field methods to be used in these decisions, the technology must be
capable of informing the user, with known precision and accuracy, that soil concentrations are greater than
or less than 50 ppm, and that wipe samples are greater than or less than 100 (jg/100 cm2 [3]. The samples
selected for analysis in the demonstration study were chosen with this objective in mind.
The experimental design is summarized in Table 3-1. This design was approved by all participants prior to
the start of the demonstration study. In total, the developers analyzed 208 soil samples, 104 each at both
locations (outdoors and chamber). The 104 soil samples comprised 68 environmental samples (17 unique
environmental samples prepared in quadruplicate) ranging in PCB concentration from 0.1 to 700 ppm and
36 PE soils (9 unique PE samples in quadruplicate) ranging in PCB concentration from 0 to 50 ppm. To
determine the impact of different environmental conditions on the technology's performance, each batch of
104 samples contained five sets of quadruplicate soil samples from DOE's Paducah site. These were
analyzed under both sets of environmental conditions (i.e., outdoor and chamber conditions). For the
developers participating in the extract sample portion (i.e., simulated wipe samples) of the
17
-------
Table 3-1. Summary of experimental design by sample type
Concentration
Range
Sample ID a
Outdoor Site
Chamber Site
Total #
Samples
Analyzed
PE Materials
0
2.0 ppm
2.0 ppm
5.0 ppm
10. 9 ppm
20.0 ppm
49. 8 ppm
50.0 ppm
50.0 ppm
126
118
124
120
122
119
125
121
123
226
218
224
220
222
219
225
221
223
8
8
8
8
8
8
8
8
8
Environmental Soils
0.1-2.0 ppm
2. 1-20.0 ppm
20. 1-50.0 ppm
50. 1-700.0 ppm
101,107,108,109,113,114
102,103,104,115
111,116
105,106,110,112,117
201,202,206
203,207,212,213
204,208,209,214,215
205,210,211,216,217
36
32
28
40
Extracts
0
10 |ig/mL
100 ng/mL
Grand Total
129b/132c
127/130
128/131
116
229/232
227/230
228/231
116
8
8
8
232 d
aEach sample ID was analyzed in quadruplicate.
b Extract prepared in iso-octane for Dexsil and the reference laboratory.
c Extract prepared in methanol for Electronic Sensor Technology, SDI, and the reference laboratory.
d All samples were analyzed in random order.
18
-------
demonstration, 12 extracts, ranging in concentration from 0 to 100 (jg/mL, were analyzed in each location
(chamber and outdoors). All samples were analyzed without prior knowledge of sample type or
concentration and were analyzed in a randomized order that was unique for each developer.
Environmental Conditions during Demonstration
As mentioned above, field activities were conducted both outdoors under natural environmental conditions
and indoors in a controlled environmental atmosphere chamber to evaluate the effect of environmental
conditions on technology performance. The weather outside was relatively uncomfortable during the July
demonstration, with highs approaching 100°F and 90% relative humidity (RH). Daily average temperatures
were around 85 °F with 70% RH. While outside, the developers set up canopies to provide shade and
protection from frequent late afternoon thundershowers.
In the indoor chamber tests, conditions were initially set to 55 °F and 25% RH. An independent check of the
conditions inside the chamber revealed that the temperature was closer to 68 °F with a 38% RH on the first
day of testing. A maintenance crew was called in to address the inconsistencies between the set and actual
conditions. By the middle of the third day of testing, the chamber was operating properly at 55 °F and 50%
RH.
Appendix C contains a summary of the environmental conditions (temperature and relative humidity) during
the demonstration. The SDI team performed analyses with the RaPID Assay Kit outdoors on July 25 and in
the chamber on July 22.
Sample Descriptions
PCBs (C12H10.XC1X) are a class of compounds that are chlorine-substituted linked benzene rings. There are
209 possible PCB compounds (also known as congeners). PCBs were commercially produced as complex
mixtures beginning in 1929 for use in transformers, capacitors, paints, pesticides, and inks [2]. Monsanto
Corporation marketed products that were mixtures of 20 to 60 PCB congeners under the trade name
Aroclor. Aroclor mixtures are identified by a number (e.g., Aroclor 1260) that represents the mixture's
chlorine composition as a percentage (e.g., 60%).
Performance Evaluation Materials
Samples of Tennessee reference soil [4] served as the blanks. Preprepared certified PE samples were
obtained from Environmental Resource Associates (ERA) of Arvada, Colorado, and the Analytical
Operations and Data Quality Center of EPA's Office of Solid Waste and Emergency Response. The soils
purchased from ERA had been prepared using ERA's semivolatile blank soil matrix. This matrix was a
topsoil that had been dried, sieved, and homogenized. Particle size was approximately 60 mesh. The soil
was approximately 40% clay. The samples acquired from EPA's Analytical Operations and Data Quality
Center had been prepared using contaminated soils from various sites around the country in the following
manner: The original soils had been homogenized and diluted with a synthetic soil matrix (SSM). The SSM
had a known matrix of 6% gravel, 31% sand, and 43% silt/clay; the remaining 20% was topsoil. The
dilution of the original soils was performed by mixing known amounts of contaminated soil with the SSM in
a blender for no less than 12 h. The samples were also spiked with target pesticides (a, P, A, and 6-BHC,
methoxychlor, and endrin ketone) to introduce some compounds that were likely to be present in an actual
environmental soil. The hydrocarbon background from the original sample and the spiked pesticides
19
-------
produced a challenging matrix. The PE soils required no additional preparation by ORNL and were split for
the developer and reference laboratory analyses as received.
Environmental Soil Samples
As noted in the site description above, PCB-contaminated environmental soil samples from Oak Ridge,
Portsmouth, and Paducah were used in this demonstration. The soils were contaminated with PCBs as the
result of spills and industrial processing activities at the various DOE facilities. Originally, the contaminated
soils were excavated from dikes, drainage ditches, catch basins, and organic waste storage areas. The
excavated soils were then packaged into waste containers and stored at the repositories in ETTP and
Paducah in anticipation of disposal by incineration. The environmental soil samples used in this study were
collected from these waste containers. Many of the soils contained interfering compounds such as oils, fuels,
and other chlorinated compounds, while some contained multiple Aroclors. For more information on
sampling locations and sample characteristics (soil composition, pH, and total organic carbon content), refer
to Appendices A and B, respectively.
Extract Samples
Traditionally, the amount of PCBs on a contaminated surface is determined by wiping the surface with a
cotton pad saturated with hexane. The pad is then taken to the laboratory, extracted with additional hexane,
and analyzed by gas chromatography. Unlike soil samples, which can be more readily
homogenized and divided, equivalent wipe samples (i.e., contaminated surfaces or post-wipe pads) were not
easily obtainable. Therefore, interference-free solutions of PCBs were analyzed to simulate an extracted
surface wipe pad. Extract sample analyses provided evaluation data that relied primarily on the technology's
performance rather than on elements critical to the entire method (i.e., sample collection and preparation).
Because different developers required the extract samples prepared in different solvents (e.g., methanol and
iso-octane), the reference laboratory analyzed sets of extracts in both solvents. SDI analyzed extracts
prepared in methanol. A total of 12 extracts were analyzed per site; these consisted of four replicates each of
a blank and two concentration levels (10 and 100 (jg/mL).
Sampling Plan
Sample Collection
Environmental soil samples were collected from April 17 through May 7, 1997. Portsmouth and Oak Ridge
Reservation soils were collected from either storage boxes or 55-gal drums stored at ETTP. Briefly, the
following procedure was used to collect the soil samples. Approximately 30 Ib of soil were collected from
the top of the drum or B-25 box using a scoop and placed in a plastic bag. The soil was sifted to remove
rocks and other large debris, then poured into a plastic-lined 5-gal container. All samples were subjected to
radiological screening and were determined to be nonradioactive. Similarly, soil samples were collected
from 55-gal drums stored at Paducah and shipped to ORNL in lined 5-gal containers.
Sample Preparation, Labeling, and Distribution
Aliquots of several of the environmental soils were analyzed and determined to be heterogeneous in PCB
concentration. Because this is unsatisfactory for accurately comparing the performance of the field
technology with the laboratory-based method, the environmental soils had to be homogenized prior to
sample distribution. Each Portsmouth and Oak Ridge environmental soil sample was homogenized by first
placing approximately 1500 g of soil in a glass Pyrex dish. The dish was then placed in a large oven set at
20
-------
35 °C, with the exhaust and blower fans turned on to circulate the air. After drying overnight, the soil was
pulverized using a conventional blender and was sieved using a 9-mesh screen (2-mm particle size). Last,
the soil was thoroughly mixed using a spatula. A comparison of dried and undried soils showed that a
minimal amount of PCBs (< 20%) was lost due to sample drying, making this procedure suitable for use in
the preparation of the soil samples. The Paducah samples, because of their sandy characteristics, only
required the sieving and mixing preparation steps. Extract sample preparation involved making solutions of
PCBs in methanol and iso-octane at two concentration levels (10 and 100 (jg/mL). Multiple aliquots of each
sample were analyzed using the analytical procedure described below to confirm the homogeneity of the
samples with respect to PCB concentration.
To provide the developers with soils contaminated at higher concentrations of PCBs, some of the
environmental soils (those labeled with an "S" in Appendix B) were spiked with additional PCBs. Spiked
soils samples were prepared after the soil was first dried in a 35 °C oven overnight. The dry soil was ground
using a conventional blender and sieved through a 9-mesh screen (2-mm particle size). Approximately 1500
g of the sieved soil were spiked with a diethyl ether solution of PCBs at the desired concentration. The
fortified soil was agitated using a mechanical shaker and then allowed to air-dry in a laboratory hood
overnight. A minimum of four aliquots were analyzed using the analytical procedure described below to
confirm the homogeneity of the soil with regard to the PCB concentration.
The environmental soils were characterized at ORNL prior to the demonstration study. The procedure used
to confirm the homogeneity of the soil samples entailed the extraction of 3 to 5 g of soil in a mixture of
solvents (1 mL water, 4 mL methanol, and 5 mL hexane). After the soil/solvent mixture was agitated by a
mechanical shaker, the hexane layer was removed and an aliquot was diluted for analysis. The hexane
extract was analyzed on a Hewlett Packard 6890 gas chromatograph equipped with an electron capture
detector and autosampler. The method used was a slightly modified version of EPA's SW-846 dual-column
Method 8081 [5].
After analysis confirming homogeneity, the samples were split into jars for distribution. Each 4-oz sample
jar contained approximately 20 g of soil. Four replicate splits of each soil sample were prepared for each
developer. The samples were randomized in two fashions. First, the order in which the filled jars were
distributed was randomized, such that the same developer did not always receive the first jar filled for a
given sample set. Second, the order of analysis was randomized so that each developer analyzed the same set
of samples, but in a different order. The extract samples were split into 10-mL aliquots and placed into 2-oz
j ars. The extracts were stored in the refrigerator (at < 4 ° C) until released to the developers. Each sample j ar
had three labels: (1) developer order number; (2) sample identifier (ID) number; and (3) a PCB warning
label. The developer order number corresponded to the order in which the developer was required to analyze
the samples (e.g., SDI 1001 through SDI 1116). The sample ID number was in the format of "xxxyzz,"
where "xxx" was the three-digit sample ID (e.g., 101) listed in Table 3-1, "y" was the replicate (e.g., 1 to 4),
and "zz" was the aliquot order of each replicate (e.g., 01 to 11). For example, sample ID 101101
corresponded to sample ID "101" [an Oak Ridge soil from request for disposal (RFD) 40022, drum 02], "1"
corresponded to the first replicate from that sample, and "01" corresponded to the first jar filled in that
series.
21
-------
Once the samples were prepared, they were stored at a central sample distribution center. During the
demonstration study, developers were sent to the distribution center to pick up their samples. Samples were
distributed sequentially in batches of 12 to ensure that samples were analyzed in the order specified.
Completion of chain-of-custody forms and scanning of bar code labels documented sample transfer
activities. Some of the developers received information regarding the samples prior to analysis. SDI received
information pertaining to which Aroclors were in the samples and at what ratio if multiple Aroclors were
present. This was provided at the request of SDI to simulate the type of information that would be available
during actual field testing. The developers returned the unused portions of the samples with the analytical
results to the distribution center when testing was completed. The sample bar codes were scanned upon
return to document sample throughput time.
Three complete sets of extra samples, called archive samples, were available for distribution in case the
integrity of a sample was compromised. Very few (<5) archive samples were utilized over the course of the
demonstration.
Predemonstration Study
Ideally, environmental soil samples are sent to the developers prior to the demonstration study to allow them
the opportunity to analyze representative samples in advance of the verification test. This gives developers
the opportunity to refine and calibrate their technologies and revise their operating procedures on the basis
of the predemonstration study results. The predemonstration study results can also be used as an indication
that the selected technologies are of the appropriate level of maturity to participate in the demonstration
study.
According to ORNL regulations, however, one of two conditions must exist in order to ship environmental
soils that were once classified as mixed hazardous waste. First, the recipient—in this case, the developer's
facilities—must have proper Nuclear Regulatory Commission (NRC) licensing to receive and analyze
radiological materials. Second, the soils must be certified as entirely free of radioactivity, beyond the no-rad
certification issued from radiological screening tests based on ORNL standards. Because none of the
developers had proper NRC licensing and proving that the soils were entirely free of radioactivity was
prohibitive, spiked samples of Tennessee reference soil were used for the predemonstration study. The
developers had an opportunity to evaluate the Tennessee reference soils spiked with PCBs at concentrations
similar to what would be used in the demonstration study. The developers also analyzed two performance
evaluation samples and one solvent extract. The reference laboratory analyzed the same set of samples,
which included two extracts samples, prepared in the two solvents (methanol and iso-octane) requested by
the developers.
Predemonstration Sample Preparation
Two soil samples were prepared by ORNL using Tennessee reference soil [4]. The soil was a Captina silt
loam from Roane County, Tennessee, that was slightly acidic (pH ~5) and low in organic carbons (-1.5%).
The soil composition was 7.7% sand, 29.8% clay, and 62.5% silt. To prepare a spiked sample, the soil was
first ground either using a mortar and pestle or a conventional blender. The soil was then sieved through a
16-mesh screen (1 mm particle size). Approximately 500 g of the sieved soil was spiked with a diethyl ether
solution of PCBs at the desired concentration. The soil was agitated using a mechanical shaker, then allowed
to air-dry overnight in a laboratory hood. A minimum of five aliquots were analyzed by gas chromatography
22
-------
using electron capture detection. The PCB concentration of the spiked samples was determined to be
homogeneous. The remaining two soil samples used in the predemonstration study were performance
evaluation materials acquired from ERA and EPA (see the section "Performance Evaluation Materials"
above). In addition, a solvent extract was prepared by ORNL to simulate an extracted surface wipe sample.
The extracts were prepared in two different solvents (iso-octane and methanol) to accommodate developer
requests.
Predemonstration Results
The predemonstration samples were sent to the developers and the reference laboratory on June 2, 1997.
Predemonstration results were received by June 26, 1997. Table 3-2 summarizes the RaPID Assay's results
for the predemonstration samples. Results indicated that SDFs RaPID Assay System was ready for field
evaluation.
Table 3-2. Summary of the RaPID Assay's predemonstration results
Sample Description
2 ppm of Aroclor 1260
100ppm(total)of
Aroclors 1254 and 1260
1 1 ppm of Aroclor 1260
50 ppm of Aroclor 1254
5 ppm of Aroclor 1242
Matrix
Soil
Soil
Soil
Soil
Extract
Source
ORNL
ORNL
EPA
ERA
ORNL
RaPID Assay a
Result
(ppm)
1.2
66.3
4.2
66.0
4.4
Duplicate result
(ppm)
1.3
61.9
4.8
b
4.8
Reference Laboratory
Result
(ppm)
2.2
78.0
11.0
37.0
4.7
Duplicate result
(ppm)
2.3
89.0
9.5
b
4.9
1 Results were Aroclor-adjusted (see Section 2 for more details).
' Replicate was not analyzed because of lack of adequate sample for second analyses.
Deviations from the Demonstration Plan
A few deviations from the demonstration plan occurred. In Appendix B of the technology demonstration
plan [6], the reference laboratory's procedure states that no more than 10 samples will be analyzed with
each analytical batch (excluding blanks, standards, QC samples, and dilutions). The analytical batch is also
stated as 10 samples in the Quality Assurance Project Plan of the demonstration plan. The reference
laboratory actually analyzed 20 samples per analytical batch. Because a 20-sample batch is recommended
in SW-846 Method 8081, this deviation was deemed acceptable.
Table 5 of the demonstration plan [6] delineates the environmental soils according to concentration. The
classification was based on a preliminary analysis of the soils at ORNL. Table 3-1 of this report arranges
the concentrations as characterized by the reference laboratory. The reference laboratory determined that
five sample sets (sample IDs 102, 105, 110, 111, and 210) were in the next highest concentration range,
23
-------
differing from what was originally outlined in the demonstration plan. Also, the highest concentration
determined by the reference laboratory was 700 ppm, while the preliminary analysis at ORNL found the
highest concentration to be 500 ppm.
During the demonstration study, the SDI team did not note any deviations from the procedure described in
the technology demonstration plan [6] for the RaPID Assay System.
24
-------
Section 4
Reference Laboratory Analytical Results and Evaluation
Objective and Approach
The purpose of this section is to present the evaluation of the PCB data generated by the reference
laboratory. Evaluation of the results from the analysis of PE, environmental soil, and extract samples was
based on precision, accuracy, representativeness, completeness, comparability (PARCC) parameters [7].
This section describes how the analytical data generated by the reference laboratory were used to establish a
baseline performance for PCB analysis.
Reference Laboratory Selection
The Oak Ridge Sample Management Office (SMO) has been tasked by DOE Oak Ridge Operations (DOE-
ORO) with maintaining a list of qualified laboratories to provide analytical services. The technology
demonstration plan [6] contains the SMO's standard operating procedures (SOPs) for identifying,
qualifying, and selecting analytical laboratories. Laboratories are qualified as acceptable analytical service
providers for the SMO by meeting specific requirements. These requirements include providing pertinent
documentation (such as QA and chemical hygiene plans), acceptance of the documents by the SMO, and
satisfactory performance on an on-site prequalification audit of laboratory operations. All laboratory
qualifications are approved by a laboratory selection board, composed of the SMO operations manager and
appointees from all prime contractors that conduct business with the SMO.
All of the qualified laboratories were invited to bid on the demonstration study sample analysis. The lowest-
cost bidder was LAS Laboratories, in Las Vegas, Nevada. A readiness review conducted by ORNL and the
SMO confirmed the selection of LAS as the reference laboratory. Acceptance of the reference laboratory
was finalized by satisfactory performance in the predemonstration study (see Table 3-2). The SMO
contracted LAS to provide full data packages for the demonstration study sample analyses within 30 days of
sample shipment.
The SMO conducts on-site audits of LAS annually as part of the laboratory qualification program. At the
time of selection, the most recent audit of LAS had occurred in February 1997. Results from this audit
indicated that LAS was proficient in several areas, including program management, quality management,
and training programs. No findings regarding PCB analytical procedure implementation were noted. A
second on-site audit of LAS occurred August 11-12, 1997, during the analysis of the demonstration study
samples. This surveillance focused specifically on the procedures that were currently in use for the analysis
of the demonstration samples. The audit, jointly conducted by the SMO, DOE-ORO, and EPA-Las Vegas,
verified that LAS was procedurally compliant. The audit team noted that LAS had excellent adherence to
the analytical protocols and that the staff were knowledgeable of the requirements of the method. No
findings impacting data quality were noted in the audit report.
25
-------
Reference Laboratory Method
The reference laboratory's analytical method, also presented in the technology demonstration plan [6],
followed the guidelines established in EPA SW-846 Method 8081 [5]. According to LAS's SOP, PCBs
were extracted from 30-g samples of soil by sonication in hexane. Each extract was then concentrated to a
final volume that was further subjected to a sulfuric acid cleanup to remove potential interferences. The
analytes were identified and quantified using a gas chromatograph equipped with dual electron-capture
detectors. Each extract was analyzed on two different chromatographic columns with slightly different
separation characteristics (primary column: RTX-1701, 30 m x 0.53 mm ID x 0.5 (jm; confirmatory
column: RTX-5, 30 m x 0.53 mm ID x 0.5 (jm). PCBs were identified when peak patterns from a sample
extract matched the patterns of standards for both columns. PCBs were quantified based on the initial
calibration of the primary column.
Calibration
Method 8081 states that, because Aroclors 1016 and 1260 include many of the peaks represented in the
other five Aroclor mixtures, it is only necessary to analyze two multilevel standards for these Aroclors to
demonstrate the linearity of the detector response for PCBs. However, per LAS SOPs, five-point (0.1 to 4
ppm) initial calibration curves were generated for Aroclors 1016, 1248, 1254, and 1260 and the surrogate
compounds [decachlorobiphenyl (DCB) and tetrachloro-m-xylene (TCMX)]. Single mid-level standards
were analyzed for the other Aroclors (1221, 1232, and 1242) to aid in pattern recognition. All of the multi-
point calibration data, fitted to quadratic models, met the QC requirement of having a coefficient of
determination (R2) of 0.99 or better over the calibration range specified. The detection limits for soil samples
were 0.033 ppm ((Jg/g) for all Aroclors except Aroclor 1221, which was 0.067 ppm. For extract samples,
the detection limits were 0.010 ppm (ng/mL) for all Aroclors except Aroclor 1221, which was 0.020 ppm.
Reporting detection limits were calculated based on the above detection limits, the actual sample weight, and
the dilution factor.
Sample Quantification
For sample quantification, Aroclors were identified by comparing the samples' peak patterns and retention
times with those of the respective standards. Peak height ratios, peak shapes, sample weathering, and
general similarity in detector response were also considered in the identification. Aroclor quantifications
were performed by selecting three to five representative peaks, confirming that the peaks were within the
established retention time windows, integrating the selected peaks, quantifying the peaks based on the
calibrations, and averaging the results to obtain a single concentration value for the multicomponent
Aroclor. If mixtures of Aroclors were suspected to be present, the sample was typically quantified in terms
of the most representative Aroclor pattern. If the identification of multiple Aroclors was definitive, total
PCBs in the sample were calculated by summing the concentrations of both Aroclors. Aroclor
concentrations were quantified within the concentration range of the calibration curve. If PCBs were
detected and the concentrations were outside of the calibration range, the sample was diluted and reanalyzed
until the concentration was within the calibration range. If no PCBs were detected, the result was reported as
a non-detect (i.e., "< reporting detection limit").
26
-------
Sample Receipt, Handling, and Holding Times
The reference laboratory was scheduled to analyze a total of 256 PCB samples (208 soil samples, 24 iso-
octane extract samples, and 24 methanol extract samples). Of these same samples, the developer was
scheduled to analyze a total of 232 PCB samples (208 soil samples and 24 extract samples in solvent of
choice). The samples were shipped to LAS at the start of the technology demonstration activities (July 22).
Shipment was coordinated through the SMO. Completion of chain-of-custody forms documented sample
transfer. The samples were shipped on ice in coolers to maintain <6°C temperatures during shipment.
Samples were shipped with custody seals to ensure sample integrity and to prevent tampering during
transport.
Upon receipt of the samples, the reference laboratory checked the receipt temperature and conditions of the
sample containers, assigned each sample a unique number, and logged each into its laboratory tracking
system. All samples were received at the proper temperature and in good condition. Demonstration samples
were divided into 11 analytical batches (with no more than 20 samples per batch). The samples were
analyzed in an order specified by ORNL to ensure that the analysis of sample types was randomized.
Analyses of QC samples, supplied by the reference laboratory to indicate method performance, were
performed with each analytical batch of soils.
Prior to analysis, samples were stored in refrigerators kept at 4 to 6 °C to maintain analyte integrity. The
reference laboratory was required to analyze the extract samples and to extract the soil samples within
14 days of shipment from ORNL. Once the soils were extracted, the reference laboratory had an additional
40 days to analyze the soil extracts. Maximum holding times were not exceeded for any of the demonstration
samples. The final reference laboratory data package for all samples was received at ORNL in 72 days, on
October 1, 1997. The contractual obligation was 30 days.
The remainder of this section is devoted to summarizing the data generated by the reference laboratory and
to assessing the analytical performance.
Quality Control Results
Objective
The purpose of this section is to provide an assessment of the data generated by the reference laboratory's
QC procedures. The QC samples included continuing calibration verification standards (CCVs), instrument
blanks, method blanks, surrogate spikes, [laboratory control samples (LCSs)], and matrix spike/matrix
spike duplicate (MS/MSD) samples. Each control type is described in more detail in the following text and
in the technology demonstration plan [6]. Because extraction of these liquid samples was not required,
calibration check standards and instrument blanks were the only control samples implemented for the extract
samples. The reference laboratory's implementation of QC procedures was consistent with SW-846
guidance.
Continuing Calibration Verification Standard Results
A CCV is a single calibration standard of known concentration, usually at the midpoint of the calibration
range. This standard is evaluated as an unknown and is quantified against the initial calibration. The
calculated concentration is then compared with the nominal concentration of the standard to determine
whether the initial calibration is still valid. CCVs were analyzed with every 10 samples or at least every 12.
27
-------
The requirement for acceptance was a percentage difference of less than 15% for the CCV relative to the
initial calibration. This QC requirement was met for all Aroclors and surrogates, except for one standard
that had a 16% difference for DCB. These results indicated that the reference laboratory maintained
instrument calibrations during the course of sample analysis.
Instrument and Method Blank Results
Instrument blanks (hexane) were analyzed prior to each CCV. The QC requirement was that instrument
blanks must contain less than the reporting detection limit for any analyte. All instrument blanks were
acceptable.
A method blank is an analyte-free soil matrix sample that is taken through the extraction process to verify
that there are no laboratory sources of contamination. One method blank was analyzed for each analytical
batch. The QC requirement was that method blanks must contain less than the reporting detection limit for
any Aroclor. No PCBs were detected in any of the eleven method blanks that were analyzed. These results
demonstrated that the reference laboratory was capable of maintaining sample integrity and that it did not
introduce PCB contamination to the samples during preparation.
Surrogate Spike Results
A surrogate is a compound that is chemically similar to the analyte group but is not expected to be present in
the environmental sample. A surrogate is added to test the extraction and analysis methods to verify the
ability to isolate, identify, and quantify a compound similar to the analyte(s) of interest without interfering
with the determination. Two different surrogate compounds, DCB and TCMX, were used to bracket the
retention time window anticipated in the Aroclor chromatograms. All soil samples, including QC samples,
were spiked with surrogates at 0.030 ppm prior to extraction. Surrogate recoveries were deemed to be
within QC requirements if the measured concentration fell within the QC acceptance limits that were
established by past method performance. (For LAS this was 39 to 117% for DCB, and 66 to 128% for
TCMX). The results were calculated using the following equation:
measured amount ,nnn,
percent recovery = x 100% (4-1)
actual amount
In all undiluted samples, both of the surrogates had percentage recoveries that were inside the acceptance
limits. Surrogate recoveries in diluted samples were uninformative because the spike concentration (0.030
ppm, as specified by the method) was diluted below the instrument detection limits. The surrogate recovery
results for undiluted samples indicated that there were no unusual matrix interferences or batch-processing
errors for these samples.
Laboratory Control Sample Results
A laboratory control sample (LCS) is an aliquot of a clean soil that is spiked with known quantities of target
analytes. The LCS is spiked with the same analytes and at the same concentrations as the MS. (MSs are
described in the next section.) If the results of the MS analyses are questionable (i.e., indicating a potential
28
-------
matrix effect), the LCS results are used to verify that the laboratory can perform the analysis in a clean,
representative matrix.
Aroclors 1016 and 1260 were spiked into the clean soil matrix at approximately 0.300 ppm, according to
the reference laboratory's SOP. The QC requirements (defined as percent recovery) for the LCS analyses
were performance-based acceptance limits that ranged from 50 to 158%. In all but one of the eleven LCSs
analyzed, both Aroclor percent recoveries fell within the acceptance limits. Satisfactory recoveries for LCS
verified that the reference laboratory performed the analyses properly in a clean matrix.
Matrix Spike Results
In contrast to an LCS, an MS sample is an actual environmental soil sample into which target analytes are
spiked at known concentrations. MS samples are used to assess the efficiency of the extraction and
analytical methods for real samples. This is accomplished by determining the amount of spiked analyte that
is quantitatively recovered from the environmental soil. An MSB sample is spiked and analyzed to provide a
measure of method precision. Ideally, to evaluate the MS/MSD results, the environmental soil is analyzed
unspiked so that the background concentrations of the analyte in the sample are considered in the recovery
calculation.
For the demonstration study samples, one MS and MSB pair was analyzed with each analytical batch. The
MS samples were spiked under the same conditions and QC requirements as the LCS (50 to 158%
acceptance limits), so that MS/MSD and LCS results could be readily compared. The QC requirement for
MS and MSB samples was a relative percent difference (RPD) of less than 30% between the MS/MSD
pair. RPD is defined as:
r,r.r^ I MS recovery - MSD recovery ,nf\n/
RPD = -! •£• ^— x 100% (4.2)
average recovery
A total of eleven MS/MSD pairs were analyzed. Because the MS/MSD spiking technique was not always
properly applied (e.g., a sample which contained 100 ppm of Aroclor 1254 was spiked ineffectively with
0.300 ppm of Aroclor 1260), many of the MS/MSD results were uninformative. For the samples that were
spiked appropriately, all MS/MSD QC criteria were met.
Conclusions of the Quality Control Results
The reference laboratory results met performance acceptance requirements for all of the samples where
proper QC procedures were implemented. Acceptable performance on QC samples indicated that the
reference laboratory was capable of performing analyses properly.
Data Review and Validation
Objective
The purpose of validating the reference laboratory data was to ensure usability for the purposes of
comparison with the demonstration technologies. The data generated by the reference laboratory were used
as a baseline to assess the performance of the technologies for PCB analysis. The reference laboratory data
29
-------
were independently validated by ORNL and SMO personnel, who conducted a thorough quality check and
reviewed all sample data for technical completeness and correctness.
Corrected Results
Approximately 8% of the results provided by the reference laboratory (20 of 256) were found to have
correctable errors. So as not to bias the assessment of the technology's performance, errors in the reference
laboratory data were corrected. These changes were made conservatively, based on the guidelines provided
in the SW-846 Method 8081 for interpreting and calculating Aroclor results. The errors (see Appendix D,
Table D-3) were categorized as transcription errors, calculation errors, and interpretation errors. The
corrections listed in Table D-3 were made in the final data set that was used for comparison with the
demonstration technologies.
Suspect Results
Normally, one would not know if a single sample result was "suspect" unless (1) the sample was a
performance evaluation sample, where the concentration is known or (2) a result was reported and flagged
as suspect for some obvious reason (e.g., no quantitative result was determined). The experimental design
implemented in this demonstration study provided an additional indication of the abnormality of data
through the inspection of the replicate results from a homogenous soil sample set (i.e., four replicates were
analyzed for each sample ID).
Data sets were considered suspect if the standard deviation (SD) of the four replicates was greater than 30
ppm and the percent relative standard deviation (RSD) was greater than 50%. Five data sets (sample IDs
106, 205, 216, 217, 225) contained measurements that were considered suspect using this criteria, and the
suspect data are summarized in Table 4-1. A number of procedural errors may have caused the suspect
measurements (e.g., spiking heterogeneity, extraction efficiencies, dilution, etc.). In the following
subsections for precision and accuracy, the data were evaluated with and without these suspect values to
represent the best and worst case scenarios.
Table 4-1. Suspect measurements within the reference laboratory data
Criteria
SD > 30 ppm
and
RSD > 50%
Qualitative Result
Sample ID
106
205
216
217
225
110
112
PCB Concentration (ppm)
Replicate Results
(ppm)
255.9,269.9,317.6
457.0,483.3,538.7
47.0,54.3,64.0
542.8, 549.8, 886.7
32.1,36.5,56.4
< reporting detection
limits
Suspect Result(s)
(ppm)
649.6
3,305.0
151.6
1,913.3
146.0
< 66, < 98, < 99, < 490
< 66, < 130, < 200,< 200
Data Usability
Performed data analysis with
and without this value
Used as special case for
comparison with developer
results
30
-------
Samples that did not fall into the above criteria, but were also considered suspect, were non-blank samples
that could not be quantified and were reported as "< the reporting detection limit." This was the case for
environmental soil sample IDs 110 and 112. It is believed that the reference laboratory had trouble
quantifying these soil samples because of the abundance of chemical interferences. These samples were
diluted by orders of magnitude to reduce interferences, thereby diluting the PCB concentrations to levels that
were lower than the instrument detection limits. With each dilution, the reporting detection limits values
were adjusted for sample weight and dilution, which accounts for the higher reporting detection limits (up to
490 ppm). It is believed that these samples should have been subjected to additional pre-analytical cleanup
to remove these interferences before quantification was attempted. Sample IDs 110 and 112 were collected
from the same cleanup site (see Appendix B), so it is not surprising that similar difficulties were encountered
with both sample sets. Because the results for sample IDs 110 and 112 were not quantitative, these data
were compared with the technology data only on a special case basis.
Data Assessment
Objective
The purpose of this section is to provide an evaluation of the performance of the reference laboratory results
through statistical analysis of the data. The reference laboratory analyzed 72 PE, 136 environmental soil,
and 48 extract samples. All reference laboratory analyses were performed under the same environmental
conditions. Therefore, site differentiation was not a factor in data assessment for the reference laboratory.
For comparison with the technology data, however, the reference laboratory data are delineated into
"outdoor site" and "chamber site" in the following subsections. For consistency with the technology review,
results from both sites were also combined to determine the reference laboratory's overall performance for
precision and accuracy. This performance assessment was based on the raw data compiled in Appendix D.
All statistical tests were performed at a 5% significance level.
Precision
The term "precision" describes the reproducibility of measurements under a given set of conditions. The SD
of four replicate PCB measurements was used to quantify the precision for each sample ID. SD is an
absolute measurement of precision, regardless of the PCB concentration. To express the reproducibility
relative to the average PCB concentration, RSD is used to quantify precision, according to the following
equation:
„„„ Standard Deviation inno/
RSD = x 100% (4.3)
Average Concentration
Performance Evaluation Samples
The PE samples were homogenous soils containing certified concentrations of PCBs. Results for these
samples represent the best estimate of precision for soil samples analyzed in the demonstration study. Table
4-2 summarizes the precision of the reference laboratory for the analysis of PE samples. One suspect
31
-------
measurement (sample ID 225, 146.0 ppm) was reported for the PE soil samples. The RSDs for the
combined data ranged from 9 to 33% when the suspect measurement was excluded, and from 9 to
Table 4-2. Precision of the reference laboratory for PE soil samples
Outdoor Site
Sample
ID
126 a
118
124
120
122
119
125
121
123
Average
Concentration
(ppm)
0
1.6
1.7
5.0
11.1
20.1
37.9
54.6
60.1
SD
(ppm)
n/a
0.6
0.2
1.0
0.9
3.4
6.9
3.4
4.6
RSD
(%)
n/a
39
13
20
8
17
18
6
8
Chamber Site
Sample
ID
226
218
224
220
222
219
225
221
223
Average
Concentration
(ppm)
0
2.6
1.7
5.8
12.8
23.3
41.7"
44.9
55.8
SD
(ppm)
n/a
0.2
0.5
1.8
0.3
6.1
12.9"
11.3
7.7
RSD
(%)
n/a
6
29
31
3
26
31"
25
14
Combined Sites
Average
Concentration
(ppm)
0
2.1
1.7
5.4
11.9
21.7
39. 5C
49.8
58.0
SD
(ppm)
n/a
0.7
0.4
1.4
1.1
4.9
9.2C
9.3
6.3
RSD
(%)
n/a
33
21
26
9
23
23C
19
11
a All PCB concentrations were reported as non-detects.
b Results excluding the suspect value (results including the suspect value: mean = 67.1
c Results excluding the suspect value (results including the suspect value: mean = 52.!
ppm, SD = 53.2 ppm, and RSD = 79%).
ppm, SD = 38.6 ppm, and RSD = 73%).
79%, including the suspect measurement. The overall precision, determined by the mean RSD for all PE
samples, was 21% for the worst case (including the suspect result) and 18% for the best case (excluding the
suspect result).
Environmental Soil Samples
The precision of the reference laboratory for the analysis of environmental soil samples is reported in Table
4-3. In this table, results including suspect measurements are presented in parentheses. Average
concentrations were reported by the reference laboratory as ranging from 0.5 to 1,196 ppm with RSDs that
ranged from 7 to 118% when the suspect results were included. Excluding the suspect results, the highest
average concentration decreased to 660 ppm, and the largest RSD decreased to 71%. Because the majority
of the samples fell below 125 ppm, precision was also assessed by partitioning the results into two ranges:
low concentrations (< 125 ppm) and high concentrations (> 125 ppm). For the low concentrations, the
average RSD was 23% excluding the suspect value and 26% including the suspect value. These average
RSDs were only slightly larger than the RSDs for the PE soils samples of comparable concentration (18%
for best case and 21% for worst case). Five soil sample sets (sample IDs: 106, 117, 205, 211 and 217) were
in the high-concentration category. The average precision for high concentrations was 56% for the worst
case and 19% for the best case. The precision estimates for the low and high concentration ranges were
comparable when the suspect values were excluded. This indicated that the reference laboratory's precision
32
-------
for the environmental soils was consistent (approximately 21% RSD), and comparable to the PE soil
samples when the suspect values were excluded.
The Paducah soils (indicated as bold sample IDs in Table 4-3) were analyzed by the technologies under both
outdoor and chamber conditions to provide a measure of the effect that two different environmental
conditions had on the technology's performance. Although this was not an issue for the reference laboratory
(because all the samples were analyzed under laboratory conditions), the reference laboratory's results were
delineated into the different site categories for comparison with the technologies. Sample IDs 113 and 201,
114 and 202, 115 and 203, 116 and 204, and 117 and 205 each represent a set of eight replicate samples of
the same Paducah soil. The RSDs for four of the five Paducah pairs (excluding the suspect value for sample
ID 205) ranged from 11 to 17%. The result from one pair (sample IDs 113 and 201) had an RSD of 42%,
but the reported average concentration was near the reporting limits.
Table 4-3. Precision of the reference laboratory for environmental soil samples
Outdoor Site
Sample
ID
101
102
103
104
105
106
107
108
109
110
111
112
113 c
114
115
116
117
Average
Concentration
(ppm)
0.5
2.0
2.3
9.4
59.4
281.0(373.2)a
1.3
1.8
2.0
n/ab
38.7
n/a
1.1
1.3
14.8
41.3
383.9
Standard
Deviation
(ppm)
0.1
0.3
0.6
4.0
16.5
32.4(186.2)
0.3
0.1
0.4
n/a
4.3
n/a
0.6
0.3
1.8
5.9
55.2
RSD
(%)
16
16
27
43
28
12(50)
20
8
20
n/a
11
n/a
55
20
12
14
14
Chamber Site
Sample
ID
206
207
208
209
210
211
212
213
214
215
216
217
201
202
203
204
205
Average
Concentration
(ppm)
1.9
18.8
30.5
40.2
88.6
404.5
3.2
8.1
25.2
26.7
55.1(79.2)
659.8(973.2)
0.9
1.4
13.9
44.3
493.0(1196.0)
Standard
Deviation
(ppm)
0.9
3.5
7.9
28.5
25.6
121.8
1.6
1.6
3.7
3.2
8.5 (48.7)
196.6(647.0)
0.2
0.2
1.7
2.9
41.7(1406.4)
RSD
(%)
49
19
26
71
29
30
50
20
15
12
15(62)
30 (66)
24
12
12
7
8(118)
a Data in parentheses include suspect values.
bn/a indicates that qualitative results only were reported for this sample.
c Bold sample IDs were matching Paducah sample pairs (i.e., 113/201, 114/202, 115/203, 116/204, 117/205).
33
-------
Extract Samples
The extract samples, which were used to simulate surface wipe samples, were the simplest of all the
demonstration samples to analyze because they required no extraction and were interference-free. Three
types of extract samples were analyzed: solvent blanks, spikes of Aroclor 1242 at 10 (jg/mL, and spikes of
Aroclor 1254 at 100 (jg/mL. Identical extract samples were prepared in two solvents (iso-octane and
methanol) to accommodate the developer's request. The reference laboratory analyzed both solvent sets. A
Student's t-test [8, 9] was used to compare the reference laboratory's average PCB concentrations for the
two different solvents and showed that no significant differences were observed at either concentration.
Therefore, the reference laboratory results for the two extract solvents were combined. Additionally, all
blank samples were quantified as non-detects by the reference laboratory.
Table 4-4 summarizes the reference laboratory results for the extract samples by site. RSDs for the four
replicates for each sample ID ranged from 3 to 24%. For the combined data set (16 replicate
measurements), the average RSD at the 10-(jg/mL level was 19%, while the average RSD at the 100-(jg/mL
level was 8%. For the entire extract data set, an estimate of overall precision was 14%. The overall
precision for the extract samples was comparable to the best case precision for environmental soil samples
(21%) and PE soil samples (18%).
Table 4-4. Precision of the reference laboratory for extract samples
Outdoor Site
Sample
ID
129 a
132 a
127
130
128
131
Average
Cone
(ug/mL)
0
0
10.9
12.1
67.4
63.8
SD
(ug/mL)
n/a
n/a
0.4
2.9
2.3
5.0
RSD
(%)
n/a
n/a
4
24
3
8
Chamber Site
Sample
ID
229
232
227
230
228
231
Average
Cone
(ug/mL)
0
0
9.6
8.9
65.2
57.7
SD
(ug/mL)
n/a
n/a
0.8
1.4
5.1
3.1
RSD
(%)
n/a
n/a
8
16
8
5
Combined Sites
Average
Cone
(ug/mL)
10.4
63.5
SD
(ug/mL)
n/a
1.9
5.2
RSD
(%)
n/a
19
8
* All PCB concentrations reported as non-detects by the laboratory.
Accuracy
Accuracy represents the closeness of the reference laboratory's measured PCB concentrations to the
accepted values. Accuracy was examined by comparing the measured PCB concentrations (for PE soil and
extract samples) with the certified PE values and known spiked extract concentrations. Percent recovery
was used to quantify the accuracy of the results. The optimum percent recovery value is 100%. Percent
recovery values greater than 100% indicate results that are biased high, and values less than 100% indicate
results that are biased low.
34
-------
Performance Evaluation Soil Samples
The reference laboratory's performance for the PE samples is summarized in Table 4-5. Included in this
table are the performance acceptance ranges and the certified PCB concentration values. The acceptance
ranges, based on the analytical verification data, are guidelines established by the provider of the PE
materials to gauge acceptable analytical results. As shown in Table 4-5, all of the average concentrations
were within the acceptance ranges, with the exception of sample ID 218. The average result of sample ID
225 was outside of the acceptance range only when the suspect result was included. All of the replicate
measurements in sample ID 225 were biased slightly high. Average percent recoveries for the PE samples
(excluding suspect values) ranged from 76 to 130%. Overall accuracy was estimated as the average
recovery for all PE samples. The overall percent recovery was 105% as a worst case when the suspect value
was included. Excluding the suspect value as a best case slightly lowered the overall percent recovery to
101%. A regression analysis [10] indicated that the reference laboratory's results overall were unbiased
estimates of the PE sample concentrations.
Table 4-5. Accuracy of the reference laboratory for PE soil samples
Certified
Concentration
(ppm)
(Acceptance
Range, ppm)
Oa
(n/a)
2.0
(0.7-2.2)
2.0
(0.9-2.5)
5.0
(2.1-6.2)
10.9
(4.0-12.8)
20.0
(11.4-32.4)
49.8
(23.0-60.8)
50.0
(19.7-63.0)
50.0
(11.9-75.9)
Outdoor Site
Sample
ID
126
118
124
120
122
119
125
121
123
Average
Cone
(ppm)
0
1.6
1.7
5.0
11.1
20.1
37.9
54.6
60.1
Recovery
(%)
n/a
79
85
99
102
100
76
109
120
Chamber Site
Sample
ID
226
218
224
220
222
219
225
221
223
Average
Cone
(ppm)
0
2.6
1.7
5.8
12.8
23.3
41.7"
44.9
55.8
Recovery
(%)
n/a
130
85
117
117
116
84"
90
112
Combined Sites
Average
Cone
(ppm)
0
2.1
1.7
5.4
11.9
21.7
39. 5 c
49.8
58.0
Recovery
(%)
n/a
105
85
108
109
109
79 c
100
116
a All PCB concentrations reported as non-detects by the laboratory.
b Results excluding the suspect value (results including the suspect value: average = 67.8 ppm and recovery = 136%).
'Results excluding the suspect value (results including the suspect value: average = 52.8 ppm and Recovery = 106%).
35
-------
Extract Samples
Percent recovery results for extract samples are summarized in Table 4-6 for the reference laboratory. The
average percent recoveries for extract samples ranged from 58 to 121%. In terms of concentration levels, the
average recovery at the 10-(jg/mL level (for both solvents) was 104%, compared with 64% at the 100-
(jg/mL level. The reference laboratory classified all 16 samples spiked at 10 (jg/mL as Aroclor 1016;
however, these samples were actually spiked with Aroclor 1242. Despite this misclassification, the results
did not appear to be biased. In contrast, the samples spiked at 100 (jg/mL were correctly classified as
Aroclor 1254 but were all biased low. Although these results suggested that Aroclor classification had little
effect on the quantification of the extract samples, there was an obvious, consistent error introduced into the
analysis of the 100-(jg/mL samples to cause the low bias. For the entire extract data set, the overall percent
recovery was 84%.
Table 4-6. Accuracy of the reference laboratory for extract samples
Spike
Concentration
(ug/mL)
Oa
Oa
10
10
100
100
Outdoor Site
Sample
ID
129
132
127
130
128
131
Average
Cone
(ug/mL)
0
0
10.9
12.1
67.4
63.8
Recovery
(%)
n/a
n/a
109
121
67
64
Chamber Site
Sample
ID
229
232
227
230
228
231
Average
Cone
(ug/mL)
0
0
9.6
8.9
65.2
57.7
Recovery
(%)
n/a
n/a
96
89
65
58
Combined Sites
Average
Cone
(ug/mL)
u
10.4
63.5
Recovery
(%)
n/a
104
64
a All PCB concentrations reported as non-detects by the laboratory.
Representativeness
Representativeness expresses the degree to which sample data accurately and precisely represent the
capability of the method. Representativeness of the method was assessed based on the data generated for
clean-QC samples (i.e., method blanks and laboratory control samples) and PE samples. Based on the data
assessment (discussed in detail in various parts of this section), it was determined that the representativeness
of the reference laboratory data was acceptable. In addition, acceptable performance on laboratory audits
substantiated that the data set was representative of the capabilities of the method. In all cases, the
performance of the reference laboratory met all requirements for both audits and QC analyses.
Completeness
Completeness is defined as the percentage of measurements that are judged to be usable (i.e., the result was
not rejected). Usable results were obtained for 248 of the 256 samples submitted for analysis by the
reference laboratory. Eight results (for sample IDs 110 and 112) were deemed incomplete and therefore not
valid because the measurements were not quantitative. To calculate completeness, the total number of
complete results were divided by the total number of samples submitted for analysis, and then multiplied by
36
-------
100 to express as a percentage. The completeness of the reference laboratory was 97%, where a
completeness of 95% or better is typically considered acceptable.
Comparability
Comparability refers to the confidence with which one data set can be compared with another. The
demonstration study was designed to have a one-to-one, sample-by-sample comparison of the PCB results
obtained by the reference laboratory and the PCB results obtained by the technology being evaluated. Based
on thorough examination of the data and acceptable results on the PE samples, it was concluded that the
reference laboratory's SOPs for extraction and analysis, and the data generated using these procedures,
were of acceptable quality for comparison with the field technology results. Additional information on
comparability was available because the experimental design incorporated randomized analysis of blind,
replicate samples. Evaluation of the replicate data implicated some of the individual data points as suspect
(see Table D-2). The reference laboratory's suspect data were compared with the technology data on a
special-case basis, and exceptions were noted.
Summary of Observations
Table 4-7 provides a summary of the performance of the reference laboratory for the analysis of all sample
types used in the technology demonstration study. As shown in Table 4-7, the precision of the PE soils was
comparable to the environmental soils. A weighted average, based on the number of samples, gave a best
case precision of 21% and a worst-case precision of 28% for all the soil data (PE and environmental). The
extract samples had a smaller overall RSD of 14%. Evaluation of overall accuracy was based on samples
with certified or known spiked concentrations (i.e., PE and extract samples). The overall accuracy, based on
percent recovery, for the PE samples was 105% for the worst case (which included the suspect value) and
101% for the best case (which excluded the suspect value). These results indicated that the reference
laboratory measured values were unbiased estimates of the certified PE concentrations (for samples that
contained <50 ppm of PCBs). Accuracy for the extract samples at 10 ppm was also unbiased, with an
average percent recovery of 104%. However, the accuracy for the extract samples at 100 ppm was biased
low, with an average recovery of 64%. Overall, the average percent recovery for all extract samples was
84%. The reference laboratory correctly reported all blank samples as non-detects, but had difficulty with
two soil sample IDs (110 and 112) that contained chemical interferences. In general, the reference
laboratory's completeness would be reduced, at the expense of an improvement in precision and accuracy, if
the suspect measurements were excluded from the data analysis. Based on this analysis, it was concluded
that the reference laboratory results were acceptable for comparison with the developer's technology.
37
-------
Table 4-7. Summary of the reference laboratory performance
Sample Matrix
Blank
Environmental soil with
interferences
Soil
Best Case
(excluding suspect data)
Soil
Worst Case
(including suspect data)
Extract
Sample Type
Soil
Extract
Sample ID 110
Sample ID 112
PE
Environmental
< 125 ppm
> 125 ppm
overall
PE
Environmental
< 125 ppm
> 125 ppm
overall
10 ppm
100 ppm
overall
Number of Samples
8
16
4
4
63
107
17
187
64
108
20
192
16
16
32
Precision
(Average % RSD)
n/aa
n/aa
18
23
19
21
21
26
56
28
19
8
14
Accuracy
(Average %Recovery)
All samples were
reported as non-detects.
All samples were
reported as non-detects.
101
n/ab
n/ab
101
105
n/ab
n/ab
105
104
64
84
a Because the results were reported as non-detects, precision assessment is not applicable.
b Accuracy assessment calculated for samples of known concentration only.
38
-------
Section 5
Technology Performance and Evaluation
Objective and Approach
The purpose of this section is to present the evaluation of data generated by SDFs RaPID Assay System.
The technology's precision and accuracy performance are presented for the data generated in the
demonstration study. In addition, an evaluation of comparability, through a one-to-one comparison with the
reference laboratory data, is presented. An evaluation of other aspects of the technology (such as detection
limits, cost, sample throughput, hazardous waste generation, and logistical operation) is also presented in
this section.
Data Assessment
The purpose of the data assessment section is to present the evaluation of the performance of SDFs RaPID
Assay System through a statistical analysis of the data. PARCC parameters were used to evaluate RaPID
Assay's ability to measure PCBs in PE, environmental soil, and extract samples. The developer analyzed
splits of replicate samples that were also analyzed by the reference laboratory (72 PE soil samples, 136
environmental soil samples, and 24 extract samples). See Section 4 for a more detailed analysis of the
reference laboratory's results. Replicate samples were analyzed by the developer at two different sites
(under outdoor conditions and inside an environmentally controlled chamber) to evaluate the effect of
environmental conditions on performance; see Section 3 for further details on the different sites. Evaluation
of the measurements at each site indicated that there were no significant differences in the accuracy of the
measurements made at each site. There was a significant "site effect" (i.e., significant differences in the data
generated under the outdoor and chamber conditions), however, in the precision of the measured
concentrations. In cases where the environmental conditions did not affect the results significantly (i.e., for
accuracy), data from both sites were combined to determine overall performance. All statistical tests were
performed at the 5% significance level. Appendix D contains the raw data that were used to assess the
performance of the RaPID Assay System.
Precision
Precision, as defined in Section 4, is the reproducibility of measurements under a given set of conditions.
The standard deviation and relative standard deviation of four replicate measurements were used to quantify
the technology's precision. The average PCB concentration for a replicate set was used to calculate the
percent RSD for each Sample ID (see Equation 4-3). For comparative information on the reference
laboratory's precision, refer to the data presented in Section 4 under the heading "Precision."
Performance Evaluation Samples
Table 5-1 summarizes the precision of the RaPID Assay System for the analysis of PE samples. Operating
under the outdoor conditions, the RSDs ranged from 11 to 42%. RSDs ranged from 2 to 33% while
39
-------
operating inside the chamber. Because there was a significant site effect for the samples that were analyzed
under outdoor conditions, the precision data for both sites were not combined.
Table 5-1. Precision of the RaPID Assay System for PE soil samples
Outdoor Site
Sample
ID
126
118
124
120
122
119
125
121
123
Average
Concentration
(ppm)
<0.5
1.5
2.9
6.3
6.7
15.1
74.5
70.8
54.8
SD
(ppm)
n/aa
0.2
0.7
1.5
1.2
6.2
9.4
24.0
22.8
RSD
(%)
n/a
11
25
24
18
41
13
34
42
Chamber Site
Sample
ID
226
218
224
220
222
219
225
221
223
Average
Concentration
(ppm)
<0.5
1.3
2.3
5.4
6.2
10.8
77.3
63.7
44.5
SD
(ppm)
n/a
0.4
0.4
1.3
0.1
0.9
10.0
7.9
5.7
RSD
(%)
n/a
33
17
23
2
9
13
13
13
' SD and RSD cannot be calculated because all results were reported as < 0.5 ppm.
Environmental Soil Samples
The precision of the RaPID Assay for the analysis of environmental soil samples is reported in Table 5-2.
Operating under the outdoor conditions, the RSDs ranged from 1 to 51%. RSDs ranged from 7 to 22%
while operating inside the chamber. Because the majority of measurements fell below 125 ppm, precision
was also assessed by partitioning the results into two ranges: low concentrations (reference laboratory
values <125 ppm) and high concentrations (reference laboratory values >125 ppm). See Section 4 for
delineation of Sample IDs in each concentration range. For the low-concentration range, the average RSD
was 24% at the outdoor site compared with 11% at the chamber site. The same trend was observed at the
high-concentration range where the average RSD was 25% at the outdoor site compared with 6% at the
chamber site. The site effect for the precision data is illustrated in Figure 5-1, which shows the standard
deviations for the RaPID Assay's measurements vs the measured average concentrations for the
environmental soil samples less than 100 ppm. The figure distinctly depicts how the variability of the
measurements was significantly higher under the outdoor conditions compared with the chamber.
40
-------
Table 5-2. Precision of the RaPID Assay System for environmental soil samples
Outdoor Site
Sample
ID
101
107
108
109
102
103
104
110
111
105
112
106
113"
114
115
116
117
Average
Concentration
(ppm)
0.5 a
1.9
3.2
3.4
3.4
5.1
15.3
31.0
49.1
91.4
123.8
>200
1.4
1.1
15.1
75.9
223.9
Standard
Deviation
(ppm)
n/a
0.3
0.9
0.8
0.3
1.2
7.8
9.3
12.6
19.9
19.4
0
0.2
0.4
3.2
29.3
56
RSD
(%)
n/a
16
27
23
8
24
51
30
26
22
16
0
13
36
21
39
25
Chamber Site
Sample
ID
206
212
213
207
208
215
214
209
216
210
211
217
201
202
203
204
205
Average
Concentration
(ppm)
1.5
4.0
5.7
19.3
29.1
29.8
31.1
48.4
57.1
80.8
>200
>200
0.8
0.9
17.7
88.5
198.0
Standard
Deviation
(ppm)
0.3
0.7
0.4
4.2
3.1
2.1
2.0
6.9
8.3
10.5
0
0
0.1
0.2
2.7
8.3
12.1
RSD
(%)
17
18
6
22
11
7
7
14
14
13
0
0
18
22
15
9
6
a Three results reported as < 0.5 ppm; one result reported as 0.6 ppm.
b Sample IDs in bold were matching Paducah sample pairs (i.e., 113/201,
114/202, 115/203, 116/204, 117/205).
Q.
_ 20 -
C
o
13
> i
a
•a
10 20 30 40 50 60 70 80 90 100
Average PCB Concentration ( ppm )
Figure 5-1. Standard deviation vs average PCB concentration for the
RaPID Assay measurements with average values less than 100 ppm.
Upper line and closed circles are for outdoor conditions and lower
line and open circles are for chamber conditions.
41
-------
The Paducah soils (indicated by bold Sample IDs in Table 5-2) were analyzed at both sites to provide an
assessment of the RaPID Assay's performance under different environmental conditions. Sample IDs 113
and 201, 114 and 202, 115 and 203, 116 and 204, and 117 and 205 represented replicate Paducah soil
sample sets, where the 100 series were samples analyzed under the outdoor conditions and the 200 series
were samples analyzed inside the chamber. An analysis of variance test (ANOVA) was used to compare the
effect of the two environmental conditions on the average measurements. Results from this analysis showed
that there were significant differences in the data generated at each site, further confirming that
environmental conditions had an effect on the precision of the RaPID Assay. Because there was a significant
site effect, the Paducah sample results were not combined to determine the overall precision for the Paducah
samples.
Extract Samples
Table 5-3 summarizes the RaPID Assay's results for the extract samples that were used to simulate surface
wipe samples. Refer to Section 3, "Extract Samples," for further clarification of this sample type. Operating
under the outdoor conditions, the RSDs ranged from 5 to 8%. RSDs were both 6% while operating inside
the chamber. In terms of concentration level, the average RSD at the 10 (jg/mL level was 16%, while the
average RSD at the 100 (jg/mL level was 8%.
Table 5-3. Precision of the RaPID Assay System for extract samples
Outdoor Site
Sample
ID
132
130
131
Average
Concentration
(ug/mL)
<0.5
12.5
96.9
SD
(ug/mL)
n/aa
0.9
5.2
RSD
(%)
n/a
8
5
Chamber Site
Sample
ID
229
227
228
Average
Concentration
(ug/mL)
<0.3
9.5
86.3
SD
(ug/mL)
n/a
0.6
5.3
RSD
(%)
n/a
6
6
Combined Sites
Average
Concentration
(ug/mL)
<0.4
11.0
91.6
SD
(ug/mL)
n/a
1.8
7.5
RSD
(%)
n/a
16
8
1 SD and RSD cannot be calculated because results were reported either as < 0.3 ppm or < 0.5 ppm.
Precision Summary
The overall precision was characterized by three summary values for the RSD: mean (i.e., average), median
(i.e., 50th percentile value at which 50% of all individual RSD values are below and 50% are above), and
95th percentile (i.e., the value at which 95% of all individual RSD values are below and 5% are above).
These values are summarized in Table 5-4 for each of the sample types. Because there was a significant site
effect for the soil samples (i.e., PE and environmental) that were analyzed under the outdoor conditions, the
precision data for both sites were not combined. Under the outdoor conditions, the mean RSD for the PE
sample results was 26% and lower (at 15%) under the chamber conditions. Similarly, the mean RSD for the
environmental soil samples was 24% under the outdoor conditions and 13% under the chamber conditions.
42
-------
Table 5-4. Overall precision of the RaPID Assay System for all sample types
Statistic
Mean
Median
95th
percentile
PE Samples
%RSD
Outdoor Chamber
26
24
41
15
13
29
Environmental Soil Samples
%RSD
Outdoor Chamber
24
23
42
13
14
22
Extract Samples
%RSD
Outdoor Chamber Combined
7
n/aa
n/a
13
n/a
n/a
12
n/a
n/a
1 Median and 95th percentile statistics were not applicable to extract samples.
The extract results did not indicate a significant site effect. This was most likely because the replicate
extract measurements were performed over a shorter period of time outdoors (i.e., less temperature
variability from the time the first and last extract sample were analyzed). All extract sample results were
combined to determine an overall precision of 12% RSD.
Accuracy
Accuracy, as defined in Section 4, represents the closeness of the technology's measured PCB
concentrations to the accepted values. Accuracy was examined in terms of percent recovery (see Equation 4-
1), and average percent recoveries were calculated by averaging the four replicates within a Sample ID. For
comparative information on the performance of the reference laboratory, refer to Section 4 under the
heading of "Accuracy."
Performance Evaluation Soil Samples
The RaPID Assay's performance for the PE samples is summarized in Table 5-5. Included in this table are
the performance acceptance ranges and the certified PCB concentration values. Average percent recoveries
ranged from 61 to 150% while operating under the outdoor conditions. Under chamber conditions, average
percent recoveries ranged from 54 to 155%. Regression analyses indicated that there was not a significant
site effect in terms of the accuracy of the measurements. However, three of the combined average
concentrations (Sample IDs 121/221, 124/224, and 125/225) were outside of the acceptance ranges, and all
were biased high. When the data for the two sites were combined, the average recoveries ranged from 60 to
152%, indicating both negative and positive bias. While the results were biased, the RaPID Assay data did
correlate with the certified PE values. Although the correlation between the certified values and the RaPID
Assay's results in this study was most accurately described by a quadratic equation (which draws a curved
line through the data points), a simpler linear equation (which draws a straight line through the data points)
can be used with minimal loss of predictive capability. Further discussion on how the RaPID Assay data
might be used in a decision-making process is presented in Appendix E.
The RaPID Assay's bias appeared to be influenced by the type of Aroclor identified. An ANOVA test was
used to determine which experimental factors (concentration level, Aroclor type, environmental conditions)
affected the bias of the results. This analysis indicated that environmental conditions did not affect the
accuracy, which confirmed the regression analysis results. Concentration level was a
43
-------
Table 5-5. Accuracy of the RaPID Assay System for PE soil samples
Certified
Concentration
(ppm)
(Acceptance
Range, ppm)
0
(n/a)
2.0
(0.7-2.2)
2.0
(0.9-2.5)
5.0
(2.1-6.2)
10.9
(4.0-12.8)
20.0
(11.4-32.4)
49.8
(23.0-60.8)
50.0
(19.7-63.0)
50.0
(11.9-75.9)
Outdoor Site
Sample
ID
126
118
124
120
122
119
125
121
123
Average
(ppm)
<0.5
1.5
2.9
6.3
6.7
15.1
74.5
70.8
54.8
Recovery
n/a
75
145
126
61
76
150
142
110
Chamber Site
Sample
ID
226
218
224
220
222
219
225
221
223
Average
(ppm)
<0.5
1.3
2.3
5.4
6.2
10.8
77.3
63.7
44.5
Recovery
n/a
65
115
108
57
54
155
127
89
Combined Sites
Average
(ppm)
<0.5
1.4
2.6
5.9
6.5
12.9
75.9
67.3
49.6
Recovery
n/a
70
130
118
60
65
152
135
99
significant factor that was shown to affect the bias, as evidenced by the positive and negative bias described
previously. More significantly, Aroclor type appeared to have a strong influence on the bias. Those samples
quantified as Aroclor 1254 were biased particularly high at the higher (i.e., 50 ppm) PE concentrations.
Extract Samples
Percent recovery results for the extract samples are summarized in Table 5-6 for the RaPID Assay System.
The average percent recoveries for extract samples ranged from 97 to 125% when the RaPID Assay was
used under the outdoor conditions and from 86 to 95% inside the chamber. In terms of concentration levels
(i.e., for the combined site data), the average recovery at the 10 (jg/mL level was 110%, compared with 92%
at the 100 (jg/mL level.
Accuracy Summary
The overall accuracy was characterized by three summary values for percent recovery: mean, median, and
95th percentile. These values are summarized in Table 5-7 for the PE and extract samples. For the PE
samples, despite a mean percent recovery of 103%, the overall accuracy of the RaPID Assay can be
44
-------
Table 5-6. Accuracy of the RaPID Assay System for extract samples
Spike
Concentration
(ug/mL)
0
10
100
Outdoor Site
Sample
ID
132
130
131
Avg
Cone.
(ug/mL)
<0.5
12.5
96.9
Recovery
(%)
n/a
125
97
Chamber Site
Sample
ID
232
230
231
Avg
Cone.
(ug/mL)
<0.3
9.5
86.3
Recovery
(%)
n/a
95
86
Combined Sites
Avg
Cone.
(ug/mL)
<0.4
11.0
91.6
Recovery
(%)
n/a
110
92
Table 5-7. Overall accuracy of the RaPID Assay System for all sample ty
Statistic
Mean
Median
95th percentile
PE Samples
%Recovery
Outdoor Chamber Combined
111
115
185
96
88
157
103
94
173
pes
Extract Samples
%Recovery
Outdoor Chamber Combined
111
n/aa
n/a
90
n/a
n/a
101
n/a
n/a
1 Median and 95th percentile statistics were not applicable to extract samples.
characterized as biased. As discussed previously in the "Accuracy" section, this is because of a statistically
significant effect on recovery caused by Aroclor type. The result is that the recoveries ranged from 60%
(biased low) to 152% (biased high), depending on the specific Aroclor(s) in the sample. Averaging across all
samples produces an overall recovery close to 100%.
The overall accuracy for all extract samples was a mean percent recovery of 101% (which did not involve
significant bias in either direction); the 95th percentile and median data were not presented because of the
limited number of data points.
False Positive/False Negative Results
A false positive result (fp) [11] is one in which the technology detects PCBs in the sample when there
actually are none. A false negative result (fn) [11] is one in which the technology indicates that there are no
PCBs present in the sample when there actually are. Both fp and fn results are influenced by the method
detection limit (MDL) of the technology. Of the eight blank soil samples analyzed, none were reported as
having detectable levels of PCBs (i.e., fp = 0%). Of the 192 non-blank soil samples analyzed, two were
reported as non-detects. Therefore, the percentage of fn results was 1%. For the extract samples, the
percentage of fp and fn results were both 0%.
45
-------
Representativeness
Representativeness expresses the degree to which sample data accurately and precisely represent the
capability of the technology. The performance data were accepted as being representative of the technology
because the RaPID Assay was capable of analyzing diverse sample types (PE, simulated surface wipe
extract, and actual field environmental soil samples) under multiple environmental conditions. When using
this technology, QC samples should be analyzed to assess the performance of RaPID Assay System under
the testing conditions.
Completeness
Completeness is defined as the percentage of measurements that are judged to be useable (i.e., the result was
not rejected). Useable results were obtained by the technology for all 232 samples. Therefore, the
completeness of the RaPID Assay System was 100%.
Comparability
Comparability refers to the confidence with which one data set can be compared with another. A one-to-one
sample comparison was performed to assess the comparability of the PCB concentrations found in all soil
samples (PE and environmental) for the RaPID Assay measured values vs the reference laboratory results.
Additional statistical analysis of the PCB soil concentrations for paired samples showed that the RaPID
Assay's results were significantly different (higher) than the reference laboratory results for low PCB
concentrations (i.e., < 125 ppm). At higher concentrations (i.e., > 125 ppm), 12 measurements were
reported semiquantitatively (i.e., > 200 ppm) by the RaPID Assay. This is illustrated in Figure 5-2, which is
a plot of the RaPID Assay's measured PCB soil concentrations vs the corresponding reference laboratory
(a)
0 100 200 300 400 500 600 700 800 900
Reference Laboratory PCB Concentration (ppm)
25 50 75 100 125
Reference Laboratory PCB Concentration (ppm)
(b)
Figure 5-2. Paired PCB measurements for RaPID Assay and reference laboratory for (a) all soil samples (RaPID Assay
results reported as > 200 ppm not included) and (b) soil samples where the reference laboratory results were less than
or equal to 125 ppm. Lines denote perfect correlation.
46
-------
measured concentrations (excluding the suspect values listed in Table 4-1). Figure 5-2 (a) is a plot of all of
the soil data, and (b) is a plot of the concentration region from 0 to 125 ppm, where most of the variation
can be viewed. Note that the diagonal lines drawn in Figure 5-2 represent the line of theoretically perfect
correlation (R2 =1.0) between the reference laboratory data set (plotted along the x axis) and the RaPID
Assay data set (plotted along the y axis). A value above the diagonal line indicates that the RaPID Assay's
measurement was higher than the reference laboratory's measurement, while those below the diagonal line
indicated a lower result.
Coefficients of determination (R2) [10] were computed using a linear model fitted to the plot of the RaPID
Assay PCB concentrations vs the reference laboratory PCB concentrations. Excluding the reference
laboratory's suspect measurements, the coefficient of determination (R2) was 0.754 when all soil samples (0
to 900 ppm) were considered. As shown in Figure 5-2 (b), the majority of the soil samples were in the
concentration range of 0 to 125 ppm. The R2 value for this concentration range was 0.716.
A direct comparison between the RaPID Assay and reference laboratory data was performed by evaluating
the percent difference (%D) between the measured concentrations, defined as:
0/r. [RaPID] - [RefLab] inno/
%D = — !-^ — x 100% (5_n
[RefLab] ^ '
Figure 5-3 provides a summary of the range of %D values for the soil samples, as calculated using Equation
5-1. The graph represents the percentage of samples that fall within each range of %D values but does not
reflect any grouping according to the actual concentrations of the replicate sets. Results for Sample IDs 110,
112, 126, and 226 were not included because the reference laboratory did not report quantitative results for
these samples. Results for samples that the RaPID Assay Kit reported as >200 ppm were also not included.
As shown in Figure 5-3, the %D values were evenly distributed between -75% to 100%. Approximately
40% of the samples were biased low (%D <-1%) relative to the reference laboratory results. Approximately
36% of the soil sample results had %D values within the range of ±25%.
Comparability was also assessed for the extract samples. Figure 5-4 is a plot of the RaPID Assay measured
PCB extract concentrations vs the corresponding reference laboratory measured concentrations. These data
indicated that the RaPID Assay was biased high for the 100-ppm extract sample results relative to the
reference laboratory. However, compared with the actual spike concentrations, the RaPID Assay's results
for these extract samples were accurate because the reference laboratory results were biased low. The
coefficient of determination (R2) for a line fitted to this data was 0.977, indicating near perfect correlation.
The %D values for the extract samples were also assessed and are shown in Figure 5-5. Because the
reference laboratory was biased low on the 100-ppm extract measurements relative to the actual spiked
concentrations, when compared with the RaPID Assay measurements, approximately half of the %D values
were between 50 and 75%. Approximately 56% of the extract sample results had %D values within ±25%.
47
-------
-75%to- -50%to- -25%to- 1%to 25%to 50%to 75%to
51% 26% 0% 24% 49% 74% 100%
Range of percent difference values
> 101%
Figure 5-3. Range of percent difference values for the comparison of the RaPID Assay's soil sample
results with the reference laboratory results.
Reference Laboratory PCB Concentration (ppm)
Figure 5-4. Paired extract PCB measurements for the
RaPID Assay and reference laboratory. Measurements
above the diagonal line indicate that the RaPID Assay's
measurements are higher than the reference laboratory's
measurements.
48
-------
-25% to -0% 1 % to 24% 25% to 49% 50% to 74%
Range of percent difference values
75% to 100%
Figure 5-5. Range of percent difference values for the comparison of the RaPID Assay's extract sample
results with the reference laboratory results.
The soil data not included in previous comparability evaluations (because the replicate data for the reference
laboratory were considered suspect) are shown in Table 5-8. Refer to Section 4, in particular Table 4-1, for
more information on the reference laboratory's suspect measurements. The reference laboratory's suspect
data were compared with the RaPID Assay's matching results. For sample IDs 110 and 112, the reference
laboratory obtained qualitative results only, while SDI reported quantitative PCB concentrations for the four
replicates that were precise. For two of the other five suspect measurements (sample IDs 106 and 217), SDI
reported qualitative results only. For the other three suspect measurements, the RaPID Assay generated
results for all four replicates that were consistent and that were comparable with the reference laboratory's
replicate means that excluded the suspect value. These comparisons demonstrate the RaPID Assay's ability
to successfully analyze some of the samples that were troublesome for the reference laboratory.
Summary of PARCC Observations
Table 5-9 provides a summary of the performance of SDI's RaPID Assay System for the analysis of all
sample types used in this demonstration. The reference laboratory's performance (excluding suspect data) is
also presented in this table for comparison. In terms of precision, the overall average RSD for the RaPID
Assay, weighted for the number of samples, was 12% for the soil data generated under the chamber
conditions, compared with 25% under outdoor conditions. This data indicated that the RaPID Assay System
exhibited a significant site effect in terms of the precision of the measurements. While the variability was
lower under the controlled chamber conditions, the outdoor precision was still comparable to the overall
precision forthe reference laboratory (21% RSD). For the extract samples, the overall RSD of the RaPID
Assay was comparable to the reference laboratory (12% and 14%, respectively).
49
-------
Table 5-8. Comparison of the reference laboratory's suspect data with the RaPID Assay System data
Sample ID
110
112
106
205
216
217
225
Reference Laboratory
Suspect Measurement
(ppm)
200
185.4
55.2
>200
73.2
Replicate Mean
(ppm)
31.0
123.8
>200
198.0
57.1
>200
77.3
aMean result excluding the suspect measurement.
b Measurement reported qualitatively as less than or equal to the reporting detection limit (125 ppm c
Sample ID 110
Sample ID 112
Overall
10 ppm
100 ppm
Overall
RaPID Assay
Number of
Samples
8
8
32
32
52
56
4
4
4
4
96
92
8
8
16
Precision (Average % RSD)
RaPID Assay
n/a
26 (outdoor)
1 5 (chamber)
24 (outdoor)
1 1 (chamber)
25 (outdoor)
6 (chamber)
30 (outdoor)
16 (outdoor)
25 (outdoor)
12 (chamber)
16
8
12
Reference
Laboratory
n/a
18a
23 a
19a
not quantified
not quantified
21 a
19
8
14
Accuracy (Average % Recovery)
RaPID Assay
All reported as
< detection
limits.
103
103
110
92
101
Reference
Laboratory
All reported as
non-detects.
101 a
101 a
104
64
84
a Average result excluding the suspect measurements.
b Samples where the reference laboratory values were <125 ppm.
c Samples where the reference laboratory values were >125 ppm, excluding measurements where SDI reported semiquantitative
results (i.e., >200 ppm).
50
-------
In terms of accuracy, the RaPID Assay's PE soil measurements were biased both high (152% recovery) and
low (60% recovery), such that the overall average percent recovery was 103%. However, the results are
considered biased because the direction of bias was dependent on the Aroclor type present in the sample, as
discussed in the "Accuracy" section. In comparison, the reference laboratory reported unbiased PCB
concentrations (101% recovery) for the PE soil samples. Extract measurements by the RaPID Assay were
unbiased at both 10 ppm (110% recovery) and 100 ppm (92% recovery) concentration levels. The reference
laboratory results were unbiased at 10 ppm (104% recovery) but were biased low at 100 ppm (64%
recovery) for the extract samples.
SDI correctly reported all blanks analyses as "less than the detection limit," as did the reference laboratory
(i.e., 0% fp). Two fn were reported by the RaPID Assay. Overall, the performance of the RaPID Assay
System for the PCB demonstration samples was characterized as biased and precise.
Regulatory Decision-Making Applicability
One of the objectives of this demonstration was to assess the technology's ability to perform at regulatory
decision-making levels for PCBs, specifically 50 ppm for soils and 100 (jg/100cm2 for surface wipes. To
assess this, the RaPID Assay's performance for soil samples (both PE and environmental soil samples)
ranging in concentration from 40 to 60 ppm can be used, and the data are provided in Table 5-10. The
performance of the RaPID Assay System for this concentration range showed comparable precision (21%
RSD) and slightly higher percent recovery (129%) compared with the entire PE and environmental soil
sample data set. The mean %D value was 39% when compared with the corresponding reference laboratory
result. The RaPID Assay System's performance on extract samples is provided in Tables 5-4 and 5-7.
Assuming a 10-mL extract volume, extract samples (at 10 and 100 (jg/mL) represented surface wipe sample
concentrations of 100 and 1000 (jg/100 cm2. For the simulated wipe extract samples, the RaPID Assay was
precise (12% RSD) and accurate (101% recovery).
Table 5-10. Performance of the RaPID Assay System for soil samples between 40 and 60 ppm
Overall Performance
Mean
Median
95th percentile
Precision (% RSD)
21
14
40
Accuracy (% Recovery)
129
129
175
Comparability (% Difference)
39
37
77
Additional Performance Factors
Detection Limits
The MDL is often defined as the minimum concentration of a substance that can be measured and reported
with 99% confidence that the analyte concentration is higher than zero. An MDL is determined from
repeated analyses of a sample in a given matrix containing the analyte [12]. The reported MDL for the
RaPID Assay Kit was 0.5 ppm. An MDL, calculated from the data for the PE samples, was 1.5 ppm.
51
-------
Sample Throughput
Sample throughput is representative of the average amount of time required to extract the PCBs, perform
appropriate reactions, and to analyze the sample. SDFs sample throughput rate was relatively consistent
under both environmental conditions at 10 to 11 samples/h.
Cost Assessment
The purpose of this economic analysis is to provide an estimation of the range of costs for an analysis of
PCB-contaminated soil samples using the RaPID Assay System and a conventional analytical reference
laboratory method. The analysis was based on the results and experience gained from this demonstration,
costs provided by SDI, and representative costs provided by the reference analytical laboratories who
offered to analyze these samples. To account for the variability in cost data and assumptions, the economic
analysis was presented as a list of cost elements and a range of costs for sample analysis by the RaPID
Assay System and by the reference laboratory.
Several factors affected the cost of analysis. Where possible, these factors were addressed so that decision-
makers can independently complete a site-specific economic analysis to suit their needs. The following
categories were considered in the estimate:
• sample shipment costs,
• labor costs,
• equipment costs,
• waste disposal costs.
Each of these cost factors is defined and discussed in the following and serves as the basis for the estimated
cost ranges presented in Table 5-11. This analysis assumed that the individuals performing the analyses
were fully trained to operate the technology. SDI recommends that new users attend a training session that
they offer on the use of the RaPID Assay System. Sample acquisition and preanalytical sample preparation
costs, tasks common to both methods, are not included here.
RaPID Assay System Costs
Because the samples were analyzed on-site, no sample shipment charges were associated with the cost of
operating the RaPID Assay System. Labor costs included mobilization/demobilization, travel, per diem, and
on-site labor.
• Labor mobilization/demobilization: This cost element included the time for one person to
prepare for and travel to each site. The estimate ranged from 5 to 8 h, at a rate of $50/h.
Travel: This element was the cost for the analyst(s) to travel to the site. If the analyst is
located near the site, the cost of commuting to the site (estimated to be 50 miles at
$0.30/mile) would be minimal ($15). The estimated cost of an analyst traveling to the site
forthis demonstration ($1,000) included the cost of airline travel and rental car fees.
52
-------
• Per diem: This cost element included food, lodging, and incidental expenses and was
estimated ranging from zero (for a local site) to $150/day per analyst.
Rate: The cost of the on-site labor was estimated at a rate of $30 to $75/h, depending on the
required expertise level of the analyst. This cost element included the labor involved with
the entire analytical process, comprising sample preparation, sample management, analysis,
and reporting.
Table 5-11. Estimated analytical costs for PCB soil samples
RaPID Assay System
Strategic Diagnostics Inc.
Sample throughput rate: 10-11 samples per hour
Cost Category
Sample Shipment
Labor
Mobilization/demobilization
Travel
Per diem
Rate
Equipment
Mobilization/demobilization
Kit rental fee
Kit purchase price
Photometer purchase price
Training
Reagents/supplies
Waste Disposal
Cost (S)
0
250-400
15- 1,000 per analyst
0-150 per day per analyst
30 - 75 per hour per analyst
0-150
450 per week
1,665
595-3,985
<935
21 per sample
75-1,060
EPA SW-846 Method 8080/8081/8082
Reference Laboratory
Typical turn-around time: 14
Cost Category
Sample Shipment
Labor
Overnight shipping charges
Labor
Mobilization/demobilization
Travel
Per diem
Rate
Equipment
Mobilization/demobilization
Rental/purchase of system
Reagents/supplies
Waste Disposal
- 30 days
Cost (S)
100-200
50-150
Included a
Included
Included
44 - 239 per sample
included
Included
Included
Included
a"Included" indicates that the cost is included in the labor rate.
Equipment costs included mobilization/demobilization, rental fees or purchase of equipment, and the
reagents and other consumable supplies necessary to complete the analysis.
• Equipment mobilization/demobilization: This included the cost of shipping the equipment to
the test site. If the site is local, the cost would be zero. For this demonstration, the cost of
shipping equipment and supplies was estimated at $150.
• Rental/Purchase: The fee to rent the RaPID Assay System at the time of the demonstration
study was $450 per week. At the time of the demonstration, the cost of purchasing the
equipment accessory kit was $1,665. The price of the photometer was $595 or $3,985
(depending on the model).
53
-------
Reagents/Supplies: These items are consumable and are purchased on a per-sample basis.
At the time of the demonstration, the cost of the reagents and supplies needed to prepare
and analyze PCB soil samples using the RaPID Assay was $21 per sample. This cost
included the sample preparation supplies, assay supplies, and consumable reagents.
Waste disposal costs were estimated based on the 1997 regulations for disposal of PCB-contaminated
waste. Using the RaPID Assay, SDI generated approximately 20 Ib of vials containing soils and liquid
solvents (classified as solid PCB waste suitable for disposal by incineration) and approximately 20 Ib of
other solid PCB waste (i.e., used and unused soil, gloves, paper towels, ampules). The disposal costs for the
PCB solid waste by incineration at a commercial facility was estimated at $1.50/lb. For comparison, the
cost for PCB waste disposal at ETTP was estimated at $18/lb for solids. The RaPID Assay also generated
approximately 19 Ib of liquid waste. The cost for liquid PCB waste disposal at a commercial facility was
estimated at $0.25/lb; the cost at ETTA was estimated at $11/lb.
Reference Laboratory Costs
Sample shipment costs to the reference laboratory included overnight shipping charges, as well as labor
charges associated with the various organizations involved in the shipping process.
Labor: This cost element included all of the tasks associated with the shipment of the
samples to the reference laboratory. Tasks included packing the shipping coolers,
completing the chain-of-custody documentation, and completing the shipping forms.
Because the samples contained PCBs, the coolers were inspected by qualified personnel to
ensure acceptance with the U.S. Department of Transportation's shipping regulations for
PCBs. The estimate to complete this task ranged from 2 to 4 h at $50/h.
• Overnight Shipping: The overnight express shipping service cost was estimated to be $50
for one 50-lb cooler of samples.
The labor bids from commercial analytical reference laboratories who offered to perform the PCB analysis
for this demonstration ranged from $44/sample to $239/sample. The bid was dependent on many factors,
including the perceived difficulty of the sample matrix, the current workload of the laboratory, and the
competitiveness of the market. In this case, the wide range in bids may also be related to the cost of PCB
waste disposal in a particular laboratory's state. LAS Laboratories was awarded the contract to complete
the analysis as the lowest qualified bidder ($44/sample). This rate was a fully loaded analytical cost,
including equipment, labor, waste disposal, and report preparation.
Cost Assessment Summary
An overall cost estimate for the RaPID Assay System vs the reference laboratory was not made because of
the extent of variation in the different cost factors, as outlined in Table 5-11. The overall costs for the
application of each technology will also be based on the number of samples requiring analysis, the sample
type, and the site location and characteristics. Decision-making factors, such as turnaround time for results,
must also be weighed against the cost estimate to determine the value of the field technology vs the reference
laboratory.
54
-------
General Observations
The following are general observations regarding the field operation of the RaPID Assay System:
• The system was light, easily transportable, and rugged. It took about 1 h for the SDI team
to prepare to analyze samples on the first day of testing. While working at the outdoor site,
the SDI team completely disassembled their work station and brought everything inside at
the close of each day. It took the SDI team less than 1 hour each morning to prepare for
sample analysis.
• Three operators were used for the technology demonstration because of the number of
samples and working conditions, but the technology can be run by a single person. With
three SDI technologies (D TECH, EnviroGard, and RaPID Assay) being demonstrated,
SDI elected to work as a team to complete the analyses for each technology (as opposed to
three SDI people working with three different technologies).
• Operators generally require 2 to 4 h of training and should have a basic knowledge of field
techniques.
The measurement system (RPA-I RaPID Analyzer) required 120-V ac power.
Alternatively, it can be operated using a car battery.
• Using the RaPID Assay, SDI generated approximately 20 Ib of vials containing soils and
liquid solvents (classified as solid PCB waste suitable for disposal by incineration) and
approximately 20 Ib of other solid PCB waste (i.e., used and unused soil, gloves, paper
towels, ampules). The RaPID Assay also generated approximately 19 Ib of liquid waste.
Performance Summary
A summary of the performance characteristics of SDFs RaPID Assay System, presented previously in this
section, is shown in Table 5-12. The overall performance of the RaPID Assay System was characterized as
biased (with the direction of the bias dependent on Aroclor type) and precise for a given set of environmental
conditions.
55
-------
Table 5-12. Performance Summary for the RaPID Assay System
Feature/Parameter
Blank samples
Method detection limit
Precision
Accuracy
False positive results
False negative results
Comparison with reference laboratory results
Regulatory decision-making applicability
Sample throughput
Operator requirements
Power requirements
Cost
Hazardous waste generation
Performance Summary
Soils: No PCBs detected
Extracts: No PCBs detected
SDI specified: 0.5 ppm
Calculated: 1.5 ppm
Average RSD
PE soils: 26% (outdoor); 15% (chamber)
Environmental soils: 24% (outdoor); 13% (chamber)
Extracts: 14%
Average Percent Recovery
PE soils: 103% (both positive and negative Aroclor-dependent biases)
Extracts: 101%
Blank Soils: 0% (0 of 8 samples)
Blank Extracts: 0% (0 of 8 samples)
PE and Environmental Soils: 1% (2 of 192 samples)
Spiked Extracts: 0% (0 of 16 samples)
PE and Environmental Soil Samples
Percent difference: 40% of samples were ± 25 %D
Coefficients of determination (R2): 0.754 (all data)
0.716 (<125 ppm)
Extract Samples
Percent difference: 56% of samples were ± 25%D
Coefficient of determination (R2): 0 . 977
40 to 60 ppm PE and Environmental Soil Samples
precision: 21% average RSD
accuracy: 129% average recovery
comparability: 39% average percent difference
100 ug/100cm2 and 100 ug/100cm2 Extract Samples
precision: 12% average RSD
accuracy: 101% average recovery
comparability: 46% average difference
10-11 samples/h
Basic knowledge of field chemical techniques; 2-4 h technology-specific
training
120V AC or car battery (for RPA-I Analyzer)
Consumables: $21 per sample
Instrument: $450 weekly rental; $1,665 purchase accessories; $595 - 3,985
purchase photometer
Approximately 20 Ib of solid/liquid
(classified as solids suitable for disposal by incineration)
Approximately 19 Ib of liquid
Approximately 20 Ib of solid (used gloves, pipettes, paper towels, etc.)
Approximately 19 Ib of liquid waste (aqueous with trace methanol)
56
-------
Section 6
Technology Update and Representative Applications
Objective
In this section, SDI describes new technology developments that have occurred since the demonstration
activities. In addition, the developer has provided a list of representative applications where the SDI RaPID
Assay System has been or is currently being utilized.
Technology Update
Reconfiguration of Soil Extraction (Sample Preparation) Products
SDI is in the process of commercializing a common extraction kit for three of SDFs four remediation
immunoassay test kit product lines. The affected product lines include the EnviroGard, EnSys (not
demonstrated), and RaPID Assay Test Kit Systems. The new "Universal Extraction Kit" will be used with
assay kits of these three product lines, with extraction solvents or dilution reagents specifically formulated
to match individual kits available as kit component options where required. The new test kit configuration
will provide increased user convenience and simplify the product specification and ordering process without
affecting test kit analytical performance. Commercialization of the new Universal Extraction Kit was
initiated in April 1998. The new kits are not for use with the D TECH product line, which will continue to
use the existing SDI Soil Extraction Pac products.
Instrument Consolidation
Associated with the incorporation of several independently developed product lines into SDFs product
offerings, some consolidation of equipment and instrumentation is anticipated in the near future. This will
consist primarily of reducing the number of pipet types and photometers used to perform the assays. While
pipet types and procedures for pipeting reagents and reading and interpreting assay results may change
slightly, no effect of assay performance will result.
Please note that the previously listed product improvements are ongoing projects; therefore, the information
presented is subject to change.
57
-------
Representative Applications
In a 1997 report entitled, "Field Analytical and Site Characterization Technologies: Summary of
Applications" [13], the use of SDI immunoassay kits is documented at more than 30 remediation sites under
state or federal oversight. Contact information is provided for many of the immunoassay kit users at these
sites. The summary report can be obtained from the National Center for Environmental Publications and
Information (NCEPI). Hard copies of the report can be ordered, free of charge, by telephone, 513-
891-6561; by fax 513-891-6685; or through the NCEPI home page on the Internet at http://www.epa.gov
/ncepihom/. The summary report is available for viewing or downloading as a Portable Document Format
(PDF) file from the CLU-IN Internet Web site: http://clu-in.com/pubichar.htm.
Data Quality Objective Example
This application of SDFs RaPID Assay System is based on data quality objective (DQO) methods for
project planning advocated by the American Society for Testing and Materials (ASTM) [14, 15] and EPA
[16]. ORNL derived a DQO example from the performance results in Section 5. This example, which is
presented in Appendix E, illustrates the use of the RaPID Assay's performance data from the ETV
demonstration in the DQO process to select the number of samples and to quantify the action level for the
decision rule.
58
-------
Section 7
References
[1] Timothy Lawruk et al., "Quantitative Determination of PCBs in Soil and Water by a Magnetic
Particle-Based Immunoassay," Environmental Science & Technology 30(2), 695-700 (1996).
[2] M. D. Erickson, Analytical Chemistry of PCBs, 2nd ed., CRC Press/Lewis Publishers, Boca Raton,
Fla, 1997.
[3] "Polychlorinated Biphenyls (PCBs) Manufacturing, Processing, Distribution in Commerce, and Use
Prohibitions," Code of Federal Regulations, 40 CFRPt. 761, rev. 7, December 1994.
[4] Stability of Volatile Organics in Environmental Soil Samples, ORNL/TM-12128, Lockheed Martin
Energy Systems, Inc., Oak Ridge Natl. Lab., Oak Ridge, Tenn., November 1992.
[5] U.S. Environmental Protection Agency, "Organochlorine Pesticides and PCBs as Aroclors by Gas
Chromatography: Capillary Column Technique," Method 8081, in Test Methods for Evaluating Solid
Waste, Physical/Chemical Methods, Final Update II, U.S. EPA SW846, EPA, Washington, D.C.,
September 1994.
[6] Oak Ridge National Laboratory, Technology Demonstration Plan: Evaluation of Polychlorinated
Biphenyl (PCB) Field Analytical Techniques, Lockheed Martin Energy Research Corp., Oak Ridge
Natl. Lab., Oak Ridge, Tenn., July 1997.
[7] U.S. Environmental Protection Agency, Data Quality Objectives for Remedial Response Activities,
EPA 540/G-87/003, EPA, Washington D.C., March 1987.
[8] Lothar Sachs, Applied Statistics: A Handbook of Techniques, 2nd ed., Springer-Verlag, New York,
1984.
[9] G. W. Snedecor and William G. Cochran, Statistical Methods, Iowa State University Press, Ames,
Iowa, 1967.
[10] N. R. Draper and H. Smith, Applied Regression Analysis, 2nd ed., John Wiley & Sons Inc.,
New York, 1981.
[11] Walter Berger, Harry McCarty, and Roy-Keith Smith, Environmental Laboratory Data Evaluation,
Genium Publishing Corp., Schenectady, New York, 1996.
59
-------
[12] "Definition and Procedure for the Determination of the Method Detection Limit," Code of Federal
Regulations, CFR 40, Pt. 136, Appendix B, rev. 1.11.
[13] Field Analytical and Site Characterization Technologies: Summary of Applications, EPA-542-R-97-
011, Office of Solid Waste and Emergency Response, Washington, D. C., November 1997.
[14] American Society for Testing and Materials (ASTM), Standard Practice for Generation of
Environmental Data Related to Waste Management Activities Quality Assurance and Quality
Control Planning and Implementation, D5283-92, 1997.
[15] American Society for Testing and Materials (ASTM), Standard Practice for Generation of
Environmental Data Related to Waste Management Activities Development of Data Quality
Objectives, D5 792-95, 1997.
[16] U.S. Environmental Protection Agency, Guidance for Data Quality Assessment, EPA QA/G-9;
EPA/600/R-96/084, EPA, Washington, D.C., July 1996.
60
-------
Appendix A
Description of Environmental Soil Samples
61
-------
-------
Table A-l. Summary of soil sample descriptions
Location
Oak Ridge
Oak Ridge
Oak Ridge
Oak Ridge
Oak Ridge
Paducah
Portsmouth
Tennessee
Reference Soil
Request for
Disposal Drum #
(RFD)#
40022 02
40267 01
02
03
04
24375 01
02
03
43275 01
02
134555 03
97002 01
02
03
04
7515 858
1069
1096
1898
2143
2528
3281
538
940
4096
n/a n/a
Description
Soil from spill cleanup at the Y-12 Plant in Oak Ridge, Tennessee. This
soil is PCB-contaminated soil excavated in 1992.
Soil from the Elza Gate area, a DOE Formerly Utilized Sites Remedial
Action Program site in Oak Ridge, Tennessee. This soil is PCB-
contaminated soil that was excavated in 1992.
Catch-basin sediment from the K-71 1 area (old Powerhouse Area) at the
DOE East Tennessee Technology Park (formerly known as Oak Ridge
Gaseous Diffusion Plant) in Oak Ridge, Tennessee. This soil is PCB-
contaminated storm drain sediment that was excavated in 1 99 1 .
Soil from the K-25 Building area at the DOE East Tennessee
Technology Park (formerly known as Oak Ridge Gaseous Diffusion
Plant) in Oak Ridge, Tennessee. This soil is PCB-contaminated soil that
was excavated in 1993.
Soil from the K-707 area at the DOE East Tennessee Technology Park
(formerly known as Oak Ridge Gaseous Diffusion Plant) in Oak Ridge,
Tennessee. This soil is PCB-contaminated soil from a dike spillage that
was excavated in 1995.
Soil from the DOE Paducah Gaseous Diffusion Plant in Kentucky. This
soil is PCB-contaminated soil from a spill cleanup at the C-746-R
(Organic Waste Storage Area) that was excavated in 1989.
Soil from the DOE Portsmouth Gaseous Diffusion Plant in Ohio. This
soil is PCB-contaminated soil from a probable PCB oil spill into the
East Drainage Ditch that was excavated in 1 986.
Captina silt loam from Roane County, Tennessee; used as a blank in
this study (i.e., not contaminated with PCBs)
63
-------
-------
Appendix B
Characterization of Environmental Soil Samples
65
-------
-------
Table B-l. Summary of environmental soil characterization
Location
Oak Ridge
Paducah
Portsmouth
Sample
ID
101
102
103
104
105
106
107
108
109
110
111
112
126, 226
113,201
114,202
115,203
116,204
117,205
206
207
208
209
210
216
211
217
212
213
214
215
RFD
Drum # a
40022-02
40267-03
40267-01
40267-04
40267-0 IS"
24375-03
24375-01
40267-02
24375-02
43275-01
134555-03S"
43275-02
non-PCB soil
97002-04
97002-01
97002-03
97002-02
97002-02S b
7515-4096
7515-1898
7515-1096
7515-2143
7515-0940
7515-0538
7515-0538S"
7515-0538S"
7515-2528
7515-3281
7515-0858
7515-1069
Composition
% gravel
0
0.5
0.2
0.6
0.5
0.5
2.5
0.4
0.3
0
0.5
0.1
0
0
0.2
0.1
0.4
0
0.2
0.4
0
0.3
0.5
0.5
0.5
0
1.3
% sand
91.8
99.3
96.7
98.2
94.8
87.8
92.5
94.2
93.1
89.2
88.1
91.4
85.6
92.4
87.6
83.6
93.7
87.1
78.0
74.4
74.3
73.0
73.3
70.4
72.6
65.8
75.0
% silt + clay
8.2
0.2
3.1
1.2
4.7
11.7
5.0
5.4
6.6
10.8
11.4
8.5
14.4
7.6
12.2
16.3
5.8
12.9
21.8
25.2
25.7
26.7
26.3
29.1
26.8
34.2
23.7
Total Organic
Carbon
(mg/kg)
5384
13170
13503
15723
14533
19643
1196
9007
1116
14250
10422
38907
9249
1296
6097
3649
4075
3465
3721
3856
10687
7345
1328
5231
5862
6776
4875
PH
7.12
7.30
7.21
7.07
7.28
7.36
7.26
7.30
7.48
7.57
7.41
7.66
7.33
7.71
7.64
7.59
7.43
7.72
7.66
7.77
7.71
7.78
7.78
7.92
7.67
7.85
7.56
a Request for disposal drum number (see Table A-l).
b "S" indicates that the environmental soil was spiked with additional PCBs.
67
-------
-------
Appendix C
Temperature and Relative Humidity Conditions
69
-------
-------
Table C-l. Average temperature and relative humidity conditions during testing periods
Date
7/22/97
7/23/97
7/24/97
7/25/97
7/26/97
7/27/97
7/28/97
7/29/97
Outdoor Site
Average
Temperature
(°F)
85
85
85
80
85
80
79
b
Average
Relative Humidity
(%)
62
70
67
70
55
75
88
b
Chamber Site
Average
Temperature
(°F)
70 a
60 a
58
56
57
55
57
55
Average
Relative Humidity
(%)
38 a
58 a
66
54
51
49
52
50
a The chamber was not operating properly on this day. See discussion in Section 3.
b No developers were working outdoors on this day.
120 T-
7/22/97 7/23/97 7/24/97 7/25/97 7/26/97 7/27/97 7/28/97
Figure C-l. Summary of temperature conditions for outdoor site.
71
-------
120
100 --
60 -
40
20 --
7/22/97 7/23/97 7/24/97 7/25/97 7/26/97 7/27/97 7/28/97
Figure C-2. Summary of relative humidity conditions for the outdoor site.
7/22/97 7/23/97 7/24/97 7/25/97 7/26/97 7/27/97 7/28/97 7/29/97
Figure C-3. Summary of temperature conditions for chamber site.
72
-------
c*u -
80
70
:=• 60 -
0^
£
2 50 -
E
3
X
at 40 -
_>
IS
"3
U. 30 -
20
10
n
_
_
— i —
—
_
_
— i —
"
_
_
—i —
-
— i —
—
i— i
—i —
_
PI
— i —
—
— i —
~
—
_
—
_
• High RH
• Low RH
DAvg RH
J
7/22/97 7/23/97 7/24/97 7/25/97 7/26/97 7/27/97 7/28/97 7/29/97
Figure C-4. Summary of relative humidity conditions for chamber site.
73
-------
-------
Appendix D
RaPID Assay System for PCB Analysis
PCB Technology Demonstration Sample Data
75
-------
-------
Legend for Appendix D Tables
Table Heading
Obs
Sample ID
Rep
RaPID Result
Ref Lab Result
Reference Aroclor
Type
Order
Definition
Observation
Sample identification
101 to 126 = outdoor site soil samples
127 to 130 = outdoor site extract samples
201 to 226 = chamber site soil samples
227 to 230 = chamber site extract samples
Replicate of sample ID (1 through 4)
RaPID Assay System measured PCB concentration (ppm)
LAS reference laboratory measured PCB concentration (ppm)
Values with "<" are samples that the reference laboratory reported as
"< reporting detection limit"
Aroclor(s) identified by the reference laboratory
Sample = environmental soil
1242, 1248, 1254, 1260 = Aroclor in PE samples
Blank = non-PCB-contaminated sample
Order of sample analysis (started with 2001-21 16, then 1001-1 116)
77
-------
Table D-l. RaPID Assay System PCB technology demonstration soil sample data
Obs
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
31
32
33
34
35
36
37
38
39
40
Sample
ID
101
101
101
101
102
102
102
102
103
103
103
103
104
104
104
104
105
105
105
105
106
106
106
106
107
107
107
107
108
108
108
108
109
109
109
109
110
110
110
110
Rep
1
2
3
4
1
2
3
4
1
2
3
4
1
2
3
4
1
2
3
4
1
2
3
4
1
2
3
4
1
2
3
4
1
2
3
4
1
2
3
4
RaPID
Result
(ppm)
<0.5
<0.5
<0.5
0.6
3.2
3.5
3.2
3.8
5.8
3.8
4.3
6.4
25.6
9.2
9.3
17.2
74.4
107.2
110.0
74.0
>200.0
>200.0
>200.0
>200.0
1.8
1.5
2.2
2.0
2.5
2.9
4.5
3.0
4.1
3.9
2.8
2.6
26.4
30.8
44.0
22.8
Ref Lab
Result
(ppm)
0.6
0.4
0.5
0.5
2.2
2.1
1.7
2.5
3.0
2.4
2.0
1.6
6.8
6.0
14.8
9.9
49.7
84.1
50.6
53.2
269.6
255.9
317.6
649.6
1.0
1.6
1.2
1.2
1.7
2.0
1.7
1.9
1.5
2.1
1.8
2.4
<490.0
<99.0
<66.0
<98.0
Reference
Aroclor
1254
1254
1254
1254
1254
1254
1260
1260
1254
1254
1260
1260
1260
1254
1254
1254
1260
1260
1260
1260
1254
1254
1254
1254
1254
1254
1254
1254
1254
1254
1254
1254
1254
1254
1254
1254
Non- Detect
Non- Detect
Non- Detect
Non- Detect
Type
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Order
1079
1069
1020
1025
1016
1055
1022
1024
1021
1059
1103
1049
1029
1094
1004
1028
1099
1031
1040
1005
1001
1102
1088
1014
1071
1043
1042
1050
1018
1077
1046
1101
1053
1058
1067
1082
1066
1091
1009
1097
78
-------
Obs
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
67
68
69
70
71
72
73
74
75
76
77
78
79
80
Sample
ID
111
111
111
111
112
112
112
112
113
113
113
113
114
114
114
114
115
115
115
115
116
116
116
116
117
117
117
117
118
118
118
118
119
119
119
119
120
120
120
120
Rep
1
2
3
4
1
2
3
4
1
2
3
4
1
2
3
4
1
2
3
4
1
2
3
4
1
2
3
4
1
2
3
4
1
2
3
4
1
2
3
4
RaPID
Result
(ppm)
38.8
47.6
67.2
42.8
97.2
125.2
143.6
129.2
1.3
1.3
1.4
1.7
0.9
1.2
0.7
1.6
12.0
13.4
15.4
19.4
119.0
58.5
69.3
56.6
296.0
230.6
205.6
163.3
1.7
1.5
1.3
1.6
9.0
11.8
23.1
16.5
4.8
6.3
8.4
5.7
Ref Lab
Result
(ppm)
44.5
36.0
39.3
35.1
<66.0
<200.0
<130.0
<200.0
0.7
1.1
0.6
1.9
1.1
1.2
1.3
1.7
14.9
12.4
15.0
16.9
41.4
41.2
48.5
34.0
431.6
406.3
304.7
392.8
2.1
1.9
0.7
1.6
21.2
17.2
17.4
24.4
4.5
4.0
6.3
5.0
Reference
Aroclor
1254
1254
1254
1254
Non- Detect
Non- Detect
Non- Detect
Non- Detect
1260
1260
1260
1248/1260
1260
1260
1260
1260
1248
1016
1248
1248
1248
1016
1248
1016
1016
1016
1016
1016
1248
1016
1248
1248
1016
1248
1248
1248
1254
1254
1254
1254
Type
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
1248
1248
1248
1248
1248
1248
1248
1248
1254
1254
1254
1254
Order
1063
1052
1083
1070
1096
1081
1032
1078
1027
1017
1036
1038
1045
1100
1011
1104
1089
1092
1057
1061
1035
1015
1047
1075
1034
1048
1003
1076
1026
1013
1044
1023
1093
1006
1019
1010
1074
1073
1039
1060
79
-------
Obs
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
114
115
116
117
118
119
120
Sample
ID
121
121
121
121
122
122
122
122
123
123
123
123
124
124
124
124
125
125
125
125
126
126
126
126
201
201
201
201
202
202
202
202
203
203
203
203
204
204
204
204
Rep
1
2
3
4
1
2
3
4
1
2
3
4
1
2
3
4
1
2
3
4
1
2
3
4
1
2
3
4
1
2
3
4
1
2
3
4
1
2
3
4
RaPID
Result
(ppm)
60.4
44.8
100.8
77.2
6.0
5.8
6.7
8.4
33.3
82.1
64.6
39.0
2.5
2.7
4.0
2.5
86.4
76.8
70.4
64.4
<0.5
<0.5
<0.5
<0.5
0.8
0.7
1.0
0.7
0.9
1.0
0.6
1.0
20.7
14.9
19.2
15.9
97.6
78.0
86.6
91.7
Ref Lab
Result
(ppm)
58.7
55.7
53.2
50.9
12.2
10.9
11.3
10.0
59.2
56.9
66.8
57.5
1.8
1.4
1.9
1.8
32.0
41.3
46.0
32.2
<0.1
<0.1
<0.2
<1.3
1.0
1.0
1.1
0.6
1.4
1.6
1.2
1.5
14.0
12.8
16.2
12.4
43.1
45.3
41.0
47.7
Reference
Aroclor
1254
1254
1254
1254
1260
1260
1260
1260
1260
1260
1260
1260
1254
1260
1254
1254
1254
1254
1254
1260
Non- Detect
Non- Detect
Non- Detect
Non- Detect
1016/1260
1016/1260
1016/1260
1260
1260
1260
1260
1260
1248
1248
1248
1248
1248
1248
1248
1248
Type
1254
1254
1254
1254
1260
1260
1260
1260
1260
1260
1260
1260
1254/1260
1254/1260
1254/1260
1254/1260
1254/1260
1254/1260
1254/1260
1254/1260
Blank
Blank
Blank
Blank
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Order
1065
1054
1012
1002
1090
1072
1084
1087
1095
1030
1041
1062
1007
1086
1037
1085
1080
1098
1064
1051
1008
1033
1068
1056
2037
2056
2006
2104
2040
2025
2050
2002
2022
2102
2045
2100
2068
2039
2019
2036
80
-------
Obs
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
156
157
158
149
150
151
152
153
154
155
156
157
158
159
160
Sample
ID
205
205
205
205
206
206
206
206
207
207
207
207
208
208
208
208
209
209
209
209
210
210
210
210
211
211
211
211
212
212
212
212
213
213
213
213
214
214
214
214
Rep
1
2
3
4
1
2
3
4
1
2
3
4
1
2
3
4
1
2
3
4
1
2
3
4
1
2
3
4
1
2
3
4
1
2
3
4
1
2
3
4
RaPID
Result
(ppm)
185.4
210.4
206.1
190.1
1.4
1.3
1.5
1.9
23.6
16.8
22.0
14.8
29.2
28.4
33.2
25.6
40.0
46.8
56.4
50.5
75.6
92.8
69.2
85.6
>200
>200
>200
>200
3.0
4.1
4.0
4.7
6.2
5.5
5.8
5.4
33.2
28.8
32.4
30.0
Ref Lab
Result
(ppm)
3305.0
538.7
457.0
483.3
2.9
1.1
1 . 1
2.5
17.8
14.3
21.6
21.6
42.0
27.7
24.0
28.4
32.7
79.3
11.0
37.9
123.2
61.5
84.1
85.5
387.8
581.4
330.0
318.7
3.8
3.9
4.3
0.8
6.9
7.3
7.8
10.5
26.0
25.6
29.1
20.2
Reference
Aroclor
1016/1260
1016
1016
1016
1260
1260
1016/1260
1260
1260
1260
1260
1254
1260
1016/1260
1254
1260
1260
1260
1260
1260
1260
1260
1260
1260
1254
1254
1254
1254
1260
1260
1260
1260
1260
1260
1260
1260
1260
1260
1260
1260
Type
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Order
2038
2001
2030
2073
2083
2008
2087
2027
2043
2029
2062
2009
2061
2103
2096
2054
2059
2094
2041
2021
2090
2047
2015
2065
2075
2055
2084
2042
2086
2032
2093
2095
2076
2057
2028
2089
2018
2070
2016
2007
81
-------
Obs
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
Sample
ID
215
215
215
215
216
216
216
216
217
217
217
217
218
218
218
218
219
219
219
219
220
220
220
220
221
221
221
221
222
222
222
222
223
223
223
223
224
224
224
224
Rep
1
2
3
4
1
2
3
4
1
2
3
4
1
2
3
4
1
2
3
4
1
2
3
4
1
2
3
4
1
2
3
4
1
2
3
4
1
2
3
4
RaPID
Result
(ppm)
30.8
30.0
26.8
31.6
55.2
53.2
69.2
50.8
>200.0
>200.0
>200.0
>200.0
0.9
1.0
1.8
1.5
11.7
11.2
10.7
9.5
5.8
4.2
4.6
7.0
62.4
74.4
62.8
55.2
6.4
6.1
6.2
6.2
52.8
41.5
43.1
40.5
2.6
2.4
2.4
1.7
Ref Lab
Result
(ppm)
25.1
24.1
26.2
31.2
151.6
47.0
54.3
64.0
886.7
549.8
542.8
1913.3
2.8
2.4
2.6
2.6
22.4
26.0
29.4
15.2
8.5
4.9
4.7
5.2
32.0
44.1
43.8
59.6
13.2
12.4
12.7
12.7
56.6
50.3
49.9
66.4
2.2
1.2
1.4
2.1
Reference
Aroclor
1260
1260
1260
1016/1260
1260
1260
1260
1260
1254
1254
1254
1016/1260
1248
1248
1248
1248
1248
1016
1248
1248
1254
1254
1254
1254
1016/1260
1016/1260
1254
1254
1260
1260
1260
1260
1260
1260
1260
1260
1254
1260
1260
1254
Type
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
1248
1248
1248
1248
1248
1248
1248
1248
1254
1254
1254
1254
1254
1254
1254
1254
1260
1260
1260
1260
1260
1260
1260
1260
1254/1260
1254/1260
1254/1260
1254/1260
Order
2014
2020
2034
2097
2017
2079
2010
2063
2099
2067
2024
2074
2049
2091
2046
2066
2013
2004
2053
2085
2081
2052
2080
2072
2064
2088
2058
2069
2078
2031
2051
2077
2044
2005
2092
2011
2035
2033
2012
2098
82
-------
Sample RaPID Ref Lab Reference
Obs ID Rep Result Result Aroclor Type Order
(ppm) (ppm)
201 225 1 65.2 56.4 1260 1254/1260 2060
202 225 2 83.6 36.5 1016/1260 1254/1260 2026
203 225 3 87.2 32.1 1260 1254/1260 2048
204 225 4 73.2 146.0 1254 1254/1260 2003
205 226 1 <0.5 <0.1 Non-Detect Blank 2071
206 226 2 <0.5 <0.8 Non-Detect Blank 2082
207 226 3 <0.5 <0.1 Non-Detect Blank 2101
208 226 4 <0.5 <0.1 Non-Detect Blank 2023
83
-------
Table D-2. RaPID Assay System PCB technology demonstration extract sample data
DBS
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
Sample
ID
130
130
130
130
131
131
131
131
132
132
132
132
230
230
230
230
231
231
231
231
232
232
232
232
Rep
1
2
3
4
1
2
3
4
1
2
3
4
1
2
3
4
1
2
3
4
1
2
3
4
RaPID
Result
(ppm)
13.0
11.9
11.5
13.5
101.2
101.2
94.4
90.8
<0.5
<0.5
<0.5
<0.5
9.5
10.0
9.7
8.6
86.0
91.2
79.0
89.0
<0.3
<0.3
<0.3
<0.3
Ref Lab
Result
(ppm)
16.4
10.9
10.3
10.7
67.1
57.1
62.8
68.2
<0.1
<0.1
<0.1
<0.1
9.8
10.4
7.6
7.9
55.2
55.0
61.3
59.1
<0.1
<0.1
<0.1
<0.1
Reference
Aroclor
1016
1016
1016
1016
1254
1254
1254
1254
Non- Detect
Non- Detect
Non- Detect
Non- Detect
1016
1016
1016
1016
1254
1254
1254
1254
Non- Detect
Non- Detect
Non- Detect
Non- Detect
Type
1242
1242
1242
1242
1254
1254
1254
1254
blank
blank
blank
blank
1242
1242
1242
1242
1254
1254
1254
1254
blank
blank
blank
blank
Spikea
(ppm)
10
10
10
10
100
100
100
100
0
0
0
0
10
10
10
10
100
100
100
100
0
0
0
0
Order
1111
1114
1108
1109
1105
1112
1113
1115
1107
1106
1110
1116
2114
2112
2109
2115
2111
2116
2110
2105
2113
2108
2107
2106
"Nominal spike concentration of the extract sample prepared by ORNL.
84
-------
Table D-3. Corrected reference laboratory data
Error
Transcription
Calculation
Interpretation
Sample ID
106
130
205
207
210
118
119
209
214
219
101°
101a
107
109
1136
1136
119
127
201
219
Reported Result
(ppm)
<490
5.6
32,000
180
160
3.6
4.3
2.3
43.0
29.0
<0.7
<0.7
<1.3
18.0
<0.9
<1.0
18.0
7.2
< 1.0
21.0
Corrected Result
(ppm)
255.9
10.3
3,305.0
17.8
123.2
2.1
17.4
37.9
26.0
22.4
0.5
0.6
1.2
1.5
0.6
0.7
21.2
10.9
0.6
26.0
a Two of four measurements in sample ID 101 were corrected.
b Two of four measurements in sample ID 113 were corrected.
85
-------
86
-------
Appendix E
Data Quality Objective Example
87
-------
-------
Disclaimer
The following hypothetical example serves to demonstrate how the information provided in this report may
be used in the DQO process. This example serves to illustrate the application of quantitative DQOs to a
decision process, but cannot attempt to provide a thorough education in this topic. Please refer to other
educational or technical resources for further details. In addition, since the focus of this report is on the
analytical technology, this example makes the simplifying assumption that the contents of these drums will
be homogeneous. In the real world, however, this assumption is seldom valid, and matrix heterogeneity
constitutes a source of considerable uncertainty which must be adequately evaluated if the overall certainty
of a site decision is to be quantified.
Background and Problem Statement
An industrial company discovered a land area contaminated with PCBs from an unknown source. The
contaminated soil was excavated into waste drums. Preliminary characterization determined that the PCB
concentration in a single drum was homogenous, but PCB concentrations varied greatly from drum to drum.
The company's DQO team was considering the use of SDFs RaPID to measure the PCB concentration in
each drum. The DQO team decided that drums will be disposed of by incineration if the PCB concentration
is greater than or equal to 50 ppm ("hot"). A concentration of 50 ppm is the Toxic Substances Control Act
(TSCA) regulatory threshold (RT) for this environmental problem. Those drums with PCB concentrations
less than 50 ppm will be put into a landfill because incineration of soil is very expensive. With regulator
agreement, the DQO team determined that a decision rule for disposal would be based on the average
concentration of PCBs in each drum.
General Decision Rule
If average PCB concentration < than action level, then send the soil drum to the landfill.
If average PCB concentration action level, then send the soil drum to the incinerator.
Data Quality Objective Goals
EPA's Guidance for Data Quality Assessment [16] states in Section 1.2: "The true condition that occurs
with the more severe decision error . . . should be defined as the null hypothesis." The team decided that the
more severe decision error would be for a drum to be erroneously sent to a landfill if the drum's PCB
concentration actually exceeded 50 ppm. Therefore, the null hypothesis is constructed to assume that a
drum's true PCB concentration is greater than 50 ppm; and as a "hot" drum, it would be sent to an
incinerator. Drums would be sent to the landfill only if the null hypothesis is rejected and it is concluded that
the "true" average PCB concentration is less than 50 ppm.
With the null hypothesis defined in this way, a false positive decision is made when it is concluded that a
drum contains less than 50 ppm PCBs (i.e., the null hypothesis is rejected), when actually the drum is "hot"
(i.e., the null hypothesis is true). The team required that the error rate for sending a "hot" drum to the
landfill (i.e., the false positive error rate for the decision) could not be more than 5%. Therefore, a sufficient
89
-------
number of samples must be taken from each drum so that the false positive decision error rate (FP) is 0.05
(or less) if the true drum concentration is 50 ppm. This scenario represents a 5% chance of sending a drum
containing 50 ppm or more of PCBs to the landfill.
The DQO team did not want to send an excessive number of drums to the incinerator if the average PCB
concentration was less than 50 ppm because of the expense. In this situation, a false negative decision is
made when it is concluded that a drum is "hot" (i.e., the null hypothesis is not rejected), when in actuality,
the drum contains soil with less than 50 ppm PCBs (i.e., the null hypothesis is actually false). After
considering the guidelines presented in Section 1.1 of EPA's Guidance for Data Quality Assessment [16] for
developing limits on decision errors, the team selected the false negative decision error rate (FN) to be 0.10
if the true drum concentration was 40 ppm. That is, there would be a 10% probability (Pr) of sending a
drum to the incinerator (denoted as Pr[Take Drum to Incinerator]) if the true PCB concentration for a drum
was 40 ppm.
Permissible FP and FN Error Rates and Critical Decision Points
FP: Pr[Take Drum to Landfill] < 0.05 when true PCB concentration = 50 ppm
FN: Pr[Take Drum to Incinerator] < 0.10 when true PCB concentration = 40 ppm
Use of Technology Performance Information to Implement the Decision Rule
Technology performance information is used to evaluate whether a particular analytical technology can
produce data of sufficient quality to support the site decision. Because the DQO team is considering the use
of the RaPID Assay Kit, the performance of this technology (as reported in this ETV report) was used to
assess its applicability to this project. Two questions arise:
1. How many samples are needed from a single drum to permit a valid estimation of the true average
concentration of PCBs in the drum to the specified certainty? Recall that the simplifying assumption
was made that the PCB distribution throughout the soil within a single drum is homogeneous, and thus,
matrix heterogeneity will not contribute to overall variability. The only variability, then, to be
considered in this example is the variability in the RaPID Assay's method, which is determined by
precision studies.
2 What is the appropriate action level (AL) for using the RaPID Assay Kit to make decisions in the
field? After the required number of samples have been collected from a drum and analyzed, the results
are averaged together to get an estimate of the "true" PCB concentration of the drum. When using the
RaPID Assay Kit, what is the value (here called "the action level for the decision rule") to which that
average is compared to decide if the drum is "hot" or not? This method-specific or site-specific action
level is derived from evaluations of the method's accuracy using an appropriate quality control
regimen.
90
-------
RaPID Assay Accuracy
In the ETV study, the overall data generated by SDFs
RaPID Assay Kit was slightly biased when compared
with the certified PCB values for the PE samples,
which had concentrations ranging from 0 to 50 ppm.
As summarized in Table 5-5, environmental conditions
(temperature and humidity) showed no effect on the
accuracy achieved by the RaPID Assay Kit; therefore,
the data from both sites can be plotted to create Figure
E-l. Similarly, environmental soil data in the
concentration range of 0 to 60 ppm, which was
generated by the reference laboratory, can be plotted
against the corresponding SDI results to create Figure
E-2.
The lines on these graphs depict the lines of best fit
(with 95% confidence intervals), and can be used to
predict the results that the RaPID Assay would
produce for a particular "true" PCB concentration.
The arrows on the plots demonstrate how to quickly
estimate this. The same task can be performed
mathematically, if desired, by using equations which to
define the lines in Figures E-l and E-2.
Table E-l presents the prediction performance of the
linear model for the SDI RaPID Assay kit at the 40 and
50 ppm critical decision points. Table E-l shows that
the RaPID Assay results tend to be slightly biased high
within this range of PCB values.
The 95% confidence intervals (CIs) for the
environmental soils and PE data sets overlap for a
single nominal value. At a nominal value of 40 ppm, for
example, the 95% CI predicted for RaPID's results
from comparison with the PE sample set is 43 to 55
ppm, which overlaps the 95% CI predicted from the
environmental soils sample set (53 to 65 ppm). Since
these ranges overlap, they can be considered to be
statistically similar. Although this similar
Certified PCB Concentration (ppm)
Figure E-l. A linear model for predicting RaPID Assay
PCB concentrations from certified PCB concentrations
with 95% confidence intervals (dashed lines).
10 20 30
40 50 60
Reference Laboratory PCB Concentration (ppm)
Figure E-2. A linear model for predicting RaPID Assay
PCB concentrations from reference laboratory PCB
concentrations for environmental soils up to 60 ppm with
95% confidence intervals (dashed lines).
91
-------
Table E-l. Predicting SDI's RaPID Assay Kit results from graphs of performance data
Soil Matrix
(concentration range)
Certified PE Soil (0 to 50 ppm)
from Figure E-l
Environmental Soils (0 to 60 ppm)
from Figure E-2
Critical Decision Point
40 ppm
50 ppm
40 ppm
50 ppm
Predicted SDI Results (95% CI)
49 (43 - 55)
63 (65 - 70)
59 (53-65)
74 (66 - 82)
performance with a variety of soil types in the ETV study increases the confidence that the RaPID Assay
Kit may perform equivalently on other soil types, site-specific quality control is always advisable to detect
unforeseen matrix effects or interferences.
After considering the matrix characteristics and Aroclor types of the PE and environmental soils in this ETV
study and comparing these to the soils that had been placed into the drums that the DQO team planned to
test, the DQO team reasoned that there was a strong likelihood that the SDI Kit would perform in a similar
manner for their testing. However, they decided to design an appropriate QC regimen (which included
matrix spikes, split samples sent for confirmatory laboratory analysis, and duplicates) to ensure that the
SDI Kit was performing as expected under their site-specific conditions.
Additionally, to address the possibility that these QC samples would reveal that the kit's performance was
different from that expected, the team created a backup plan (which became part of the Sampling and
Analysis Plan) that would permit them to document and account for deviations in expected performance.
The team reasoned that, even if the kit's performance varied somewhat from their expectations, being able to
account for any deviations would permit them to produce verifiable and defensible data that might still be
able to support decision-making at the site, without the need for resampling.
The DQO team also decided not to compensate for the positive bias they expected to see based on the ETV
study. Until the members of the team had more field experience with the kit's performance under a variety of
conditions, they believed that the bias would provide a margin of safety in their project that would be
valuable for regulator and stakeholder acceptance. Therefore, the critical decision points of 40 and 50 ppm
(which correspond to CFN and RT in the following equations) were selected for use with the SDI RaPID
Assay Kit when calculating the sample size and action level to meet the project DQO goals.
Determining the Number of Samples
With the critical decision points selected for the kit, the team could determine the number of samples needed
from each drum to calculate its "true" average PCB concentration. For a homogeneous matrix, the number
of samples required depends on the precision of the analytical method.
The ETV demonstration results indicated the standard deviation increased with concentration levels and that
the RSD would be a more appropriate precision measurement than the standard deviation. For SDI's RaPID
92
-------
Assay measurements, the mean RSDs were 15% for controlled environmental conditions and 26% for
outdoor conditions (see Table 5-4). The DQO team could set up a controlled environment at the remediation
site with additional effort and cost. The DQO team decided to calculate the number of samples for both
environmental conditions to judge whether the additional effort would be cost effective. A formula (Equation
E-l) is provided in EPA's Guidance for Data Quality Assessment [16] (pp. 3.2-3, Box 3.2-1) that can be
adapted to this example for calculating the number of samples required to meet the false positive and false
negative error rate for the decision. This formula uses a constant SD for the analytical method's precision
but can be modified to use RSD by dividing the numerator and denominator by (RT)2 and multiplying by
(100%)2, as shown in Equation E-2. The final form of the formula appears as Equation E-3.
(SD)2 ( Z^Fp + Z^m )2
(E_2)
RSD ( Z, ™ + Z,
, „,, ?
'
-------
(15%)2 (1.645 + 1.282)2 m,.n,A^2 ^0-7 , A ;•*•
n = — — + (0.5)(1.645)z = 6.2 7 chamber conditions
(20%)2
(26%)2 (1.645 + 1.282)2 /n ,W1 ,,«2 i«o i* j j- •
n = '—± '— + (0.5)(1.645)2 = 15.8 16 outdoor conditions
(20%)2
Note that, to be conservative, the sample size was rounded up to the next integer. The DQO team was able
to compare the cost of an additional 9 samples per drum vs the cost of establishing a controlled environment
(such as an air-conditioned camper). Because of the large number of drums to be tested during a hot
summer, the DQO team opted for a controlled analytical environment to reduce the total number of samples.
The action levels for the decision rule will then be calculated based on taking 7 samples from each drum.
Determining the Action Level
Since the DQO team decided not to compensate for the kit's positive bias, seven sample results from each
drum will be averaged (arithmetic mean) to produce an estimate of the drum's "true" PCB concentration.
This average PCB concentration will be compared with the AL for the decision rule. The AL for the decision
rule is calculated based on regulation-driven requirements (the TSCA regulatory threshold of 50 ppm) and
on controlling the FP established in the DQO process. Recall that the team set the permissible FP error rate
ASTM D5283-92 [13] shows the formula for the AL based on a constant SD over a relevant concentration
range (Equation E-4). Since the RaPID Assay Kit did not produce data with a constant standard deviation,
this formula must be adapted to this example by using the relationship between SD and RSD, which is SD =
(Concentration) x RSD/100%. Thus Equation E-4 becomes Equation E-5, and the regulatory threshold (RT
= 50 ppm) is the concentration used in the formula.
SD
AL = RT - Zv_Fp x (E.4)
AT VT 7
AL = RT - Zv_Fp x (E.5)
100% x^
The AL for the decision rule using SDI's RaPID Assay Kit to satisfy a 5% FP and a 10% FN for seven
samples under chamber conditions is
94
-------
AL = 50ppm - (1.645) x —EE. 2 = 45,3ppm chamber conditions.
100%x/7
Note that if the analytical testing was done under uncontrolled environmental conditions with a sample size
of 16, the AL for the decision rule would become
AL = 50 ppm - (1.645) x —EE. 2 = 44.7ppm outdoor conditions.
100%x/L6
Decision Rule for 5% FP and 10% FN
If the average PCB concentration of 7 (or 16) random soil samples on a drum is less than 45.3 ppm (or
44.7 ppm), then send the drum to the landfill.
If the average PCB concentration of 7 (or 16) random soil samples on a drum is greater or equal to
45.3 ppm (or 44.7), then consider the drum "hot," and send it to the incinerator.
The decision performance curve (see ref 16) calculates the probability of sending a drum to the incinerator
for different values of true PCB concentration in a drum. Figure E-3 illustrates that the decision
performance curves for the controlled and uncontrolled environments both have the value of Pr[Take Drum
to Incinerator] = 0.95 for True = 50 ppm. This indicates that the decision rule meets the FP requirement of
5% for both environmental conditions. The Pr[Take Drum to Incinerator] = 0.02 and 0.04 at True = 40 ppm
for the controlled (RSD = 15%) and uncontrolled environment (RSD = 25%), respectively. These false
negative probabilities are better than the FN = 10% that the DQO team had specified. This improved
performance is caused by rounding up the number of samples to the next integer in the calculation of the
number of samples required.
Alternative FP Error Rate
Because of random sampling and analysis error, there is always some chance that analytical results will not
accurately reflect the true nature of a decision unit (such as a drum, in this example). Often, 95% certainty
(a 5% FP) is customary and sufficient to meet stakeholder comfort. But suppose that the DQO team wanted
to be even more cautious about limiting the possibility that a drum might be sent to a landfill when its true
value is 50 ppm. If the team wanted to be 99% certain that a drum was correctly sent to a landfill, the
following discussion describes how changing the FP from 5% to 1% would affect the decision rule. Using
FP = 0.01, the sample sizes are calculated to be 9 and 23, and the ALs for the decision rule are 44.2 and
95
-------
43.7 ppm for the controlled and uncontrolled
environments, respectively. The decision
performance curves has the value of Pr [Take
Drum to Incinerator] = 0.99 for True =
50 ppm. This indicates that the decision rule
meets the FP of 1%. The Pr [Take Drum to
Incinerator] = 0.02 and 0.04 at True = 40 ppm
for the controlled and uncontrolled
environment, respectively. These probabilities
are better than the FN of 10% than the DQO
team had specified. This improved
performance is caused by rounding up the
number of samples to the next integer in the
calculation of number of samples required.
The decision rule for the lower FP would be:
3
E
D
^
Q
------- |