United States
Environmental Protection
Agency
Office of Research and
Development
Washington, D.C. 20460
EPA/600/R-97/149
December 1997
Environmental Technology
Verification Report
Field Portable Gas
Chromatograph/Mass
Spectrometer
Bruker-Franzen Analytical Systems,
Inc. EM640™
E
:SERDP
Environmental Technology
Verification Program
-------
-------
Environmental Technology
Verification Report
Field Portable Gas Chromatograph/ Mass
Spectrometer
Bruker-Franzen Analytical Systems, Inc.
EM640™
Prepared By
Wayne Einfeld
Susan F. Bender
Michael R. Keenan
Steven M. Thornberg
Michael M. Hightower
Environmental Characterization
and Monitoring Department
Sandia National Laboratories
Albuquerque, New Mexico
Sponsored by
U.S. ENVIRONMENTAL PROTECTION AGENCY
OFFICE OF RESEARCH AND DEVELOPMENT
NATIONAL EXPOSURE RESEARCH LABORATORY
ENVIRONMENTAL SCIENCES DIVISION
LAS VEGAS, NEVADA
-------
Notice
The information in this document has been funded wholly or in part by the U.S. Environmental Protection
Agency (EPA) under an Interagency Agreement number DW89936700-01-0 with the U.S. Department of
Energy's Sandia National Laboratories. This verification effort was supported by the Consortium for Site
Characterization Technology, a pilot operating under the EPA's Environmental Technology Verification
(ETV) Program. It has been subjected to the Agency's peer and administrative review, and it has been
approved for publication as an EPA document. Mention of corporation names, trade names, or commercial
products does not constitute endorsement or recommendation for use of specific products.
In 1995, the U. S. Environmental Protection Agency established the Environmental Technology
Verification Program. The purpose of the Program is to promote the acceptance and use of innovative
environmental technologies. The verification of the performance of the Bruker-Franzen Analytical
Systems, Inc. EM640™ field transportable gas chromatograph/mass spectrometer (GC/MS) system
represents one of the first attempts at employing a testing process for the purpose of performance
verification. One goal of this process is to generate accurate and credible data that can be used to verify the
characteristics of the technologies participating in the program. This report presents the results of our first
application of the testing process. We learned a great deal about the testing process and have applied what
we learned to improve upon it. We expect that each demonstration will serve to improve the next and that
this project merely represents the first step in a complex process to make future demonstrations more
efficient, less costly, and more useful.
-------
UNITED STATES ENVIRONMENTAL PROTECTION AGENCY
Office of Research and Development
Washington, D.C. 20460
ENVIRONMENTAL TECHNOLOGY VERIFICATION PROGRAM
VERIFICATION STATEMENT
TECHNOLOGY TYPE: FIELD PORTABLE GAS CHROMATOGRAPH/MASS
SPECTROMETER
APPLICATION: MEASUREMENT OF VOLATILE ORGANICS IN SOIL, WATER, AND
SOIL GAS
TECHNOLOGY NAME: EM640™
COMPANY: BRUKER-FRANZEN ANALYTICAL SYSTEMS, INC.
ADDRESS: 19 FORTUNE DRIVE, MANNING PARK
BILLERICA, MASSACHUSETTS 01821
PHONE: (508) 667-9580
The U.S. Environmental Protection Agency (EPA) has created a program to facilitate the deployment of innovative
environmental 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 is intended to assist
and inform those involved in the design, distribution, permitting, and purchase of environmental technologies. This
verification statement provides a summary of the demonstration and results for the Bruker-Franzen Analytical Systems,
Inc. EM640™ field portable gas chromatograph/mass spectrometer (GC/MS) system.
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. The EPA's National Exposure Research
Laboratory, which conducts demonstrations of site characterization and monitoring technologies, selected Sandia
National Laboratories, Albuquerque, New Mexico, as the testing organization for field portable GC/MS systems.
DEMONSTRATION DESCRIPTION
In July and September 1995, the performance of two field transportable GC/MS systems was determined under field
conditions. Each system was independently evaluated by comparing field analysis results to those obtained using
approved reference methods. Performance evaluation (PE), spiked, and environmental samples were used to
independently assess the accuracy, precision, and comparability of each instrument.
The demonstration was designed to detect and measure a series of primary target analytes in water, soil, and soil gas.
The primary target analytes at the U.S. Department of Energy's Savannah River Site in Aiken, South Carolina, were
trichloroethene and tetrachloroethene. The primary analytes at Wurtsmith Air Force Base in Oscoda, Michigan, were
EPA-VS-SCM-l 1 The accompanying notice is an integral part of this verification statement December 1997
iii
-------
benzene, toluene, and xylenes. Secondary analytes at the Michigan site included a variety of chlorinated organic
solvents. The sites were chosen because they exhibit a wide range of concentrations for most of the analytes and
provided different climatic and geological conditions. The conditions at each of these sites represent typical, but not
all inclusive, conditions under which the technology would be expected to operate. Details of the demonstration,
including a data summary and discussion of results may be found in the report entitled "Environmental Technology
Verification Report, Field Portable Gas Chromatograph/Mass Spectrometer, Bruker-Franzen Analytical Systems, Inc.
EM640™." The EPA document number for this report is EPA/600/R-97/149.
TECHNOLOGY DESCRIPTION
GC/MS is a proven laboratory analytical technology that has been used in environmental laboratories for many years.
The combination of gas chromatography and mass spectrometry enables the rapid separation and identification of
individual compounds in complex mixtures. The gas chromatograph separates the sample extract into individual
components. The mass spectrometer then ionizes each component which provides the energy to fragment the molecules
into characteristic ions. These ion fragments are then separated by mass and detected as charged particles, which
constitutes a mass spectrum. This spectrum can be used in the identification and quantitation of each component in the
sample extract. For nontarget or unknown analytes the mass spectrum is compared to a computerized library of
compounds to provide identification of the unknown. Field transportable GC/MS is a versatile technique that can be
used to provide rapid screening data or laboratory quality confirmatory analyses. In most systems, the instrument
configuration can also be quickly changed to accommodate different inlets for media such as soil, soil gas, and water.
As with all field analytical studies, it may be necessary to send a portion of the samples to an independent laboratory
for confirmatory analyses.
The EM640™ is a commercially available GC/MS system that provides laboratory-grade performance in a field
transportable package. The instrument is ruggedized and may be operated during transport. It weighs about 140 Ibs and
can be transported and operated in a small van. The EM640™ used in the demonstration used a Spray-and-Trap Water
Sampler, direct injection for soil gas, and heated headspace analysis for soil samples. The minimum detection limit
is 1 ppb for soil gas, IjWg/L for water, and 50 //g/kg for soil. The instrument requires a skilled operator; recommended
training is one week for a chemist with GC/MS experience. At the time of testing, the baseline cost of the EM640™
was $170,000 plus the cost of the inlet system.
VERIFICATION OF PERFORMANCE
The observed performance characteristics of the EM640™ include the following:
• Throughput: The throughput was approximately 5 samples per hour for all media when the instrument was
operated in the rapid analysis mode. Throughput would decrease if the instrument were operated in the
analytical mode.
• Completeness: The EM640™ detected greater than 99 percent of the target compounds reported by the
reference laboratory.
• Precision: Precision was calculated from the analysis of a series of duplicate samples from each media. The
results are reported in terms of relative percent difference (RPD). The values compiled from both sites
generally fell within the range of 0 to 40 percent RPD for soil and 0 to 50 percent for the water and soil gas
samples.
EPA-VS-SCM-l 1 The accompanying notice is an integral part of this verification statement December 1997
iv
-------
• Accuracy: Accuracy was determined by comparing the Bruker GC/MS analysis results with performance
evaluation and spiked samples of known contaminant concentrations. Absolute percent accuracy values from
both sites were calculated for five target analytes. For soil, most of the values are scattered in the 0-90 percent
range with a median of 39 percent. For water, most of the values fall in the 0-70 percent range with a median
of 36 percent. The soil gas accuracy data generally fall in the 0-70 percent range with a median of 22 percent.
• Comparability: This demonstration showed that the EM640™ produced water and soil gas data that were
comparable to the reference laboratory data (median absolute percent difference less than 50 percent). The soil
data were not comparable. This was due, in part, to difficulties experienced by the reference laboratory and
other problems associated with sample handling and transport.
• Deployment: The system was ready to analyze samples within 60 minutes of arrival at the site. The instrument
was operated in a van. Warmup and calibration checks were completed in transit to the site.
The results of the demonstration show that the Bruker-Franzen EM640™ can provide useful, cost-effective data for
environmental problem-solving and decision-making. The deviation of EM640™ and reference laboratory results for
the soil samples, while statistically significant, is not so great as to preclude the effective use of the EM640™ GC/MS
system in many field screening applications. We were unable to determine whether the Bruker GC/MS soil data or that
of the reference laboratory or both were problematic. Undoubtedly, this instrument 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 what
is appropriate for the application and the project data quality objectives.
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-11 The accompanying notice is an integral part of this verification statement December 1997
V
-------
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.
EPA's Superfund Innovative Technology Evaluation (SITE) Program evaluates technologies for the
characterization and remediation of Superfund and Resource Conservation and Recovery Act corrective
action sites. The SITE Program was created to provide reliable cost and performance data to speed the
acceptance of innovative remediation, characterization, and monitoring technologies. One component of
SITE, the Monitoring and Measurement Technologies Program, evaluates new and innovative
measurement and monitoring technologies. Effective measurement and monitoring technologies are
needed to (1) assess the degree of contamination at a site, (2) provide data to determine the risk to public
health or the environment, (3) be cost effective, and (4) monitor the success or failure of a remediation
process. This program is administered by NERL's Environmental Sciences Division in Las Vegas,
Nevada.
Candidate technologies for these programs originate from the private sector and must be market ready.
Through the ETV and SITE Programs, 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
VI
-------
Acknowledgment
The authors wish to acknowledge the support of all those who helped plan and conduct the demonstrations,
analyze the data, and prepare this report. In particular we recognize the technical expertise of Susan
Bender, Jeanne Barrera, Dr. Steve Thornberg, Dr. Mike Keenan, Grace Bujewski, Gary Brown, Bob
Helgesen, Dr. Curt Mowry, and Dr. Brian Rutherford of Sandia National Laboratories. The contributions
of Gary Robertson, Dr. Stephen Billets, and Eric Koglin of the EPA's National Exposure Research
Laboratory, Environmental Sciences Division in Las Vegas, Nevada, are also recognized in the various
aspects of this project.
Demonstration preparation and performance also required the assistance of numerous personnel from the
Savannah River Technology Center and University of Michigan/Wurtsmith Air Force Base. The
contributions of Joe Rossabi and co-workers at the Savannah River Technology Center and Mike
Barcelona and co-workers at the University of Michigan are gratefully acknowledged. The Wurtsmith site
is a national test site funded by the Strategic Environmental Research and Development Program.
Cooperation and assistance from this agency is also acknowledged.
Performance evaluation (PE) samples provided a common reference for the field technologies. Individuals
and reference laboratories who analyzed water and soil samples included Alan Hewitt, of the U.S. Army
Cold Regions Research and Engineering Laboratory, for soil PE samples; and Michael Wilson, of the U.S.
EPA Office of Emergency and Remedial Response, Analytical Operations and Data Quality Center, for the
water PE samples.
We also acknowledge the participation of Bruker-Franzen Analytic GMBH, in particular, Ms. Nolke and
Mr. Zey who operated the Bruker instrument during the demonstrations.
For more information on the Bruker GC/MS demonstrations, contact:
Gary Robertson, Project Technical Leader
Environmental Protection Agency
National Exposure Research Laboratory
Human Exposure and Atmospheric Sciences Division
P.O. Box 93478
Las Vegas, Nevada 89193-3478
(702) 798-2215
For more information on the Bruker GC/MS technology, contact:
Paul Kowalski or Mark Emmons
Bruker Instruments, Inc.
19 Fortune Drive, Manning Park
Billerica, MA01821
(508) 667-9580
fax (508) 667-5993
vn
-------
Contents
Notice ii
Verification Statement iii
Foreword vi
Acknowledgment vii
Figures xii
Tables xiii
Abbreviations and Acronyms xiv
Sections
1. Executive Summary 1
Technology Description 2
Demonstration Objectives and Approach 2
Demonstration Results 2
Performance Evaluation 3
2. Introduction 4
Site Characterization Technology Challenge 4
Technology Verification Process 4
Needs Identification and Technology Selection 5
Demonstration Planning and Implementation 5
Report Preparation 5
Information Distribution 6
The GC/MS Demonstration 6
3. Technology Description 9
Theory of Operation and Background Information 9
Operational Characteristics 9
Performance Factors 10
Detection Limits 11
Dynamic Range 11
Sample Throughput 11
Advantages of the Technology 11
Limits of the Technology 12
Vlll
-------
4. Site Descriptions and Demonstration Design 14
Technology Demonstration Objectives 14
Qualitative Assessments 14
Quantitative Assessments 14
Site Selection and Description 15
Savannah River Site Description 15
Wurtsmith Air Force Base Description 17
Overview of the Field Demonstrations 21
Overview of Sample Collection, Handling, and Distribution 21
SRS Sample Collection 21
WAFB Sample Collection 24
Reference Laboratory Selection and Analysis Methodology 25
General Engineering Laboratory 26
Traverse Analytical and Pace Environmental Laboratories 26
SRS and WAFB On-Site Laboratories 26
Pre-demonstration Sampling and Analysis 26
Deviations from the Demonstration Plan 27
Pre-demonstration Activities 27
SRS Soil Spike Samples 27
SRS Soil Gas Survey Evaluation 27
Soil Gas Samples at WAFB 27
Water Samples at WAFB 28
Calibration Check Sample Analysis 28
5. Reference Laboratory Analysis Results and Evaluation 29
Laboratory Operations 29
General Engineering Laboratories 29
SRS On-Site Laboratory 29
Traverse Analytical Laboratory 29
Pace Inc. Environmental Laboratories 30
WAFB On-Site Laboratory 30
Laboratory Compound Detection Limits 30
Laboratory Data Quality Assessment Methods 31
Precision Analysis 31
Accuracy Analysis 31
Laboratory Internal Quality Control Metrics 32
Laboratory Data Quality Levels 33
IX
-------
Laboratory Data Validation for the SRS Demonstration 33
GEL Data Quality Evaluation 33
GEL Data Quality Summary 35
SRS On-Site Laboratory Data Quality Evaluation 35
SRS Laboratory Data Quality Summary 36
Laboratory Data Validation for the WAFB Demonstration 36
Traverse Data Quality Evaluation 37
Traverse Laboratory Data Quality Summary 39
Pace Data Quality Evaluation 39
Pace Data Quality Summary 41
Summary Description of Laboratory Data Quality 41
6. Technology Demonstration Results and Evaluation 43
Introduction 43
Pre-Demonstration Developer Claims 43
Field Demonstration Data Evaluation Approach 44
Instrument Precision Evaluation 44
Instrument Accuracy Evaluation 45
Instrument Comparison with Reference Laboratory Data 46
Summary of Instrument Performance Goals 49
Accuracy 49
Precision 50
Bruker to Reference Laboratory Comparison 51
Field Operation Observations 52
SRS Demonstration 52
WAFB Demonstration 52
Bruker Accuracy and Precision Results 54
Bruker Accuracy ~ SRS Demonstration 54
Bruker Accuracy ~ WAFB Demonstration 54
Overall Bruker Accuracy Performance 55
Bruker Precision ~ SRS Demonstration 57
Bruker Precision ~ WAFB Demonstration 58
Overall Bruker Precision Performance 59
Bruker to Reference Laboratory Data Comparison 59
Scatter Plots/Histograms ~ SRS Demonstration 61
Scatter Plots/Histograms ~ WAFB Demonstration 61
Overall Bruker to Laboratory Comparison Results 67
Summary of Bruker Accuracy, Precision, and Laboratory Comparison Performance 68
-------
Other Broker GC/MS Performance Indicators 69
Target Compound Identification in Complex Mixtures 69
Field Handling and Operation 69
Overall Bruker GC/MS Performance Conclusions 70
7. Applications Assessment 72
Applicability to Field Operations 72
Capital and Field Operation Costs 72
Advantages of the Technology 72
Rapid Analysis 72
Sampling and Sample Cost Advantages 73
Transportability 74
Field Screening of Samples 74
Sample Size 74
Interferences 74
Conclusions 74
8. Developer's Forum 75
9. Previous Deployments 77
10. References 78
Appendix
A: Environmental Monitoring Management Council (EMMC) Method A-l
XI
-------
Figures
2-1 Example total ion chromatogram of a complex mixture 7
3-1 Block Diagram of Bruker-Franzen EM640™ GC/MS 10
4-1 Location of the Savannah River Site 16
4-2 SRS M-Area Well Locations 18
4-3 Location of Wurtsmith Air Force Base 19
4-4 WAFB Fire Training Area 2 Sampling Locations 20
6-1 Example scatter plots with simulated data 48
6-2 Example histograms with simulated data 50
6-3 Plot of daily temperatures during the SRS demonstration 53
6-4 Plot of daily temperatures during the WAFB demonstration 53
6-5 Absolute percent accuracy histogram for Bruker soil samples 56
6-6 Absolute percent accuracy histogram for Bruker water samples 56
6-7 Absolute percent accuracy histogram for Bruker soil gas samples 56
6-8 Relative percent difference histogram for Bruker soil samples 60
6-9 Relative percent difference histogram for Bruker water samples 60
6-10 Relative percent difference histogram for Bruker soil gas samples 60
6-11 Bruker vs. Laboratory data for SRS low concentration water samples 62
6-12 Bruker vs. Laboratory data for SRS high concentration water samples 62
6-13 Percent difference histogram for SRS water samples 63
6-14 Bruker vs. Laboratory data for SRS soil gas samples 63
6-15 Percent difference histogram for SRS soil gas samples 63
6-16 Bruker vs. Laboratory data for WAFB soil samples 64
6-17 Relative percent difference histogram for WAFB soil samples 64
6-18 Bruker vs. Laboratory data for WAFB low concentration water samples 65
6-19 Bruker vs. Laboratory data for WAFB high concentration water samples 65
6-20 Relative percent difference histogram for WAFB water samples 66
6-21 Bruker vs. Laboratory data for WAFB soil gas samples 66
6-22 Relative percent difference histogram for WAFB soil gas samples 66
6-23 Absolute percent difference histogram for soil samples 67
6-24 Absolute percent difference histogram for water samples 68
6-25 Absolute percent difference histogram for gas samples 68
6-26 Bruker GC/MS reconstructed chromatogram of target analytes in a WAFB water sample 69
xn
-------
Tables
3-1 Bruker-Franzen EM640™ GC/MS Instrument Specifications 11
4-1 PCE and TCE Concentrations in SRS M-Area Wells 17
4-2 Historical Ground Water Contamination Levels at WAFB 18
4-3 VOC Concentrations in WAFB Fire Training Area 2 Wells 20
4-4 Sample Terminology and Description 22
4-5 SRS Demonstration Sample Type and Count 23
4-6 WAFB Demonstration Sample Type and Count 24
5-1 Reference Laboratory Practical Quantitation Limits 30
5-2 GEL Laboratory Accuracy Data 34
5-3 GEL Laboratory Precision Data 35
5-4 SRS Laboratory Accuracy Data 36
5-5 SRS Laboratory Precision Data 36
5-6 Traverse Laboratory Accuracy Data 37
5-7 WAFB Water and Soil PE/Spike Sample Reference Concentrations 38
5-8 Traverse Laboratory Precision Data 38
5-9 WAFB Water and Soil Duplicate Sample Concentrations 38
5-10 Pace Laboratory Accuracy Data 40
5-11 WAFB Soil Gas PE/Spike Sample Reference Concentrations 40
5-12 Pace Laboratory Precision Data 41
5-13 WAFB Soil Gas Duplicate Sample Concentrations 41
5-14 SRS Demonstration Laboratory Data Quality Ranking 41
5-15 WAFB Demonstration Laboratory Data Quality Ranking 42
6-1 Bruker Recoveries at SRS 54
6-2 Bruker Recoveries at Wurtsmith 55
6-3 Bruker and Reference Laboratory Accuracy Summary 57
6-4 Bruker Precision for SRS Demonstration 58
6-5 Bruker Precision for Wurtsmith Demonstration 58
6-6 Bruker and Reference Laboratory Precision Summary 61
6-7 Bruker-Laboratory Comparison Summary 67
6-8 Summary Performance of the Bruker GC/MS 69
6-9 Identified Target Compounds from a Wurtsmith Water Sample Analysis 70
6-10 Summary of Bruker Performance Goals and Actual Performance 71
7-1 Bruker EM640™ GC/MS Capital and Field Operation Costs 73
xni
-------
Abbreviations and Acronyms
AC
amu
amp
APA
APD
BTEX
CSCT
DNAPL
DCE
DIP
DoD
DOE
DOT
EPA
ESD-LV
ETV
ETVR
g
GC/MS
GEL
Hz
kg
kW
L
^
mg
mL
MS
NCIBRD
NA
ND
NERL
NETTS
ng
NP
PAH
PCE
PE
ppb
ppm
ppt
PQL
QA
QC
REC
RPD
RSD
Alternating current
Atomic mass unit
Ampere
Absolute percent accuracy
Absolute percent difference
Benzene, toluene, ethylbenzene, xylenes
Consortium for Site Characterization Technology
Dense nonaqueous phase liquid
Dichloroethylene
Percent difference
Department of Defense
Department of Energy
Department of Transportation
Environmental Protection Agency
Environmental Sciences Division of the National Exposure Research Laboratory
Environmental Technology Verification Program
Environmental Technology Verification Report
Gram
Gas chromatograph/mass spectrometer
General Engineering Laboratories
Hertz
Kilogram
Kilowatt
Liter
Microgram
Milligram
Milliliter
Mass spectrometer
National Center for Integrated Bioremediation Research and Development
Not analyzed
Not detected or no determination
National Exposure Research Laboratory
National Environmental Technology Test Sites Program
nanogram
Not present
Polycyclic aromatic hydrocarbons
Tetrachloroethene
Performance evaluation
Parts per billion
Parts per million
Parts per trillion
Practical quantitation limit
Quality assurance
Quality control
Percent recovery
Relative percent difference
Relative standard deviation
xiv
-------
SERDP Strategic Environmental Research and Development Program
SIM Single ion monitoring
SNL Sandia National Laboratories
SRS Savannah River Site
SUMMA® (Registered trademark for Passivated Canister Sampling Apparatus)
TCA Trichloroethane
TCE Trichloroethene
v Volts
VOA Volatile organic analysis
VOC Volatile organic compound
WAFB Wurtsmith Air Force Base
W Watt
xv
-------
Section 1
Executive Summary
The performance evaluation of innovative and alternative environmental technologies is an integral part of
the U.S. Environmental Protection Agency's (EPA) 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 demonstrate and evaluate the cost and performance of remediation and monitoring
technologies under the Superfund Innovative Technology Evaluation (SITE) program (in response to the
mandate in the Superfund Amendments and Reauthorization Act of 1987). 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, purchase, and use 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, Advanced Monitoring Systems, EvTEC (an
independent, private-sector approach), Wet Weather Flows Technologies, Pollution Prevention/Metal
Finishing, Source Water Protection Technologies, Site Characterization and Monitoring Technology (a.k.a.
Consortium for Site Characterization Technology (CSCT)), and Climate Change Technologies. The
performance verification contained in this report is based on the data collected during a demonstration of a
field portable gas chromatograph/mass spectrometer (GC/MS) system. The demonstration was
administered by the Consortium for Site Characterization Technology.
For each pilot, EPA utilizes the expertise of partner "verification organizations" to design efficient
procedures for conducting performance tests of environmental technologies. EPA selects its partners from
both the public and private sectors including Federal laboratories, states, 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. The U.S. Department of Energy's Sandia National Laboratories, Albuquerque, New Mexico, served
as the verification organization for this demonstration.
In 1995, the Consortium conducted a demonstration of two field portable gas chromatograph/mass
spectrometer systems. These technologies can be used for rapid field analysis of organic-contaminated soil,
ground water, and soil gas. They are designed to speed and simplify the process of site characterization and
to provide timely, on-site information that contributes to better decision making by site managers. The two
system developers participating in this demonstration were Bruker-Franzen Analytical Systems, Inc.: and
Viking Instruments Corporation. The purpose of this Environmental Technology Verification Report
(ETVR) is to document demonstration activities, present demonstration data, and verify the performance of
The company is now known as Bruker Instruments, Inc.
-------
the Bruker-Franzen EM640™ field transportable GC/MS. Demonstration results from the other system are
presented in a separate report.
Technology Description
The Bruker-Franzen EM640™ GC/MS consists of a temperature-programmable gas chromatograph
coupled to a mass spectrometer. This field transportable system uses a small gas chromatographic column
and accompanying mass spectrometer to provide separation, identification, and quantification of volatile
and semi-volatile organic compounds in soils, liquids, and gases. In the demonstration, the system used a
spray-and-trap technique for water analysis, as well as direct injection and head space analysis for soil gas
and soil analyses, respectively. The column enables separation of individual analytes in complex mixtures.
As these individual analytes exit the column, the mass spectrometer detects the analytes, providing a
characteristic mass spectrum that identifies each compound. An external computer system provides
quantitation by comparison of detector response with a calibration table constructed from standards of
known concentration. The system provides very low detection limits for a wide range of volatile and semi-
volatile organic contaminants.
Demonstration Objectives and Approach
The GC/MS systems were taken to two geologically and climatologically different sites: the U. S.
Department of Energy's Savannah River Site (SRS), near Aiken, South Carolina, and Wurtsmith Air Force
Base (WAFB), in Oscoda, Michigan. The demonstration at the Savannah River Site was conducted in July
1995 and the Wurtsmith AFB demonstration in September 1995. Both sites contained soil, ground water,
and soil gas that were contaminated with a variety of volatile organic compounds. The demonstrations
were designed to evaluate the capabilities of each field transportable system.
The primary objectives of this demonstration were: (1) to evaluate instrument performance; (2) to
determine how well each field instrument performed compared to reference laboratory data; (3) to evaluate
instrument performance on different sample media; (4) to evaluate adverse environmental effects on
instrument performance; and, (5) to determine logistical needs and field analysis costs.
Demonstration Results
The demonstration provided adequate analytical and operational data with which to evaluate the
performance of the Bruker-Franzen EM640™ GC/MS system. Accuracy was determined by comparing the
Bruker GC/MS analysis results with performance evaluation and spiked samples of known contaminant
concentrations. Absolute percent accuracy values from both sites were calculated for five target analytes.
For soil, most of the values are scattered in the 0-90% range with a median of 39%. For water, most of the
values fall in the 0-70% range with a median of 36%. The soil gas accuracy data generally fall in the 0-
70% range with a median of 22%. Precision was calculated from the analysis of a series of duplicate
samples from each media. The results are reported in terms of relative percent difference (RPD). The
values compiled from both sites generally fell within the range of 0 to 25% RPD for soil and 0 to 50% for
the water and soil gas samples. The EM640™ produced water and soil gas data that were comparable to
the reference laboratory data. However, the soil data were not comparable. This was due in part to
difficulties experienced by the reference laboratory in analyzing soil samples and other problems
associated with sample handling and transport.
Considerable variability was encountered in the results from reference laboratories, illustrating the degree
of difficulty associated with collection, handling, shipment, storage, and analysis of soil gas, water, and
soil samples using off-site laboratories. This demonstration revealed that use of field analytical methods,
with instruments such as the Bruker GC/MS, can eliminate some of these sample handling problems.
-------
Performance Evaluation
Overall, the results of the demonstration indicated that most of the performance goals were met by the
Broker GC/MS system under field conditions, and that the system can provide good quality, near-real-time
field analysis of soil, water, and soil gas samples contaminated by organic compounds. The system was
easily transportable in a van and required only two technicians for operation. A limited analysis of capital
and field operational costs for the Bruker system shows that field use of the system may provide some cost
savings when compared to conventional laboratory analyses. Based on the results of this demonstration, the
Bruker EM640™ GC/MS system was determined to be a mature field instrument capable of providing on-
site analyses of water and soil gas samples comparable to those from a conventional fixed laboratory.
-------
Section 2
Introduction
Site Characterization Technology Challenge
Rapid, reliable, and cost-effective field screening and analysis technologies are needed to assist in the
complex task of characterizing and monitoring hazardous and chemical waste sites. Environmental
regulators and site managers are often reluctant to use new technologies which have not been validated in
an objective EPA-sanctioned testing program or similar process which facilitates acceptance. Until field
characterization technology performance can be verified through objective evaluations, users will remain
skeptical of innovative technologies, despite their promise of better, less expensive, and faster
environmental analyses.
The Environmental Technology Verification (ETV) Program was created by the U. S. Environmental
Protection Agency (EPA) 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, purchase, and use 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,
Advanced Monitoring Systems, EvTEC (an independent, private-sector approach), Wet Weather Flows
Technologies, Pollution Prevention/Metal Finishing, Source Water Protection Technologies, Site
Characterization and Monitoring Technology (a.k.a. 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 field transportable gas chromatograph/mass
spectrometer (GC/MS) systems. The demonstration was administered by the Consortium for Site
Characterization Technology. The mission of the Consortium is to identify, demonstrate, and verify the
performance of innovative site characterization and monitoring technologies. The Consortium also
disseminates information about technology performance to developers, environmental remediation site
managers, consulting engineers, and regulators.
For each pilot, EPA utilizes the expertise of partner "verification organizations" to design efficient
procedures for conducting performance tests of environmental technologies. EPA selects its partners from
both the public and private sectors including Federal laboratories, states, 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. The U.S. Department of Energy's Sandia National Laboratories, Albuquerque, New Mexico, served
as the verification organization for this demonstration.
Technology Verification Process
The technology verification process is intended to serve as a template for conducting technology
demonstrations that will generate high-quality data which 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; and,
• Information Distribution.
-------
Each component is discussed in detail in the following paragraphs.
Needs Identification and Technology Selection
The first aspect of the technology verification process is to determine technology needs of the EPA and the
regulated community. EPA, the U.S. Department of Energy, 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 the 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.
• Applicable to a variety of environmentally impacted sites.
• High potential for resolving problems for which current methods are unsatisfactory.
• Costs are competitive with current methods.
• Performance is 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.
• 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
responsibilities for conducting the demonstration and evaluating the technology. The following issues are
addressed at this time:
• Identifying demonstration sites that will provide the appropriate physical or chemical
attributes, in the desired environmental 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://clu-in.com).
The GC/MS Demonstration
In late 1994, the process of technology selection for the GC/MS systems was initiated by publishing a
notice to conduct a technology demonstration in the Commerce Business Daily. In addition, active
solicitation of potential participants was conducted using manufacturer and technical literature references.
Final technology selection was made by the Consortium based on the readiness of technologies for field
demonstration and their applicability to the measurement of volatile organic contaminants at
environmentally impacted sites.
GC/MS is a proven laboratory analytical technology that has been in use in environmental laboratories for
many years. The instruments are highly versatile with many different types of analyses easily performed on
the same system. Because of issues such as cost and complexity, the technology has not been fully adopted
for use by the field analytical community. The purpose of this demonstration was to provide not only an
evaluation of field portable GC/MS technology results compared to fixed laboratory analyses, but also to
evaluate the transportability, ruggedness, ease of operation, and versatility of the field instruments.
For this demonstration, three instrument systems were initially selected for verification. Two of the systems
selected were field portable GC/MS systems, one from Viking Instruments Corporation and the other from
Bruker-Franzen Analytical Systems, Inc. The other technology identified was a portable direct sampling
device for an ion trap mass spectrometer system manufactured by Teledyne Electronic Technologies.
However, since the direct sampling inlet for this MS system was not commercially available, its
performance has not been verified. In the summer of 1995, the Consortium conducted the demonstration
which was coordinated by Sandia National Laboratories.
The versatility of field GC/MS instruments is one of their primary features. For example, an instrument
may be used in a rapid screening mode to analyze a large number of samples to estimate analyte
concentrations. This same instrument may be used the next day to provide fixed-laboratory-quality data on
selected samples with accompanying quality control data. The GC/MS can also identify other contaminants
that may be present that may have been missed in previous surveys. Conventional screening instruments,
such as portable gas chromatographs, would only indicate that an unknown substance is present.
An example of compound selectivity for a GC/MS is shown in Figure 2-1. The upper portion of the figure
is a GC/MS total ion chromatogram from a water sample containing numerous volatile organic
compounds. The total ion chromatogram is a plot of total mass detector response as a function of time from
sample injection into the instrument. Many peaks can be noted in the retention time window between 7 and
11 minutes. In many cases the peaks are not completely resolved as evidenced by the absence of a clear
baseline. The inset figure shows a reconstructed ion chromatogram for ion mass 146. This corresponds to
the molecular ion peak of the three isomers of dichlorobenzene. The relative intensities of these peaks are
at a level of about 60,000 with the background considerably higher at an intensity level between 500,000
and 1,000,000. This is an example of the ability of the GC/MS to detect and quantitate compounds in the
midst of high background levels of other volatile organic compounds.
-------
8
co
•a
4500K
4000K
3500K
3000K
2500K
2000K
1500K
1000K
500K
4 5
100K
90K
80K
70K
g 60K
B
•g 50K
< 40K
30K
20K
10K
Time, minutes
Ion 146.00
(Dichlorobenzene)
tvj
9.40
9.60 9.80 10.00 10.20
Figure 2-1. Example total ion chromatogram of a complex mixture. The inset shows the ability of the GC/MS
system to detect the presence of dichlorobenzenes in a high organic background.
The objectives of this technology demonstration were essentially five-fold:
• To evaluate instrument performance;
• To determine how well each field instrument performed compared to reference laboratory data;
• To evaluate developer goals regarding instrument performance on different sample media;
• To evaluate adverse environmental effects on instrument performance; and,
• To determine the logistical and economic resources needed to operate each instrument.
-------
Section 3
Technology Description
Theory of Operation and Background Information
Gas chromatography/mass spectrometry (GC/MS) is a proven laboratory technology that has been in use in
fixed analytical laboratories for many years. The instruments are highly versatile, with many different types
of analyses easily performed on the same instrument. The combination of gas chromatography and mass
spectrometry enables rapid separation and identification of individual compounds in complex mixtures.
One of the features of the GC/MS is its ability to detect and quantitate the compounds of interest in the
presence of large backgrounds of interfering substances. Using GC/MS, an experienced analyst can often
identify every compound in a complex mixture.
The varying degrees of affinity of compounds in a mixture to the GC column coating makes their
separation possible. The greater the molecular affinity, the slower the molecule moves through the column.
Less affinity on the other hand causes the molecule to elute from the column more rapidly. A portion of the
GC column effluent is directed to the MS ion source where the molecules are fragmented into charged
species. These charged species are in turn passed through a quadrupole filter which separates them on the
basis of their charge-to-mass ratio. The charged fragments are finally sensed at an electron multiplier at the
opposite end of the quadrupole filter. The array of fragments detected for each eluting compound is known
as a mass spectrum and provides the basis for compound identification and quantitation. The GC/MS mass
spectrum can be used to determine the molecular weight and molecular formula of an unknown compound.
In addition, characteristic fragmentation patterns produced by sample ionization can be used to deduce
molecular structure. Typical detection limits of about 10~12g can be realized with MS.
Operational Characteristics 1
The Bruker-Franzen EM640™ shown in Figure 3-1 is a complete GC/MS system that provides laboratory-
grade performance in a field transportable package. The system is based on transferring VOCs in liquid or
solid samples to the gas phase. General instrument specifications are presented in Table 3-1. VOCs
extracted from air, liquid, or solid samples are introduced in the gas phase into a gas chromatograph (GC)
for separation. Compounds eluting from the GC column permeate through an inlet membrane into the
vacuum chamber of the MS. The molecules are ionized by electron impact and subsequently pass through
a mass selective filter. The ions are detected in an electron multiplier that generates an electrical signal
proportional to the number of ions. The data system records these electrical signals and converts them into
a mass spectrum. The sum of all ions in a mass spectrum at any given instant corresponds to one point in
the total detector response (total ion chromatogram) that is recorded as a function of time. A mass spectrum
is like a fingerprint of a compound. These fingerprints are compared with stored library spectra and used
together with the GC retention times for the identification of the compounds. The signal intensity of
selected mass peaks is used for quantitation of pre-selected target compounds.
Recommended ancillary analysis equipment is the Spray-and-Trap Water Sampler (Bruker Analytical
Systems Inc., Billerica, MA). The Spray-and-Trap Water Sampler device consists of a mechanical pump to
inject a continuous flow of an aqueous sample into a sealed extraction chamber through a spray nebulizer.
The droplet formation enormously increases the total interfacial area between the sprayed water and the
carrier gas, which supports the transfer of the VOCs into the gas phase. The steadily flowing carrier gas is
transferred to a suitable sorbent tube which collects the extracted VOCs. In contrast to the purge-and-trap
The information presented in the remainder of Section 3 was provided by Bruker. It has been minimally edited. This information is solely that of
Bruker and should not be construed to reflect the views or opinions of EPA.
-------
method, spray-and-trap utilizes a dynamic equilibrium. During water spray, an equilibrium VOC transfer rate
between the droplet surfaces and flowing carrier gas is established.
Irtpoiw
GC
Gas
Supply
Module
\
Figure 3-1. Block Diagram of Bruker-Franzen EM640™GC/MS.
Performance Factors
The following sections describe the Bruker-Franzen EM640™ GC/MS performance factors. These factors
include detection limits, sensitivities, and sample throughput.
10
-------
Table 3-1. Bruker-Franzen EM640™ GC/MS Instrument Specifications.
Parameter
Practical Quantitation Limits (scan
mode)
Mass range
Dynamic Range
Sample throughput
Maximum scan speed
Temperature range
Power requirements
Weight
Size
Operator and training required
Support equipment
Computer requirements
Cost
Developer's Specification
20 ppb air (soil gas), 0.1 //g/L water, and 50 mg/kg soil
1 - 650 amu
4-5 orders of magnitude
10 minutes per sample including analysis time
2000 amu/sec
-10to45°C
500 W
ca. 65 kg
750x450x350 mm
Full chemist (1 week operation, method development, evaluation), lab
operator (3 weeks execution of methods, protocol)
Spray-and-trap extractor, batteries, power generator (as an alternative to
batteries)
PC with OS/2 multitask software
Baseline $170K + cost of inlet system
Practical Quantitation Limits
Detection limits vary depending on compound, media, operation mode of the MS ("scan" or "single ion
monitoring"), and sample volume. Generally, for thirty-six of the most common VOCs, the practical
quantitation limits (PQL) in the "scan mode" are: 20 ppb for soil gas (100 mL sample volume); 0.1 //g/L for
water samples (250 mL sample volume); and, 50 mg/kg for soil samples (6 g sample weight). The "single ion
monitoring" (SIM) mode of operation increases the sensitivity by a factor of 10. To express this in absolute
values, the mass spectrometer needs 1 ng of a compound to produce a signal-to-noise ratio of 10 in the scan
mode.
Dynamic Range
Approximately 4-5 orders of magnitude linear dynamic range are possible with the Bruker-Franzen EM640™
depending upon the analyte and the analysis conditions.
Sample Throughput
Sample throughput measures the amount of time required to prepare and analyze one field sample. Bruker-
Franzen claims that the complete analysis time is as follows: air and water samples, 8-10 minutes per sample or
6 samples per hour, soil samples, 7-10 minutes or 7 - 8 samples per hour. This does not include sample
handling, data documentation, or difficult dilutions and concentrations.
Advantages of the Technology
The EM640™ offers the following advantages:
11
-------
• It is a ruggedized instrument, built for reliability and ease of operation. It is shock and vibration proof
and can be successfully transported in a four wheel drive vehicle in rough terrain (a special damping
bed with quick release connector is used to mount the instrument).
• The instrument can be calibrated during transport to the site, therefore increasing overall analysis time
on site.
• The application of fast analysis runs results in 6 to 8 sample analyses per hour, as a result of the short-
column GC analysis technique applied. Incomplete GC separation is compensated for by mathematical
separation routines.
• The analysis report for a sample is available within a few minutes after start of the analysis, making it
possible to evaluate and direct the sampling strategy in the field. With one or two EM640™
instruments in a small van, the analysis speed can be adapted to the sampling speed of a sampling team.
Sampling and analysis can easily progress simultaneously.
• The EM640™ analytical procedures can be optimized with respect to a variety of parameters, e.g.
highest analysis speed, safest substance identification, maximum precision, or lowest detection limits.
• The EM640™ GC/MS technology offers low cost sample analysis. Costs should be considerably lower
than 25% of those incurred using conventional laboratory analysis.
• The high sample throughput rate allows for the analysis of many QA/QC samples during the day,
providing better quality control for the analyses.
• A calibration gas stored inside a small container inside the instrument is the only consumable of the
EM640™. The GC column is operated using an ambient air as the carrier gas. There are no pump oils,
lubricants, or other maintenance materials. Little maintenance is necessary. No ion source cleaning is
required. The high vacuum pump inside the EM640™ does not contain any moving parts, and
there is no roughing pump at all. To aid in trouble-shooting, the EM640™ features internal
monitoring of all electric functions.
• The preparation of samples is simplified by the use of a large dynamic measuring range, featuring
a linear calibration curve over four to five orders of magnitude.
• For soil extraction, a special battery-operated ultra sound extraction method with acetone has been
developed, minimizing the use of chlorinated solvents that must be treated as hazardous waste.
Limits of the Technology
Some limitations associated with the EM640™ are listed below:
• Detection limits in air. By sampling 500 mL of air on a sorption tube, the limit of detection for
toluene is approximately 10 ppb, using the instrument in full-scan mode. The limit of detection for
toluene in air is 1 ppm, if measured with the instrument's flexible probe in full-scan mode without
any enrichment.
• Detection limits in water: Spraying 300 mL of water by the Spray and Trap Water Sampler, which
takes about two minutes, a detection limit of 0.1 //g/L is measured for most volatile substances like
trichloroethene and perchloroethene. Less polar substances have lower detection limits; more polar
compounds have higher detection limits.
• GC limitations: The GC usually operates with air as the carrier gas, therefore the maximum
temperature of the column is restricted to 240°C. Most analytical separations can be achieved
within this temperature limitation by selection of the right type of GC column. Nitrogen can be
used to extend the useful temperature range to 300° C if high boiling point semi-volatiles are to be
analyzed.
12
-------
Analyte limitations: The membrane inlet system limits the analytes that can be analyzed. Extremely
polar compounds cannot be analyzed with the same sensitivity as non-polar compounds. Some
classes of compounds are not easily analyzed.
Sample Media Effects: In general, air and water samples are more easily analyzed than soil by
GC/MS instruments. Therefore, accuracy and precision for soil is expected to be lower.
Additionally, soil is often more difficult to homogenize, giving rise to additional analytical
variation.
Spectral Interference: With GC/MS technology in general, interference can occur with excessive
water vapor and with contamination. Water vapor may increase some detection levels;
contamination may reside in sampling equipment which must be periodically checked; cross
contamination may occur with sequential high and low concentration samples. This can be
checked and eliminated by periodically analyzing reagent blanks.
13
-------
Section 4
Site Descriptions and Demonstration Design
This section provides a brief description of the sites used in the demonstration and an overview of the
demonstration design. Sampling operations, reference laboratory selection, and analysis methods are also
discussed. A comprehensive demonstration plan entitled "Demonstration Plan for the Evaluation of Field
Transportable Gas Chromatograph/Mass Spectrometer" [SNL, 1995] was prepared to help guide the
demonstration. The demonstration plan was designed to ensure that the demonstration would be
representative of field operating conditions and that the sample analytical results from the field GC/MS
technologies under evaluation could be objectively compared to results obtained using conventional
laboratory techniques.
Technology Demonstration Objectives
The purpose of this demonstration was to thoroughly and objectively evaluate field transportable GC/MS
technologies during typical field activities. The primary objectives of the demonstration were to:
• To evaluate instrument performance;
• To determine how well each field instrument performed compared to reference laboratory data;
• To evaluate developer goals regarding instrument performance on different sample media;
• To evaluate adverse environmental effects on instrument performance; and,
• To determine the logistical and economic resources needed to operate each instrument.
In order to accomplish these objectives, both qualitative and quantitative assessments of each system were
required and are discussed in detail in the following paragraphs.
Qualitative Assessments
Qualitative assessments of field GC/MS system capabilities included the portability and ruggedness of the
system and its logistical and support requirements. Specific instrument features that were evaluated in the
demonstration included: system transportability, utility requirements, ancillary equipment needed, the
required level of operator training or experience, health and safety issues, reliability, and routine
maintenance requirements.
Quantitative Assessments
Several quantitative assessments of field GC/MS system capabilities related to the analytical data produced
by the instrument were conducted. Quantitative assessments included the evaluation of instrument
accuracy, precision, and data completeness. Accuracy is the agreement between the measured
concentration of an analyte in a sample and the accepted or "true" value. The accuracy of the GC/MS
technologies was assessed by evaluating performance evaluation (PE) and media spike samples. Precision
is determined by evaluating the agreement between results from the analysis of duplicate samples.
Completeness, in the context of this demonstration, is defined as the ability to identify all of the
contaminants of concern in the samples analyzed. Sites were selected for this demonstration with as many
as fifteen contaminants to identify and analyze and with high background hydrocarbon concentrations.
Additional quantitative capabilities assessed included field analysis costs per sample, sample throughput
rates, and the overall cost effectiveness of the field systems.
Site Selection and Description
Sandia National Laboratories and the EPA's National Exposure Research Laboratory/Environmental
Sciences Division-Las Vegas (NERL/ESD-LV) conducted a search for suitable demonstration sites
between January and May 1995. The site selection criteria were guided by logistical demands and the need
14
-------
to demonstrate the suitability of field transportable GC/MS technologies under diverse conditions
representative of anticipated field applications. The site selection criteria were:
• Accessible to two-wheel drive vehicles;
• Contain one or more contaminated media (water, soil, and soil gas);
• Provide a wide range of contaminant types and concentration levels to truly evaluate the
capabilities and advantages of the GC/MS systems;
• Access to historical data on types and levels of contamination to assist in sampling activities;
• Variation in climatological and geological environments to assess the effects of environmental
conditions and media variations on performance; and,
• Appropriate demonstration support facilities and personnel.
Several demonstration sites were reviewed and, based on these selection criteria, the U. S. Department of
Energy's Savannah River Site (SRS) near Aiken, South Carolina, and Wurtsmith Air Force Base (WAFB)
in Oscoda, Michigan, were selected as sites for this demonstration.
The Savannah River Site is a DOE facility, focusing on national security work; economic development and
technology transfer initiatives; and, environmental and waste management activities1. The SRS staff have
extensive experience in supporting field demonstration activities. The SRS demonstration provided the
technologies an opportunity to analyze relatively simple contaminated soil, water, and soil gas samples
under harsh operating conditions. The samples contained only a few chlorinated compounds (solvents)
with little background contamination, but high temperatures and humidity offered a challenging operating
environment.
WAFB is one of the Department of Defense's (DoD) National Environmental Technology Test Site
(NETTS) test sites. The facility is currently used as a national test bed for bioremediation field research,
development, and demonstration activities. The WAFB demonstration provided less challenging
environmental conditions for the technologies but much more difficult samples to analyze. The soil, water,
and soil gas samples contained a complex matrix of fifteen target VOC analytes along with relatively high
concentration levels of jet fuel, often about 100 times the concentration levels of the target analytes being
measured.
Savannah River Site Description
Owned by DOE and operated under contract by the Westinghouse Savannah River Company, the
Savannah River Site complex covers 310 square miles, bordering the Savannah River between western
South Carolina and Georgia as shown in Figure 4-1.
The Savannah River Site was constructed during the early 1950's to produce the basic materials used in the
fabrication of nuclear weapons, primarily tritium and plutonium-239. Weapons material production at SRS
has produced unusable byproducts such as intensely radioactive waste. In addition to these high-level
wastes, other wastes at the site include low-level solid and liquid radioactive wastes; transuranic waste;
Much of this site descriptive material is adapted from information available at the Savannah River Site web page (http://www.srs.gov/general/srs-
home.html)
15
-------
South Carolina
Georgia
A
N
Figure 4-1. Location of the Savannah River Site.
hazardous waste; mixed waste, which contains both hazardous and radioactive components; and sanitary
waste, which is neither radioactive nor hazardous. Like many other large production facilities, chemicals
have been released into the environment during production activities at SRS. These releases and the
common disposal practices of the past have resulted in subsurface contamination by a variety of
compounds used in or resulting from production processes.
SRS Geologic andHydrologic Characteristics
The facility is located on the upper Atlantic coastal plain on the Savannah River, approximately 30 miles
southeast of Augusta, Georgia and about 90 miles north of the Atlantic coast. The site is underlain by a
thick wedge (approximately 1,000 feet) of unconsolidated Tertiary and Cretaceous sediments that overlay
the basement which consists of Precambrian and Paleozoic metamorphic rocks and consolidated Triassic
sediments (siltstone and sandstone). The younger sedimentary section consists predominantly of sand,
clayey sand, and sandy clay.
Ground water flow at the site is controlled by hydrologic boundaries. Flow at or immediately below the
water table is predominately downward and toward the Savannah River. Ground water flow in the shallow
aquifers in the immediate vicinity of the demonstration site is highly influenced by eleven pump-and-treat
recovery network wells.
SRS Demonstration Site Characteristics
Past industrial waste disposal practices at the Savannah River Site, like those encountered at other DOE
weapons production sites, often included the release of many chemicals into the local environment. These
releases and early disposal practices have resulted in the contamination of the subsurface of many site areas
by a number of industrial solvents used in, or resulting from the various weapons material production
processes. The largest volume of contamination has been from chlorinated volatile organic compounds.
16
-------
The primary VOCs encountered at SRS include: tetrachloroethene (PCE), trichloroethene (TCE),
trichloroethane (TCA), Freon 11, and Freon 113.
The area selected for the demonstration is designated the M-Area. The M-Area is located in the northwest
section of SRS and consists of facilities that fabricated reactor fuel and target assemblies for the SRS
reactors, laboratory facilities, and administrative support facilities. Operations at these and other facilities
resulted in the release of the chlorinated solvents previously mentioned. The releases have resulted in the
contamination of soil and ground water within the area. The technology staging site was located near an
abandoned process sewer line which carried waste water from M-Area processing facilities to a settling
basin for 27 years, beginning in 1958. Site characterization data indicate that several leaks existed in the
sewer line, located about 20 feet below the surface, producing localized sources of contamination.
Although the use of the sewer line was discontinued in 1985, estimates are that over 2 million pounds of
these solvents were released into the subsurface during its use.
Typical PCE and TCE concentrations are listed in Table 4-1 for the demonstration wells identified in
Figure 4-2. The soil and underlying sediments at the demonstration site are highly contaminated with
chlorinated solvents at depths in excess of 50 feet. Identification of the contaminant concentration levels in
the soil and sediments has been complicated by the nature of these media at SRS. They have very low
organic content, resulting in significant contaminant loss during typical sampling operations. These
sampling concerns and limitations, and their influence on the demonstration, are discussed in detail later in
this section.
Table 4-1. PCE and TCE Concentrations in SRS M Area Wells.
Cone. Level
Low
Medium
High
Water
Well
MHT-11C
MHT-12C
MHT-17C
PCE (Mg/L)
12
110
3700
TCE (Mg/L)
37
100
2700
Soil Gas
Well
MHV-2C
CPT-RAM 15
CPT-RAM 4
PCE (ppm)
10
80
800
TCE (ppm)
5
50
350
Wurtsmith Air Force Base Description
Wurtsmith Air Force Base covers approximately 7.5 square miles and is located on the eastern side of
Michigan's lower peninsula on Lake Huron, about 75 miles northeast of Midland, Michigan, near the town
of Oscoda (Figure 4-3). It is bordered by three connected open water systems; Lake Huron to the east,
shallow wetlands and the Au Sable River to the south, and Van Etten Lake to the north. State and National
Forest lands surround much of the base. WAFB began operations as an Army Air Corps facility, known as
Camp Skeel, in 1923. It was originally used as a bombing and artillery range and as a winter training
facility. The WAFB was decommissioned in 1993 and is currently being used as a national test bed for
bioremediation field research, development, and demonstration. The National Center for Integrated
Bioremediation Research and Development (NCIBRD) of the University of Michigan coordinates these
bioremediation activities. Several contaminant features consistent with its history as an Air Force base have
17
-------
MHT-11C*
MHT-12C *
MHV-2C *
North
<-
sewer line
CPT-RAM15*
MHT-17C *
CRT-RAM 4 *
Figure 4-2. SRS M-Area Well Locations.
been identified at WAFB. These include landfills with mixed leachate, gasoline and jet fuel spills, a fire
fighting training area, leaking underground storage tanks, an airplane crash site, and pesticide
contamination.
Contamination has spread to soil and ground water under approximately 20 percent of the base. A number
of VOC contaminants, some of which are identified in Table 4-2, are commingled at the site. The ground
water contaminants include: chlorinated solvents such as DCE, TCE, PCE and chlorobenzenes; polycyclic
aromatic hydrocarbons (PAHs); aromatic hydrocarbons such as benzene, toluene, ethylbenzene, and
xylenes (BTEX); and, other hydrocarbons such as aldehydes, ketones, gasoline, and jet fuel. Many of the
VOC contaminants are found in the capillary fringe at the water table as part of a non-aqueous or free
phase hydrocarbon medium. Contaminant concentration levels in this medium can be several orders of
magnitude higher than in the ground water. Current remediation efforts at WAFB include three pump-and-
treat systems using air strippers.
Table 4-2. Historical Ground Water Contamination Levels at WAFB.
Cone. Level
Low
Med
Hish
DCE
<1
200
700
TCE
<1
<1
2
PCE
<1
<1
<1
Benz.
<1
20
250
Ethyl Benz.
<1
300
1200
Tol.
<1
10
400
Xyl.
<1
200
600
Chlorobenzene
<1
5
30
DCB
<1
5
20
Note: Concentration levels in units of fj,g/L.
WAFB Geologic and Hydrologic Characteristics
The WAFB site rests on a 30-80 foot thick layer of clean, medium-grained sand and gravel sediments
formed by glacial meltwater, channel, deltaic and upper shore face-beach depositional processes. This
surface layer is underlain by a 100-250 foot thick layer of silty-clay deposited through settlement of the silt
18
-------
A
N
\Van Etten
Lake
WAFB
Fire Train!
Au Sable River
River Road
Lake Huron
Old Highway 23
Figure 4-3. Location of Wurtsmith Air Force Base.
and clay-sized particles from glacial meltwater following glacier retreat after the glacial episodes of the
Pleistocene Epoch. This layer lies on top of bedrock that consists of Mississippian sandstone and shale
formations that have a structural dip to the southwest into the Michigan Basin. The water table ranges from
about 5 feet below land surface in the northern regions to 20 feet below land surface in the southern
regions. A ground water divide runs diagonally across the base from northwest to southeast. South of the
divide, ground water flows toward the Au Sable River, and north of the divide, toward Van Etten Creek
and Van Etten Lake. Eventually, all water from WAFB reaches Lake Huron.
WAFB Demonstration Site Characteristics
The demonstration area selected is located at the former Fire Training Area 2, near the southern boundary
of the base (Figure 4-4). A wide range of organic contaminants from former fire training and other
19
-------
Figure 4-4. WAFB Fire Training Area 2 Sampling Locations. The cross-hatched region shows the approximate location
of the below-ground contaminant plume. A number of deep (D), medium (M), and shallow (S) well locations
are also shown.
Table 4-3. VOC Concentrations in WAFB Fire Training Area 2 Wells.
Cone. Level
Low
Medium
High
Water
Well
FT5S
FT3
FT8S
Benzene
(Mg/L)
0.24
20
225
Toluene
(Mg/L)
0.20
15
2
Xylenes
(Mg/L)
20
400
1800
Soil Gas
Well
SB3at4'
SB3at7'
SB3 at 10'
Total VOCs
(ppm)
30
55
62
activities exist in the soil and ground water at the site. Based on historic data, over fifteen organic
contaminants exist at the site. Additionally, high background levels of petroleum hydrocarbons such as jet
and diesel fuel exist at the site. Historic contaminant concentration levels are listed in Table 4-3 for the
monitoring wells at the Fire Training Area. The monitoring wells at this site are often clustered together
with one well screened at a shallow depth, denoted by an (S), and one screened at a deeper depth denoted
by a (D). No historical data regarding the expected soil contamination levels were available for the site.
20
-------
Overview of the Field Demonstrations
The demonstrations were designed to evaluate both the analytical and operational capabilities of the field
GC/MS technologies under representative field conditions. The analytical method for the operation of the
Bruker GC/MS is provided in Appendix A. The SRS field demonstration was conducted in July 1995 and
lasted three days. The technologies arrived at the demonstration site on Monday, July 17. As is typically
the case for this part of the country in mid-summer, the weather was hot (up to 95 °F) and humid but with
no rain. Each day the technologies arrived at the site about 6:30 a.m. They were set-up, calibrated, and
ready for sample analysis by about 7:30 a.m. Sample analysis typically lasted through mid-afternoon. Soil
vapor samples were prepared and analyzed on-site by the participants on Tuesday, July 18. The water and
soil samples were collected and analyzed by the participants on Wednesday and Thursday, respectively.
Each developer provided their own transportation, personnel, and equipment needed to conduct their
analyses. At SRS, the developers were required to provide their own electrical power as part of their field
operations. The field demonstration was completed by Friday, July 21.
The WAFB field demonstration was conducted in September 1995. The participants arrived at the
demonstration site on Sunday, September 10. The weather was generally cool, typically 40°F in the
mornings, warming to about 70°F during the afternoons. No appreciable precipitation was encountered
during the demonstration. Each participant arrived with their respective instrument early in the morning.
Following set up and calibration, instruments were ready for sample analysis by 7:30 a.m. Sample
collection and on-site analysis took three days, one day for each media. A fourth day was used as a "media
day" to showcase the participating technologies. As at SRS, each developer provided their own
transportation, personnel, equipment, etc., to conduct the sample analyses.
Overview of Sample Collection, Handling, and Distribution
Soil gas, water, and soil samples were collected during the demonstrations at both sites. Sample splits were
provided to the technology developers for on-site analysis the day of the sampling and shipped to reference
analytical laboratories for analysis using conventional methods. Formal chain-of-custody forms were used
for distribution of the samples to each of the reference laboratories. The samples were collected, numbered,
stored, and shipped to the laboratories in accordance with laboratory procedures that incorporate EPA
sampling guidelines. Somewhat less formal chain-of-custody records were maintained for distribution of
the samples analyzed on site. An overview of the site-specific sampling plans and the procedures for
collecting, handling, and distributing the samples is presented below. Additional sampling details can be
found in the demonstration plan referenced earlier. A description of the sampling terminology used in the
context of this demonstration is presented in Table 4-4.
SRS Sample Collection
A total of 33 samples were collected and analyzed in the SRS demonstration. The samples were distributed
among the three sample media, soil gas, water, and soil, as identified in Table 4-5. Sample collection and
on-site analysis took place over a three day period in July 1995. Water and soil gas samples were obtained
from the six M-Area wells identified in Table 4-1. The principal analytes were TCE and PCE at
concentration ranges noted in the table, but other contaminants such as TCA, Freon 11, Freon 113, and
their degradation products were sometimes present at lower concentrations in the wells.
SRS Soil Gas Survey
Wells MHV-2C, CPT-RAM 15, and CPT-RAM 4, shown in Figure 4-2, were sampled using Tedlar™
bags and SUMMA™ canisters. The Tedlar™ bags were used for on-site analyses and the SUMMA™
21
-------
Table 4-4. Sample Terminology and Description.
Term
Method Blanks
Spike Samples
Performance
Evaluation
Samples
Duplicate or Split
Samples
Description
Method blanks are samples which do not contain the target analytes. Water blanks
consisted of deionized water; Soil blanks consisted of uncontaminated soil representative
of the site being sampled; Soil gas blanks consisted of dry nitrogen gas.
Spike samples are generated by adding a known amount of analyte to a sample matrix.
Spike samples are used to evaluate the accuracy of an instrument by comparing the
concentration measured to the prepared reference concentration (spike recovery).
Performance evaluation (PE) samples are samples having a certified concentration for
specific analytes of interest. PE samples may include dilutions of a certified sample where
so noted. PE samples are also used to evaluate the accuracy of a technology or laboratory
during sample analysis by comparing the measured concentration to the defined reference
concentration.
At both SRS and WAFB, water PE samples were obtained, diluted to appropriate
concentrations, and submitted for analysis to the developers and the reference laboratories.
At SRS, a soil vapor PE sample was generated using a VOC vapor standard from SRS. At
WAFB, the soil PE samples were acquired in sealed vials and were submitted to the
developers and laboratories. Each laboratory did their own dilutions as appropriate.
A duplicate sample is a split of an initial sample. Duplicate samples are used to evaluate
the precision of an instrument by comparing the relative difference between the duplicate
measurements. For water and soil gas samples, a duplicate sample is often considered a
second sample taken sequentially from the same well.
Table 4-5. SRS Demonstration Sample Type and Count.
Media
Soil gas
Water
Soil
Concentration
Level
Blank
Low
Medium
High
Blank
Low
Medium
High
Blank
Low
Medium
High
Samples
2
1
1
1
2
1
1
1
1
Duplicates
1
1
1
1
1
1
Spikes
1
1
1
1
1
1
3
3
PE Samples
1
1
2
Total
13
13
7
canisters were sent to the reference laboratory for analysis. For the soil gas survey, soil vapor2 from each
A soil gas survey is conducted to measure the vapor phase concentration of VOC contaminants in a soil sample.
This vapor phase contaminant concentration is commonly referred to as the soil gas concentration. The terms soil
vapor and soil gas are used interchangeably in this report.
22
-------
well was pumped sequentially into three Tedlar™bags. The first bag was used to fill a SUMMA™ canister
while the other two bags were used for analysis by the developers. A sample aliquot was taken directly
from the Tedlar™ bags by each developer for analysis on a round-robin basis.
Additional aliquots were taken by the SRS on-site laboratory prior to and following drawing developer
samples from each bag in order to monitor the stability of both TCE and PCE in the bags during the
analysis. A blank sample, a Tedlar™ bag filled with nitrogen, was provided to participants at the beginning
and another blank provided at the end of the analyses. Spike samples were prepared by the SRS on-site
laboratory by injecting a known volume of TCE and PCE into a Tedlar™ bag filled with a known volume
of nitrogen. Two gas PE samples from certified cylinders were metered into Tedlar™bags for analysis by
the participants. Sample aliquots were also taken by the developers from each of the PE and blank sample
bags on a round-robin basis.
SRS Water Sampling
Water samples were collected from wells MHT-11C, MHT-12C, and MHT-17C (Figure 4-2). Each well
was initially purged and a 2 liter sample collected. Immediately after collection, the sample was sealed and
stirred for 10 minutes. The homogenized sample was then split into individual sample vials for distribution
and analysis. Two blank samples consisting of deionized water were provided to the participants for
analysis. Two water PE samples, prepared for the EPA's Hazardous Substances Evaluation Division in
Washington, DC for use in the Contract Laboratory Program, were also provided to the participants for
analysis. For on-site analysis, the PE sample ampules were mixed with the appropriate volume of water to
obtain the defined reference concentration. For the reference laboratory analysis, the ampules were
provided directly to the laboratories without prior dilution.
SRS Soil Spike Samples
Soils and sediments at the demonstration site are highly contaminated with PCE and TCE. Accurate
analyses of the sediments at depths greater than 50 feet, where contaminant levels appropriate to the
requirements of this demonstration exist, have been difficult because the SRS soils have low organic
content and the VOCs do not bind well to the soil matrix. The expense of drilling and sampling at these
depths and the composition of the soil at the site indicated that collecting standard soil core samples for
analysis was inappropriate for this demonstration. In order to maximize the amount of data that could be
derived from the demonstration under these circumstances, the demonstration plan contained a procedure
for using spiked soil samples in the place of soil core samples. This procedure called for soil to be collected
from an erosion pit at SRS, homogenized, spiked with solutions of TCE and PCE, separated into 5-gram
portions, and placed in 40 mL VOA vials equipped with screw-top lids and septa. This vial configuration
allowed participants the option of either purge and trap or head space sample introduction and analysis. Actual
sample preparation was done differently from that presented in the demonstration plan. These deviations are
discussed later in this section.
SRS Sample Handling, Storage, and Shipping
Developers analyzed the samples as soon as practical following collection or preparation, but generally within an
hour of sampling. Formal chain of custody protocol was maintained for the reference laboratory samples. The
field-analyzed samples had less formal custody procedures, but all transfers were recorded in log books.
Samples collected for laboratory analysis were transported to the reference laboratory at the end of each day.
Possible loss of VOCs in the samples was a major concern; therefore, all water, soil vapor, and soil samples
were stored and shipped in coolers maintained at approximately 40°F.
23
-------
WAFB Sample Collection
A total of 37 samples were collected and analyzed in the WAFB demonstration. The distribution of the samples
from each media, soil, water, and soil gas are presented in Table 4-6. Sample collection and on-site analysis took
place over a three day period in September 1995. Water and soil gas samples were obtained from the wells
identified in Table 4-3. Fiistorical sampling and analysis data show VOC concentrations ranging from 0.2 to
1800 //g/L in water and total VOC concentrations in soil gas ranging from 30 to 62 ppm.
Table 4-6. WAFB Demonstration Sample Type and Count.
Media
Soil gas
Water
Soil
Concentration Level
Blank
Low
Medium
High
Blank
Low
Medium
High
Blank
Low
Medium
High
Samples
2
1
1
1
2
1
1
1
2
1
1
1
Duplicates
1
1
1
1
1
1
2
2
2
Spikes
1
1
1
1
1
1
PE Samples
2
2
Total
11
13
13
WAFB Soil Gas Survey
Well SB3, shown in Figure 4-4, was sampled at three depths using Tedlar™bags. At each depth, two bags were
sequentially filled. The bags were used for on-site developer analyses and the residual gas in each bag was used
to fill SUMMA™ canisters for analysis by the reference laboratory. For on-site analysis, sample aliquots were
drawn from the bags by each developer in a round-robin format like that used at SRS. A blank sample, a
Tedlar™ bag filled with nitrogen, was used at the beginning and at the end of the soil gas analytical sequence. A
spiked sample for each concentration level was made by injecting a known volume of liquid into a Tedlar™bag
filled with a known volume of nitrogen.
WAFB Water Sampling
Water samples were collected from wells FT5S, FT3, and FT8S (Figure 4-4). Each well was purged for ten
minutes. Water samples were then drawn to sequentially fill standard 40 mL VOA vials. A blank sample
consisting of deionized water was provided to each developer at the beginning and at the end of the analysis run.
Two water PE samples were provided to each developer for analysis. Two samples were provided to each
participant from each well for duplicate analysis.
WAFB Soil Sampling
Three soil samples were obtained as sub-cores from a sediment boring taken with a two-inch diameter Geoprobe
at a location 100 feet south of well FT6, identified as SS-1 in Figure 4-4. On-site photoionization detector
readings taken while drilling allowed the core sample to be subdivided into segments having varying levels of
VOC contamination. The three segments, 8 to 9, 9 to 10, and 10 to 11 feet below the surface, were each
homogenized, split, and placed in vials. All soil samples were weighed using a calibrated balance. The reference
laboratory and each developer received splits from the homogenized samples for analysis. Two soil PE samples,
24
-------
prepared by the Army's Cold Regions Research and Engineering Laboratory in Hanover, New Hampshire were
also provided to the participants for analysis.
WAFB Sample Handling, Storage, and Shipping
After the water and soil samples were collected, they were placed in a cooler containing double-bagged ice to
maintain an approximate 40 °F temperature. The water samples sent to the reference laboratory were preserved
with 1 percent sodium bisulfate (NaHSO,), and placed in an ice-filled cooler for storage and shipment. The soil
gas samples for reference laboratory use were transferred to SUMMA™ canisters for shipment to the laboratory.
All reference laboratory samples were picked up by the laboratories at the end of each day. All samples collected
for on-site analysis during the demonstration were presented to the developers for analysis as soon as practical.
Analysis always occurred on the day of sample collection and often within an hour of collection.
Reference Laboratory Selection and Analysis Methodology
One objective of this demonstration was to determine how well each developer's field instrument performed in
comparison to conventional laboratory methods and protocols. Standard analytical methods applicable to the
sample media and analytes of interest in these demonstrations were selected as the standard of comparison.
These include EPA SW-846 Method 8260 Gas Chromatography/Mass Spectroscopy for Volatile Organics:
Capillary Column Technique for water and soil analyses, and EPA Compendium Method TO-14 The
Determination of Volatile Organic Compounds in Ambient Air Using Summa Passivated Canister Sampling
and GC/MS Analysis for soil gas analyses.
The selection of reference analytical laboratories for this demonstration was based on consideration of several
criteria including:
• Certification in one or more states;
• Recommendation from the site manager of prior use;
• Proximity to the site (generally within 3 hours driving time) to minimize sample transport and
handling;
• Proven capability to measure VOCs at the required concentration ranges in the appropriate media
and in accordance with the selected analytical methodologies as determined from a review of
quality assurance (QA)/quality control (QC) data results;
• Proven capability to provide an analytical data package consistent with the requirements of Method
8260 and TO-14 as determined by a review of QA/QC data results; and,
• Passing a pre-demonstration audit by Sandia that included a review of facilities, personnel, QA/QC
procedures, protocols, and overall operations.
Based on recommendations from each of the sites, several analytical laboratories were identified for possible use
as reference laboratories for the demonstration. Each laboratory identified was asked to provide information on
its QA/QC procedures and sample analysis data, using the same methods, for review. Based on a review of this
information, discussions with the site managers, and discussions with other users, three laboratories were
identified for further evaluation. These were the General Engineering Laboratory, Inc. for the SRS
demonstration, and Traverse Analytical Laboratory and Pace Incorporated Environmental Laboratories for the
WAFB demonstration.
Further evaluation included a pre-demonstration audit of each of the laboratories. Each pre-demonstration audit
included meeting with laboratory personnel; touring the facility; and reviewing laboratory operations, personnel
qualifications, and laboratory QC procedures. Chain of custody procedures, sample holding areas and
procedures, and analytical equipment operation were also reviewed. During the Traverse pre-demonstration
audit, it was determined that soil gas analysis was not routinely conducted at their laboratory and that Traverse
25
-------
commonly subcontracted soil gas analyses to Pace. Therefore, Pace was considered for soil gas analysis for the
WAFB demonstration and was audited in the same manner as GEL and Traverse.
General Engineering Laboratory
General Engineering Laboratory (GEL) is located in Charleston, South Carolina, and is certified in South
Carolina, Georgia, Alabama, and Florida. This laboratory is close to the SRS facility and has been used
extensively by SRS to analyze environmental site characterization and monitoring samples. GEL provided
QA/QC documentation of their operations for review. Based on the results of the pre-demonstration audit, GEL
was found to conduct their analyses in accordance with the analytical methods identified for this demonstration
and was selected as the reference laboratory for the SRS demonstration. GEL personnel picked up and
transported samples daily to their laboratory in Charleston and performed the analyses for all three media. The
final data packages provided also included the corresponding QC results.
Traverse Analytical and Pace Environmental Laboratories
Traverse Analytical Laboratory is located in Traverse City, Michigan, and is certified by the state of Wisconsin.
Michigan has no laboratory certification program. This laboratory is close to WAFB and had been previously
used by WAFB for environmental sample analysis. Traverse provided QA/QC documentation of their operations
for review. The pre-demonstration audit by Sandia determined that their analyses were conducted in accordance
with the analytical methods identified for this demonstration. Based on these audit results, Traverse was selected
as the reference laboratory for soil and water analyses for the WAFB demonstration.
Traverse suggested that Pace in Camarillo, California, conduct the soil gas analyses for the WAFB
demonstration. Pace provided QA/QC documentation on their analytical procedures for review. Based on the
pre-demonstration audit of the laboratory by Sandia, they were found to conduct their analyses in accordance
with the analytical methods identified and were selected as the reference laboratory for soil gas analysis for the
WAFB demonstration. Traverse personnel picked up and transported samples daily to their laboratory and
conducted the analyses on the water and soil samples. The soil gas samples were shipped from Traverse to Pace
for analysis. Both laboratories provided data packages with accompanying QC results.
SRS and WAFB On-Site Laboratories
In addition to the selected reference analytical laboratories, the on-site laboratories at both SRS and WAFB
provided rapid analyses to confirm general sample integrity and to ensure that the samples collected contained
the expected levels of contamination. Both on-site laboratories use gas chromatograph systems for routine
sample analysis. The laboratory analytical methods and corresponding QC protocol were reviewed by Sandia to
confirm their use of acceptable laboratory procedures.
Pre-demonstration Sampling and Analysis
Pre-demonstration sampling and analysis were conducted at both demonstration sites to establish that samples
from the sites were appropriate for analysis by the GC/MS technologies and that the technology results could be
objectively compared to reference laboratory data. The pre-demonstration activities allowed the technology
developers an opportunity to refine their systems, revise operating procedures as necessary, and evaluate media
effects or interferences that could influence their analytical results. The pre-demonstration sampling events
required one field day at each site and took place on June 5, 1995 at SRS and on July 30, 1995 at WAFB.
The pre-demonstration samples consisted of one SUMMA® canister of soil gas from a medium-concentration
soil gas well at each site (CPT-RAM 15 at SRS and SB3 at WAFB), and three water samples from a medium-
concentration well (MFTT-12C at SRS and FT4S at WAFB). No pre-demonstration soil samples were provided
for developer analysis from either site. The soil gas and water samples were split and sent to the developers and
the respective reference laboratories for analysis. The reference laboratories used EPA Compendium Method
TO-14 for soil gas analysis and EPA SW-846, Method 8260 for water analysis. Unfortunately, the results from
these laboratory analyses were not available prior to the SRS demonstration to help guide the technology
26
-------
developers. The SRS and WAFB pre-demonstration sample analytical results were available to assist the
developers prior to the WAFB demonstration. No interpretation or analysis of the analytical results was
conducted since these data were intended primarily to assist the developers in refining their operations and
procedures for this demonstration.
Deviations from the Demonstration Plan
Several deviations from the demonstration plan occurred as the demonstrations progressed and are discussed
below.
Pre-demonstration Activities
The pre-demonstration activities identified in the demonstration plan called for analysis of pre-demonstration
samples to allow the technology developers to refine their methodologies, revise operating parameters, and identify
matrix effects or interferences. Predemonstration sample analytical results from the reference laboratory were not
available prior to the SRS demonstration. Also, no soil samples from either site were provided for evaluation of
media effects. The omission of pre-demonstration soil sampling at SRS led to soil sampling problems at SRS
during the demonstration.
SRS Soil Spike Samples
A number of deviations occurred in the soil sample preparation procedures used during the SRS demonstration.
The demonstration plan deviations were not discovered until the day of the sample analysis. Consequently, new
soil spike samples could not be prepared in the allotted time. These deviations in sample preparation were
judged to be significant. Consequently, no assessment or comparison can be made between the Bruker GC/MS
and the reference laboratory on the results of soil sample analyses.
SRS Soil Gas Survey Evaluation
Of the thirteen soil vapor samples analyzed by GEL, ten were reported as estimated values. In particular, these
data were obtained by extrapolation of the calibration curve beyond the normal calibration range of the
laboratory instrument used for analysis. This deviation is significant in that no quantitative information on
laboratory precision or accuracy can be derived from such data. Consequently, the GEL soil gas data were not
used as a reference data set for comparison with the Bruker GC/MS data.
Soil Gas Samples at WAFB
No initial or final analysis of each Tedlar™bag gas sample was conducted by the field laboratory as called for in
the demonstration plan. Therefore, no data are available on the stability of the vapor sample in these bags during
the round-robin sampling and analysis by the developers. Based on the results from a similar analysis at SRS
where the vapor sample was determined to be very stable, this is not considered a significant deviation.
Condensation was noted by the participants in the Tedlar™bags from the well samples. It was not possible to
determine if this was liquid contaminant or simply water condensation from the samples given that the ambient
air temperatures during the morning of the sampling was below the soil temperatures in the sampling well. In
either case, the analytical results from these samples could differ as a result of variation in the vapor phase
sample constituent concentrations over the sample handling and transfer interval. No determination could be
made of the significance of the condensation and its effect on the analytical results.
Appropriate soil gas PE samples could not be obtained from suppliers in time for the WAFB demonstration.
Consequently, the number of samples to evaluate laboratory and developer accuracy was limited. To minimize
the significance of this deviation, the number of spike samples prepared on-site was increased.
Water Samples at WAFB
27
-------
The compound 1,4-difluorobenzene was included in the spike mixture used for water samples at WAFB. This
compound is normally used as a Method 8260 internal standard by the Traverse reference laboratory. Inclusion
of this compound in the spike mixture required the reference laboratory analyst to revert to an external standard
method for quantitative measurements of 1,4-difluorobenzene in the spike samples. Reference laboratory data
quality was not adversely impacted by use of the external standard method.
Calibration Check Sample Analysis
The demonstration plan called for the participants to run calibration check samples throughout the day in order
to facilitate the assessment of instrument stability. Given the intensity of the schedule for analyzing the samples
as quickly as they were distributed, periodic analysis of calibration check samples was not completed. The
calibration data were not used in evaluating instrument performance, but were only for use by the instrument
operator. This deviation was not considered significant.
28
-------
Section 5
Reference Laboratory Analysis Results and Evaluation
An important objective of this demonstration was to provide the technology developers with a validated
data set from conventional laboratory analyses of water, soil, and soil gas samples. Validated laboratory
results are essential for direct comparison with the analytical results from the field methods under
evaluation. This section describes a number of qualitative and quantitative data quality indicators that were
used to evaluate and validate the analytical results from the reference laboratories. Qualitative factors
reviewed included adequacy of laboratory QA/QC procedures and deviations from standard procedures.
Quantitative factors reviewed included accuracy and precision of the reference laboratories' analyses of
reference samples. The laboratories evaluated were the General Engineering Laboratories for the Savannah
River Site demonstration, and the Traverse Analytical Laboratory and Pace Inc. Environmental
Laboratories for the Wurtsmith demonstration. The on-site SRS laboratory was evaluated to a limited
extent as a result of some data limitations encountered with GEL. The National Center for Integrated
Bioremediation Research and Development laboratory was used as an on-site screening laboratory and not
in a reference capacity at the Wurtsmith demonstration. Consequently, no formal data quality evaluation
was done for this laboratory.
Laboratory Operations
General Engineering Laboratories
Prior to the demonstration, GEL provided a quality assurance plan that described laboratory and personnel
capabilities, analytical methods, and internal quality control procedures. A complete description of the
analytical methods used in the soil, water, and soil gas sample analysis was also included in their plan. The
laboratory quality assurance plan was prepared using EPA guidance [U.S. EPA, 1991]. Water and soil
analyses were done using EPA SW-846, Method 8260 for purge and trap GC/MS. Soil gas analysis
followed EPA Compendium Method TO-14. A number of laboratory performance quality control
indicators were provided in the data package including: daily mass spectrometer tuning results, daily
calibration check results, daily blank check results, continuing calibration check results, and surrogate
compound recovery results.
SRS On-Site Laboratory
The SRS on-site laboratory was intended to provide rapid on-site analysis to assist in determining sample
integrity during the demonstration, and was not identified as a reference laboratory in the demonstration
plan. Analyses were conducted using a Hewlett Packard 5890 GC with flame ionization and electron
capture detectors. Water samples were analyzed using headspace methods, while the soil gas analysis
followed EPA Method TO-14. All samples were analyzed on-site the day of the sampling. No formal
QA/QC plan was obtained from the SRS laboratory prior to the demonstration. When SRS data were
recognized as a possible replacement for GEL data, an informal quality control package was obtained from
laboratory personnel that documented the calibration performance of the GC system for several months
prior to the demonstration.
Traverse Analytical Laboratory
Prior to the demonstration, Traverse provided a complete quality assurance plan much like that submitted
by GEL. Water and soil analyses were conducted using EPA SW-846, Method 8260 for purge and trap
GC/MS and samples were analyzed within 14 days of receipt. As a result of equipment limitations,
Traverse did not conduct any soil gas analyses. Instead, this activity was subcontracted to Pace Inc.
Environmental Laboratories. A number of laboratory performance quality control indicators were provided
in the Traverse data package including: mass spectrometer tuning results, daily calibration check results,
29
-------
blank check results, continuing calibration check results, internal duplicate analysis data, and surrogate
recovery analysis data on selected compounds.
Pace Inc. Environmental Laboratories
Pace conducted soil gas analysis for the Wurtsmith demonstration under contract to Traverse. Prior to the
demonstration, Pace provided a complete QA package similar to those provided by the other reference
laboratories. Complete method descriptions were also given in the QA/QC plan. The analyses were carried
out in accordance with EPA Method TO-14. A number of internal laboratory quality control indicators,
including blank and internal laboratory gas spike analytical results, were provided in the data package.
WAFB On-Site Laboratory
The on-site laboratory at WAFB is the NCIBRD Field Laboratory. The function of this laboratory was
much like that of the on-site SRS laboratory during the SRS demonstration. The laboratory was used
primarily for on-site, quick-turnaround analyses to help in the assessment of field sample integrity prior to
distribution to the participants. Water analyses were conducted using a Perkin-Elmer gas chromatograph
with a Tekmar 2016 purge and trap system following EPA Method 502. Samples were analyzed on-site the
day of sample collection. A photoionization detector was used for on-site soil gas screening prior to sample
distribution to the participants. Soil sample analysis was not conducted by this laboratory.
Laboratory Compound Detection Limits
Detection limits for various compounds are identified for each EPA method. These limits vary both by
compound and sample media type. Each of the data packages provided by the reference laboratories
defined practical quantitation limits (PQL) for each media and analyte. The PQL is defined as the level at
which instrument noise and method inaccuracies have negligible effects on the accuracy and precision of
the analytical results. This value is commonly considered to be about three times higher than the method
detection limit. The PQLs for each of the laboratories are presented in Table 5-1. Although the PQLs are
analyte and method specific, this table is provided to illustrate the range of PQLs for all target analytes in
the three media for each of the four laboratories generating reference data used in this demonstration.
Table 5-1. Reference Laboratory Practical Quantitation Limits.
Laboratory
GEL
SRS
Traverse
Pace
Media
Water
Soil Gas
Soil
Water
Soil Gas
Soil
Water
Soil
Soil Gas
PQL
2-g/L
lOppb
100 • g/kg
2-g/L
lOppb
100 • g/kg
l'g/L
100 • g/kg
lOppb
Laboratory Data Quality Assessment Methods
All analytical data are subject to some level of inaccuracy and imprecision. This section discusses the
methods used in determining the level of confidence placed in the analytical results from the reference
laboratories participating in these demonstrations.
30
-------
Precision Analysis
Precision is a measure of the degree to which repeated analytical measurements of the same sample agree
with one another. In the context of this study, precision is indicative of the random errors associated with
the measurement process and is intended to yield a measure of the variability encountered in the normal
operation of a laboratory or field instrument. In the absence of any inaccuracy or bias in a measurement,
repeated determinations of a given analyte in a single sample will be evenly distributed above and below
the analyte's true concentration, and the average of several measurements will be a better estimation of the
true value than is any individual measurement. A simple way to express precision for duplicate
measurements is relative percent difference (RPD), which is defined as follows:
x -x I
X
where Xj and x2 are the duplicate measurements and x is the average of Xj and x2. The precision of
reference laboratory measurements is assessed by using analytical results from duplicate field, PE, or spike
samples. However, caution is warranted where sequential samples, often called duplicates, are drawn from
a well. These sequential samples may not be equivalent and care must be taken in their evaluation as
duplicates.
The standard methods employed in this demonstration generally call for RPD values of 20% or less as an
indicator of acceptable analytical precision. In some sampling media these criteria are relaxed to values as
high as 50% for selected compounds. The specific acceptance criteria are discussed in more detail in the
sections dealing with the laboratory data evaluation.
Accuracy Analysis
In the context of this demonstration, accuracy is defined as the agreement between the measured
concentration of a reference sample and the accepted or "true" concentration of the sample. Bias is a term
that is related to accuracy. Bias can be either positive or negative depending on whether the measured
values are consistently higher or lower, respectively, than the true value, whereas accuracy is normally
given in terms of absolute variation with no reference to positive or negative direction. An observed bias
indicates the presence of systematic errors in the measurement process. For example, a calibration error in
the setup of an instrument may produce a consistent negative or positive bias in the measurement results.
Consistently lower values from a particular method may be indicative of evaporation losses, chemical
reactions, biological degradation or other analyte loss mechanisms.
Accuracy is often reported in terms of percent recovery. The analysis result from a sample run on an
instrument can be compared with the "true" or "reference" value of the sample and expressed in terms of
percent recovery (REC). The percent recovery is computed as follows:
x
SZ = — • D
x
where ~x.instrument is the measured concentration by a field instrument and xre^reBce is the true concentration.
The evaluation of accuracy is considerably more difficult than the evaluation of precision since there
always exists some uncertainty in the "true" value of the reference material's concentration. While
precision can be measured in the absence of information about the true concentrations, accuracy cannot be
assessed without some level of confidence in the reference value used in the determination.
The accuracy of reference laboratory measurements is assessed by analyzing two types of reference
materials, namely, performance evaluation (PE) samples and media spike samples. PE samples are
typically purchased chemical standards with an accompanying certification of the chemical composition of
31
-------
the sample. In some cases, the PE samples require additional preparation in the field, for example, sample
dilution prior to distribution. Media spike samples are prepared by adding known quantities of the pure
chemicals of interest to uncontaminated samples of the various media. For these demonstrations, the spike
samples were prepared by the on-site laboratories and provided to all participants, including the reference
laboratories, for analysis. The quality of the prepared media spikes must always be reviewed carefully
because of the potential errors that can occur during their preparation. Since they are prepared in the field,
a certificate of analysis is not available to confirm the composition of the sample.
Standard EPA methods such as TO-14 and 8260 indicate that appropriate recovery levels for the
compounds of interest in this demonstration should generally be in the range of 80-120%. In some cases,
the acceptable recovery level range is extended to 50-150% or greater depending on the nature of the
media and the analyte. In most cases, these recovery levels are empirically derived by the laboratory during
routine method use and are incorporated into the laboratory's QA plan. The specific acceptance criteria for
each laboratory are discussed in more detail in the sections dealing with laboratory data evaluation.
Laboratory Internal Quality Control Metrics
Each of the reference laboratories provided internal quality control data along with their analytical results.
These data were used as one of several indicators of laboratory data quality. Specific laboratory internal
quality control indicators that were evaluated are discussed below.
Blank Analysis
The results from the analysis of blank samples are used primarily as a measure of instrument contamination
and as a secondary check on compound detection limits for the laboratory instruments.
Continuing Calibration Check
A continuing calibration verification procedure uses a calibration solution containing target analytes that is
periodically analyzed during a sample batch analysis. The analysis results are recorded as a series of
percent recoveries relative to the starting calibration value. The procedure gives an indication of the
calibration or accuracy drift of the instrument over time. Control limits of ±25% are normally applied for
Method 8260 for water and soil and Method TO-14 for soil gas. Values falling outside these limits are
suggestive of inadequate analytical process control and questionable data quality.
Internal Duplicate Analysis
The standard methods for water, soil, and soil gas analysis require periodic duplicate analysis of both
standards and field samples. These data provide a measure of laboratory precision, often expressed in terms
of RPD, as described earlier. The methods used in this demonstration generally call for RPD values of less
than 20% in a specified concentration range. For example, Method TO-14 indicates that the RPD for
duplicate measurements must be within 20% only for those compounds detected at a level of 5 times
greater that the instrument detection level for the compound of interest. Significant variations in duplicate
sample measurements are indicative of inadequate analytical process control and questionable data quality.
Laboratory Data Quality Levels
Each of the reference laboratories data were evaluated and assigned one of three levels of data quality
based on laboratory internal quality control, accuracy, and precision results. This ranking method identifies
those laboratory data that do not meet commonly accepted data quality criteria and therefore are unsuitable
or inappropriate for comparison with field technology data. The data quality levels are further described
below:
32
-------
Good Data Quality
• Good laboratory internal quality control5 results (e.g., internal blanks, duplicates, continuous
calibration, and control samples);
• Analytical accuracy results, based on external (supplied by the project) PE and spike samples,
consistently within +30% of the reference concentrations; and,
• Analytical precision results, as determined by RPD on duplicate samples, consistently less
than 30%.
Satisfactory Data Quality
• Satisfactory laboratory internal quality control6 results;
• Analytical accuracy results, based on PE and spike samples, consistently within +50% of the
references concentrations; and,
• Analytical precision results, as determined by RPD on duplicate samples, consistently less
than 50%.
Unacceptable Data Quality
• Poor or missing internal quality control results; or
• Analytical accuracy results, based on PE and spike samples, consistently exceeding +50% of
the reference concentrations; or,
• Analytical precision results, as determined by RPD on duplicate samples, consistently
exceeding 50%.
Laboratory Data Validation for the SRS Demonstration
GEL Data Quality Evaluation
GEL QA/QC procedures were audited by Sandia personnel prior to the demonstration and were found to
be operating in accordance with accepted good laboratory practice and the requirements outlined in the
standard methods used in analysis of these demonstration samples. GEL analyzed all three sample media
types from the SRS demonstration and provided a quality control data package with their analysis results.
These results and a discussion of the analytical data are presented below.
Two out-of-limit conditions were identified in the quality control data package sent along with the analysis
results. Eight water samples were flagged as missing the maximum holding time by one day. This
occurrence was judged not to have significant impact on data quality. Also, ten of thirteen soil gas sample
analyses were reported as estimated values since they were outside the calibration range for the species of
interest. These reported values were judged to be of unacceptable data quality as further discussed below.
5 Good internal lab quality control indicates that a complete QC package was received with the sample analysis data
and that the QC data were within method or laboratory guidelines.
6 Satisfactory internal lab quality control indicates that an incomplete QC package was received with the sample
analysis data but that the available QC data were within method or laboratory guidelines.
33
-------
GEL Internal Quality Control Data
The quality control data package revealed that the GEL GC/MS passed daily calibration and internal blank
checks during analysis of SRS samples. Surrogate chemical spikes were used for all seven soil samples
analyzed and recoveries were within the 80-120% range prescribed by Method 8260. Internal laboratory
duplicates of soil and water spikes gave RPD values that were less than 20%. Overall, the internal quality
control data reveal good laboratory procedures and results.
GEL Accuracy Data
The accuracy results for GEL are presented in Table 5-2 and derived from the analysis of spike and PE
samples containing TCE and PCE. EPA Method 8260 calls for empirical derivation of acceptable
compound recovery ranges by each lab as they routinely conduct analyses.
Table 5-2. GEL Laboratory Accuracy Data.
Sample Media/Description
Water Low Spike
Water Medium Spike
Water High Spike
Water PE Sample 1
Water PE Sanrole 2
Reference Concentration in • g/L
PCE
2.7
40.4
270
NP
198
TCE
2.6
38.3
256
NP
46
Percent Recovery
PCE
116
106
130
-
127
TCE
220
210
276
-
128
Notes: NP = Not Present, TCE or PCE not present in sample
The GEL water analyses results reveal percent recoveries for PCE for all samples within the acceptable
range of 64-148% as provided in GEL's implementation of Method 8260 in their QA plan. However, the
spike recoveries for TCE in water are in excess of 200%. Recovery limits for TCE, stated in GEL's
implementation of Method 8260, are 71-157%. On the other hand, the TCE recovery values for PE Sample
No. 2 is within prescribed limits. The result for the PE sample is given precedence over the results from the
spike samples since the water spikes were prepared in the field by SRS laboratory personnel and their
reference values were not independently certified.
As noted previously in Section 4, problems were encountered in soil spike preparation at the SRS
demonstration and the overall quality of the spikes was judged to be unacceptable as reference material.
No accuracy determinations were made for the soil gas samples because all the analytical values reported
by GEL were estimates.
GEL Precision Data
The precision analysis results for GEL are presented in Table 5-3 and are based on the results of duplicate
sample analyses.
Table 5-3. GEL Laboratory Precision Data.
Sample Media/Description
Water Low
Water Medium
Water High
Reference Concentration in • g/L
PCE
10
150
12,200
TCE
60
160
6,000
Relative Percent Difference
PCE
3
2
5
TCE
22
5
2
34
-------
Precision results for the water duplicate samples show good precision for both PCE and TCE for all three
samples. Precision results for the soil gas duplicate samples could not be evaluated since all the reported
values were estimated.
GEL Data Quality Summary
The soil gas analysis results from GEL indicate that this data set is unsuitable for comparison with the field
technology results. The soil gas analyses are unacceptable because the sample analysis values were
estimated values only and cannot be used for comparisons with the field technologies. The precision data
for the water analyses were judged to be good. The accuracy data for the water analyses were also good if
the recovery data for the TCE spike samples are discounted. The general data quality for the GEL water
analyses was considered good and suitable for comparison with the different technologies.
SRS On-Site Laboratory Data Quality Evaluation
As noted previously, the SRS Laboratory analyzed soil, water, and soil gas samples during the SRS
demonstration; however, their data were not originally intended for reference use. Because of the semi-
quantitative nature of the GEL soil gas data, the SRS on-site soil gas data were evaluated as a possible
replacement. A post-demonstration evaluation of their soil gas data and accompanying quality control data
was carried out in the hope that the SRS laboratory data were of sufficient quality for comparison with
field technology results.
SRS Internal Quality Control Data
The SRS Laboratory Hewlett-Packard Model 5890 gas chromatograph, equipped with dual flame
ionization and electron capture detectors, was calibrated daily and internal blank checks showed acceptable
performance in terms of detection levels and instrument contamination. The SRS Laboratory also provided
a record of calibrations performed on their system for seven chlorinated compounds. These data give an
indication of the day-to-day variability of the GC system. Multiple analyses of standard solutions gave
relative standard deviations in the range of 2 to 11% for high (1,000 ppm) vapor concentrations; in the
range of 5 to 9% for medium (100 ppm) concentrations, and in the range of 6 to 11 % for low (10 ppm)
concentrations. These data indicate that the SRS GC system meets the 30% precision criteria, indicating
good overall quality control procedures and instrument performance.
SRS Accuracy Data
The accuracy data for the SRS Laboratory soil gas analyses are summarized in Table 5-4. The results are
based on the laboratory analyses of TCE and PCE spike and PE samples.
Table 5-4. SRS Laboratory Accuracy Data.
Sample Media/Description
Soil Gas Low Spike
Soil Gas Medium Spike
Soil Gas High Spike
Soil Gas PE Sample 1
Soil Gas PE Sample 2
Reference Concentration in ppm
PCE
1.18
118
1,182
1
93
TCE
1.34
134
1,340
1.1
98
Percent Recovery
PCE
56
76
99
61
94
TCE
78
77
91
91
95
Accuracy results for the SRS Laboratory soil gas sample analyses show that of the ten analyses conducted,
all but two fall within the accepted limits of 75-125% specified in method TO-14, while all of the results
are within +50%. Low recovery results are observed for the spike samples nearer the 10 ppm instrument
PQL. The SRS Laboratory GC-flame ionization detector data were used for this evaluation since this
-------
system was best suited for analysis of the medium and high level spike and PE sample concentrations.
Although the recoveries for these samples are consistently low, they generally indicate satisfactory
performance for both PCE and TCE over this wide concentration range.
SRS Precision Data
The soil gas precision data for the SRS Laboratory are presented in Table 5-5. The precision analysis of the
soil gas data shows RPD values less than 30% for PCE and TCE at all three concentration levels. This
indicates that the performance of the laboratory was good and these data can be used in the verification.
Table 5-5. SRS Laboratory Precision Data.
Sample Media/Description
Soil Gas Low
Soil Gas Medium
Soil Gas High
Reference Concentration in ppm
PCE
1
80
250
TCE
0.1
100
500
Relative Percent Difference
PCE
8
3
1
TCE
<1
3
3
SRS Laboratory Data Quality Summary
The accuracy and precision results for the soil gas samples consistently fall within 30% of the reference
values. Overall, the soil gas analyses results reveal satisfactory data quality, and are a suitable replacement
for the GEL soil gas data.
Laboratory Data Validation for the WAFB Demonstration
At least 15 VOC contaminants are known to exist at the WAFB demonstration site. Typical concentration
levels of these major contaminants in soil, water, and gas media were provided to the developers in the
demonstration plan [SNL, 1995]. As stated in Section 4, the contaminants at the WAFB Site include
BTEX, chlorinated hydrocarbons, and other organics. High background levels of petroleum hydrocarbons
(jet fuel) are encountered as well. Based on the information in the demonstration plan and the pre-
demonstration activities, each developer chose at least ten of the identified contaminants for analysis. Not
all the target contaminants were detected in all of the samples collected and not all of the developers chose
to analyze the same contaminants. Therefore, the data quality evaluation of the analytical laboratories
participating in the WAFB demonstration were based on the analytical results from five compounds that
were analyzed by each of the laboratories and the field technologies. The compounds used for evaluation
were benzene, toluene, total xylenes, PCE, and TCE. In a few cases, dichloroethene (DCE) was also
included where TCE or PCE was not detected.
Traverse Data Quality Evaluation
The laboratory QA/QC plan had been audited by Sandia prior to the demonstration. Laboratory operations
were found to be in accordance with good laboratory practice guidelines and the requirements stated in the
various standard methods used in the analysis of samples.
Traverse Analytical Laboratory analyzed soil and water samples from the WAFB demonstration and
provided Sandia with complete analysis results and accompanying quality control data package. A
complete description of the analytical methods used in the analysis was also included in the quality control
package. No out-of-limit quality control conditions were reported in the data package. The internal quality
control results and a discussion of the analytical data with respect to accuracy and precision are presented
below.
Traverse Internal Quality Control Data
36
-------
Daily quality control results for the GC/MS instrument used for the sample analyses included mass
spectrometer tuning, blank checks, and initial and continuing calibration checks. The quality control results
reveal good instrument performance throughout the course of the WAFB demonstration sample analyses.
Surrogate chemical spikes were also used for all seven soil samples analyzed and surrogate chemical
recoveries were within the 80-120% that is considered acceptable according to Method 8260. Internal
laboratory duplicates of selected soil and water spike samples also met the RPD criteria of less than 20%.
Overall, the internal quality control results indicate good quality control and instrument performance
during the analyses.
Traverse Accuracy Data
The accuracy data for Traverse is presented in Table 5-6 and is based on the results of spike recoveries and
PE sample recoveries of the five target analytes for the soil and water media analyzed. Table 5-7 provides
the reference concentration levels for each of these analytes.
Table 5-6. Traverse Laboratory Accuracy Data.
Sample Media/
Description
Soil PE No. 1
Soil PE No. 2
Water Low Spike
Water Medium Spike
Water High Spike
Water PE Sample 1
Water PE Sample 2
Percent Recovery
Benzene
89
66
97
119
76
78
81
Toluene
98
71
95
93
79
NA
89
Xylenes
87
65
112
106
81
101
101
PCE
154
102
105
96
75
NA
76
TCE
85
62
120
84
70
NA
85
Note: NA = not analyzed; analyte not present in sample
The percent recoveries for eight out often soil analyses fall within accepted recovery levels of 65-135%, as
given in the Traverse QA documentation. On the basis of these results, overall laboratory accuracy
performance is judged to be good. Likewise, the percent recoveries for the water analyses reveal recoveries
for the five target analytes, over a wide range of concentrations, well within the laboratory's acceptance
limits for all 22 analyses.
Table 5-7. WAFB Water and Soil PE/Spike Sample Reference Concentrations.
Sample Media/
Description
Soil PE No. 1 (• g/kg)
Soil PE No. 2 (• g/kg)
Water Low Spike (• g/L)
Water Medium Spike (• g/L)
Water High Spike (• g/L)
Water PE Sample 1 (• g/L)
Water PE Sample 2 (• g/L)
Reference Concentrations
Benzene
61,000
64,000
59
1,180
66,140
66
20
Toluene
55,000
59,000
45
904
50,620
NP
20
Xylenes
76,000
81,000
190
3,790
212,300
158
50
PCE
91,000
98,000
63
1,256
70,340
NP
20
TCE
7,900
8,600
27
715
40,040
NP
46
Note: NP = not present in sample
Traverse Precision Data
The precision data for Traverse is presented in Table 5-8 and is based on the results of duplicate and in
some cases triplicate analysis of the analytes in the two media. Table 5-9 shows the reference concentration
levels for each of the target analytes evaluated.
37
-------
Table 5-8. Traverse Laboratory Precision Data.
Sample Media/
Description
Soil PE No. 1
Soil PE No. 2
Soil Low
Soil Medium
Soil High
Water Low
Water Medium
Water High
Relative Percent Difference
Benzene
16
<1
*
*
*
<1
7
9
Toluene
8
9
*
*
13
*
11
*
Xylenes
6
8
*
122
14
<1
8
22
PCE
7
9
*
*
*
<1
*
*
TCE
6
8
*
*
*
*
*
*
Note: * = No evaluation as a result of non-detectable levels in one or more samples
Table 5-9. WAFB Water and Soil Duplicate Sample Concentrations.
Sample Media/
Description
Soil PE Sample No. 1 (• g/kg)
Soil PE Sample No. 2 (• g/kg)
Soil Low (• g/kg)
Soil Medium (• g/kg)
Soil High (• g/kg)
Water Low (• g/L)
Water Medium (• g/L)
Water High (• s/L)
Reference Concentrations
Benzene
61,000
64,000
ND
ND
ND
2
40
20
Toluene
55,000
59,000
ND
ND
600
ND
35
2
Xylenes
76,000
81,000
ND
5,000
55,000
20
385
50
PCE
91,000
98,000
ND
ND
ND
2
ND
ND
TCE
7,900
8,600
ND
ND
ND
ND
ND
ND
Note: ND = not detected
A number of non-detects were reported for the five target analytes in the soil duplicate samples. Since the
soil samples were prepared in triplicate, the RPD values shown are an average of the two or three RPD
values. For the available data, except for one high xylene RPD value, the RPD results for the soil analyses
are less than 30%.
Low RPD values were observed for the eight water samples that could be evaluated. Unfortunately, most
of the samples had non-detectable levels of TCE and PCE and could not be evaluated in terms of precision.
Overall, the observed precision for the water samples, where precision determinations were possible, was
less than 30% and judged to be good.
Traverse Laboratory Data Quality Summary
The Traverse internal quality control results revealed good laboratory procedures and instrument
performance. Accuracy and precision data for the water samples are consistently at values of 30% or less.
On the basis of these considerations, the Traverse water data set was judged to be of good quality and
suitable for use as reference data. Likewise, the soil accuracy and precision analysis data also are
consistently (with one exception concerning the precision of a xylene analysis) within 30% of the reference
values; however, evaluations were carried out only at relatively high (>1 mg/kg) compound concentration
levels. These data are judged to be of good quality and suitable for reference use, with the caution that soil
matrix precision and accuracy are not determined at lower concentrations ranges.
Pace Data Quality Evaluation
38
-------
The Pace QA plan was audited by Sandia personnel prior to the demonstration. Laboratory operations were
found to be in accordance with good laboratory practice guidelines and the requirements stated in EPA
Method TO-14.
As noted earlier, Pace analyzed soil gas samples from the WAFB demonstration and provided analysis
results and a quality control data package. The results and a discussion of the analytical data are provided
below. No out-of-limit quality control conditions were reported in the analysis results data package. As
with Traverse, the assessment of precision and accuracy of the analytical laboratory are considered using
the data from the five target compounds. Because the results for the soil gas analyses for both PCE and
TCE generally showed non-detects, another chlorinated solvent that was detected in the analysis, DCE,
was included in the soil gas accuracy and precision analyses.
Pace Internal Quality Control Data
Blank soil gas samples were analyzed in the laboratory and the results were in accordance with
performance specified in the TO-14 method. Spiked vapor samples were also run on two different days.
Calibration check recoveries for the five target compounds ranged from a low of 94% to a high of 110%,
all within the 75-125% acceptance criteria called for in Method TO-14. The quality control data provided
in the analysis report indicated good instrument performance.
Pace Accuracy Data
The accuracy data for Pace analysis of Summa™ canisters are presented in Table 5-10 and is based on the
results of spike recoveries of the target analytes. Table 5-11 provides the associated reference concentration
values for the compounds evaluated. Data from Tedlar bag samples are not included in the analyses since
the TO-14 method requires the use of passivated steel canisters for better sample stability and recovery.
Table 5-10. Pace Laboratory Accuracy Data.
Sample Media/
Description
Soil Gas Low Spike1
Soil Gas Medium Spike
Soil Gas High Spike
Percent Recovery
Benzene
112
93
44
Toluene
96
NA
NA
Xylenes
88
82
43
PCE
107
NA
NA
TCE
61
48
20
DCE
95
65
32
Note: ' - recovery values shown corrected by a factor often, see text for discussion.
NA = not analyzed, contaminants not present in spike mixture
Table 5-11. WAFB Soil Gas PE/Spike Sample Reference Concentrations.
Sample Media/
Description
Soil Gas Low Spike1
Soil Gas Medium Spike
Soil Gas High Spike
Reference Concentrations in ppm
Benzene
3
50
250
Toluene
2
NA
NA
Xylene
8
73
364
PCE
1
NA
NA
TCE
1
50
250
DCE
2
58
291
Note: ' - reference values shown corrected by factor of 10, see text for further discussion
NA = Not applicable; analyte not present in spike mixture
Initial evaluation of the low spike sample recoveries yielded values in the range of 1,000% and were
suggestive of a factor often error. A calculation error in computation of the reference values was
suspected; however, a definite error was not found. Some uncertainty exists, as recorded in laboratory
notebooks, as to whether the air volume into which the spike was injected was 1 or 10 liters. A review of
the data from each of the field participants showed a similar very high recovery trend, giving further
support for the factor often error. In view of these combined results, the low spike reference value was
increased by a factor often as shown in Table 5-11 with resultant changes in Table 5-10. Compound
39
-------
recovery for Pace is marginal with satisfactory recovery at the low and medium ranges and poor recovery at
higher concentration ranges. Recovery is also generally better for non-chlorinated compounds than for
chlorinated compounds.
The soil gas spikes were transferred from bags to canisters prior to shipment to the Pace. These transfers,
as well as dilution required for the high spike sample analysis, may have caused significant changes in the
sample composition. Previous WAFB soil gas survey results presented in Table 4-3 show that the
concentration ranges of interest at this demonstration site are generally less than 100 ppm for most VOC
compounds. As noted in Table 5-10, recovery data for Pace at contaminant concentration levels less than
100 ppm are generally good. Since the field sample concentration levels are reasonably well matched to the
concentration levels at which Pace performance is satisfactory, the laboratory results are considered
suitable as reference data for soil gas contaminant concentrations less than 100 ppm.
Pace Precision Data
The soil gas samples used for this analysis were taken sequentially over approximately a two minute period
from the monitoring well at each of the three levels selected for sampling. As stated previously, these
sequential samples may not be true duplicate samples. As with the other field duplicates taken sequentially
from monitoring wells during this demonstration, an assumption is made for the purposes of the precision
evaluation that the sequential samples are equivalent. Calculation of RPD values based on these samples
then gives an upper limit of the laboratory instrument RPD since some portion of the RPD could be
attributable to sample differences. Precision determinations, based on these assumptions for the Pace soil
gas analyses, are given in Table 5-12, while Table 5-13 provides the reference concentration levels for the
soil gas samples.
The precision data in Table 5-12 show that eight of 11 values fall within 30% margin. Nine of the 11 RPD
values fall within the 0-50% range. On the basis of these data, Pace precision performance is judged to be
satisfactory.
Table 5-12. Pace Laboratory Precision Data.
Sample Media/Description
Soil Gas Low
Soil Gas Medium
Soil Gas High
Relative Percent Difference
Benzene
76
9
3
Toluene
NA
22
12
Xylenes
66
5
7
PCE
NA
NA
NA
TCE
NA
NA
NA
DCE
47
9
7
Note: NA = Not analyzed, contaminant not detected in one or both samples
Table 5-13. WAFB Soil Gas Duplicate Sample Concentrations.
Sample Media/Description
Soil Gas Low
Soil Gas Medium
Soil Gas High
Reference Concentrations in ppm
Benzene
2
7
9
Toluene
0.1
0.5
1
Xylenes
20
30
50
PCE
ND
ND
ND
TCE
ND
ND
ND
DCE
8
10
13
Note: ND = not detected
Pace Data Quality Summary
Based on the accuracy and precision results and the concentration ranges shown in Tables 5-11 and 5-13,
the Pace soil gas analysis data quality can be regarded as satisfactory for soil gas contaminant
concentration ranges between 1 and 100 ppm and are suitable for comparison with the various technologies
only within this range.
40
-------
Summary Description of Laboratory Data Quality
The data quality from each of the laboratory analyses was systematically evaluated for each of the three
sampling media selected for study in this demonstration. The results of these evaluations have been
previously discussed in detail and are summarized in Table 5-14 for the SRS demonstration and Table 5-15
for the WAFB demonstration. Because of the number of laboratories, evaluation criteria, and media, as
previously discussed, an overall data quality grade of good, satisfactory, or unacceptable has been
assigned to each of the reference laboratory data sets. Data sets falling into the good or satisfactory
categories are considered suitable for comparison with field technologies. An unacceptable data quality
ranking indicates that these data are unsuitable for use as reference data.
For SRS each sample media type except soil was determined to have a satisfactory or better reference data
set. The data quality for the WAFB demonstration was determined to have a satisfactory or better reference
data set for comparison to field analytical results.
41
-------
Table 5-14. SRS Demonstration Laboratory Data Quality Ranking.
Sample Media
Soil
Water
Soil Gas
Laboratory
GEL
No Determination
Good
Unacceptable
SRS
No Determination
No Determination
Satisfactory
Table 5-15. WAFB Demonstration Laboratory Data Quality Ranking.
Sample Media
Soil
Water
Soil Gas
Laboratory
Traverse
Good
Good
No Determination
Pace
No Determination
No Determination
Satisfactory
42
-------
Section 6
Technology Demonstration Results and Evaluation
Introduction
Analytical results and an evaluation of the Bruker GC/MS data collected during the SRS and WAFB
demonstrations are presented in this section. Both demonstrations provided an opportunity for analysis of
soil, water, and soil gas media by the Bruker GC/MS system. Data from the Bruker GC/MS system are
compared to the previously discussed reference laboratory data. Following presentation of the Bruker
sample analysis data, instrument performance is assessed using a number of performance goals also
described in this section.
Pre-Demomtration Developer Claims
Before the actual field demonstration, the Consortium requested GC/MS instrument performance claims
from Bruker. The performance claims provided by Bruker were, to a limited extent, discussed with the
Consortium; however, they were not significantly altered prior to their inclusion in the demonstration plan.
Since this was a pilot demonstration, the various proposed methodologies for field instrument comparison
with reference laboratories were largely unproved. Consequently, the claims made by the developer,
although loosely defined at the outset, were accepted with the expectation that specific performance
evaluation goals would be developed as experience was gained carrying out the demonstration and the
subsequent data analysis. The initial claims made by Bruker for the GC/MS system taken from the
demonstration plan prepared prior to the field demonstrations [SNL, 1995] are as follows:
• Accuracy: GC/MS data within ±35% of reference laboratory values for soil, water, and soil gas
analyses.
• Precision: GC/MS relative percent differences less than 30% for water and soil gas analyses; less
than 35% for soil analyses.
• Completeness: For all samples analyzed, 95% of target VOC compounds detected by reference
laboratory also detected by GC/MS system.
• Sample throughput: Water and Soil Gas: 8-10 min/sample, 6 sample/hour; Soil: 7-9 min/sample,
7-8 sample/hour.
• Methodology: Soil via headspace analysis; Water via spray and trap accessory; Soil gas via
sorbent trapping/thermal desorption.
• Reported Data: Quantitative results submitted at the end of each run.
• Deployment: The GC/MS system can be set up and ready for sample runs within 60 minutes.
The performance claims stated above address the critical areas of instrument performance and provide a
framework to evaluate the capabilities and utility of the technology. However, the claims as stated do not
take such issues as statistical variation in the reference laboratory and the field technology data into
consideration. An evaluation based solely on these claims may not fairly evaluate technology performance.
For example, the following questions illustrate problems that can arise when considering the above claims:
• Is the instrument accuracy or precision claim considered met if one or two outliers or extreme
values do not meet the claim?
• Should reference laboratory data be used for comparison with technology data if significant
inaccuracies are encountered in the reference laboratory data?
43
-------
As a result of these considerations, the original developer claims were used as a basis in formulating more
specific performance goals such that instrument performance could be more fairly evaluated. The
performance goals incorporated a consideration of the statistical characteristics of the reference laboratory
and technology measurements, rather than simply evaluating individual measurements. This approach
serves to decrease the effect of an occasional outlier data point on the overall evaluation of the technology
or the reference laboratory. The rationale and approach used in setting the performance goals for each of
the identified criteria, such as accuracy and precision, generally follows that stated in EPA Method 8260
for water and soil analyses. The development of these performance goals and evaluation criteria are
discussed in further detail in the following sections.
Field Demonstration Data Evaluation Approach
A discussion of the methodology and its underlying rationale used for Bruker instrument performance
assessment is given in the following paragraphs. The methodology is based on instrument performance in
three specific areas: precision, accuracy, and comparability of Bruker data to reference laboratory data. The
evaluation methodology uses instrument performance claims made by Bruker as a starting point in the
formulation of specific instrument performance goals.
One of the limitations encountered in this particular study is the existence of a limited number of data
points for each target analyte in the sample media. For example, analysis of water samples for
dichlorobenzene at the Wurtsmith demonstration by Bruker and the reference laboratory produced only
five sample pairs that could be compared. A small sample size of five pairs significantly limits the ability
to draw conclusions about the performance of the Bruker instrument for this particular compound. One
method of dealing with small sample sizes and their associated uncertainties is to pool the data for all
analytes in a particular sampling medium and apply statistical techniques to the pooled data set in order to
gain an understanding of the overall performance of a field instrument relative to reference data. Many of
the factors contributing to measurement uncertainty are random and thus tend to average out when many
analysis results are considered together. However, in pooling these data, the assumption must be made that
the GC/MS responds to the various target compounds in a similar fashion. As an example, the instrument
accuracy and precision for benzene is not assumed to be significantly different than its accuracy and
precision for trichloroethene. This is a reasonable assumption for the compounds under investigation in
this study. All of the target compounds were either aliphatic or aromatic compounds with a subset of these
being chlorinated species with similar chromatographic and detection properties. Compounds with
significantly different chromatographic properties, such as alcohols, ethers, ketones, etc., which have
different GC/MS response characteristics, were not included in the target analyte list.
Instrument Precision Evaluation
The precision of the field instrument is obtained by the analysis of duplicate samples and is based on the
percent difference between the two analytical results. The definition for relative percent difference (RPD)
is as follows:
'x -x I
where Xj and x2 are duplicate measurements and x is the average of the two measurements. Precision
provides a measure of the stability of the instrument under actual field operations and is one of the key
indicators of instrument performance. The overall precision claim made by Bruker was that RPD values of
30% or less would be achieved for water and soil gas and less than 35% for soil. Refinement of these
claims into more specific performance goals was done following the demonstration in order to incorporate
statistical considerations of the data. For a first test, the Bruker RPD data are pooled by sampling media
and a determination is made as to whether the median value of the distribution falls in the respective range.
44
-------
This first precision performance criterion is consistent with that specified in Methods 82607 and TO-14. If
this first test criteria is not met, a second evaluation is done. In the second test, the determination is made
as to whether the median Bruker RPD value is less than or equal to the 95th percentile of RPD values
similarly pooled from the reference laboratory data. The underlying rationale is that Bruker precision is
adequate if it is comparable to that observed in the reference laboratory data in this demonstration. Thus, if
either precision criteria are met, Bruker precision performance is judged acceptable.
Instrument Accuracy Evaluation
Instrument data accuracy is evaluated by comparing the Bruker GC/MS analysis results from PE and spike
samples with known VOC contaminant levels. The spike samples were prepared in the field in soil, water,
and gas media from pure compounds or known mixtures. Performance evaluation samples, on the other
hand, were purchased or obtained from independent vendors and had undergone extensive analysis by
multiple laboratories. They were accompanied by a certificate of analysis in which the concentration levels
of the sample components were specified. Often an uncertainty or confidence interval, based upon the mul-
tiple laboratory results, was also provided. One way of expressing instrument accuracy, relative to the
reference concentration of the PE or spike sample, is by the use of the term percent recovery (REC), as
described below:
x
JE = — • D
x
where Ttinstrument is the measured concentration by a field instrument and xre/ereBce is the true concentration. For
example, if the benzene content in a PE sample was 100 ppm and the field instrument analytical result was
110 ppm, the percent recovery would be 110%. The absolute percent accuracy (APA), also used in these
evaluations, is defined as follows:
APA = \REC-WO\
Acceptable limits of recovery are empirically derived by a particular laboratory during routine use of
Methods 8260 and TO-14. Consequently, they vary somewhat among laboratories and compounds.
However, recovery values falling within the range of 70 - 130% are considered acceptable in terms of
instrument and method quality control [EPA, 1987]. Recoveries that fall outside this range may still be
acceptable for some compounds and must be evaluated on an individual compound and laboratory basis.
Initial Bruker claims regarding accuracy were made only in the context of reference laboratory data.
Strictly speaking, accuracy assessments should only be made against samples for which a true or certified
value is available. Reference laboratory results do not necessarily represent the true VOC contaminant
content of the samples. Bruker accuracy performance goals are stated in the context of Bruker and
reference laboratory analytical results on PE and spike samples for which reference values were available.
They are much like those developed for precision assessment, in the sense that they also include statistical
considerations. As with RPD, the absolute percent accuracy values for Bruker are pooled by sampling
medium. An initial determination is then made as to whether the median APA value is less than 35%. This
initial test criteria closely follows that specified in Methods 8260 and TO-14 for the precision evaluation.
For example, Method 8260 specifies that the APA should be less than 30% for most compounds covered
by the method. A second evaluation is done if the first criterion is not met. In this test, a determination is
7 Paragraph 8.5.5 of Method 8260 states in part: "Results are comparable if the calculated percent relative standard
deviation (RSD) does not exceed 2.6 times the single laboratory RSD or 20% whichever is greater and the mean
recovery lies within the interval R ± 3S or R ± 30%, whichever is greater."
45
-------
made as to whether the median Bruker APA value is less than or equal to the 95th percentile of APA
values similarly pooled from the reference laboratory data. As with the precision assessment, the
understanding is that Bruker accuracy should be judged acceptable as long as it is comparable to that
observed in the reference laboratory data in this demonstration. Thus, if either criteria are met, Bruker
accuracy performance is judged acceptable.
Instrument Comparison with Reference Laboratory Data
A third approach for evaluation of Bruker GC/MS instrument performance entails a comparison of Bruker
results with reference laboratory results for paired sample analyses. As described in Section 5, the
reference laboratory data were ranked good, satisfactory and poor in terms of overall quality. Only those
laboratory results that were ranked good or satisfactory were used for comparison with the field instrument
data.
To varying degrees, reference laboratory analytical results possess inaccuracies and uncertainties~a fact
which complicates comparisons between laboratory and field instrument data. Uncertainties in analytical
results are caused by factors such as sample transport and storage, improper instrument calibration,
operator technique, instrument noise, to name a few. Even in cases where sample transportation and
storage operations are performed correctly, the sample that reaches the fixed-laboratory analytical
instrument may be different in chemical composition from that analyzed by the field instrument because of
unavoidable heterogeneities in the sample matrix.
The Bruker to laboratory comparison takes these uncertainties into account by computing the percent
difference between the Bruker and reference laboratory results for each duplicate sample pair and
examining the distribution of these percent differences for each sample media relative to an absolute
accuracy standard—in this case ±50%. The percent difference (DIP) for each sample pair is mathematically
expressed as follows:
[x - x ]
W = — — • D
x
where xinstniment is the measured concentration by the field instrument and xlab is the measured concentration
of the same sample by the reference laboratory. The absolute percent difference (APD) ignores the sign of
the percent difference value. For this demonstration, the absolute percent difference criterion was set at
50%. Methods 8260 and TO-14 typically specify ±30% as a tolerable range for percent differences of an
instrument relative to a PE or spike reference value; however, the reference value (the laboratory result)
also has inherent inaccuracy. Consequently the ±30% margin is widened to ±50% to account for this
variability. The Bruker to laboratory comparison performance goal stipulates that the median absolute
percent difference of the distribution for each sampling media should be in the range of 0-50%. Failure to
meet this goal suggests that a significant bias between the Bruker system and the laboratory may exist. An
additional bias significance test is performed as described more fully below.
Where a large bias is suspected between the field method and reference laboratory, a statistical technique
known as the Wilcoxon Matched Pair test was used to assess the differences encountered between field
technology and laboratory data. Both laboratory and field method data include measurement uncertainty as
a result of random variability encountered in the sample collection, distribution, and analysis process. A
statistical comparison was carried out to determine whether the range of differences encountered between
the two methods could be explained by random variability or, alternatively, whether a significant or true
bias exists between the two methods. The Wilcoxon Matched Pairs test is a non-parametric test, meaning
that no assumptions are made concerning the distribution of the population from which the field instrument
and reference laboratory samples are drawn [Iman, 1994; Conover, 1980]. The test produces a test statistic
which can be interpreted as the ratio of observed differences in the two data sets to expected random
46
-------
differences in the same two data sets. Thus, in very general terms, a large test statistic indicates a
significant difference between two instruments or methods. The Wilcoxon test statistic is influenced to a
greater extent by the larger values in the data sets being compared, particularly when the data span many
orders of magnitude such as encountered in this study. In order to eliminate this undesired effect, the data
pairs were normalized by using the percent difference determination for each sample pair, as described
earlier. The percent difference computed for each Bruker measurement relative to the paired sample
laboratory value is compared against a reference percent difference value of zero in the statistical test.
The quantitative aspect of the Wilcoxon test is given by the p-value, or probability, associated with a
computed test statistic. For example, a test result with a p-value of 0.05 indicates that the probability of two
equivalent techniques producing the observed differences as a result of random variability alone would be
0.05. By convention, a p-value of 0.05 is often used as the decision point as to whether a statistically
significant bias exists between the two sets of measurements. A p-value less than 0.05 is generally
understood to indicate differences between two methods that cannot be explained by random variation
alone. On the other hand, p-values greater than 0.05, indicate that observed differences between two
methods can be explained by random variation alone and that the methods can be considered comparable.
The outcome of the Wilcoxon test is used to make a final decision as to whether one is justified in calling
the Bruker field measurements comparable or not comparable to reference laboratory measurement. A
Wilcoxon test result with a p-value less than 0.05 indicates that Bruker data are not comparable to
reference laboratory data for a particular sampling medium.
In summary then, two criteria are used for assessing Bruker to laboratory accuracy. First a determination is
made as to whether the median absolute percent difference is in the range of 0-50%. If this criteria is not
met, the Wilcoxon test is performed on the data set to test the significance of the observed bias. If the test
result indicates significance, the overall Bruker to laboratory comparison goal is judged not met.
Alternatively, if the test indicates no significance, the goal is judged to have been met.
Scatter Plots
Another way to evaluate the performance of the of the Bruker GC/MS with respect to the reference
laboratory is through the use of scatter plots. The plots are prepared in log-log format since the data
concentrations generally spanned many orders of magnitude. Two solid lines are positioned on each graph
which mark the ±50% difference about the zero bias line. As noted in earlier discussion, the ±50% margins
are chosen as an indicator of acceptable instrument performance relative to reference laboratory data. The
value is derived as follows: If the assumption is made that the uncertainty on field technology and
laboratory measurements is ±20%, then a worst case percent difference between two measurements at
either extremes of the ±20% range would be ±50%. As an example, consider a sample with a true
concentration of a target analyte of 100 mg/L. The field method reports a value 20% high at 120 mg/L and
the laboratory reports the value 20% low at 80 mg/L. The percent difference between these two values is
50%. Thus, the ±50% bias lines denote acceptable tolerances on field instrument comparisons to reference
laboratory data.
47
-------
In some cases where the data span five or more orders of magnitude, two plots are used. The plots give the
reader an indication of instrument bias and correlation8 relative to reference laboratory data. Here, bias is
defined as a systematic difference of one method relative to another across a portion of or the entire
measurement range of the instrument. Bias is directly related to accuracy. A method with low bias is one
with high accuracy. A method with a -10% bias is accurate to within 10%. As an example, Figure 6-1
shows four sets of simulated data with various amounts of bias and random "noise" added. The zero bias
line, not shown in the plots, extends from the lower left to the upper right corners of each plot. These
findings are based on the assumption that any bias is a constant fraction of the actual value of the result,
and that instrument performance is linear over several orders of magnitude.
Lab Value
(a) High correlation; low bias
Lab Value
(b) High correlation; positive bias
X X
Lab Value
(c) Low correlation; low bias
Lab Value
(d) Low correlation; positive bias
Figure 6-1. Example scatter plots with simulated data. The four plots illustrate various degrees of
measurement correlation and method bias or accuracy. The reference laboratory value is
plotted on the x-axis and the paired field technology value on the y-axis. The solid lines
mark the ±50% interval about the zero bias line.
Data with low bias and high correlation are shown in Figure 6-la. The data are closely clustered between
the ±50% lines near the zero bias line. A low bias and low correlation example is shown in Figure 6-lc.
Most of the data fall within the ±50% lines; however, the scatter is larger than that observed in Figure 6-la.
Correlation is a measure of the degree of linear relationship between two variables.
48
-------
Positive bias is illustrated in Figures 6-lb and 6-ld for high and low correlation data, respectively. In each
case, the data are grouped about a line that is shifted well above the zero bias line. As in the other
examples, the degree of data scatter is greater for the low correlation case.
Histograms
A final means to evaluate performance is accomplished by computing the percent difference between the
field and laboratory value for each paired sample analysis and pooling these data across all compounds by
sample medium. These percent difference computations were then tabulated into frequency histograms,
examples of which are shown in Figure 6-2. The histogram gives the frequency of occurrence as a function
of the percent difference interval and enables a visual evaluation of the field technology data when
compared with reference laboratory data. As an example, in Figure 6-2a, nearly all the percent difference
computations fall within the ±30% interval, indicating relatively good data correlation. The center of mass
of the histogram falls near zero, indicating an average bias near zero. A low correlation, low bias example
is shown in Figure 6-2c. In this example, 12 of the 42 total observations fall outside the ±30% range.
Although the overall average percent difference falls near zero, the width of the distribution reveals
diminished instrument correlation. Figures 6-2a and 6-2c illustrate the same type of data shown in scatter
plot format in Figures 6-la and 6-lc, respectively. Positive bias examples with high and low correlation are
illustrated in Figure 6-2b and 6-2d, respectively. In these examples, the center of mass of the histogram
falls at a point other than zero, thereby revealing an overall measurement bias. Figures 6-2b and 6-2d
illustrate the same type of data shown in the scatter plots in Figures 6-lb and 6-ld, respectively.
Summary of Instrument Performance Goals
The original instrument performance claims, although relevant to the objectives of the demonstration, were
determined to be somewhat vague following compilation and evaluation of all demonstration data. For
example, should all measurements fall within ±35% of reference laboratory results or should the average
results for a particular sample medium meet the 35% criterion? To maintain objectivity in the process of
instrument evaluation, instrument performance goals were modified and restated in such a manner that a
simple yes are no answer describes whether the goals were met. A summary description of the performance
goals, developed in the preceding paragraphs, and how data are evaluated in the context of those goals is
presented below.
Accuracy
Two goals are stated. If the first goal is met, no evaluation is done relative to the second.
Accuracy Goal 1: Median Bruker absolute percent accuracy for each sampling medium is in the
range of 0-35%.
Accuracy Goal 2: Median Bruker absolute percent accuracy for each sampling medium is less
than or equal to the 95th percentile of the pooled reference laboratory absolute percent accuracy
for each sampling medium9.
9 The specific accuracy evaluation procedure is as follows:
• Pool all the Bruker PE and spike sample analytical results, in terms of absolute percent accuracy, for each
sample medium for which a reference value is known. Data from SRS and WAFB are combined in this
analysis. All sample data are excluded whose reference values are less than 2«PQL Bmker
• Compile these data into a frequency histogram and compute the median (APA 5)Bmker, 80th percentile
(APA8)Bmker, and 95th percentile (AP A 95)Bn]ker values.
• Do the same compilations for the reference laboratory data combined from both demonstration sites by
sampling media.
• Apply the following assessment criteria to the absolute percent accuracy distributions:
49
-------
-70 -60 -50-40-30-20-10 0 10 20 30 40 50
Percent Difference Interval
(a) High correlation; low bias
-30 -20 -10 0 10 20 30 40 50
Percent Difference Interval
(b) High correlation; positive bias
70 80 90
-70 -60 -50 -40 -30 -20 -10 0 10 20 30 40 50
Percent Difference Interval
(c) Low correlation; low bias
-30 -20 -10 0 10 20 30 40 50 60 70
Percent Difference Interval
(d) Low correlation; positive bias
Figure 6-2. Example histograms with simulated data. Various combinations of measurement correlation and
bias are shown.
Precision
In a similar manner as described for accuracy evaluation, two precision goals are stated. If the first goal is
met, no evaluation is done relative to the second.
Precision Goal 1: Median Bruker relative percent difference for each sampling medium is in the
range of 0-30% for water and soil gas samples and 0-35% range for soil samples.
(1) If (APA 5)Braker <35%: Accuracy Goal 1 Met ~ Bruker accuracy performance is better than or
equal to that specified in Methods 8260 and TO-14.
(2) If Bruker (APA5)Bruker < (APA95)Lab: Accuracy Goal 2 Met - Bruker performs comparably to
conventional laboratory using accepted analytical methodologies.
(3) If (APA 5)Bri]ker > (APA 95)Lab. Accuracy Goal 2 Not Met - Bruker data does not compare with the
reference laboratory data.
50
-------
Precision Goal 2: Median Broker relative percent difference for each sampling medium is less
than or equal to the 95th percentile of the pooled reference laboratory relative percent differences
for each sampling medium10.
Bruker to Reference Laboratory Comparison
Two goals are stated. If the first goal is met, no evaluation is done relative to the second.
Comparison Goal 1: Median absolute percent difference is in the range of 0-50%.
Comparison Goal 2: If Goal 1 not met, the Wilcoxon test result between Bruker and reference
laboratory data should indicate no significant bias (p > 0.05)11.
Field Operation Observations
SRS Demonstration
The SRS demonstration consisted of three days of sample analysis by the Bruker analytical team.
Following pre-demonstration calibration, the instrument was shipped to the USA from Germany by air
freight with no degradation of instrument performance noted following its transport. The instrument was
10 The specific precision evaluation procedure is as follows:
• Pool all the Bruker duplicate sample results, in terms of relative percent difference (RPD) for each sample
medium (soil, water, and gas) for which a reference value is known. Data from SRS and WAFB are
combined in this analysis. All sample data are excluded whose reference values are less than 2«PQL Bruker
• Compile these data into a frequency histogram and compute the median (RPD 5)Bruker, 80th percentile
(RPD8)Braker, and 95th percentile (RPD 95)Braker values.
• Do the same compilations for the reference laboratory data combined from both demonstration sites by
sampling media.
• Apply the following assessment criteria to the absolute percent accuracy distributions:
(1) If (RPD 5)Bn]ker < 30% (water and soil gas), 35% (soil): Precision Goal 1 Met ~ Bruker precision
performance within the range identified in original developer claims and is very near that specified in
Methods 8260 and TO-14.
(2) If Bruker (RPD 5)Bruker < (RPD 95)Lab: Precision Goal 2 Met - Bruker performs comparably to
conventional laboratory using accepted analytical methodologies.
(3) If (RPD 5)Bruker > (RPD 95)Lab: Precision Goal 2 Not Met ~ Bruker precision performance is worse
than that reported by reference laboratories despite considerable allowance given for field variability.
11 The specific Bruker to laboratory evaluation procedure is as follows:
• Compute the percent difference and the absolute percent difference for each set of Bruker-laboratory paired
sample results.
• Compile the percent difference and absolute percent difference data into frequency histograms and compute
the median (APD 5) Bmker.Lab, 80th percentile (APD 8) Bn]ker.Lab, and 95th percentile (APD 95) Bn]ker.Lab values for
each distribution.
• Apply the following assessment criteria to the compiled absolute percent differences:
(1) If (APD 5) Bruker.Lab < 50%: Comparison Goal 1 Met ~ Bruker results are consistently within ±50%
of reference laboratory results.
(2) If (APD 5) Braker.Lab > 50%: Proceed to Step (3) for further evaluation.
(3) Perform Wilcoxon test to determine if the observed bias is statistically significant. The percent
difference values (not absolute percent difference values since the sign of the difference is necessary
information in the Wilcoxon Test) are compared to a value of zero percent difference.
(a) If p-value for the test > 0.05: Comparison Goal 2 Met
(b) If p-value for the test < 0.05: Comparison Goal 2 Not Met
51
-------
installed in a small air conditioned van and driven to the demonstration site. The instrument was operated
from batteries and was also set up for operation during transit from the hotel to the measurement site, a
time-saving feature of interest in these field investigations. The Bruker GC/MS was also operated for some
portion of the time under ambient air temperatures during the hot afternoons to demonstrate its field
versatility. Typical temperatures during the SRS demonstration are shown in Figure 6-3.
On the first day of the demonstration, each developer was provided soil gas samples for analysis. Eighteen
GC/MS runs, that included blanks, calibration checks and field samples were completed and data reports
were submitted for 16 of these. Two runs were not submitted because power supply problems affected the
instruments' performance.
The second day at SRS was dedicated to water analysis. Fifteen blank, calibration and sample runs were
completed and data reports were submitted for all runs.
The third day was devoted to soil analysis. Eight runs were completed and data reports were submitted for
all runs completed. The Bruker analytical team did their soil analysis using headspace vapor measurements
and had calibrated the system using 10 mL vials.
The Bruker team included internal standards in most of their analyses. In nearly all cases, recovery of these
compounds was good. However, in some cases, significant changes were noted in the surrogate
compounds in water samples. Since much of the sample preparation was done under the changing ambient
conditions, the Bruker analysts suspected that the high ambient temperature influenced sample preparation
or storage stability. Sample run times were on the order of 12 minutes for all media.
WAFB Demonstration
Analysis at WAFB proceeded much like that at SRS, with soil gas analysis done on the first day, water
analysis on the second, and soil analysis on the third. Typical temperatures encountered during the WAFB
demonstration are shown in Figure 6-4. Fifteen soil gas runs, including calibration checks, blank analyses,
and field sample analyses, were completed on the first day of the demonstration. An instrument breakdown
was noted during one run. The problem was resolved and the instrument was brought back on-line in about
60 minutes. Blank soil gas sample analyses results for target analytes were less than 10 ppb.
A total of 17 water sample analysis runs were completed on the second day of the WAFB demonstration.
Power problems were encountered during one run which were related to the fact that input voltage to the
instrument exceeded the 28 volt maximum for which the instrument was configured. This caused some
shifts in calibration data; however, the data were still usable from the run. Blank water sample analyses
produced results for the target analytes in the range of 0.1 to 8 //g/L.
Eighteen runs were completed on the third day of the demonstration which was devoted to soil sample
analysis. Blank soil sample analyses produced results of about 50 //g/kg for the target analytes. The Bruker
analysts noted more scatter in the soil analysis results for their internal standards and suspected that to be
true for the target analytes as well.
52
-------
06:00 AM 08:00 AM 10:00 AM 12:00 PM 02:00 PM 04:00 PM 06:00 PM
Time of Day
Figure 6-3. Plot of daily temperatures during the SRS demonstration. Note that two data points
are missing from July 20,1995.
u.
9
•a
-------
Bruker Accuracy and Precision Results
Bruker Accuracy ~ SRS Demonstration
As discussed in Sections 4 and 5, SRS is predominately contaminated with the chlorinated solvents TCE
and PCE. Consequently, all of the Bruker GC/MS data and corresponding laboratory data are limited to
these two contaminants. Recovery data for the Bruker GC/MS system for SRS spike and PE samples for
water and soil gas are presented in Table 6-1. Data from reference values lower than the Bruker PQL are
not shown in the table.
Table 6-1. Bruker Recoveries at SRS.
Sampling Medium / Description
Water Medium Spike
Water High Spike
Water PE No. 2
Soil Gas Low Spike
Soil Gas Medium Spike
Soil Gas High Spike
Soil Gas PE No. 1
Soil Gas PE No. 2
Percent Recovery
TCE
102
145
78
134
52
36
127
78
PCE
64
71
152
195
80
44
265
51
Because of significant deviations from the demonstration plan, discussed earlier in Section 4, no data are
available for assessment of Bruker instrument accuracy on soil samples at SRS.
Percent recoveries for the water spike and PE samples are provided in Table 6-1. Five of the six data points
are in the 50-150% range. A recovery value of 145% was obtained for the high TCE spike. The percent
recoveries from the reference laboratories for this sample showed similar high recovery results leading to
questions about the validity of the reported reference value for this particular sample. The other high value
of 152% was for PCE in a PE sample at a relatively low level of 30 //g/L.
Recoveries for soil gas measurements are also shown in Table 6-1. The level of confidence in the reference
values for the soil gas spike samples is high since they were drawn from gas cylinders with accompanying
certificates of analysis. The low spike recoveries are shown since the TCE and PCE content of this sample
was above the reported PQL of the Bruker instrument. Recoveries generally fall into two categories: they
are high (>100%) for low level spikes and low (<100%) for high level spikes and nearly all the reported
values fall outside the 70-130% recovery range. These data suggest questionable accuracy performance for
the Bruker system during the SRS demonstration.
Bruker Accuracy ~ WAFB Demonstration
The sampling media at the WAFB demonstration site are contaminated with a wide range of VOCs
including petroleum hydrocarbons and chlorinated solvents. Approximately fifteen different VOCs were
analyzed by both the reference laboratories and the field technologies, though several of these compounds
were found in only trace amounts in the different media. To most efficiently evaluate the Bruker GC/MS, a
subset of analytes was chosen for accuracy and precision evaluation using the spike and PE samples. The
subset included benzene, toluene, ethylbenzene, and xylenes, as well as TCE and PCE. Unless otherwise
noted, ethylbenzene is grouped with xylenes in the total xylenes category. Recovery data for the WAFB
54
-------
demonstration from Bruker GC/MS analysis of PE and spike samples is presented for soil, water, and soil
gas samples in Table 6-2.
Two soil PE samples were available for accuracy determinations. These samples were laboratory-prepared
in sealed ampules and were judged to be reliable standards. Bruker GC/MS recoveries for the five target
compounds fall in the range of 90 to 190%. Computed recoveries from spike and PE water samples, also
shown in Table 6-2, are consistently above 100%, and fall in the range of 112 to 177%. The recovery
results are similar for all target analytes and indicate no obvious instrument response changes across the
range of target compounds.
Soil gas sample recoveries shown in Table 6-2 are all in the range of 60-108% and, in contrast to the SRS
recovery data, reveal good accuracy performance during the WAFB demonstration12
Table 6-2. Bruker Recoveries at Wurtsmith.
Sampling Medium/Description
Soil PE No. 1
Soil PE No. 2
Water Low Spike
Water Medium Spike
Water High Spike
Water PE No. 1
Water PE No. 2
Soil Gas Low Spike
Soil Gas Medium Spike
Soil Gas High Spike
Percent Recovery
TCE
190
174
153
140
132
ND
128
72
108
82
PCE
143
133
123
112
132
ND
136
91
ND
ND
Benzene
153
152
119
152
130
ND
131
79
87
62
Toluene
93
134
142
177
156
ND
134
89
ND
ND
Total
Xylenes
101
94
160
145
161
ND
144
83
104
99
Notes: ND = Not detected, compound not present in spike; no recovery data available.
See Tables 5-7 and 5-11 for approximate target analyte concentrations in the various PE/Spike samples.
Overall Bruker Accuracy Performance
The absolute percent accuracy values from both sites were compiled for the target analytes by sampling
media and are shown as histograms in Figures 6-5, 6-6 and 6-7 for soil, water, and soil gas, respectively.
For soil, most of the values are scattered in the 0-90% range with no clear clustering of results. For water,
most of the values fall in the 0-70% range. The soil gas accuracy data generally fall in the 0-70% range.
12 The Bruker recovery for the lowest soil gas spike sample is shown in Table 6-2 since it is above the reported
Bruker PQL. Initially, the computed Bruker recoveries for this particular sample were consistently high by a
factor often for all target analytes. Results similar to those from Bruker were obtained by other participants,
suggesting an error in the reference sample preparation. A review of the laboratory and field data and log books
strongly suggest that a factor often error occurred in the reference sample concentration computation. The
reference values were changed accordingly and the reported Bruker recovery data are based on the corrected
values.
55
-------
Bruker Soil Accuracy
20 30 40 50 60 70 80 90 100
Absolute Percent Accuracy Interval
Figure 6-5. Absolute percent accuracy histogram for Bruker
soil samples. The number of observations are
shown on the y-axis.
Bruker Water Accuracy
20 30 40 50 60 70 80 90 100
Absolute Percent Accuracy Interval
Figure 6-6. Absolute percent accuracy histogram for Bruker
water samples.
Bruker Soil Gas Accuracy
40 60 80 100 120 140 160 180 200
Absolute Percent Accuracy Interval
Figure 6-7. Absolute percent accuracy histogram for Bruker
soil gas samples.
56
-------
Summary data from the absolute percent accuracy distributions are shown in Table 6-3 for all media. In
general, the lower the number the better the performance. For example, the median absolute percent
accuracy was 22% for the pooled reference laboratory soil data. The laboratory's 95th percentile value is
47%. In comparison, the Bruker median absolute percent accuracy for soil samples was 39% and less than
the 95th percentile value for the reference laboratories.
Table 6-3. Bruker and Reference Laboratory Accuracy Summary.
Data Set/Media Type
Ref. Lab - Soil
Bruker - Soil
Ref. Lab - Water
Bruker - Water
Ref. Lab - Soil Gas
Bruker - Soil Gas
n
10
10
30
26
24
21
Absolute Percent Accuracy
X5
22
39
19
36
20
22
x.s
36
57
27
52
47
49
X95
47
83
116
61
66
95
Note: x5 = 50th percentile (median); x8 = 80th percentile; x95 = 95th percentile
Assessment of Bruker performance in terms of the accuracy goals stated previously results in the following
determinations:
Soil:
(APA.5)Bruker.S
35%:
Accuracy Goal 1 Not Met
(APA5)Bruker_Soil < (APA95)Lab_Soil: Accuracy Goal 2 Met
Water: (APA5)
.5/Bruker-Water
> 35%:
Accuracy Goal 1 Not Met
(APA5)Bruker_Water < (APA95)Lab_Water: Accuracy Goal 2 Met
Soil Gas: (APA 5)Bruker_Gas < 35%: Accuracy Goal 1 Met
Bruker accuracy performance is acceptable in terms of established performance goals for all media. The
most stringent performance goals for accuracy were not met for soil and water; however, median Bruker
performance was as good as or better than reference laboratory results from similar analyses. Soil gas
accuracy performance was better than the most restrictive goal of 35% absolute percent accuracy.
Bruker Precision ~ SRS Demonstration
Bruker GC/MS RPD values from the SRS demonstration are presented in Table 6-4 and are based on the
results of duplicate field sample analysis. No precision determination was done on SRS soil sample data as
discussed previously.
Relative percent differences for water samples, shown in Table 6-4, are from split water samples. The
calculated RPD values range from 5-38%. Low concentration samples near the Bruker PQL exhibit the
same level of precision as higher concentration samples.
Soil gas duplicate samples were actually gas samples drawn sequentially from each well within a time
interval of about 2 minutes. The degree of chemical equivalence of these sequential samples could not be
determined since a real-time measure of soil gas species was not made at the wellhead when the
57
-------
Table 6-4. Bruker Precision for SRS Demonstration.
Sampling Media / Description
Water Low Sample
Water Medium Sample
Water High Sample
Soil Gas Low Sample
Soil Gas Medium Sample
Soil Gas High Sample
Relative Percent Difference
TCE
34
37
24
22
5
10
PCE
38
29
24
25
10
14
sequential samples were collected. In this analysis, a worst case precision determination was carried out by
assuming that the sequential samples were chemically equivalent. The RPD values for the low, medium,
and high vapor samples, provided in Table 6-4, are less than 25%. These data are indicative of good
Bruker precision performance.
Bruker Precision ~ WAFB Demonstration
The Bruker precision data for the WAFB demonstration are presented in Table 6-5. The tabulated RPD
values are based on the analysis of a number of duplicate samples for each media.
Table 6-5. Bruker Precision for Wurtsmith Demonstration.
Sampling Medium / Description
Soil Low Sample
Soil Medium Sample
Soil High Sample
Water Low Sample
Water Medium Sample
Water High Sample
Soil Gas Low Sample
Soil Gas Medium Sample
Soil Gas High Sample
Relative Percent Difference
TCE
ND
ND
14
5
47
120
29
6
108
PCE
ND
ND
ND
5
21
99
ND
ND
ND
Benzene
ND
ND
21
10
12
15
26
7
17
Toluene
ND
ND
12
27
9
119
43
<1
46
Total
Xylenes
ND
12
8
8
8
21
52
10
11
Notes: ND = not detected; Analyte was either below detection or not present in the sample.
For soil sample analyses, Bruker reported a number of results that were below its PQL of 50 mg/kg.
Precision analysis was performed using only those concentrations that were greater than the PQL. The
results are shown in Table 6-5.
The low, medium, and high concentration water samples collected at WAFB generally contained only low
concentrations of the target analytes (Table 5-9). The RPDs in Table 6-5 are attributed to the fact that the
Bruker GC/MS was operating at or very near its PQL. The Bruker operator was told to expect samples that
would have analyte concentrations in predetermined ranges, therefore, he performed dilutions based on the
expectations. This caused the operator to perform dilutions on low concentration samples, which may have
impacted the precision of the measurement.
As noted earlier in the discussion of SRS demonstration results, the duplicate soil gas samples were
actually samples drawn sequentially from different levels of the well. The assumption is made that the
sequential samples are chemically equivalent. Thus, the RPD value represents a worst-case estimate of
58
-------
Broker GC/MS precision since some of the variability encountered in the analysis result may in fact be
attributable to variation between two sequential samples. The data in Table 6-5 must be interpreted in light
of the target analyte concentrations provided previously in Table 5-13. For example, the reference
laboratory reported non-detectable levels of TCE and PCE in all the soil gas samples (low, medium, and
high). The Bruker system was able to detect TCE in the samples; however, the levels are near the PQL of
the system where precision is not expected to be as good. Taking these considerations into account, the
overall Bruker precision for soil gas analyses is satisfactory.
Overall Bruker Precision Performance
The relative percent differences for both sites were compiled by sample media and are shown as
histograms in Figures 6-8, 6-9 and 6-10 for soil, water, and soil gas, respectively. For soil, most of the
RPD values fall in the 0-25% range. For water and soil gas, nearly all reported values fall in the 0-50%
range.
Summary data from the relative percent difference distributions are shown in Table 6-6 for the three
sampling media.
Bruker performance with respect to precision goals are as follows:
Soil: (RPD5)Bruker.Soil < 35%: Precision Goal 1 Met
Water: (RPD.5)Bruker.Water < 30%: Precision Goal 1 Met
Soil Gas: (RPD 5)Bruker_Gas < 30%: Precision Goal 1 Met
Bruker precision performance, relative to these performance goals, is judged acceptable for soil, water, and
soil gas analyses.
Bruker to Reference Laboratory Data Comparison
As discussed earlier in this section, comparisons of the Bruker GC/MS analytical data with the reference
laboratory analytical data for water, soil, and soil gas samples are presented in a number of formats that
include log-log scatter plots, percent difference histograms, and formal performance assessment in light of
established goals. All reported values that were less than two times the Bruker GC/MS PQL, although
plotted, are not included in any of the statistical analyses. Analytical results from only one of the duplicate
or triplicate sample set is included in the statistical analyses since their inclusion would violate
requirements for a random sample.
59
-------
at
.£>
o
«^
o
o
111 I
Figure 6-8. Relative percent difference histogram for Bruker
soil samples.
in
J2
O
i^
O
o
Figure 6-9. Relative percent difference histogram for Bruker
water samples.
Figure 6-10. Relative percent difference histogram for Bruker
soil gas samples.
60
-------
Table 6-6. Bruker and Reference Laboratory Precision Summary.
Data Set/Media Type
Ref. Lab Soil
Bruker Soil
Ref. Lab Water
Bruker Water
Ref. Lab Soil Gas
Bruker Soil Gas
n
13
5
14
21
17
18
Relative percent difference
X.5
8
12
5
24
7
16
X.8
13
15
10
38
20
37
X 9.5
58
20
22
119
68
60
Note: x5 = 50th percentile (median); x8 = 80th percentile; x95 = 95th percentile
Scatter Plots/Histograms ~ SRS Demonstration
Target analytes at this site were limited to TCE and PCE. The GEL water data are used as the reference
data set. The on-site SRS laboratory soil gas data were judged to be an acceptable reference data set.
Plots of Bruker GC/MS water analysis results for TCE and PCE are given versus the laboratory reference
data in Figure 6-11 and Figure 6-12 for low (1-100 //g/L) and high (0.1-100 mg/L) concentration ranges,
respectively. All but six of the 21 points fall within the ±50% interval about the zero bias line on both the
low and high concentration plots.
The distribution of percent difference values is shown as a histogram in Figure 6-13. Differences range
from a low of -60 % to a high of 80%.
A plot of the Bruker soil gas data for TCE and PCE relative to the SRS reference laboratory data are
shown in Figure 6-14. Half of the data points fall either above or below the 50% bias lines. The
accompanying histogram, shown in Figure 6-15, reveals no obvious positive or negative data bias.
Scatter Plots/Histograms ~ WAFB Demonstration
As discussed in Section 5, the Traverse laboratory data were given a good ranking for soil sample analyses
and a good ranking for water sample analyses. These data are used as reference data for comparison with
Bruker GC/MS results. The Pace laboratory data were given a satisfactory ranking for soil gas sample
analyses and are used as the reference data set for the soil gas measurements.
Considerably more compounds were detected in the various samples at Wurtsmith AFB than at the
Savannah River Site. For the Wurtsmith demonstration, all target analytes have been combined in the plots
and statistical analysis. However, duplicate samples are not included in the statistical analysis.
A plot of Bruker data versus Traverse data for all soil analyses is presented in Figure 6-16. Eleven
observations fall outside the ±50% margins and 8 fall inside. In general, the Bruker measurements are high
with respect to the laboratory data. The percent difference histogram, shown in Figure 6-17, shows nearly
all observations on the positive bias side with a significant number in the 120-200% range.
61
-------
E
m
Figure 6-11. Bruker vs. Laboratory data for SRS low concentration water
samples. The solid lines show the ±50% range about a zero-bias
(dashed) line.
m
Figure 6-12. Bruker vs. Laboratory data for SRS high concentration water
samples. The solid lines show the ±50% range about a zero-bias
(dashed) line.
62
-------
Bruker vs. Laboratory - SRS Water
£1
o
2
o
-80 -70 -60 -50 -40 -30 -20-10 0 10 20 30 40 50 60 70 80 90 100
Percent Difference Interval
Figure 6-13. Percent difference histogram for SRS water
samples.
Q.
Q.
z
m
Figure 6-14. Bruker vs. Laboratory data for SRS soil gas samples
Bruker vs. Laboratory - SRS Soil Gas
o
•5
-150 -100 -50 0 50 100 150 200 250 300 350 400 450
Percent Difference Interval
Figure 6-15. Percent difference histogram for SRS soil gas
samples.
63
-------
I
D) •
Figure 6-16. Bruker vs. Laboratory data for WAFB soil samples.
Bruker vs. Laboratory - WAFB Soil
o
•5
o
-80 -60 -40 -20 0 20 40 60 80 100 120 140 160 180 200 220 240 260
Percent Difference Interval
Figure 6-17. Relative percent difference histogram for WAFB
soil samples.
Plots of Bruker data versus Traverse data for all water analyses are presented in Figures 6-18 and 6-19 for
the low Gug/L) and high (mg/L) ranges, respectively. Most Bruker values fall within the ±50% margins on
the low concentration plot. In the high concentration plot, 6 observations fall on the high side of the 50%
margin and 11 fall inside the margins. The percent difference histogram for the combined data, shown in
Figure 6-20, reveals most values in the -20 to 100% range.
A plot of Bruker GC/MS data versus Traverse data for all WAFB soil gas analyses is presented in Figure
6-21. Most of the observations fall inside the ±50% margin lines on the plot, revealing relatively good
agreement with the reference laboratory. The histogram, shown in Figure 6-22, shows most observations
falling in the 0 to -50% range.
64
-------
i
«
Im
•
it
m
^S
Figure 6-18. Bruker vs. Laboratory data for WAFB low concentration water
samples.
,„
"w -
Bruker Value, n
X
x^
•• ,--
M^
_X
X
^ ^
^x
X
X
X
^-
•
X
x"
,x
x
x
x
x
X
x
X.-'
x"^
•
>
X
X
,xj
•-•
x
1— I
-r
i
x
^
X
X"
x
x"
x
x
x
x
•
X
xX^
-X^
x^
x^
jX
x
X
X
x
X*
X
x
X
x
x"
x
x'
Figure 6-19. Bruker vs. Laboratory data for WAFB high concentration water
samples.
65
-------
Bruker vs. Laboratory - WAFB Water
.Q
2.10
o
i i i n i
-100 -80 -60 -40 -20 0 20 40 60 80 100 120 140 160 180 200
Percent Difference Interval
Figure 6-20. Relative percent difference histogram for WAFB
water samples.
Bruker vs. Laboratory • WAFB Soil Gas
10 100
Laboratory Value, ppm
Figure 6-21. Bruker vs. Laboratory data for WAFB soil gas
samples.
Bruker vs. Laboratory - WAFB Soil Gas
100 200 300 400 500 600 700
Percent Difference Interval
Figure 6-22. Relative percent difference histogram for WAFB
soil gas samples.
66
-------
Overall Bruker to Laboratory Comparison Results
Absolute percent differences from both sites were compiled by sampling media and are shown as
histograms in Figures 6-23, 6-24 and 6-25 for soil, water, and soil gas, respectively. For soil, absolute
percent differences occur with equal frequency across the 0 to 200% range. Water results are similar with
approximately equal distribution across the 0 to 100% range. As noted earlier, soil gas differences
generally fall in the 0-50% range.
Summary data from the three absolute percent difference distributions for the three sampled media are
shown in Table 6-7.
Table 6-7. Bruker-Laboratory Comparison Summary.
Media Type
Soil
Water
Soil Gas
n
17
71
49
Absolute Percent Difference
X5
72
39
40
X.8
159
72
73
X 9.5
186
95
275
Note: x5 = 50th percentile (median); x8 = 80th percentile; x95 = 95th percentile
Bruker performance assessment in terms of the comparison goals result in the following determinations:
Soil: (APD.5)Bmker.Lab
Wilcoxon test result:
Water: (APD5) Bmker.Lab
Soil Gas: (APD 5) Bmker.Lab
> 50%:
p< 0.05:
< 50%:
< 50%:
Comparison Goal 1 Not Met
Comparison Goal 2 Not Met
Comparison Goal 1 Met
Comparison Goal 1 Met
Overall comparison results, in light of established performance goals, indicate that the first goal was met
for water and soil gas media. However, the Bruker to laboratory comparison did not meet either the first or
second comparison goal for soil. The Wilcoxon test results point to the presence of a significant positive
bias of the Bruker data relative to that of the laboratory.
a
o '
•5
o
Figure 6-23. Absolute percent difference histogram for soil
samples. Data are from WAFB demonstration
only.
67
-------
Bruker vs. Laboratory - Water, All Sites
.a
o
o
I I
II II
II II
-10 0 10 20 30 40 50 60 70 80 90 100 110 120 130 140
Absolute Percent Difference Interval
Figure 6-24. Absolute percent difference histogram for water
samples.
Bruker vs. Laboratory - Soil Gas, All Sites
50
100 150 200 250 300 350 400 450 500 550 600 650
Absolute Percent Difference Interval
Figure 6-25. Absolute percent difference histogram for gas
samples.
Summary of Bruker Accuracy, Precision, and Laboratory Comparison Performance
A summary of Bruker GC/MS performance for both demonstration sites is given in Table 6-8. The results
of the foregoing evaluation are summarized with respect to accuracy, precision, and comparison of Bruker
and reference laboratory data. The summary information in the table shows that the Bruker GC/MS met
performance goals in eight of the nine evaluation categories. Precision was judged acceptable for soil
analyses; however, only a very limited sample size (n=5) was available for evaluation. Accuracy of soil
analyses also met established goals. Accuracy, precision, and the Bruker to reference laboratory
comparison goals were also met for water analyses. For additional insight and evaluation, the data quality
rankings of the reference laboratories are given in the table as well. Traverse received a good data quality
ranking for soil analyses. The Wilcoxon test result for the soil media indicates that the observed difference
between Bruker and reference laboratory data cannot be explained by random variation alone and it is
likely that some other factor was causing the Bruker data to be high relative to the laboratory data. Possible
Bruker or laboratory factors could be a calibration, sample handling, or injection error by either Bruker or
the reference laboratory. For example, the Traverse results on the soil PE samples were consistently low,
whereas the Bruker results were much closer to the reference values.
68
-------
Table 6-8. Summary Performance of the Bruker GC/MS.
Sample
Medium
Soil
Water
Soil Gas
Accuracy1 via
PE/Spike Samples
Goal Met
Goal Met
Goal Met
Precision2 via
Duplicate Analyses
Goal Met
Goal Met
Goal Met
Bruker-Lab Data
Comparison3
Goal Not Met
Goal Met
Goal Met
Reference Lab
Data Quality
SRS: Undetermined
WAFB: Good
SRS: Good
WAFB: Good
SRS: Satisfactory
WAFB: Satisfactory
Other Bruker GC/MS Performance Indicators
Target Compound Identification in Complex Mixtures
The ability of the instrument to identify a broad range of compounds in complex mixtures was also amply
demonstrated with many of the WAFB samples. An example of a Bruker reconstructed ion chromatogram
for a selection of target compounds in a water sample containing numerous volatile organic compounds is
shown in Figure 6-26. A listing of the identified target compounds, corresponding to the peak numbers
shown in Figure 6-26, is presented in Table 6-9. An on-site computerized library search of Bruker GC/MS
analysis results from a similar WAFB water sample produced over 100 tentatively identified compounds
revealing the complex nature of this particular water matrix.
-I te
ec« rt», to:
Figure 6-26. Bruker GC/MS reconstructed chromatogram of target analytes in
a WAFB water sample.
Field Handling and Operation
The Bruker GC/MS is designed to be shipped to the field and is shock mounted for durability in handling
and field use. Prior to the WAFB demonstration, the system was transported by air freight from Germany
to the USA. The ruggedized design contributed to stable instrument calibrations over the duration of each
demonstration, despite several days of travel on dirt roads during the course of the demonstration. At each
site, the system was operated for several days in a row, typically 8-10 hours per day. The instrument was
69
-------
exposed to ambient temperatures ranging from 40° to 95 °F. During both demonstrations, the instrument
operated properly and without significant breakdowns or mechanical problems.
Table 6-9. Identified Target Compounds from a Wurtsmith Water Sample Analysis.
Peak No.
1
2
3
4
5
6
1
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
Compound
Ethene, 1,1-dichloro-
Methylene chloride
Ethene, 1,2-dichloro-, trans
Ethene, 1,2-dichloro-, cis
Chloroform
Methane, dibromofluoro-
Benzene, pentafluoro-
Ethane, 1,2-dichloro-
Ethane, 1,1,1-trichloro-
Benzene
Benzene, 1,4-difluoro-
Trichloroethene
Methyl isobutyl ketone
Toluene-d8
Toluene
Tetrachloroethene
B enzene-d5 -, chloro -
Benzene, chloro-
Ethylbenzene / m-Xylene
p-Xylene
Styrene
o-Xylene
p-Bromofluorobenzene
Benzene-d4, 1,4-dichloro
Benzene, 1,2(1, 3)-dichloro
Benzene, 1,4-dichloro
Naphthalene
Note: The peak numbers refer to those given in Figure 6-26.
System set-up was simple and uncomplicated. Set-up procedures involved connecting the ancillary
equipment and checking the instrument calibration. Sample analysis time varied according to the sample
media and the mode of injection; however, a typical time interval was on the order of 12 minutes per
sample. Hard copy data were available at the end of each run.
Overall Bruker GC/MS Performance Conclusions
The Bruker GC/MS system was tested at two locations during this technology demonstration in order to
evaluate its capabilities and performance during on-site analysis of VOC contaminated soil, water and soil
gas samples. The objectives of the demonstration were: (1) to evaluate instrument performance; (2) to
determine how well each field instrument performed compared to reference laboratory data; (3) to evaluate
instrument performance on different sample media; (4) to evaluate adverse environmental effects on
instrument performance; and, (5) to determine logistical needs and field analysis costs. The system
performance goals are summarized in Table 6-10 along with an assessment, based on the data produced in
this demonstration, as to whether these goals were met.
As shown in Table 6-10, Bruker GC/MS performance goals were met with two exceptions. Instrument
performance in terms of accuracy, precision, and comparison with reference laboratory data met
70
-------
established performance goals for water and soil gas analyses. The comparison of the Bruker data to the
laboratory data was below established goals for soil analyses. Performance goals related to data
completeness and data availability were met. The initial Bruker claims of sample throughput rates of 8
samples per hour for soil samples and 6 samples per hour for water and soil gas samples were met during
the SRS demonstration. Throughput rates at the WAFB demonstration were about 5 per hour and were not
significantly less than the original claims. The sample matrix at WAFB was considerably more complex
than that encountered at SRS. This resulted in longer analysis times and reduced sample throughput at
WAFB.
Table 6-10. Summary of Bruker Performance Goals and Actual Performance.
Performance Goal
ACCURACY:
Median absolute percent accuracy within 35% of reference value; or,
Median accuracy comparable to reference laboratory accuracy.
PRECISION:
Median relative percent difference within 30% (water, soil), 35% (soil) or,
Median relative percent difference comparable to laboratory precision.
REFERENCE LABORATORY DATA COMPARISON:
Median absolute percent difference less than 50%; or,
if greater than 50% , no significant bias via statistical test.
COMPLETENESS:
At least 95% of target compounds detected by reference laboratory also
detected by Bruker.
SAMPLE THROUGHPUT: Method dependent: Soil - 8 samples/hr;
Water and soil gas - 6 samples per hour.
DATA: Quantitative results submitted at the end of sample analyses.
DEPLOYMENT: The Bruker EM640 can be set up and ready to run
within 60 minutes.
Performance
Soil gas less than 35% accuracy;
Water and soil comparable to
reference laboratories.
Water, soil, and soil gas RPD
less than 30%
Water and soil gas differences
less than 50%; soil data not
comparable
Soil > 99%
Water > 99%
Soil Gas > 99%
Sample throughput met claim at
SRS; slightly less than claim
(5/hr)atWAFB.
Data were submitted at the end
of each run
The system was ready to run
samples in 60 min.
Goal Met?
Water: Yes
Soil Gas: Yes
Soil: Yes
Water: Yes
Soil Gas: Yes
Soil: Yes
Water: Yes
Soil Gas: Yes
Soil: No
Yes
Yes (SRS)
No (WAFB)
Yes
Yes
71
-------
Section 7
Applications Assessment
The Bruker-Franzen EM640™ field portable GC/MS instrument was demonstrated during this verification
effort at two geologically and climatologically different sites with a wide range of volatile contaminants.
The instrument was used to analyze soil gas, water, and soil samples on-site and in near-real-time. As
demonstrated, the Bruker field portable GC/MS system has an application in several field screening and
analysis scenarios.
Applicability to Field Operations
From a logistical viewpoint the Bruker-Franzen EM640™ GC/MS has been ruggedized for installation in a
van or truck for use in field operations. Since the instrument is not totally self-contained, the van is
required to carry the additional equipment needed to support the GC/MS instrument. Required ancillary
equipment includes a separate computer system, batteries, or alternatively, a power generator, and a large
cylinder of compressed carrier gas. In this configuration, the system can be reliably operated in the field
over a wide range of temperatures and relative humidity. Additionally, the system can be operated in a
moving vehicle thus increasing the amount of time for on-site sample analysis. System set-up and
operation may require a two-person operation for maximum sample throughput. The system uses unique
data handling and analysis software.
Capital and Field Operation Costs
On-site field analysis of samples has the potential to reduce overall site characterization and cleanup costs.
Real-time and on-site analysis of samples can provide immediate direction to a sampling team during site
characterization, thus reducing both the number of sampling trips to the site and the number of samples to
be analyzed. Additionally, real-time analysis of samples during site remediation can often minimize the
amount of material treated, thus reducing both remediation costs and the time required for site cleanup.
The actual cost savings that can be realized from the field analysis of samples depends on many factors.
These include: the capital costs of the field analysis system; field operation costs for equipment, supplies,
travel, and per diem; labor and overhead costs; sample analysis requirements; and the overall utilization
rate of the field instrument.
Capital and field operation costs for the Bruker EM640™ were determined during the demonstration and
are presented in Table 7-1. Estimates of sample throughput rates for the single ion monitoring and
scanning analysis modes for the Bruker instrument are also provided. Actual sample analysis rates will
vary as a function of the media and the contaminants being analyzed. The values provided in Table 7-1 can
be used as a guide in assessing the utility and cost effectiveness of using this system for various
applications.
Discussion of the Technology
Rapid Analysis
The use of the Bruker field transportable GC/MS system provides near-real-time analysis of samples on-
site. This approach is significantly faster than laboratory methods and expedites real-time decision making
in the field. This is especially important in guiding sampling activities. Near-real-time analysis of samples
on-site may eliminate the need and cost of return trips to the field to collect additional samples. Calibration
checks can be done quickly in the field using 24 internal compound standards. As many as five to six
laboratory quality sample analyses can be conducted in an hour with this system.
72
-------
Table 7-1. Bruker EM640™ GC/MS Capital and Field Operation Costs.
Capital Costs
Bruker-Franzen EM640™ GC/MS (data system and 1 yr. warranty)
Ancillary Equipment - Pumps, water sampler, generator
Four-year service contract
Vehicle
Equipment installation
Training (2 people for two weeks)
Total
Annualized system costs for five years ($350K + 5 years)
Maintenance (annually 10% of capital equipment costs)
Includes new columns, preventive maintenance, software and hardware
upgrades, etc:
Five-year annual system and maintenance costs
Field Operation Costs
Field chemist ($60K/year including overhead)
Chemical technician ($50K/year including overhead)
Per diem (lodging and meals - 2 people)
Supplies and consumables (standards, syringes, vials, gas, etc.)
Total
Sample Analysis Rates
Single ion monitoring analysis
Scanning monitoring analysis
$185K
$25K
$84K
$35K
$15K
$6K
$350K
$70K/year
$2.5K/year
$72.5K/year
60 samples/day
100 samples/day
Sampling and Sample Cost
A major cost savings obtained by the use of field analyses arises from the reduction in time required to
obtain the analytical results needed for decision making. In its configuration during this demonstration, the
Bruker GC/MS, in most cases, produced good quality data in the field. The ability to generate good quality
data in near-real-time allows decisions to be made concerning the extent and completeness of cleanup
operations while field equipment is still mobilized. This can result in significant cost reductions by
eliminating re-mobilizations or the removal of extra materials to be assured that the cleanup is complete.
Similar savings can be achieved during site characterization by eliminating the need for multiple sampling
mobilizations as additional sampling efforts can be directed based on real-time data.
73
-------
Transportability
The ruggedness of the Bruker GC/MS was illustrated by its daily transportation in a vehicle to and from
the demonstration sites. Mounted in a small van or truck for field operations, the system is self contained
and does not need additional support equipment. This ruggedness and portability provides the system with
the ability to perform good quality sample analysis.
Field Screening of Samples
The system's capabilities for transportability and real-time analysis make the instrument useful as a tool for
site characterization and monitoring activities. The instrument may be used as a high volume screening tool
(scan mode) to guide sampling and remediation efforts or to provide higher quality analyses on selected
samples (single ion monitoring mode). The capability to identify unknown compounds with the portable
GC/MS enables a site manager to investigate a site for a wide range of contaminants at a single time. The
use of the system as a screening tool to guide sampling efforts in the field can provide significant cost
savings in terms of the number of samples needed and need for return trips to the field.
Sample Size
The preferred method for water analysis using this system is spray and trap. This process requires a larger
than normal sample, at least 250 mL, for water analysis. Since this is significantly larger than normal
sample needs, standard sample collection procedures may have to be modified for this system.
Interferences
The presence of contamination can be periodically checked through the analysis of reagent blanks~a
procedure normally followed with most laboratory and field GC/MS systems. Since calibration of the
system takes only 30 minutes, the sample throughput is not significantly affected by the need to re-
calibrate.
Conclusions
The Bruker-Franzen EM 640™ GC/MS can provide good quality sample analysis on-site and in near-real-
time. The technology may offer time and cost saving advantages over conventional sampling and
laboratory analysis strategies. The system or others like it may never entirely replace conventional
laboratory analysis, but the technology can add significant benefits in terms of defining the nature and
extent of contamination at a site. The limitations of the system are generally related to the underlying
operational considerations associated with the basic use of GC/MS systems.
74
-------
Section 8
Developer's Forum
These comments were provided by Bruker. Their purpose is to provide the technology developer with the
opportunity to share their comments about the verification program. The comments have been minimally
edited. The views are solely those of Bruker Instruments Inc. and should not be construed to reflect the
views of EPA.
1. Lessons learned:
• It can easily happen in the field, that samples are mistakenly switched, even by official sampling
personnel. Therefore: Do not trust the labels on sample tubes, and if you run into saturation with a
sample marked "low concentration", do not immediately start to dilute the samples marked
"medium concentration" and "high concentration", you may have to multiply measured
concentrations near the detection limit by extremely high numbers! (It's a wonder that we came
out sufficiently well for these samples).
• Don't be proud to deliver results much below the practical detection limit defined by yourself for
your method. You may not fulfill the accuracy you claimed with these numbers because these are
pitilessly included into the evaluation, even if the reference laboratory claimed these samples as
not applicable (non-detectable levels).
• We should not be shy to apply not established but field practical evaluation methods. Example:
The calibration check samples for water analysis at WAFB shows average recoveries over 135%.
If we had used these recovery rates for continuous corrections of field analytical results, all our
results had fulfilled our own accuracy and precision claims (±35%). We are considering to apply
this method in the future.
2. How to do better:
• The analytical task should be more specifically and more uniformly defined for all participants in
the test. The substances for quantitative measurements should be precisely listed, not given as
vague recommendations in order to let the participants select.
• At least two independent reference laboratories should be used. (We simply don't believe in
results from reference laboratories, field analytics is better).
• More samples and more duplicates should have been analyzed, especially for Spike and PE
samples, which were used to determine the accuracy of each instrument. For example: If 2 of the 5
samples used for accuracy determination are mistakenly switched, this has a tremendous impact on
the overall test results.
3. And some criticisms:
• We did not take part in the sampling process, neither in the definition, nor in the real sampling or
in the documentation. For example: There were droplets in some of the Tedlar bags used to bring
soil air. We doubt that anyone has documented this fact.
• The high cost of the laboratory analyses was defined in the kickoff meeting as the primary reasons
for this test of field transportable GC/MS systems. However, later in the evaluation and assess-
ments, costs (and therefore short analysis cycle times) were listed under "secondary objectives".
75
-------
Continuously performed quality assurance measures were not merited in the assessment:
Several internal standards and surrogates in all samples
Monitoring of mass 40 amu (Argon) throughout all measurements for monitoring the
instrument performance.
PE samples, used to measure the accuracy, should generally be delivered in duplicates.
76
-------
Section 9
Previous Deployments
Fenstheen, Germany (October 1994). Evaluation of the ability of field GC/MS, in conjunction with other
sensor techniques, to detect trace amounts of hazardous contaminants. Bruker contacts John Wronka,
USA, Phone: (508) 667-9580, BirgitNolke, Germany, Phone: 001-49-421-2205-0.
77
-------
Section 10
References
Bury, K. V. 1975. Statistical Models in Applied Science, John Wiley & Sons, New York.
Conover, W.J. 1980. Practical Nonparametric Statistics, 2nd Ed., John Wiley & Sons, New York, pp 281-
288.
Iman, R.L. 1994. A Data-Based Approach to Statistics, Duxbury Press, Belmont, California, pp 349-382.
National Center for Integrated Bioremediation Research and Development. 1995. "Draft Field Sampling
Plan," University of Michigan, Ann Arbor, Michigan.
Rossabi, J. 1996. Personal Communication, April 1996.
Sandia National Laboratories. 1995. "Demonstration Plan for the Evaluation of Field Transportable Gas
Chromatography/Mass Spectrometers," Albuquerque, New Mexico.
SAS Institute Inc. 1985. "SAS Users Guide: Statistics, Version 5 Edition," Gary, North Carolina.
U.S. Environmental Protection Agency. 1987. "Method 8260: Gas Chromatography/Mass Spectrometry
for Volatile Organics: Capillary Column Technique, as contained in Test Methods for Evaluating Solid
Waste, Physical/Chemical Methods, SW-846, 3rd Edition." Research Triangle Park, North Carolina. EPA
SW-846.3.1.
U.S. Environmental Protection Agency. 1988. "Method TO-14: The Determination of Volatile Organic
Compounds in Ambient Air Using SUMMA® Passivated Canister Sampling and Gas Chromatograph
(GC) Analysis, as contained in Compendium of Methods for the Determination of Toxic Organic
Compounds in Ambient Air, 2nd Supplement." Research Triangle Park, North Carolina. EPA/600-
489018.
U.S. Environmental Protection Agency. 1991. "Preparation Aids for the Development of Category II
Quality Assurance Project Plans. Washington, D.C. EPA/600/8-91/004.
U.S. Environmental Protection Agency. 1994. "Guidance Manual for the Preparation of Site
Characterization Technology Demonstration Plans," Las Vegas, Nevada.
Westinghouse Savannah River Company. 1992. "Assessing DNAPL Contamination, A/M-Area, Savannah
River Site: Phase I Results (U)," Aiken, SC.
78
-------
Appendix A
Analytical Method for the Operation of the Bruker EM640™ Gas
Chromatograph/Mass Spectrometer System
Prepared by
Bruker Instruments, Inc.
(Formerly Bruker-Franzen Analytical Systems, Inc.)
1.0 Scope and Application
Using the EM640™ GC/MS system, organic compounds in air (soil gas), water, and soil are qualitatively
and quantitatively analyzed. This method applies to the use of the EM640™ for the analysis of volatile
compounds in all these three environmental matrices. The compounds in Table A can be determined by
applying this method.
Table A. Volatile Organic Compounds
Analyte
Detected with the EM640™
CAS No. (a)
Chloromethane
Bromomethane
Vinyl Chloride
Chloroethane
Methylene Chloride
Acetone
Carbon Bisulfide
1,1 -Dichloroethene
1,1 -Dichloroethane
1,2-Dichloroethene (cis)
1,2-Dichloroethene (trans)
Chloroform
1,2-Dichloroethane
2-Butanone
1,1,1 -Trichloroethane
Carbon Tetrachloride
Vinyl Acetate
Bromodichloromethane
1,2-Dichloropropane
cis-1,3 -Dichloropropene
Trichloroethene
Dibromochloromethane
1,1,2-Trichloroethane
Benzene
74-87-3
74-83-9
75-01-4
75-00-3
75-09-2
67-64-1
75-15-0
75-35-4
75-34-3
156-69-4
156-60-5
67-66-3
107-06-2
78-93-3
71-55-6
56-23-5
108-05-4
75-27-4
78-87-5
10061-01-5
79-01-6
124-83-1
79-00-5
71-43-2
A-l
-------
TABLE A Continued
trans-l,3-Dichloropropene 10061-02-6
Bromoform 75-25-2
4-Methyl-2-Pentanone 108-10-1
2-Hexanone 591-78-6
Tetrachloroethene 127-18-4
1,1,2,2-Tetrachloroethane 79-34-5
Toluene 108-88-3
Chlorobenzene 108-90-7
Ethyl Benzene 100-41-1
Styrene 100-42-5
p-&m-Xylene 106-42-3,108-38-3
o-Xylene 95-47-6
(a) Chemical Abstract Services Registry Number.
1.1 Principle of Operation
Volatile organic compounds (VOCs) in liquid or solid samples have to be transferred to the gas phase. The
gas phase carrying the VOCs extracted from liquid or solid samples, as well as gaseous samples, are
introduced into a gas chromatograph (GC) for separation. Compounds eluting out of the GC column
permeate through an inlet membrane into the vacuum chamber of the mass spectrometer (MS). In the ion
source, the molecules are ionized by electron impact. All ions pass through the quadrupole which filters
them depending on their mass-to-charge ratios (m/z). Finally, the ions are detected in an electron multiplier
and an electrical signal is generated proportional to the number of ions. The data system records these
electrical signals and converts them into a mass spectrum. The sum of all ions in a mass spectrum
corresponds to one point in the gas chromatogram (total ion chromatogram). A mass spectrum is like a
fingerprint of a compound. These fingerprints are compared with stored library spectra and used together
with the GC retention times for the identification of the compounds. The integrated area under a GC peak
is used for quantitation.
1.2 Detection Limits
Detection limits vary depending on compound, media, operation mode of the MS ("scan" or "single ion
monitoring") and sample volume. Generally, for most of the compounds in Table A the practical
quantitation limits (PQL) in the "scan mode" are 100 ng/L for air (100 mL sample volume) and water
samples (250 mL sample volume). For soil samples (6 g) the PQL is approximately 50 mg/kg. The
operation mode "single ion monitoring" (SIM) of the MS gives a factor of 10 in sensitivity. To express this
in absolute values, the mass spectrometer needs 1 ng of a compound to produce a signal-to-noise ratio of
10 in the scan mode.
2.0 Summary of Method
2.1 Water Samples
Water samples of 250 ML volume are sprayed in a spray chamber. During the spray process volatile
compounds are extracted into the gas phase (purified air), transferred to a Tenax tube, and trapped. In the
desorption port of the EM640™ the sorbent tube is heated and backflushed with air to desorb trapped
A-2
-------
sample components. The analytes are desorbed directly to a capillary for GC/MS analysis (8 m DBS
column; phase: 5 (im; I.D.: 0.32 (im). The GC-capillary is temperature-programmed. Analytes separated in
the capillary are then detected with the quadrupole mass spectrometer interfaced to the gas chromatograph.
The mass spectrometer (EM640™) is equipped with a membrane separator, which can be used with short
capillary columns with high carrier gas flows and protects the ion source from contamination. Finally, the
detected compounds are automatically identified (library comparison), quantified (internal standard
method), and reported.
2.2 Air Samples
100 mL air is drawn through a Tenax tube, where the organic compounds are collected. Desorption and the
analysis processes are analogous to those for the water samples.
2.3 Soil Samples
Soil samples are analyzed by a headspace technique. Six grams of soil are weighed in a scrubbed vial. The
vial is heated up in a water bath after sealing. Volatile compounds establish an equilibrium between solid
and gas phase. The headspace gas is then drawn into the GC/MS. The procedures for separation and
detection of the compounds are similar to those for the air and water samples.
Sample sizes are flexible and depend primarily on the contaminant concentration. However, the calibration
should be done with the same sample size as in the analysis.
3.0 Interferences
If isomers elute from the GC capillary at the same time and have an equal mass spectrum, they cannot be
identified as different compounds. In such cases it is only possible to give the sum of the isomers for
quantitation. Another quantitation problem can appear if the target mass of a compound is also found in the
media compound that elutes at the same time (for example Benzene-d6 with the target ion m/z 84 and at the
same position in the gas chromatogram is a hydrocarbon that has also an ion with the m/z 84).
Major contaminant sources are volatile substances present on site as well as organic solvents that may be
used for sample extraction. The use and frequent replacement of carbon filters on the GC reduces the
levels of the on-site contamination during sample analyses.
Interfering contamination may occur when a sample containing low concentrations of volatile organic
compounds is analyzed immediately after a sample containing high concentrations of volatile organic
compounds. The preventive technique is rinsing of the purging apparatus (for water samples) one time
between samples. After analysis of a sample containing high concentrations of volatile organic compounds,
one or more calibration blanks should be analyzed to check for cross contamination.
4.0 Equipment and Supplies
4.1 The Spray Extractor
The recommended spray equipment is the Spray-and-Trap Water Sampler (Bruker Analytical Systems Inc.,
Billerica, MA) or for very sludgy samples the Spray-and-Trap suitcase (ecb ONLINE; Schwerin,
Germany). The Bruker Spray-and-Trap Water Sampler device consists of a mechanical pump to inject a
A-3
-------
continuous flow of an aqueous sample into a sealed extraction chamber through a spray nebulizer. The
droplet formation enormously increases the total interfacial area between the sprayed water and the carrier
gas, which supports the transfer of the VOCs into the gas phase. The steadily flowing carrier gas is
transferred to a suitable sorbent tube, which collects the extracted VOCs. In contrast to the purge-and-trap
method, spray-and-trap utilizes a dynamic equilibrium.
4.1.1 Purge and Trap
Alternative to the spray-and-trap, a standard purge-and-trap system can be used to load a sorbent tube.
4.2 The Sorbent Tube
The trap recommended must be a glass tube 110 mm long with an 0.8 mm OD and 0.6 mm ID. The glass
should be sealed prior to use to avoid contamination. The trap must contain the following amounts of
adsorbents: 1/3 of the trap 2,6-Diphenyl phenylene oxide polymer (Tenax), another 1/3 silica gel, and the
last 1/3 Carbopack B. Silanized glass wool is used to separate each adsorbent and also to plug each end to
retain the packing material. If it is not necessary to analyze for dichlorodifluoromethane or other
fluorocarbons of similar volatility, the charcoal can be eliminated and the polymer increased to fill the 2/3
of the trap. If only compounds boiling above 35°C are to be analyzed, both the silica gel and Carbopack
can be eliminated and the polymer Tenax increased to fill the entire trap.
The polymer alone also can be used if the concentrations are large enough. In this case the breakthrough
volumes of the above compounds can easily be reached. A dynamic equilibrium will be formed, where the
same amount of a compound that enters the sorbent tube leaves it at the other end. Thus, the sampled
amount of the compound does not change with increased sampling time. The only way to use this special
method is to always analyze an equal volume and to have a constant flow rate.
4.3 The Injector Systems
4.3.1 TheDesorber Module
The desorber module is used for air and water analysis. A loaded sorbent tube is inserted into this module
and heated up rapidly (400°C/min) to 250°C. After the split injection the tube is backflushed with purified
air. Desorption and injection time, desorption temperature, backflush time and temperature and the carrier
gas pressure are programmable and controlled by the software .
4.3.2 The Headspace Module
At one side of this module is an injection needle. This needle sticks through the septum of a headspace
vial. The needle is coupled to an "Ultimetal" capillary, which is at its other end directly connected to the
GC column. At the downstream side from the inlet membrane of the MS is a pump which regulates the
pressure inside this hollow fiber membrane. This pump is used to draw vapor out of the headspace vial
(time and flow are programmable) into the GC column in the MS. The Ultimetal capillary is heated and
backflushed after injection.
4.4 Gas Chromatograph
The GC is temperature programmable and equipped with a flow controller to maintain a constant flow of
the carrier gas through the column during operation. Filtered ambient air or nitrogen is used as the carrier
gas. The GC is interfaced to the MS with a polysiloxane membrane.
A-4
-------
4.4.1 The GC column
A 8 m long DB-5 (J&W Scientific) narrow bore capillary GC column with 0.32 mm i.d. and 5 (im film
thickness is normally used. The temperature of the column is held at 50 =C as starting temperature, then
increased to 240 =C at 20 =C per minute, and held until all expected compounds have eluted. If it is
acceptable to have a lesser degree of chromatographic separation (or only a few compounds to be
analyzed), the analysis time can be shortened by increasing the slope of the temperature ramp or vice versa.
As carrier gas, charcoal filtered air or nitrogen is used.
4.5 The Mass Spectrometer
Mass spectral data are obtained with electron impact ionization at a nominal electron energy of 70eV. The
mass spectrometer is capable of scanning from 2 to 640 u in less than half a second (max speed 2000 u/s).
To ensure sufficient precision of mass spectral data, the desirable MS scan rate allows acquisition of at
least five spectra while a sample component elutes from the GC.
4.6 The Data System
The core of the data system is a ruggedized computer, that allows measurements during moving and has an
OS/2 multi-tasking software. Thus, data storage and data analysis can take place simultaneously. The
software is capable of automated peak picking, comparison with stored library spectra, quantitation, and
printing a final report. In addition, all analysis parameters are stored for quality control purposes.
4.7 Microsyringes - 10, 25,100, 500, and 1,000 jiL.
4.8 Balance - Top loading, capable of accurately weighing 0.001 g.
4.9 Disposable pipets
4.10 Vials - 2.0 mL, 5.0 mL, and 22 mL with cap and septum.
4.11 A flow-controlled pump for air sampling.
5.0 Reagents and Standards
5.1 Methanol
Methanol should be of a purge-and-trap grade, demonstrated to be free of analytes. It should be stored
apart from other solvents.
5.2 Distilled Water
Distilled water must be free of interferents at the method detection limit (MDL) of the analytes of interest.
5.3 Stock Solutions
A-5
-------
Stock solutions may be prepared from pure standard materials or purchased as certified solutions. Prepare
stock standard solutions in methanol for water and air (injection in a Tedlar bag) analysis and in ethylene
glycol for soil analysis.
5.4 Secondary dilution standards
Using stock standard solutions, use methanol (ethylene glycol for soil) to prepare secondary dilution
standards containing the compounds of interest mixed together. Secondary dilution standards must be
stored with minimal headspace and should be checked frequently for signs of degradation or evaporation,
especially just prior to preparing calibration standards from them. Store in a vial with no headspace for a
maximum of one to two weeks.
5.5 Surrogate standard
A surrogate standard for all described methods is a mixture of three compounds (Dibromofluoromethane,
Toluene-d8 and 4-Bromofluorobenzene) in methanol. This mixture is suggested for EPA method 8260A,
and used for the EM640™ methods.
5.6 Internal standard
The internal standard recommended is a mixture of Benzene-d6 and Xylene-d10in methanol or, if many
hydrocarbons are in the samples, the suggested mixture for EPA method 8260A (Chlorobenzene-ds 1,4-
Dichlorobenzene-d4, 1,4-Difluorobenzene and Pentafluorobenzene in methanol). An internal standard
spiking solution should be prepared from a stock at a needed concentration. The surrogate spiking solution
and the internal standard spiking solution may be combined into one vial.
5.7 Calibration standards
Calibration standards at a minimum of three, preferredly 5-10, concentration levels should be prepared
from a secondary dilution of stock standards. One of the concentration levels should be at a concentration
near, but above, the method detection limit. The remaining concentration levels should correspond to the
expected range of concentrations found in real samples but should not exceed the working range of the
GC/MS system. Each standard should contain each analyte for detection by this method. Calibration
standards must be prepared daily.
5.8 Media spike standards
Media spike standards can be purchased as certified solutions.
6.0 Procedure
6.1 Initial Calibration
6.1.1 Tune and Calibration of the MS
Before the initial calibration the automated mass scale calibration of EM640™ should be started. After
finishing this calibration, the automated tune program (tuning of ion source and detectors) can run. In
general, a mass scale calibration is only necessary once a month.
6.1.2 Analyte Calibration
A-6
-------
A set of minimum of three, preferredly 5-10, calibration standards containing the method analytes,
surrogates, and internal standard are needed. For a calibration over a high dynamic range it is necessary to
have two calibration points for each decade. The analytes should be spiked at concentrations above the
detection limit through the upper concentration range expected during sample analysis. The surrogate
standard should be added in the same amount as the analytes, but the internal standard should remain at a
fixed concentration level (6 (ig/1 for air, 8 (ig/1 for water and 1 mg/kg for soil samples). Analyze the
samples as described in Section 6.3.
6.1.3 Calibration Curves
Tabulate the area response of the characteristic ions against concentration for each compound and each
internal standard. Calculate response factors (RF) for each compound relative to the internal standard or
make a linear regression curve for all compounds also relative to the internal standard.
6.2 Daily calibration
Prior to sample analysis a standard (calibration check) should be run that contains all analytes of interest
near the midpoint concentration for the working range of the initial calibration. Then calculate the
concentrations from all analytes. The software generates a report which contains the percent recovery for
all detected and calibrated compounds. If the percent recovery for all analytes is between 65% and 135%,
the initial calibration curve are used to calculate sample analyte concentration. If the criterion is not met for
any one of the analytes corrective action must be taken. This involves the re-analysis of the continuing
calibration standard, using new stock standard solutions, or the generation of a new initial calibration
curve. The calibration check standard should be run at least after every 10th analysis. Thus, in the field,
where a system recalibration is sometimes not possible, with these standard measurements during the
whole day, corrective action to the initial calibration curve have to be made.
6.2.1 Internal Standard Control
The internal standard responses in the samples must be evaluated after or during data acquisition. If the
Extracted Ion Current Profile (EICP) area for the internal standard from the samples changes by a factor of
two (-50% to +100%) from the last daily calibration standard check, the sample must be reanalyzed and the
mass spectrometer must be inspected for malfunctions and corrections must be made, as appropriate.
Sometimes sample media interferences can adversely affect the internal standard area. If the re-analysis
confirms the deviation of a factor of two or more of the internal standard, the sample results should be
flagged as questionable due to media interferences. If the EICP is different only for one of the 2 or 4 added
internal standard compounds, then use for quantitation only the correct ones. The software allows it easily
to change the calculation in a way, that all relations from the incorrect internal standard are changed to one
of the correct ones. This is the case for the calculation of the calibration curve (or response factor), as well
as for the quantitation of the samples.
6.3 GC/MS Analysis
Prior to sample analysis, the mass scale calibration of the MS and the tune parameters of the ion source
should be checked. This is performed by a special program in the software.
A-7
-------
6.3.1 Surrogate Control
For each sample analyzed, the percent recovery of each surrogate in the sample is calculated. If the percent
recovery is greater than 65% but less than 135%, the sample analysis is assumed to be valid. If the criterion
is not met, the sample must be reanalyzed. Sometimes sample media interferences can adversely affect the
surrogate recovery. If the re-analysis confirms the percent recovery being outside the criterion range, the
sample results should be flagged as questionable due to media interferences.
6.3.2 Sample Storage
The advantage of field instruments is to analyze the samples directly after their collection, which makes
sample preservation and storage unnecessary.
6.3.3 Air Analysis
If the air samples are in a Tedlar-bag or a similar gas container, add the internal standard and the surrogate
mixtures to the gaseous samples. If the samples are collected directly, without any container, inject these
mixtures in the sorbent tube (direct on Tenax). Connect a flow-controlled pump to the sorbent tube and
sample at least 200 mL purified air (outside air filtered through charcoal) through the tube to get rid of the
methanol. Then suck the sample through the tube with a flow of 100 mL/min. The sample volume depends
on the expected concentration. A useable range is between 100 and 1000 mL. It is absolutely necessary to
use exactly the same sample volume for calibration and for samples, if the internal standard is injected in
the tube, if only Tenax is used as sorbent and the sample contains compounds with boiling points under
35 =C (Section 4.2). After sampling, insert the sorption tube in the desorption oven and press the start
button.
Now, the programmed analysis is running. The sorption tube is heated with a temperature gradient of
400°C/min up to 250°C for 90 s. The desorbed compounds were injected for 20 s. After injection, the tube
is backflushed and the GC program starts. The column used and the temperature program depends on the
analytical task. For normal fast field analyses use a short 8 m thick film (5 (im) DB 5 capillary with an i.d.
of 0.32 (im. Start at a temperature of 50 °C and heat with a gradient of 20 °C/min up to 240 °C. It is
recommended to optimize between necessary separation and shortest time, if many samples of the same
kind should be analyzed.
Such a temperature program can for example look like this: Start at 50 =C and heat with a gradient of
15
-------
through which a tubing is fitted. This tubing is connected to the water inlet of the water sampler with a
Swagelok ferrule. The sampler has to be driven with an equal carrier gas and water (sample) flow rate as
during the calibration. Recommended values are 250 mL/min for the carrier gas flow and 80 mL/min for
the water flow rate. An automated program pumps the water sample at the beginning for 15s over a bypass
to fill all sampler tubes with the new sample. Then, the aqueous sample is sprayed for 2 min into the
steadily flowing carrier gas in the extraction chamber. The extracted VOCs are trapped in a sorption tube
which is placed in the gas outlet stream of the sampler. After finishing the sampling, the tube is inserted in
the desorber module of the EM640™. The GC/MS analysis is the same as procedure described in Section
6.3.1.
6.3.5 SoilAnalysis
Weigh 6 g soil sample in a 22 mL headspace vial and seal it. Inject the internal standard and the surrogate
solution through the septum of the vial directly into the soil sample. Do not inject the solutions at the walls
of the vial. Put the vial for exactly 10 min into a waterbath heated on 80°C. Stick a needle through the
septum, to have atmospheric pressure inside the vial, immediately press the start button of the headspace
injector module from the EM640™ and with the other hand (at the same time) place the vial under the
injection needle of the injector. After 15 s injection time, the vial can be removed from the injector
module. The GC/MS analysis is the same as procedure described in Section 6.3.1.
6.3.6 Qualitative Analysis
For the evaluation of the data different methods can be selected. An evaluation file is generated with the
measurement of the standard. This file contains for every compound a time window for the retention time,
the compound's name, and the method for quantitation (for example the target mass of each compound).
The GC peak in this time window is detected and an average spectrum of 5 mass spectra is calculated.
With this spectrum a library search is activated. Only if the correct compound (the compound out of the
evaluation file) is found, the area under the target mass trace is integrated for quantitation. If the
identification is not positive, because it is another compound or because the compound is overlapped with
a media compound, the data postprocessing switches to a manual mode and the user can take a look at the
mass spectrum of this compound by himself. It is the analysts decision to determine if the target ion should
be integrated or not (if the compound is the right one or not). In the worst case, the overlapped media
compound contains the target ion another target ion has to be selected for this special compound. But, if
parameters are changed, a new generation of the calibration curve is needed with the original stored data.
The second mode is completely automated. The dissect algorithm calculates the mass spectra of each GC
peak in a defined time window. This special algorithm is capable of separating compounds, even if their
GC peaks are almost completely overlapped, as long as the compounds have different mass spectra (at least
one significant mass must be different), and as long as their retention time difference is larger than the
acquisition time of half a mass spectrum. The calculated mass spectrum of the pure compound is used for
the library search. If the identification is positive, then the area under the calculated trace is integrated and
used for quantitation of this compound. Negative identification is interpreted in such a way that the
compound found, is not the compound out of the evaluation list and therefore, a calibration curve does not
exist. Calculated quantitative results are printed out in a report.
All compounds that are not in the evaluation file (not quantified) are detected and identified by the dissect
algorithm. In addition, an automatic library search can be done. The result of the evaluation file is to have a
complete identification of all compounds in an analysis and a specific one for all compounds (target ion or
"dissect" area depends on specification).
A-9
-------
6.3.7 Quantitative analysis
The concentrations of the compounds are calculated using the determined peak areas. The quantification
will take place using the internal standard technique. In the quantitation file (stored calibration curves) it is
defined which compound is related to which internal standard.
All compounds that are found by the dissect algorithm but are not in the quantitation file are identified by a
library search. These compounds are listed including their "dissected" areas in the report as tentatively
identified. The internal standards are also listed with their "dissected" areas. Since the concentration of the
internal standards are known, rough concentrations of the compounds can be calculated.
7.0 Quality Control
7.1 Blank Analysis
Before processing any samples the analyst should demonstrate, through the analysis of a calibration blank,
that interferences from the analytical system, glassware, and reagents are under control. Each time a set of
sample is extracted or there is a change in reagents, a reagent blank should be processed as a safeguard
against contamination. The blanks should be carried through all stages of sample preparation and
measurement.
It is necessary to analyze a reagent blank following a heavily contaminated sample where saturated peaks
have occurred. This will eliminate problems of carry over from one sample to the next.
7.2 Calibration Check Standards
Calibration check standards should be run at least after every 10th analysis to determine if the GC/MS
system is operating properly.
7.3 System set up
Prior to sample analysis the mass scale calibration of the MS and the tune parameters of the ion source
should be checked. This is done by a special software program.
7.4 Initial Calibration
For quantitative analysis, an initial calibration of the GC/MS system must exist as specified in Section 6.1.
7.5 Surrogates
For each sample analyzed, the percent recovery of each surrogate in the sample is calculated. For all
samples of the same media (of one batch), the average percent recovery (p), and the standard deviation of
the percent recovery (sp) for each of the surrogates are calculated.
7.6 MS Control
In each analysis run the ion mass 40 (argon, if air is used as carrier gas), or the ion mass 28 (N 2 if nitrogen
is used as carrier gas) is measured. These ions are produced by the carrier gas and can be used as internal
standard for MS sensitivity. Thus a permanent control of the MS system is achieved.
A-10
-------
8.0 Method Performance
8.1 Accuracy
The method accuracy for all three matrices (air, water and soil) is 35%. That means 99% of all measured
results are between 65% and 135 % of the true values.
8.2 Precision
The method precision for air and water analysis is 30% RPD (relative percent difference). For soil analysis
the RPD is 35%.
Method precision and accuracy are determined in a field application for 3 surrogate compounds. The
surrogate concentrations added to the samples, were always the same independent of the concentration of
the analytes (see table B).
Table B:
Surrogates RPD1 A.A.I RPD2 A.A.2 RPD3 A.A.3
% % %
p-Bromofluorobenzene 18.2 91 12.5 105 19.1 104
Dibromofluoromethane 12.5 106 10.2 101 46.5* 135
Toluene-d8 07.6 90 09.0 110 20.7 108
RPD: Relative Percent Difference
A.A.: Average Accuracy
1: 10 Air Samples
2: 11 Water Samples
3: 13 Soil Samples
*: This value is out of range. A reason could be the media influence on this compound.
9.0 References
1. Method 8260 - GC/MS for Volatile Organics: Capillary Column Technique. Test Methods for
Evaluating Solid Waste Physical/Chemical Methods - Third Edition Proposed Update Package., US
EPA, Washington, DC, 1989.
2. Bruker Method B24 (VOA): Gas Chromatography/mass Spectrometry for Volatile Organics:
Capillary Column Technique for On Site Analysis.
3. B. Nolke, G. Baykut, "Enhanced efficiency spray extractor for sampling volatile organic compounds
in aqueous systems" Rev. Sci. Instrum. 65 (1994) 363.
A-ll
------- |