&EPA
United States
Environmental Protection
Agency
Office of Research and
Development
Washington, D.C. 20460
EPA/600/R-97/148
December 1997
Environmental Technology
Verification Report
Field Portable Gas
Chromatograph/Mass
Spectrometer
Viking Instruments Corporation
SpectraTrak™ 672
ET
:SERDP
Environmental Technology
Verification Program
-------
-------
Environmental Technology
Verification Report
Field Portable Gas Chromatograph/ Mass
Spectrometer
Viking Instruments Corporation
SpectraTrak™ 672
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 Laboratory. 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 Viking Instruments Corporation
SpectraTrak™ 672 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: SpectraTrak™ 672
COMPANY: VIKING INSTRUMENTS CORPORATION
ADDRESS: 3800 CONCORDE PARKWAY, SUITE 1500
CHANTILLY, VIRGINIA 20151
PHONE: (703) 968-0101
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 Viking SpectraTrak™ 672 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 soil gas, water, and soil.
The primary target analytes at the U.S. Department of Energy's Savannah River Site, near Aiken, South Carolina, were
trichloroethene and tetrachloroethene. The primary analytes at Wurtsmith Air Force Base, in Oscoda, Michigan, were
EPA-VS-SCM-10 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, Viking Instruments Corporation
SpectraTrak™ 672." The EPA document number for this report is EPA/600/R-97/148.
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 molecules
into characteristic ions. These ion fragments are then separated by mass and detected as charged particles, which
constitutes a mass spectrum. Quantitation is achieved by comparing the abundance of those ions which are
characteristic of a specific compound to the response received from the mass spectrum obtained from a reference
standard. 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 Viking SpectraTrak™ 672 is a commercially available GC/MS system that provides laboratory-grade performance
in a field transportable package. The instrument, including the on-board computer, is ruggedized and encapsulated in
a shock-mounted transport case. It weighs about 145 Ibs. and can be transported and operated in a small van. The
instrument used in the demonstration used a purge and trap device for water and soil analysis and direct injection for
soil gas samples. The minimum detection limit is 5 ppb for soil gas, 5 jWg/kg for soil, and 5/^g/L for water. The
instrument requires a skilled operator. Recommended training is one week for a chemist with GC/MS experience. At
the time of the demonstration, the baseline cost of the SpectraTrak™ 672 was $145,000.
VERIFICATION OF PERFORMANCE
The observed performance characteristics of the SpectraTrak™ 672 include the following:
• Throughput: Sample throughput was approximately 30 minutes for soil extracts and water samples using
purge and trap. The direct injection soil gas samples required 15 minutes each for analysis.
• Completeness: The SpectraTrak™ 672 detected 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 30 percent RPD for soil and 0 to 15 percent RPD for the water and soil
gas samples.
EPA-VS-SCM-10 The accompanying notice is an integral part of this verification statement December 1997
iv
-------
• Accuracy: Accuracy was evaluated by comparing the Viking 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 fall in the 0 to 20 percent range
with a median of 13 percent. For water, most of the values fall in the 0 to 20 percent range with a median of
14 percent. The soil gas accuracy data generally fall in the 0 to 60 percent range with a median of 28 percent.
• Comparability: The SpectraTrak™ 672 produced water and soil data that were comparable to the reference
laboratory data (median absolute percent difference was less than 50 percent). The soil gas data were not
comparable. This was due in part to difficulties experienced by the reference laboratory in analyzing soil gas
samples and other problems associated with sample handling and transport.
• Deployment: The system was ready to analyze samples within 30 minutes of arrival at the site. At the
Savannah River Site the instrument was transported in and operated from a hatchback passenger car; a van was
used at the Wurtsmith Air Force Base site.
The results of the demonstration show that the Viking SpectraTrak™ 672 field portable gas chromatograph/mass
spectrometer can provide useful, cost-effective data for environmental problem-solving and decision making. The
deviation between the Viking GC/MS and reference laboratory results for the soil gas samples, while statistically
significant, is not so great as to preclude the effective use of the Viking GC/MS system in many field screening
applications. We were unable to determine whether the Viking soil gas 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-10 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
-------
Acknowledgments
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 Viking Instruments Corporation, in particular, Ms. Lisa White,
Applications Chemist, who operated the instrument during the demonstrations.
For more information on the Viking 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 Viking GC/MS technology, contact:
Jeffrey Christenson, Director of Marketing
Viking Instruments Corporation
3800 Concorde Parkway, Suite 1500
Chantilly, Virginia 20151
(703) 968-0101
(703)968-0166
vn
-------
Contents
Notice ii
Foreword vi
Acknowledgments vii
Figures xiii
Tables xiv
Abbreviations and Acronyms xv
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 12
Detection Limits 12
Dynamic Range 12
Sample Throughput 12
Vlll
-------
Technology Advantages 13
Technology Limitations 13
Training Required 13
Sample Matrix Effects 14
Spectral Interferences 14
4. Site Descriptions and Demonstration Design 15
Technology Demonstration Objectives 15
Qualitative Assessments 15
Quantitative Assessments 15
Site Selection and Description 16
Savannah River Site Description 16
Wurtsmith Air Force Base Description 18
Overview of the Field Demonstrations 21
Overview of Sample Collection, Handling, and Distribution 22
SRS Sample Collection 22
WAFB Sample Collection 25
Reference Laboratory Selection and Analysis Methodology 26
General Engineering Laboratory 27
Traverse Analytical and Pace Environmental Laboratories 27
SRS and WAFB On-Site Laboratories 27
Pre-demonstration Sampling and Analysis 28
Deviations from the Demonstration Plan 28
Pre-demonstration Activities 28
SRS Soil Spike Samples 28
SRS Soil Gas Survey Evaluation 28
Soil Gas Samples at WAFB 29
Water Samples at WAFB 29
Calibration Check Sample Analysis 29
5. Reference Laboratory Analysis Results and Evaluation 30
Laboratory Operations 30
General Engineering Laboratories 30
SRS On-Site Laboratory 30
Traverse Analytical Laboratory 30
ix
-------
Pace Environmental Laboratories 31
WAFB On-Site Laboratory 31
Laboratory Compound Detection Limits 31
Laboratory Data Quality Assessment Methods 31
Precision Analysis 32
Accuracy Analysis 32
Laboratory Internal Quality Control Metrics 33
Laboratory Data Quality Levels 34
Laboratory Data Validation for the SRS Demonstration 34
GEL Data Quality Evaluation 34
GEL Data Quality Summary 36
SRS On-Site Laboratory Data Quality Evaluation 36
SRS Laboratory Data Quality Summary 37
Laboratory Data Validation for the WAFB Demonstration 37
Traverse Data Quality Evaluation 38
Traverse Laboratory Data Quality Summary 40
Pace Data Quality Evaluation 40
Pace Data Quality Summary 42
Summary Description of Laboratory Data Quality 42
6. Technology Demonstration Results and Evaluation 44
Introduction 44
Pre-Demonstration Developer Claims 44
Field Demonstration Data Evaluation Approach 45
Instrument Precision Evaluation 45
Instrument Accuracy Evaluation 46
Instrument Comparison with Reference Laboratory Data 47
Summary of Instrument Performance Goals 50
Accuracy 50
Precision 51
Viking to Reference Laboratory Comparison 52
Field Operation Observations 53
Viking Accuracy and Precision Results 54
Viking Accuracy ~ SRS Demonstration 54
Viking Accuracy ~ WAFB Demonstration 55
Overall Viking Accuracy Performance 56
Viking Precision ~ SRS Demonstration 57
-------
Viking Precision ~ WAFB Demonstration 58
Overall Viking Precision Performance 59
Viking to Reference Laboratory Data Comparison 60
Scatter Plots/Histograms ~ SRS Demonstration 60
Scatter Plots/Histograms ~ WAFB Demonstration 60
Overall Viking to Laboratory Comparison Results 67
Summary of Viking Accuracy, Precision, and Laboratory Comparison Performance 68
Other Viking GC/MS Performance Indicators 69
Unknown Compound Identification in Complex Mixtures 69
Field Handling and Operation 69
Overall Viking GC/MS Performance Conclusions 70
7. Applications Assessment 72
Applicability to Field Operations 72
Capital and Field Operation Costs 72
Discussion of the Technology 72
Rapid Analysis 72
Sampling and Sample Cost Advantages 73
Performance Advantages 73
Transportability 74
Field Screening of Samples 74
Interferences 74
Conclusions 74
8. Developer's Forum 75
General Comments 75
New Viking Transportable GC/MS System 76
9. Previous Deployments 79
10. References 80
XI
-------
Appendix
A: Analytical Method for the Operation of the SpectraTrak™ 672 A-l
xn
-------
Figures
2-1 Example total ion chromatogram of a complex mixture 7
3-1 Drawing of Viking SpectraTrak™ 672 GC/MS 10
3-2 Drawing of the weather-proof, shock-protected transport case for
the Viking SpectraTrak™ 672 10
4-1 Location of the Savannah River Site 17
4-2 SRS M-Area Well Locations 19
4-3 Location of Wurtsmith Air Force Base 20
4-4 WAFB Fire Training Area 2 Sampling Locations 21
6-1 Example scatter plots with simulated data 49
6-2 Example histograms with simulated data 51
6-3 Plot of daily temperatures during the SRS demonstration 54
6-4 Plot of daily temperatures during the WAFB demonstration 54
6-5 Absolute percent accuracy histogram for Viking soil samples 56
6-6 Absolute percent accuracy histogram for Viking water samples 56
6-7 Absolute percent accuracy histogram for Viking soil gas samples 57
6-8 Relative percent difference histogram for Viking soil samples 59
6-9 Relative percent difference histogram for Viking water samples 59
6-10 Relative percent difference histogram for Viking soil gas samples 60
6-11 Viking vs. Laboratory data for SRS low concentration water samples 62
6-12 Viking vs. Laboratory data for SRS high concentration water samples 62
6-13 Percent difference histogram for SRS water samples 63
6-14 Viking vs. Laboratory data for SRS soil gas samples 63
6-15 Percent difference histogram for SRS soil gas samples 63
6-16 Viking vs. Laboratory data for WAFB soil samples 63
6-17 Relative percent difference histogram for WAFB soil samples 64
6-18 Viking vs. Laboratory data for WAFB low concentration water samples 65
6-19 Viking vs. Laboratory data for WAFB high concentration water samples 65
6-20 Relative percent difference histogram for WAFB water samples 66
6-21 Viking 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 67
6-25 Absolute percent difference histogram for gas samples 68
6-26 Viking GC/MS total ion chromatogram from a WAFB water sample 70
8-1 A breakaway view of the Viking SpectraTrak™ 572 Portable GC/MS 78
Xlll
-------
Tables
3-1 Viking SpectraTrak™ 672 Instrument Specifications 11
3-2 Detection Limits for the Viking SpectraTrak™ 672 GC/MS 12
4-1 PCE and TCE Concentrations in SRS M-Area Wells 18
4-2 Historical Ground Water Contamination Levels at WAFB 19
4-3 VOC Concentrations in WAFB Fire Training Area 2 Wells 21
4-4 Sample Terminology and Description 23
4-5 SRS Demonstration Sample Type and Count 24
4-6 WAFB Demonstration Sample Type and Count 25
5-1 Reference Laboratory Practical Quantitation Limits 31
5-2 GEL Laboratory Accuracy Data 35
5-3 GEL Laboratory Precision Data 36
5-4 SRS Laboratory Accuracy Data 37
5-5 SRS Laboratory Precision Data 37
5-6 Traverse Laboratory Accuracy Data 38
5-7 WAFB Water and Soil PE/Spike Sample Reference Concentrations 39
5-8 Traverse Laboratory Precision Data 39
5-9 WAFB Water and Soil Duplicate Sample Concentrations 39
5-10 Pace Laboratory Accuracy Data 41
5-11 WAFB Soil Gas PE/Spike Sample Reference Concentrations 41
5-12 Pace Laboratory Precision Data 42
5-13 WAFB Soil Gas Duplicate Sample Concentrations 42
5-14 SRS Demonstration Laboratory Data Quality Ranking 42
5-15 WAFB Demonstration Laboratory Data Quality Ranking 43
6-1 Viking GC/MS Recoveries at SRS 55
6-2 Viking Recoveries at Wurtsmith 55
6-3 Viking and Reference Laboratory Accuracy Summary 57
6-4 Viking Precision for SRS Demonstration 58
6-5 Viking Precision for Wurtsmith Demonstration 58
6-6 Viking and Reference Laboratory Precision Summary 60
6-7 Viking-Laboratory Comparison Summary 68
6-8 Summary Performance of the Viking GC/MS 69
6-9 Tentatively Identified Compounds from a Wurtsmith Water Sample Analysis 70
6-10 Summary of Viking Performance Goals and Actual Performance 71
7-1 Viking SpectraTrak™ 672 GC/MS Capital and Field Operation Costs 74
xiv
-------
Abbreviations and Acronyms
AC Alternating current
amu Atomic mass unit
amp Ampere
APA Absolute percent accuracy
APD Absolute percent difference
BTEX Benzene, toluene, ethylbenzene, xylenes
CSCT Consortium for Site Characterization Technology
DNAPL Dense nonaqueous phase liquid
DCE Dichloroethylene
DIP Percent difference
DoD Department of Defense
DOE Department of Energy
DOT Department of Transportation
EPA Environmental Protection Agency
ESD-LV Environmental Sciences Division
ETV Environmental Technology Verification Program
ETVR Environmental Technology Verification Report
g Gram
GC/MS Gas chromatograph/mass spectrometer
GEL General Engineering Laboratories
Hz Hertz
kg Kilogram
kW Kilowatt
L Liter
//g Microgram
mg Milligram
mL Milliliter
MS Mass spectrometer
NCIBRD National Center for Integrated Bioremediation Research and Development
NA Not analyzed
ND Not detected or no determination
NERL National Exposure Research Laboratory
NETTS National Environmental Technology Test Sites Program
ng nanogram
NP Not present
PAH Polycyclic aromatic hydrocarbons
PCE Tetrachloroethene
PE Performance evaluation
ppb Parts per billion
ppm Parts per million
ppt Parts per trillion
PQL Practical quantitation limit
QA Quality assurance
QC Quality control
REC Percent recovery
RPD Relative percent difference
xv
-------
RSD Relative standard deviation
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
xvi
-------
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 transportable 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 hasten 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.J
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 Viking Instruments Corporation SpectraTrak™ 672 field-transportable GC/MS. Demonstration results
from the other system are presented in a separate report.
Technology Description
The Viking SpectraTrak™ 672 GC/MS is an integrated system combining a temperature programmable
gas chromatograph combined with a Hewlett Packard quadrupole mass spectrometer. This self-contained,
field transportable system, whose design has been adapted from laboratory technology, uses a
chromatographic column and accompanying mass spectrometer to provide separation, identification, and
quantification of volatile and semi-volatile organic compounds in sample matrices that include soil, liquid,
and gas. The column enables the separation of individual analytes in complex mixtures. A mass spectrum,
produced for each compound in the sample, can be used for compound identification and quantification.
An integrated computer and data acquisition system enables identification and quantification of the
analytes by comparison of detector response with a calibration table constructed from standards of known
concentration. The system provides detection limits that range from about 5 ppm for direct gas injection to
as low as 5 ppb for many volatile and semi-volatile organic contaminants in soils and liquids.
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 Viking SpectraTrak™ 672 GC/MS system. Accuracy was evaluated by comparing the
Viking 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 fall in the 0 to 20 percent range with a median of 13 percent. For water, most
of the values fall in the 0 to 20 percent range with a median of 14 percent. The soil gas accuracy data
generally fall in the 0 to 60 percent range with a median of 28 percent. 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 30
percent RPD for soil and 0 to 15 percent RPD for the water and soil gas samples. The SpectraTrak™ 672
produced water and soil data that were comparable to the reference laboratory data (median absolute
percent difference was less than 50 percent). However, the soil gas data were not comparable. This was
due in part to difficulties experienced by the reference laboratory in analyzing soil gas 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 Viking system can eliminate some of these sample handling problems.
-------
Performance Evaluation
Overall, the results of the demonstration indicated that most of the performance goals for the Viking
GC/MS system were met under field conditions, and that the system can reliably provide good quality,
near-real-time field analysis of soil, water, and soil gas samples contaminated by organic compounds. The
system was easily transported in an automobile and required only one technician for operation. A limited
analysis of capital and field operational costs for the Viking system shows that field use of the system may
provide some cost savings when compared to fixed-laboratory analyses. Based on the results of this
demonstration, the Viking SpectraTrak™ 672 instrument was determined to be a mature field instrument,
capable of providing on-site analyses of soil and water samples comparable to those from a 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 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
0
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 2
The SpectraTrak™ 672 is a complete GC/MS system that provides laboratory-grade performance in a
transportable package (Figure 3-1). The system is equipped with a temperature-programmable mini-GC
and a Hewlett Packard 5972A quadrupole MS. Scan rates up to 1,800 atomic mass units (amu) per second
are possible over a mass range of 1.6 to 700 amu with unit-mass resolution. General instrument
specifications are presented in Table 3-1. A diagram of the instrument in a weather-proof, shock-mounted
case is shown in Figure 3-2.
Sample introduction techniques include the following:
• Split/splitless injection.
• Ambient air sampling, concentration, and thermal desorption.
• Purge-and-trap on-line sampling of volatile organic compounds in water or soils using either a
compact single-sample sparger or commercially available autosamplers.
• Direct MS, using a membrane inlet, which allows for very rapid screening of target compounds or
simple unknowns.
The information presented in the remainder of Section 3 was provided by Viking. It has been minimally edited. The information is solely that of
Viking and should not be construed to reflect the views or opinions of the EPA.
-------
C'piijuJi
»nri :
Carritf 'JES
I'i«l
an; I-'
Stow ' ""*•'• ,, *""••: v
X* I I ""S^5^ -S^^
flO^flfe^'CB,",
I Ix^' ^----•^^"^
«>^^ '
^->f^^" ^'/^Ss. /
*.•<*&* -%j__;-\ x ,•
€^'
-" -'^""l^^^^^^ ,' •>»
^;:"'" -" --"^
Figure 3-1. Drawing of Viking Spectral"rak™ 672 GC/MS. (Courtesy of Viking Instruments)
Figure 3-2. Drawing of the weather-proof, shock-protected
transport case for the Viking Spectral"rak™ 672.
(Courtesy of Viking Instruments)
10
-------
The technology is designed to provide separation, identification, and quantification of volatile and
semivolatile organic compounds in solids, liquids, or gases. The GC separates a complex combination of
analytes and isolates individual analytes based on vapor pressures and chemical affinities for the stationary
phase of the chromatographic column. As the individual analytes exit the column, the MS detects the
analytes and provides a characteristic mass spectrum that positively identifies each compound. The
Windows-based computer system provides quantitation and interprets the detector response. The computer
system then compares the detector response with a calibration table constructed from standards of known
concentration.
Table 3-1. Viking Spectralrak™ 672 Instrument Specifications. (Supplied by Viking Instruments
Corporation.)
Parameter
Developer's Specification
Detection limits
Linear dynamic range
Mass range
Analysis time
Weight
Size
Operator
Power requirement
Support equipment
On-Board Computer
Cost
Down to low ppm air for direct injection (depending on analyte)
Down to 5 ppb for air preconcentration
Down to 5 f^g/L purge and trap water samples
Down to 5 //g/kg soil samples
4 orders of magnitude
1.6-700amu
Direct injection, soil vapor, 10-15 min.
Purge and trap for soil and water samples 30 to 40 min.
145 Ibs
14 in. high x 21 in. wide x 32 in. deep
1 operator (general GC/MS background), 1 week of training
1,300 W startup; 1,000 W during analyses
External roughing pump, 120 volts on AC power
Hewlett Packard 486 Personal Computer with
Windows 3.1
$145K
The unit requires an external mechanical roughing pump (or rough vacuum source) and an alternating
current (AC) electrical power source. The AC source must be 110 v or 220 v, 50 to 60 Hz, with continuous
power demand of approximately 1,000 W. The instrument requires about 1,300 W at start-up. An on-board
helium carrier gas supply is contained in a pressurized cylinder that meets Department of Transportation
(DOT) regulations for transportation. A fully pressurized cylinder typically operates for one week,
depending on instrument usage.
The analytical system, including a high performance turbomolecular pump, is encapsulated in a shock-
mounted transport case. When closed, the transport case protects the instrument from excess humidity and
shock vibrations encountered during normal handling and shipment. The encased unit weighs
approximately 145 Ibs and is 14 in. high, 21 in. wide, and 32 in. deep. The closing of a vacuum isolation
valve on the back of the unit allows the system to be transported with the vacuum intact. Transporting
under vacuum greatly minimizes start-up pump-down time, permitting rapid deployment from site to site.
Performance Factors
The following sections describe the Viking Instruments SpectraTrak™ 672 GC/MS performance factors.
These factors include detection limits, dynamic range, and sample throughput.
11
-------
Detection Limits
Typically, detection limits, as summarized in Table 3-2, are in the low //g/kg range for soil and the low
//g/L for water samples using relatively small sample volumes (1 to 5 g). Parts per trillion detection limits
are possible for gaseous samples due to the ease of concentrating from a larger sample volume (1 to 10 L).
Table 3-2. Detection Limits for the Viking SpectraTrak™ 672 GC/MS. (Supplied by Viking Instruments
Corporation.)
Analyte
Chloromethane
Vinyl chloride
Chloroethane
Bromomethane
Acetone
1 , 1 -Dichloroethene
Methylene chloride
Carbon disulfide
Trans-l,2-Dichloroethene
1 , 1 -Dichloroethane
2-Butanone
Chloroform
1 ,2-Dichloroethane
1,1,1 -Trichloroethane
Carbon Tetrachloride
Benzene
1 ,2-Dichloropropane
Detection
Limit *
10
10
10
10
100
5
5
100
5
5
100
5
5
5
5
5
5
Analyte
Trichloroethene
Bromodichloromethane
cis- 1 , 3 -Dichloropropene
trans- 1,3- Dichloropropene
1, 1,2-Trichloroethane
Toluene
Dibromochloromethane
Tetrachloroethene
Chlorobenzene
Ethylbenzene
m+p-xylenes
Styrene
1, 1,2,2-Tetrachloroethane
o-xylene
1,3- Dichlorobenzene
1,4-Dichlorobenzene
1 ,2-Dichlorobenzene
Detection
Limit *
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
* Estimated detection limits. Units depend on sample preparation:
Mg/kg For 5 gram soil samples concentrated by purge and trap.
//g/L For 5 mL water samples concentrated by purge and trap.
ppm For 250 //L of air by direct injection.
ppb For 0.25 to 1.0 L of air concentrated onto sorbent tubes.
Dynamic Range
Approximately 4 orders of magnitude linear dynamic range are possible with the Viking SpectraTrak1
depending upon the analyte and analysis conditions.
'672
Sample Throughput
Sample throughput is a measure of the amount of time required to prepare and analyze one field sample.
This, in turn, defines the number of samples that can be analyzed in one work day. Viking claims the
complete analysis times as follows: direct injection for soil vapor, 10 to 15 minutes; purge and trap for soil
and water samples, 30 to 40 minutes. This does not include sample handling, data documentation, or
difficult dilutions and concentrations.
Technology Advantages
12
-------
Reliable on-site analyses provides timely information that is representative of actual site conditions. The
timely results assist decision making and eliminate the need for costly and time consuming laboratory
analyses.
The narrow-bore columns used on the SpectraTrak™ 672 provide faster VOC analysis than analyses
performed on typical megabore columns. (Example: 12 minutes versus 30 minutes for EPA Method 8260
compounds.)
Other advantages of the technology include:
• Calibrations remain stable even after transporting the instrument, so more samples are
analyzed per day;
• The dynamic range is linear across 4 orders of magnitude, depending on the compounds and
experimental conditions;
• Columns can be installed without venting, so applications can be substituted with minimal
time delays;
• Laboratory-quality performance in the field can be achieved; and,
• The technology is shock-mounted for field ruggedness and ease of transport.
Technology Limitations
Limitations of the technology include:
• GC columns with an inside diameter greater than 0.32 mm are not recommended for use in the
SpectraTrak™ 672. Oven dimensions require that the columns be wound on a 3.5 inch
diameter column cage. Wider bore columns may become brittle and break when wound to 3.5
inches in diameter;
• The system requires approximately 1.3 kW of power for operation. This may require the use of
a portable generator for operation in the field; and,
• High initial capital costs. System costs with training are about $150K3.
Training Required
A one-week training course is usually sufficient to learn the operations and effectively analyze samples on
the SpectraTrak™ 672. However, training duration depends on the user's previous knowledge and
experience. Users familiar with GC/MS and Hewlett Packard software can often run samples after one day
of training.
Sample Matrix Effects
Sample matrix effects are deviations in observed instrument output caused by variations in the composition
of the sample media. This effect can be pronounced when comparing different sample media such as gases,
liquids, or soils. This effect can also be significant in a single media where the composition of the media
3 The new SpectraTrak™ 572 model, introduced after this study, provides equivalent or better performance at a lower cost,
starting from $90,000. The new system is described in the Developer's Forum (Section 8).
13
-------
can vary significantly. For example, the mineral composition of soils can vary greatly at a site, possibly
affecting sample analysis results.
Spectral Interferences
Interferences can occur in the presence of excessive water vapor or contamination resulting in altered
detection limits and contamination build up. Instrumentation must be periodically checked by analyzing
blanks for contamination to ensure there is no residual background contamination from sample to sample.
14
-------
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
15
-------
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 activities4. 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)
16
-------
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.
17
-------
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 (ng/L)
12
110
3700
TCE (ng/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
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.
18
-------
MHT-11C*
MHT-12C *
MHV-2C *
North
<-
sewer line
CPT-RAM15*
MHT-17C *
CRT-RAM 4 *
Figure 4-2. SRS M-Area Well Locations.
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 ,ag/L.
WAFB Geologic and Hydrologic Characteristics
The WAFB site rests on a 30-80 ft. 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 ft. thick layer of silty-clay deposited through settlement of the silt
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
19
-------
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.
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
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. Fiistoric 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
20
-------
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
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.
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
SpectraTrak™ 672 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
21
-------
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.
The Viking system is compact and rugged and was moved daily to and from each site in the back of a
passenger vehicle. Since it is essentially self-contained, it did not require a dedicated support vehicle. The
Viking system required an external roughing pump and printer for the field operations. At the SRS site,
where no field power was available, a small generator was used. At both demonstration sites the system
was operated by one technician, who analyzed as many as thirteen samples a day and provided the results
of the analyses at the end of each day. The system was able to analyze all the samples provided each day at
both demonstration sites with no operational or mechanical problems.
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® canisters were
22
-------
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
o
J
o
J
PE Samples
1
1
2
Total
13
13
7
sent to the reference laboratory for analysis. For the soil gas survey, soil vapor5 from each well was
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.
23
-------
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. The type of sample depended on the requirements of each technology, with the Viking
receiving their samples in 40 mL volatile organic analysis (VOA) vials. 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.
24
-------
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,
25
-------
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
26
-------
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 SRS 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
27
-------
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
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. Pre-demonstration 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 Viking 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 Viking 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.
28
-------
Water Samples at WAFB
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. These data
are used by the instrument operator to ensure optimum instrument performance. The calibration data were not
used in the evaluation of instrument performance. This deviation was not considered significant.
29
-------
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,
30
-------
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
2Mg/L
lOppb
100 Mg/kg
2//g/L
lOppb
100 //g/kg
lMg/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.
Precision Analysis
31
-------
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
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
JE = — • D
x
where ~x.instrument is the measured concentration by a field instrument and ^reference 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
the sample. In some cases, the PE samples require additional preparation in the field, for example, sample
32
-------
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:
33
-------
Good Data Quality
• Good laboratory internal quality control6 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 control7 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, 10 of 13 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.
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.
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.
34
-------
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
water 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 (//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
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.
Table 5-3. GEL Laboratory Precision Data.
Sample Media/Description
Water Low
Water Medium
Water High
Reference Concentration Gug/L)
PCE
10
150
12,200
TCE
60
160
6,000
Relative Percent Difference
PCE
3
2
5
TCE
22
5
2
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.
-------
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 (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
36
-------
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 (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 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 were 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
37
-------
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 are presented in Table 5-6 and are 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 Cwg/kg)
Soil PE No. 2 Cug/kg)
Water Low Spike C"g/L)
Water Medium Spike C"g/L)
Water High Spike (//g/L)
Water PE Sample 1 (//g/L)
Water PE Sample 2 (Mg/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
Traverse Precision Data
The precision data for Traverse are presented in Table 5-8 and are 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.
38
-------
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 (Mg/kg)
Water Low C"g/L)
Water Medium (//g/L)
Water High Gug/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.
39
-------
Pace Data Quality Evaluation
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 are 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
o
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
40
-------
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
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 8 of 11 RPD values fall within the 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
41
-------
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.
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 good 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.
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
42
-------
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
43
-------
Section 6
Technology Demonstration Results and Evaluation
Introduction
Analytical results and an evaluation of the Viking SpectraTrak™ 672 GC/MS data collected during the
SRS and WAFB demonstrations are presented in this section. The demonstrations provided an
opportunity for analysis of soil, water, and soil gas media by the Viking GC/MS system. Data from the
Viking system are compared to the previously discussed reference laboratory data. Following
presentation of the Viking sample analysis data, instrument performance is assessed using a number of
performance goals also described in this section.
Pre-Demonstration Developer Claims
Before the actual field demonstration, the Consortium requested instrument performance claims from the
Viking Instruments Corporation. The performance claims provided by Viking 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 Viking for the GC/MS
system taken from the demonstration plan prepared prior to the field demonstrations [SNL, 1995] are as
follows:
• Accuracy: Instrument data within ±30% of reference laboratory values.
• Precision: Instrument relative percent differences less than or equal to 25%.
• Completeness: For all samples analyzed, 95% of target VOC compounds detected by
reference laboratory also detected by the instrument.
• Sample throughput: (Dependent upon analysis methodology.) Samples requiring pre-
concentration: 30 minutes per sample; Direct injection: 10 to 15 minutes per sample.
Sample input via membrane inlet: 20 to 30 seconds per sample.
• Methodology: EPA Method 8260, a fixed-laboratory procedure.
• Calibration Stability: Instrument multi-point calibration completed before arriving at
the demonstration site. Daily one-point calibration checks during demonstration.
• Reported Data: Quantitative results submitted at the end of each day of the
demonstration.
• Deployment: Instrument can be set up and ready for sample runs within 30 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?
44
-------
• Should reference laboratory data be used for comparison with technology data if
significant inaccuracies are encountered in the reference laboratory data?
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 and EPA Method TO-14 for soil gas analyses. The development
of these performance goals and evaluation criteria are further discussed in the following sections.
Field Demonstration Data Evaluation Approach
A discussion of the methodology and its underlying rationale used for Viking 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 Viking data to reference laboratory
data. The evaluation methodology uses instrument performance claims made by Viking 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 Viking 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 Viking 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 the GC/MS system. 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 the 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 evaluated by computing the relative percent difference of
duplicate sample results. The definition for relative percent difference (RPD) is as follows:
RPD = —^=- • 100
x
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 precision claim, made by Viking prior to the start of the
demonstrations, was RPD values of 25% or less. Refinement of these initial claims into specific
performance goals was done following the demonstration in order to incorporate statistical
45
-------
considerations of the data. For example, allowance is given for occasional Viking RPD values that fall
outside the 0-25% range. The Viking RPD data are pooled by sampling media and a determination is
made as to whether the median value of the distribution falls in the 0-25% range. This first precision
performance criterion is consistent with that specified in Methods 8260s 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 Viking RPD value is less than or equal to 95th percentile of RPD values similarly
pooled from the reference laboratory data. The underlying rationale is that Viking 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, Viking precision performance is judged acceptable.
Instrument Accuracy Evaluation
Instrument data accuracy is evaluated by comparing the Viking 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 soil 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 multiple 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:
REC = X'"strument • 100
x
reference
where ^imirument is the measured concentration by a field instrument and ~x.refereiKe 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-100\
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 Viking 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. Viking accuracy performance goals are stated in the context of
Viking 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. The absolute percent accuracy values for Viking are pooled by
8 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."
46
-------
sampling medium. A determination is made as to whether the median APA value falls within the 0-30%
range. A second evaluation is done if the first criterion is not met. In this test, a determination is made as
to whether the median Viking APA value is less than or equal to 95th percentile of APA values pooled
from the reference laboratory data. As with the precision assessment, the understanding is that Viking
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, Viking accuracy performance is
judged acceptable.
Instrument Comparison with Reference Laboratory Data
A third approach for evaluation of Viking GC/MS instrument performance entails a comparison of
Viking results with reference laboratory results for paired sample analyses. As described in Section 5, the
reference laboratory data were ranked good, satisfactory and unacceptable in terms of overall quality.
Only those laboratory results that were ranked good or satisfactory are 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 influenced by factors such as sample transport and storage, improper instrument calibration,
operator technique, and instrument noise, to name a few. Even in cases where sample transportation and
storage operations are performed correctly, the sample that reaches the analytical instrument may be
different in chemical composition from that analyzed by the field instrument as a result of
heterogeneities in the sample matrix, chemical reaction, evaporative losses, microbial degradation, and
others.
The Viking to laboratory comparison takes these uncertainties into account by computing the percent
difference between the Viking 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:
[v - r 1
L instrument lab-' _ i ,-,,-,
Xlab
where ~ximtnment is the measured concentration by a field instrument and x/a6 is the measured concentration
of the same sample by the reference laboratory. A related term, the absolute percent difference, 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 call out ±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 Viking to laboratory comparison performance goal stipulates that the
median absolute percent difference for each sampling media should be in the range of 0-50%. Failure to
meet this goal suggests that a significant difference between the Viking system and the laboratory may
exist. As part of the instrument to laboratory comparison, 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
47
-------
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 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
Viking 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 to
indicates 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 between equivalent methods.
The outcome of the Wilcoxon test is used to make a final decision as to whether one is justified in calling
the Viking field measurements comparable or not comparable to reference laboratory measurement. A
Wilcoxon test result with a p-value less than 0.05 indicates that Viking data are not comparable to
reference laboratory data for a particular sampling medium.
In summary then, two criteria are used for assessing Viking 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 Viking 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 Viking 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 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 line denotes acceptable tolerances on field instrument comparisons to
reference laboratory data.
48
-------
In some cases where the data span five or more orders of magnitude, two plots are used. The plots
provide the reader an indication of instrument bias and correlation9 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 a range spanning several orders of magnitude.
£
0>
2
0>
IT
>
I
0>
2
il
^?
Lab Value
(a) High correlation; low bias
Lab Value
(b) High correlation; positive bias
>
|
I
S
"Z
SL
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.
' Correlation is a measure of the degree of linear relationship between two variables.
49
-------
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. 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 above the zero bias line.
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 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 low data 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 Viking 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 ±30% of reference laboratory
results or should the average results for a particular sample medium meet the 30% 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
quantitative 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 Viking absolute percent accuracy for each sampling medium is in the
range of 0-30%.
Accuracy Goal 2: Median Viking absolute percent accuracy for each sampling medium is less
than or equal to the 95th percentile of the pooled reference laboratory absolute percent
accuracies for each sampling medium.
The specific evaluation procedure is as follows:
• Pool all the Viking 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»PQLVlking.
50
-------
Compile these data into a frequency histogram and compute the median (APA 5)Vlking, 80th
percentile (APA 8)VMng, and 95th percentile (APA 95)VMng 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 (APA5)VMng < 30%: Accuracy Goal 1 Met ~ Viking accuracy performance is better
than or equal to that specified in Methods 8260 and TO-14.
(2) If Viking (APA5)Vlkmg < (APA95)Lab: Accuracy Goal 2 Met ~ Viking performs
comparably to conventional laboratory using accepted analytical methodologies.
(3) If (APA5)VMng > (APA 95)Lab Accuracy Goal 2 Not Met ~ Viking data do not compare
with the reference laboratory data.
-30 -20 -10 0 1(
Percent Difference Interval
10 20 30 40 50 60 70
Percent Difference Interval
(a) High correlation; low bias
(b) High correlation; positive bias
-70 -60 -50 -40 -30 -20 -10 0 10 20 30 40 50
Percent Difference Interval
-30 -20 -10
10 20 30 40 50 60 70
Percent Difference Interval
(c) Low correlation; low bias
(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 Viking relative percent difference for the pooled data from each
sampling medium is in the range of 0-25%.
Precision Goal 2: Median Viking 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 medium.
51
-------
The specific evaluation procedure is as follows:
• Pool all the Viking duplicate sample results, in terms of relative percent difference 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«PQLVlking.
• Compile these data into a frequency histogram and compute the median (RPD 5)Vlkmg, 80th
percentile (RPD 8)Vlkmg, and 95th percentile (RPD 95)Vlkmg 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 relative percent difference distributions:
(1) If (RPD 5)Vlkmg < 25%: Precision Goal 1 Met ~ Viking precision performance is
within the range identified in original developer claims and is very near that specified in
Methods 8260 and TO-14.
(2) If Viking (RPD5)Vlkmg < (RPD95)Lab: Precision Goal 2 Met -- Viking performs
comparably to conventional laboratory using accepted analytical methodologies.
(3) If (RPD 5)Vlkmg > (RPD95)Lab: Precision Goal 2 Not Met - Viking precision
performance is worse than that reported by reference laboratories despite considerable
allowance given for field variability.
Viking 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 less than or equal to 50%.
Comparison Goal 2: If Goal 1 is not met, the Wilcoxon test result between Viking and reference
laboratory data should indicate no statistically significant bias (p > 0.05).
The specific evaluation procedure is as follows:
• Compute the percent difference and the absolute percent difference for each set of Viking to
laboratory paired sample results.
• Compile the percent difference and absolute percent difference data into frequency
histograms and compute the median (APD 5) Vlkmg.Lab, 80th percentile (APD 8) VSimg.Lab, and 95th
percentile (APD 95) VMng-Lab values for each distribution.
• Apply the following assessment criteria to the compiled absolute percent differences:
(1) If (APD5) viking-Lab - 50%: Comparison Goal 1 Met ~ Viking results are consistently
within ±50% of reference laboratory results.
(2) If (APD 5) viking-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
52
-------
Field Operation Observations
Prior to the demonstration, a multi-point calibration was performed on the Viking GC/MS. This
calibration was checked daily with a one-point calibration using an appropriate method for each sample
media10. The single-point calibration samples were prepared from certified calibration standards that
were diluted to bring the compound concentrations into the working range of the instrument. Results
from the calibration check were required to meet pre-established acceptance criteria prior to proceeding
with field sample analysis. Blanks were run after high concentration samples to verify the absence of
sample contamination or carry-over prior to further sample analysis. At the end of analysis for each
sample media, a final instrument calibration check was performed. Sample analyses results were
prepared and submitted to the on-site Sandia team member at the end of each working day.
At SRS, the Viking instrument was transported from Viking corporate offices in Virginia and daily to
and from the demonstration site in a passenger vehicle. On-site it was powered by a portable 1.5 kW
generator. At the WAFB demonstration, the system was initially shipped in by air freight and
subsequently transported to and from the site in a passenger vehicle. At WAFB, the system was operated
from a line power source. At both sites the system was operated out of the back of a vehicle. An
analytical chemist was the only operator and handler for the instrument except for the initial loading of
the instrument into the vehicle. Environmental factors such as dust-laden air from dirt roads, cool
mornings, and hot, humid afternoons at SRS did not appear to adversely affect the performance of the
Viking as data accuracy and precision were generally consistent throughout the demonstrations. The
ambient temperature extremes can be seen in temperature plots from SRS and WAFB in Figure 6-3 and
6-4, respectively. Relative humidity ranged from 40 to 90% during the demonstrations. Ancillary
equipment used during the demonstrations included a roughing pump, generator, printer, and a small fan
to aid in cooling the instrument during the hot, humid afternoons. Equipment setup took one field
operator less than 30 minutes. No instrument mechanical or operating problems were observed at either
demonstration site.
The analyst demonstrated a high level of competence in field analyses and instrument calibration.
Quality control activities such as the daily calibration check, blank analysis after high concentration
samples, and calibration checks at the end of each sample media all contributed to the production of high
quality data from this particular instrument. During the analysis, a few analytical problems were
encountered. Three high concentration samples exceeded the calibration range and the analyst flagged
the data accordingly. The instrument was exposed to a large injection mass of methanol and analytes
when a high-concentration spike sample in methanol was mistakenly analyzed without prior dilution.
The large quantity of analytes introduced into the instrument overloaded the column and instrument
detector. The analyst spent the remainder of the day and evening cleaning the instrument in order to
bring instrument contamination down to tolerable levels. The analytical results for subsequent samples
were unaffected by the contamination cleanup procedure
10 One exception to this statement is that a three-point soil gas calibration was performed on-site at the SRS
demonstration.
53
-------
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
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 6-4. Plot of daily temperatures during the WAFB demonstration.
Viking Accuracy and decision Results
Viking Accuracy ~ SRS Demonstration
As discussed in Sections 4 and 5, SRS is contaminated with the chlorinated solvents TCE and PCE.
Consequently, all of the Viking GC/MS data and corresponding laboratory data are limited to these two
contaminants. Recovery data for the Viking 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 Viking PQL are not shown
54
-------
in the table. Because of significant deviations from the demonstration plan, discussed earlier in Section
4, no data are available for assessment of Viking instrument accuracy on soil samples at SRS.
Table 6-1. Viking GC/MS Recoveries at SRS.
Sample Media / Description
Water Medium Spike
Water High Spike
Water PE No. 2
Soil Gas Medium Spike
Soil Gas High Spike
Soil Gas PE No. 2
Percent Recovery
TCE
145
252
55
148
72
219
PCE
56
138
104
133
75
202
Recoveries for the water spike and PE samples are generally satisfactory with five of six data points in
the 50-150% range. 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 tabulated recoveries at the 100 ppm (medium
spike) concentration level are greater than the reference values by factors ranging from about 1.3 to 2,
revealing generally unsatisfactory accuracy performance of the Viking GC/MS for soil gas samples at
this medium soil gas concentration level. Recoveries at the 1,000 ppm level are in the acceptable range.
Viking 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. A subset of analytes was chosen for
accuracy and precision evaluation of the Viking GC/MS. 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 demonstration from Viking
GC/MS analysis of PE and spike samples are presented for soil, water, and soil gas samples in Table 6-2.
Table 6-2. Viking Recoveries at Wurtsmith.
Sample Media 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 Medium Spike
Soil Gas High Spike
Percent Recovery
TCE
89
86
120
79
89
NP
83
110
109
PCE
86
91
89
91
107
NP
111
NP
NP
Benzene
97
96
88
89
103
104
101
128
105
Toluene
86
95
80
80
90
NP
119
NP
NP
Total
Xylenes
80
86
83
85
96
106
129
91
131
Notes: NP = Not present in spike.
See Tables 5-7 and 5-11 for approximate target analyte concentrations in the various PE/Spike samples.
55
-------
Two soil PE samples were available for accuracy determinations. These samples were laboratory-
prepared in sealed ampules and were judged to be reliable standards. Viking GC/MS recoveries for the
five target compounds fall in the range of 80 to 97% and reveal good accuracy performance for both
chlorinated and non-chlorinated hydrocarbon target analytes. Computed recoveries from spike and PE
water samples, also shown in Table 6-2, fall in the range of 79 to 129%, indicating that the Viking
GC/MS performed acceptably with water sample media as well. Soil gas sample recoveries shown in
Table 6-2 fall within the range of about 90 to 130% and indicate good Viking GC/MS accuracy
performance with gas phase media.
Overall Viking Accuracy Performance
The absolute percent accuracy values from both sites were compiled for the five 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 fall in the 0-20% range. The same is true for water, with 5
values in excess of 20%. The soil gas accuracy data generally fall in the 0-60% range, with 2 values
falling outside of this range.
Viking Soil Accuracy
3
•s
i2
10
15
20
25
30
Absolute Percent Accuracy Interval
Figure 6-5. Absolute percent accuracy histogram for
Viking soil samples.
Viking Water Accuracy
22
20
18
16
14
12
10
8
6
4
2
0
180 200
0 20 40 60 80 100 120 140 160
Absolute Percent Accuracy Interval
Figure 6-6. Absolute percent accuracy histogram for
Viking water samples
Summary data from the absolute percent accuracy distributions are shown in Table 6-3 for the three
sampling media. For example, the soil median absolute percent accuracy was 22% for the pooled
reference laboratory data with the 95th percentile value at 47%. In comparison, the Viking soil median
absolute percent accuracy was 13% with a 95th percentile value of 17%.
56
-------
Viking Soil Gas Accuracy
3
•s
i2
0 20 40 60 80 100 120
Absolute Percent Accuracy Interval
Figure 6-7. Absolute percent accuracy histogram for
Viking soil gas samples.
Table 6-3. Viking and Reference Laboratory Accuracy Summary.
140
Data Set/Media Type
Ref. Lab - Soil
Viking - Soil
Ref. Lab - Water
Viking - Water
Ref. Lab - Soil Gas
Viking - Soil Gas
Absolute Percent Accuracy
n
10
10
30
28
24
12
X.5
22
13
19
14
20
28
X.8
36
14
27
26
47
45
X95
47
17
116
45
66
110
Note: x.5 = 50th percentile (median); x8 = 80th percentile; x95 = 95th percentile
Assessment of Viking performance in terms of the accuracy goals stated previously results in the
following determinations:
Soil: (APA5)VMng <30%: Accuracy Goal 1 Met
Water: (APA5)VMng <30%: Accuracy Goal 1 Met
Soil Gas: (APA5)VMng <30%: Accuracy Goal 1 Met
Viking accuracy performance is acceptable in terms of established performance goals for all sampling
media.
Viking Precision ~ SRS Demonstration
Viking GC/MS RPD values from the SRS demonstration are presented in Table 6-4 and are based on the
results of duplicate sample analysis. No precision determination was done on SRS soil sample data, as
discussed previously.
The RPD values for the water samples are generally less than 20%, indicating good precision for the
Viking GC/MS. As expected, the precision is not as good on the results of the low concentration samples
near the Viking PQL; however, in nearly all cases the RPD values are less than 20%.
57
-------
Table 6-4. Viking Precision for SRS Demonstration.
Sample Media / Description
Water Low
Water Medium
Water High
Soil Gas Medium
Soil Gas High
Relative Percent Difference
TCE PCE
21
1
6
1
20
6
6
2
4 2
Notes: The Water/Low sample is below the Viking PQL and RPD values are not used in statistical calculations.
Soil gas duplicate samples were actually soil 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 during
collection. 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 medium and high vapor samples,
given in Table 6-4, are in the range of 1 to 4%, indicating good precision performance.
Viking Precision ~ WAFB Demonstration
The 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. Viking Precision forWurtsmith Demonstration.
Sample Media / 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
NP
NP
NP
NP
NP
NP
NP
NP
NP
PCE
NP
NP
NP
NP
NP
<1
NP
NP
NP
Benzene
20
o
5
NP
13
3
4
6
<1
o
J
Toluene
29
10
13
NP
4
NP
NP
NP
NP
Total
Xylenes
59
13
1
8
3
1
11
2
5
Notes: NP = Not present; Analyte was either below detection or not present in the sample.
See Tables 5-9 and 5-13 for approximate target analyte concentrations in the listed samples.
Only limited data were available for soil precision evaluations since none of the samples had detectable
levels of TCE and PCE. A high RPD value of 59% was encountered for a duplicate measurement of total
xylenes in a low level sample. Reference laboratory data reveal that xylene content in this sample was
below the Viking PQL.
Calculated RPD values for three different concentration levels in water are less than 13%, revealing good
instrument precision for water analysis.
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
-------
Viking 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 RPD values for the vapor samples all fall
below 20% and indicate good instrument precision for the soil gas samples.
Overall Viking 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 the soil samples,
nearly all of the RPD values fall in the 0-30% range. For water and soil gas samples, all values fall in the
0-15% range.
Viking Soil Precision
0 10 20 30 40 50 60 70
Relative Percent Difference Interval
Figure 6-8. Relative percent difference histogram for
Viking soil samples.
Viking Water Precision
o
•5
10
12
14
16
Figure 6-9.
Relative Percent Difference Interval
Relative percent difference histogram for
Viking water samples.
59
-------
Viking Soil Gas Precision
J3
o
•8
Relative Percent Difference Interval
Figure 6-10. Relative percent difference histogram for
Viking soil gas samples.
Summary data from the relative percent difference distributions are shown in Table 6-6 for the three
sampling media.
Table 6-6. Viking and Reference Laboratory Precision Summary.
Data Set/Media Type
Ref. Lab - Soil
Viking - Soil
Ref. Lab - Water
Viking - Water
Ref. Lab - Soil Gas
Viking - Soil Gas
n
13
8
14
15
17
10
Relative Percent Difference
x^
8
13
5
6
7
3
X8
13
25
10
11
20
5
X9,
58
49
22
20
68
9
Note: x5 = 50th percentile (median); x8 = 80th percentile; x95 = 95th percentile
Viking performance with respect to precision goals are as follows:
Soil:
Water:
Soil Gas:
(RPD5)Vlking
(RPD5)Vlkmg
(RPD5)Vlking
<25%
<25%
<25%
Precision Goal 1 Met
Precision Goal 1 Met
Precision Goal 1 Met
Viking precision performance, relative to established performance goals, is judged acceptable for all
sampling media. The median relative percent difference values for all three sampling media were in the
range of 0-25%.
Viking to Reference Laboratory Data Comparison
As discussed earlier in this section, comparisons of the Viking GC/MS analytical data with the reference
laboratory analytical data for soil, water 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. Reported values that were less than two times the Viking GC/MS PQL (soil: 5
jWg/kg; water: 5 Aig/L; soil gas: 5 ppm) are plotted; however, they are not included in any of the statistical
analyses.
60
-------
Scatter Plots/Histograms ~ SRS Demonstration
Target analytes at this site were limited to TCE and PCE. The GEL water data were given a good
ranking, as discussed previously in Section 5, and 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 Viking GC/MS water analyses results for TCE and PCE are given versus the reference data in
Figure 6-11 and Figure 6-12 for low (1-100 Aig/L) and high (0.1-100 mg/L) concentration ranges,
respectively. All but 2 points fall within the ±50% interval about the zero bias line as shown on the plots.
The distribution of percent difference values is shown as a histogram in Figure 6-13. A clustering of
values on the negative side of the histogram suggests an overall negative bias in the Viking data relative
to the laboratory data.
A plot of the Viking soil gas data for TCE and PCE relative to the SRS reference laboratory data is
shown in Figure 6-14. Eight of the 12 plotted values fall outside the ±50% bias lines. The accompanying
histogram is shown in Figure 6-15.
Scatter Plots/Histograms ~ WAFB Demonstration
As outlined in Section 5, the Traverse laboratory data were given a good ranking for soil and water
sample analyses. These data are used as reference data for comparison with Viking GC/MS results. The
Pace laboratory data were given a satisfactory ranking for soil gas analysis and are used as the reference
data set for the soil gas measurements.
Considerably more compounds were detected in the various samples at the Wurtsmith AFB than at the
Savannah River Site, which was limited to TCE and PCE. For the Wurtsmith demonstration, all target
analytes have been combined in the plots and statistical analysis.
A plot of Viking data versus Traverse data for all soil analyses is presented in Figure 6-16. Nearly all the
data points fall within the ±50% margin lines shown on the plot. The percent difference histogram,
shown in Figure 6-17, shows most values in the -60 to 100% range.
Plots of Viking data versus Traverse data for all water analyses are presented in Figures 6-18 and 6-19
for the low (yWg/L) and high (mg/L) ranges, respectively. Nearly all Viking values fall within the ±50%
margins shown on the plots. The percent difference histogram (Figure 6-20) shows a clustering of nearly
all the values in the -50 to 50% range.
A scatter plot of Viking GC/MS data versus Traverse data for all soil gas analyses is presented in Figure
6-21. The high spike sample data which were in excess of 100 ppm are not included since the analysis of
Pace data in Section 5 showed poor results for these high concentration samples. Ten of a total of 27
plotted values fall outside the ±50% margins shown on the plot. Nearly all values outside the ±50%
margins fall on the high side. This is also indicated in the histogram shown in Figure 6-22, revealing an
overall positive bias in the Viking data relative to the reference laboratory data.
61
-------
Viking vs. Laboratory - SRS Water (Low)
Laboratory Value, |ig/L
Figure 6-11. Viking vs. Laboratory data for SRS low concentration water
samples. The solid lines show the ±50% range about a zero-bias
(dashed) line.
Viking vs. Laboratory - SRS Water (High)
100
I
a
I
> 1
100
Laboratory Value, mg/L
Figure 6-12. Viking vs. Laboratory data for SRS high concentration water samples.
The solid lines show the ±50% range about a zero-bias (dashed) line.
62
-------
Viking vs. Laboratory - SRS Water
s 3
•5
o
2 2
-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.
Viking vs. Laboratory - SRS Soil Gas
I
?
£
Laboratory Value, ppm
Figure 6-14. Viking vs. Laboratory data for SRS soil gas samples.
Viking vs. Laboratory - SRS Soil Gas
V)
Si
o
o 2
o
-60 -40 -20 0 20 40 60 80 100 120 140 160 180 200
Percent Difference Interval
Figure 6-15. Percent difference histogram for SRS soil
gas samples.
63
-------
Viking vs. Laboratory - WAFB Soil
1000
100
10
1 10 100
Laboratory Value, mg/kg
Figure 6-16. Viking vs. Laboratory data for WAFB soil samples.
Viking vs. Laboratory -WAFB Soil
.a
o
-100-75-50-25 0 25 50 75 100125150175200225250275300325350
Percent Difference Interval
Figure 6-17. Relative percent difference histogram for
WAFB soil samples.
64
-------
Viking vs. Laboratory - WAFB Water (Low)
1000
100
10
1 10 100 1000
Laboratory Value, ug/L
Figure 6-18. Viking vs. Laboratory data for WAFB low concentration water samples.
Viking vs. Laboratory - WAFB Water (High)
1000
§
I
1 10 100
Laboratory Value, mg/L
Figure 6-19. Viking vs. Laboratory data for WAFB high concentration water
samples.
65
-------
Viking vs. Laboratory - WAFB Water
14
12
« 10
o
•5 8
o
= 6
4
2
a al
-80 -60 -40 -20 0 20 40 60 80 100 120 140
Percent Difference Interval
Figure 6-20. Relative percent difference histogram for
WAFB water samples.
Viking vs. Laboratory - WAFB Soil Gas
ing Value,
1 10 100
Laboratory Value, ppm
Figure 6-21. Viking vs. Laboratory data for WAFB soil gas samples.
Viking vs. Laboratory - WAFB Soil Gas
8
7
V)
•° e
o 6
o 5
o
z 4
Z
2
1
0 50 100 150 200 250 300 350 400 450 500 550
Percent Difference Interval
Figure 6-22. Relative percent difference histogram for
WAFB soil gas samples.
66
-------
Overall Viking 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, most of the
absolute percent difference values fall in the 0-60% range. For water, most of the absolute percent
difference values fall in the 0-50% range. In the case of soil gas, nearly all of the absolute percent
difference values fall in the 0-200% range.
Viking vs. Laboratory - Soil, All Sites
O
04
3
2
1
0 20 40 60 80100120140160180200220240260280300320340360
Absolute Percent Difference Interval
Figure 6-23. Absolute percent difference histogram for
soil samples.
Viking vs. Laboratory - Water, All Sites
24
20
16
o 16
"5
o 12
n_
0 10 20 30 40 50 60 70 80 90 100
Absolute Percent Difference Interval
Figure 6-24. Absolute percent difference histogram for
water samples.
67
-------
Vking vs. Laboratory - Soil Gas, All Sites
50 100 150 200 250 300 350 400 450
Absolute Percent Difference Interval
Figure 6-25. Absolute percent difference histogram for gas
samples.
Summary data from the three absolute percent difference distributions for the three sampled media are
shown in Table 6-7.
Table 6-7. Viking-Laboratory Comparison Summary.
Media Type
Soil
Water
Soil Gas
n
18
69
25
Absolute Percent Difference
x,
32
18
57
xs
61
31
133
x,,
196
60
353
Soil:
Water:
Soil Gas:
Wilcoxon
(APD5)
(APD5)
(APD5)
test result:
Viking-Lab
Viking-Lab
Viking-Lab
<50%:
<50%:
>50%:
p<0.01
Viking performance assessment in terms of the comparison goals result in the following determinations:
Comparison Goal 1 Met
Comparison Goal 1 Met
Comparison Goal 1 Not Met
Comparison Goal 2 Not Met
Viking data are judged to be comparable to reference laboratory data for soil and water; however, the
performance criteria were not met for soil gas.
Summary of Viking Accuracy, Precision, and Laboratory Comparison Performance
An overall summary of Viking GC/MS performance for both demonstration sites is given in Table 6-8.
The results of the foregoing evaluation of the Viking GC/MS performance goals are summarized with
respect to accuracy, precision, and a comparison of the Viking GC/MS results to that of the reference
laboratory.
The summary information in the table shows that the Viking GC/MS met performance goals in eight of
the nine evaluation categories. The Viking soil gas data were determined not to be comparable to the
reference laboratory data; however, Viking accuracy and precision was determined to be acceptable on
68
-------
Table 6-8. Summary Performance of the Viking GC/MS.
Sample
Medium
Soil
Water
Soil Gas
Accuracy via
PE/Spike Samples
Goal Met
Goal Met
Goal Met
Precision via
Duplicate Analyses
Goal Met
Goal Met
Goal Met
Viking-Lab Data
Comparison
Goal Met
Goal Met
Goal Not Met
Reference Lab Data
Quality
SRS: Undetermined
WAFB: Good
SRS: Good
WAFB: Good
SRS: Satisfactory
WAFB: Satisfactory
the basis of their analytical results for PE, spike, and duplicate samples. The result of the Wilcoxon test
on the soil gas data indicates that the observed difference between Viking and reference laboratory data
cannot be explained by random variation alone and it is likely that some other factor is causing the
Viking data to be high relative to the laboratory data. Possible influential factors could be calibration,
sampling handling, or injection error by either Viking or the reference laboratory.
Other Viking GC/MS Performance Indicators
Unknown 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. In addition to identification and quantification of
target analytes, the Viking analyst also compared the mass spectra of additional non-target analyte
chromatographic peaks to a computerized mass spectrum library of many compounds. An example
Viking total ion chromatogram of water sample containing numerous volatile organic compounds is
shown in Figure 6-26. A partial listing of the library search results, called tentatively identified
compounds, for this sample is shown in Table 6-9.
Field Handling and Operation
The Viking SpectraTrak™ 672 GC/MS is designed to be shipped to the field and is shock mounted for
durability in handling and field use. At approximately 140 pounds, the system was easily transported to
and around each site in a standard passenger vehicle. For the WAFB demonstration, the system was
transported by air freight to Michigan. 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 exposed to ambient temperatures ranging from 40° to 95° F.
During both demonstrations, the instrument operated properly and without breakdowns or mechanical
problems.
System set-up was simple and uncomplicated. Set-up procedures involved connecting the ancillary
equipment and checking the instrument calibration. About 30 minutes were required for set-up and
another 30 minutes for calibration checks. Following these two activities, the instrument was ready for
sample analysis. Sample analysis time varied according to the sample media and the mode of injection;
however, a typical time interval was on the order of 30 minutes per sample. Hard copy data were
available for each day's analysis at the completion of the work day.
69
-------
Abundance
4SQOOOQ-
4000000-
350DGOO-
300DCOO-
2500000-
2000000-
1500000-
1000000-
500000 -
0 -
Time ->
TIC
9
i f
S 65
2.00 4.00 6
10S
183
e.oo
10.00
Figure 6-26. Viking GC/MS total ion chromatogram from a WAFB water
sample.
Overall Viking GC/MS Performance Conclusions
Table 6-9. Tentatively Identified Compounds from a Wurtsmith Water Sample Analysis.
Methyl-cyclopentane
5-methyl-lH-tetrazole
Ethyl-cyclobutane
1 -Methylethyl-cyclopropane
2-Methyl- 1 -pentene
Cyclohexane
2-Methyl-cyclobutanone
Methyl-cyclohexane
4 5-Dihvdro-l 5-dimethvl-lH-ovrazole
2, 3-Dimethyl-2 -pentene
4,4-Dimethyl-2 -pentene
Propyl-benzene
1 -Ethyl-3 -methyl-benzene
1 -Ethyl-2 -methyl-benzene
l-Ethyl-2 -methyl-benzene
1,2,3 -Trimethyl-benzene
1 ,2,4-Trimethyl-benzene
The Viking SpectraTrak™ 672 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 to assess
instrument performance goals and to evaluate logistical and operational capabilities such as system
reliability, ruggedness and ease of operation. 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.
70
-------
As shown in Table 6-10, the instrument performance goals of the Viking GC/MS were met with one
exception. Instrument performance in terms of accuracy and precision met the stated performance goals.
The Viking data were determined to be comparable to that of reference laboratories for soil and water;
however, the data were not comparable for soil gas analyses. Questions remain as to whether the Viking
soil gas data, or that of the reference laboratory, or both are inaccurate. In either case, the extent of
deviation of Viking and reference laboratory results for soil gas samples, while significant, is not so great
as to preclude the system's effective use in many field screening applications. Performance goals related
to sample throughput rate, data completeness, ease of field calibration, and data availability were met.
Table 6-10. Summary of Viking Performance Goals and Actual Performance.
Performance Goal
Performance
Goal Met?
ACCURACY:
Median absolute percent accuracy within 30% of reference
value; or,
Median accuracy comparable to reference laboratory
accuracy.
For all media, generally
better than 30%
accuracy; always
comparable to
reference laboratories.
Water: Yes
Soil: Yes
Soil Gas: Yes
PRECISION:
Median relative percent difference within 25%; or,
Median relative percent difference comparable to laboratory
precision.
For all media, generally
within 25% RPD;
always comparable to
reference laboratories.
Water: Yes
Soil: Yes
Soil Gas: Yes
REFERENCE LABORATORY DATA COMPARISON:
Median absolute percent difference less than or equal to
50%; or,
if greater than 50%, no significant bias via statistical test.
Less than 50% for soil
and water; not
comparable for soil
Water: Yes
Soil: Yes
Soil Gas: No
COMPLETENESS:
At least 95% of target compounds detected by reference
laboratory also detected by Viking.
Soil > 99%
Water > 99%
Soil Gas > 99%
YES
SAMPLE THROUGHPUT: Depends on methodology.
Samples requiring concentration may run for 30 minutes
apiece. Direct injections average 10-15 min. apiece. Direct
membrane analyses can be performed several times per
minute.
Sample analysis times
were less than or equal
to those claimed.
YES
CALIBRATION: The instrument will be initially
calibrated before arriving at the demonstration site and
simple calibration checks performed daily.
As stated in goal.
YES
DATA: Quantitative results will be submitted at the end of
the demonstration. The data analysis software is also
capable of full, detailed reports which include ion
chromatograms, spectra, library searches, and calibration
reports. Detailed data can be submitted when time permits.
Data were submitted on
the same day of the
analysis, all data
submitted by demo
end.
YES
DEPLOYMENT: The Viking SpectraTrak™ can be set up
and ready to run within 30 minutes.
The system was ready
to run samples in <30
min.
YES
71
-------
Section 7
Applications Assessment
The Viking SpectraTrak™ 672 field portable GC/MS instrument has been 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, on-site and in near real time, samples taken from three
media; soil gas, water, and soil. The use of field analytical instruments is emerging as a supplement to and
possible replacement for conventional laboratory off-site analysis. As demonstrated, the Viking field
portable GC/MS system has application to several field screening and analysis scenarios
Applicability to Field Operations
From a logistical viewpoint, at a weight of 145 Ib, the system is easily transportable and is ruggedized for
field handling. The Viking system is self-contained and includes an internal computer and internal carrier
gas source. These features, along with its compact design, preclude the need for a specialized vehicle
during field operations. The system can easily be operated from a rental car or van. Required external
ancillary equipment includes a roughing pump, portable printer, and power source. The system can be
operated reliably in the field over a wide range of temperature and relative humidity. System setup and
operation can be conducted by one person. The system utilizes data handling and analysis software
common to that of basic GC/MS technologies.
Capital and Field Operation Costs
On-site field analysis of samples has the potential to reduce overall site characterization and clean-up 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 Viking SpectraTrak™ 672 were determined during the
demonstration and are presented in Table 7-1. Estimates of average sample analysis rates for laboratory
quality and sample screening analysis modes for the Viking instrument are also provided. Actual sample
analysis rates will vary as a function of the media and 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 type of field
analysis system for various applications.
Discussion of the Technology
Rapid Analysis
The use of the Viking 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.
72
-------
Table 7-1. Viking Spectralrak™ 672 GC/MS Capital and Field Operation Costs.
Capital Costs
Viking SpectraTrak™ 672 GC/MS (90 day warranty)
Ancillary Equipment - Roughing pump, generator
Five-year service contract
Training (2 people for one week)
Total
Annualized system costs for five years ($215K + 5 years)
Maintenance (annually 10% of capital costs)
Includes new columns, preventive maintenance, software and hardware
upgrades, etc.
Five-year annual capital and maintenance costs
Field Operation Costs
Field Chemist labor ($60K/year including overhead)
Per diem (lodging and meals)
Vehicle rental
Supplies and consumables (standards, syringes, vials, gas, etc.)
Total
Sample Analysis Rates
Laboratory Quality Sample Analysis
Field Screening Sample Analysis
$43K/year
$ 1.5K/vear
$44.5K/year
$230/day
$100/day
$ 50/day
$ 70/dav
$450/day
15 samples/day
50 samples/day
Sampling and Sample Cost
The major cost saving obtained by the use of field analyses is the reduction in time required to obtain the
analytical results needed for decision making. As operated during this demonstration, the Viking
SpectraTrak™ 672, in most cases, produced good quality data in the field. The ability to generate good
quality analyses 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.
Performance Advantages
The developer's performance specifications for the Viking field transportable GC/MS were evaluated and,
when compared to standard laboratory methods, were statistically equivalent. This demonstration showed
that for samples whose contaminant concentration determinations are very sensitive to shipping and
analysis time delays, such as VOC soil gas samples, that on-site analysis directly after the samples are
obtained consistently provides higher results than the laboratory methods. Therefore, for some analytes like
VOCs, on-site analysis can be expected to provide results that are more representative of actual site
conditions and are more accurate than laboratory methods.
73
-------
Transportability
The ruggedness of the Viking GC/MS was illustrated by its shipment via air freight to the WAFB site and
by its daily transportation in a vehicle to and from the demonstration sites. The system is essentially self
contained and does not need a dedicated support vehicle or excessive ancillary equipment. At about 145 Ib
it can be easily transported around sites and operated out of the back of a rental car or van. This ruggedness
and portability provides the system with the capability for laboratory quality sample analysis at most
locations.
Field Screening of Samples
The system capabilities for transportability and real-time analysis make the instrument extremely useful as
a tool for site characterization and monitoring activities. The instrument may be used as a high volume
screening tool to guide sampling and remediation efforts or to provide higher quality analyses on selected
samples. 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 analyzed and will reduce the need for return trips to the field.
Interferences
Interferences from water vapor or other contaminants can affect the field transportable GC/MS systems as
well as the fixed-laboratory GC/MS systems. The presence of interferents should checked by periodic
analysis of reagent blanks and is required with both laboratory and field GC/MS systems. Since reagent
blank analysis requires about 30 minutes, the sampling speed is not significantly affected.
Conclusions
The Viking SpectraTrak™ 672 Field Transportable GC/MS provides 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 complements conventional laboratory analysis and
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 technologies.
74
-------
Section 9
Previous Deployments
Stringfellow Toxic Waste Dump Site, Southern California (July 1993). Volatile organic compounds in soil
performed by Viking for US EPA Region 9. Contact Vance Fong, Section Chief, Phone: (415) 744-1492.
Antioch, California (July 1993): Volatiles in groundwater, performed for Region 9 of the EPA at a
CERCLA site near the Sacramento River in Northern California. Contact Vance Fong, Section Chief,
Phone: (415) 744-1492.
Air and Soil Screening, EPA Region 5 (September 1993). Air screening analysis for health and safety at a
dumpsite near the Great Lakes region performed for EPA Region 5. Contact Rod Turpin, Phone:(908)
321-6762.
Superfund Site, Virginia, (August 1993). Analysis of volatiles and semivolatiles performed by the
Environmental Resources Management, Inc. with a Viking SpectraTrak™. Contact David Gallis,
Environmental Resources Management, Inc., Phone: (610) 524-3786.
Sandia National Laboratories, (1993). Soil Vapor Analysis of soil vapor samples extracted from various
depths using a GeoProbe. Contacts are Sharissa Young, Phone: (505) 845-3226 or Wyatt Booher, Phone:
(505) 269-3207.
SpectraTrak™ Transportable GC/MS U.S. Military Endurance Testing (1993) at Dugway Proving Ground.
Contact R J. Black, Phone: (801) 831-3371.
79
-------
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.
EPASW-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/600/4-89/018.
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.
80
-------
Appendix A
Analytical Method for the Operation of the SpectraTrak™ 672
Prepared by
Viking Instruments Corporation
1.0 Scope and Application
The following table lists targeted analytes and detection levels used for this evaluation. The analytes are by
no means limited to those listed here.
ANALYTE
chloromethane
vinyl chloride
chloroethane
bromomethane
acetone
1,1-dichloroethene
methylene chloride
carbon disulfide
trans-l,2-dichloroethene
1,1-dichloroethane
2-butanone
chloroform
1 ,2-dichloroethane
1,1,1 -trichloroethane
carbon tetrachloride
benzene
1 ,2-dichloropropane
DETECTION
LIMIT *
10
10
10
10
100
5
5
100
5
5
100
5
5
5
5
5
5
ANALYTE
trichloroethene
bromodichloromethane
cis- 1 ,3 -dichloropropene
trans- 1 ,3 -dichloropropene
1 , 1 ,2-trichloroethane
toluene
dibromochloromethane
tetrachloroethene
chlorobenzene
ethylbenzene
m+p-xylenes
styrene
1 , 1 ,2,2-tetrachloroethane
o-xylene
1 , 3 -dichlorobenzene
1 ,4-dichlorobenzene
1 ,2-dichlorobenzene
DETECTION
LIMIT *
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
5
* Estimated detection limits. Units depend on sample preparation:
Mg/kg
Mg/L
ppm
ppb
for 5 gram soil samples concentrated by purge and trap.
for 5 mL water samples concentrated by purge and trap.
for 250 /jL of air by direct injection.
for 0.25 to 1.0 L of air concentrated onto sorbent tubes.
Note: Lower limits (ppt and lower) can be achieved with larger sample volumes or by using selected ion
monitoring.
2.0 Summary of Method
The analytical instrumentation used for the demonstration is a Viking SpectraTrak™ 672 transportable
GC/MS system with data analysis performed by the Windows-based Hewlett-Packard Chem Station
software at SRS, and EnviroQuant at WAFB. A minimum of three concentration levels were analyzed and
used for establishing a calibration curve. An average response factor method was used for quantitation of
A-l
-------
samples. The instrument was calibrated for the known historical concentrations and contaminants at each
site: trichloroethene and tetrachloroethene at SRS, and a portion of the EPA method 8260 compound list at
WAFB (see 7.2.1 for list).
2.1 Gaseous samples
Gaseous samples were analyzed by collecting 250 //L of sample from Tedlar bags and directly introducing
the sample into the instrument by split/splitless injection. An external standard (ESTD) method of
calibration and quantitation was utilized. The instrument was calibrated at SRS for a range of 1 to 5000
ppm(v) while at WAFB a 1 to 25 ppm(v) calibration range was used. (Note: Air samples may also be
concentrated on a sorbent trap using the unit's sampling pump, then thermally desorbed for analysis. This
concentration step was not necessary at either site.)
2.2 Liquid samples
Liquid samples were analyzed by concentrating a 5 mL volume by purge and trap, then thermally
desorbing for GC/MS analysis. An internal standard (ISTD) method of calibration and quantitation was
utilized with 50 //g/L of internal standard added to each sample. The instrument was calibrated at SRS for
a range of 10 to 1000 //g/L while at WAFB a 10 to 200 //g/L calibration range was used.
2.3 Soil samples
Soil samples were analyzed at SRS by extracting a portion of soil with methanol, then injecting 2 //L of the
methanol extract into the instrument. A selected ion was monitored for each compound (m/z 130 for
trichloroethene and m/z 166 for tetrachloroethene). ESTD quantitation was applied to the extracted soil
samples within a calibration range of 50 to 1000 //g/kg.
The soils at WAFB were analyzed by purge and trap / thermal desorption GC/MS. ISTD quantitation was
used for the soils with 50 //g/kg of internal standards added to each sample. The calibration ranged from
10to200//g/kg.
3.0 Definitions
ESTD
g
GC/MS
ISTD
kg
L
mL
PCE
external standard
gram
gas chromatograph / mass spectrometer
internal standard
kilogram
Liter
milliliter
tetrachloroethene (perchloroethylene)
ppb(v)
ppm(v)
ppt
SRS
TCE
Mg
//L
WAFB
parts per billion by volume
parts per million by volume
parts per trillion
Savannah River Site
trichloroethene
microgram
microliter
Wurtsmith Air Force Base
4.0 Interferences
Interferences 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.
5.0 Safety
A-2
-------
No unusual safety practices above and beyond good laboratory practices. Knowledge of specific toxicity of
chemicals utilized and how to handle chemicals is essential.
6.0 Equipment and Supplies
The analytical GC/MS system (Viking SpectraTrak™ 672), comprised of a Hewlett Packard 5972
quadrupole mass spectrometer, a temperature-programmable mini-GC, Windows-based computer system,
and various inlets is further described in section 3 of the demonstration plan.
The GC column installed and used in this study is a 30 meter DB-VRX column supplied by J&W Scientific.
The column was remounted on a smaller column cage supplied by Viking to fit inside of the mini-GC.
Viking's purge and trap assembly is utilized for all purge and trap operations in this study. It consists of a
three-way valve, needle sparger, disposable test tube, swagelock fittings; and mounts to fit the front of the
SpectraTrak™. The unit's software and plumbing system take care of the rest of the purge and trap
operations.
Certified concentration standard mixtures, neat compounds, and solvents were purchased from vendors
such as Supelco, Aldrich, Restek, and Ultra Scientific.
7.0 Quality Control
7.1 Reagents and Standards - Savannah River Site
7.1.1 Gaseous standards of trichloroethylene and tetrachloroethylene were initially prepared from neat
standards purchased from Aldrich. A primary dilution was prepared by injecting 18 //L of trichloro-
ethylene and 21 //L of tetrachloroethylene into a 1.0 L Tedlar bag at room temperature and pressure. The
bag was filled to 1.0 L with ultrapure air to establish a final concentration of 5000 ppm(v) of each analyte.
A 1000 ppm(v) standard was also prepared from neat standards by injecting 3.6 //L of TCE and 4.2 //L of
PCE into a 1.0 L Tedlar bag. The following is an example of calculations used:
Concentration in ppm(v) = 22.4 x 106 (17273 K) (760/P) (TCE density) (volume TCE)
of trichloroethylene (TCE) (dilution volume) (molecular weight of TCE)
where T = Absolute temperature in degrees K
and P = Pressure in mmHg
„„„„ , . 22.4 x 106 (298/273) (760/760) (1.465 g/mL) (X volume of TCE)
5000ppm(v)= (1.0 L) (131)
X = 18 //L of TCE injected into 1.0 L air.
The bags were left to equilibrate for approximately 10 minutes before use. Further dilutions were prepared
from the 5000 ppm(v) standard by taking aliquots from the 5000 ppm(v) Tedlar bag and injecting them
into secondary Tedlar bags. These standards ranged from 500 ppm(v) to 1.0 ppm(v). For example:
5000 ppm(v) x 5.0 mL aliquot volume /1.0 L dilution volume = 25 ppm(v) final cone.
7.1.2 Liquid standards were prepared from liquid stock solutions purchased from Ultra Scientific and
Supelco. A 1000 //g/mL solution of volatile compounds was diluted to 50 //g/mL by a 1:20 dilution in
methanol. The 50 //g/mL solution was used for the purge and trap calibration in the range of 10 //g/L to
1000 //g/L. A 50 //g/mL solution of internal standards and added to each sample and standard for a final
A-3
-------
concentration of 50 //g/L. The internal standard 1,4-difluorobenzene was used for quantitating the
trichloroethene and tetrachloroethene. The same solution of volatiles at 50 //g/mL was diluted to
concentrations of 50 //g/L to 1000 //g/L in methanol for direct injection / calibration for soil samples.
7.2 Reagents and Standards - Wurtsmith Air Force Base
7.2.1 Gaseous standards were prepared from liquid solutions directly injected into Tedlar bags. A 1000
Atg/mL solution of volatile compounds was used as the stock solution for each calibration standard
prepared. A 5000 //g/mL solution of each isomer of dichlorobenzene was also used. Stock solutions were
purchased from Ultra Scientific and Supelco.
Standard Compounds in Gas Mixtures
Compound Name
1,1-dichloroethene
trans, 1-2-dichloroethylene
1,1-dichloroethane
cis- 1 ,2-dichloroethylene
benzene
trichloroethene
toluene
tetrachloroethene
chlorobenzene
ethylbenzene
m+p-xylenes
o-xylene
1,3 -dichlorobenzene (m-)
1,4-dichlorobenzene (p-)
1,2-dichlorobenzene (o-)
MW
96
96
98
96
78
130
92
164
112
106
106
106
146
146
146
Volume injected
inl.OL(mL)
0.01
0.01
0.01
0.01
0.01
0.01
0.01
0.01
0.01
0.01
0.01
0.01
0.001
0.001
0.001
Stock Std. cone.
Cug/mL)
1000
1000
1000
1000
1000
1000
1000
1000
1000
1000
1000
1000
5000
5000
5000
mg/cubic Meter
(or//g/L)
10
10
10
10
10
10
10
10
10
10
10
10
5
5
5
PPM(v)
2.55
2.55
2.50
2.55
3.14
1.88
2.66
1.49
2.18
2.31
4.62
2.31
0.84
0.84
0.84
ppm(v) calculation: ppm(v) = 24.46 (mg/cubic M) / Molecular Weight
Standards of approximately 10 ppm(v) and 25 ppm(v) were also prepared in similar fashion.
7.2.2 Liquid standards were prepared from the same stock solutions as used for the gaseous standards.
A 50 //g/mL working solution was prepared from the 1000 //g/mL stock standard by a 1:20 dilution in
methanol. Volumes of 1 //L to 20 //L of the working solution were injected into 5.0 mL of water for
calibration standards in the range of 10 //g/L to 200 //g/L.
Note: The calibration for liquid samples was also used for soil samples since the same analytical method
was performed on both matrices.
8.0 Sample Collection, Preservation, and Storage
See demonstration plan.
A-4
------- |