United States Office of Research and EPA/600/R-98/112
Environmental Protection Development August 1998
Agency Washington, D.C. 20460
vvEPA Environmental Technology
Verification Report
Immunoassay Kit
Strategic Diagnostics Inc.
D TECH PCB Test Kit
^H^M^HHHHItl JtaHHIH^Hi^MlP Jj^H^, J^^^^^lr ^^^^^^^^^b i^BBjjjjjH^Hj^^ft ^I^Bk Jl^^^^^lr
• v I ^^^ I? I ^^r "CHT1
-------
EPA/600/R-98/112
August 1998
Environmental Technology
Verification Report
Immunoassay Kit
Strategic Diagnostics Inc.
D TECH PCB Test Kit
By
Amy B. Dindal
Charles K. Bayne, Ph.D.
Roger A. Jenkins, Ph.D.
Oak Ridge National Laboratory
Oak Ridge Tennessee 37831-6120
Stephen Billets, Ph.D.
Eric N. Koglin
U.S. Environmental Protection Agency
Environmental Sciences Division
National Exposure Research Laboratory
Las Vegas, Nevada 89193-3478
This demonstration was conducted in cooperation with
U.S. Department of Energy
David Bottrell, Project Officer
Cloverleaf Building, 19901 Germantown Road
Germantown, Maryland 20874
Superfund Innovative Technology
Evaluation Program
on\l
-------
Notice
The U.S. Environmental Protection Agency (EPA), through its Office of Research and Development
(ORD), and the U.S. Department of Energy's Environmental Management (EM) Program, funded and
managed, through Interagency Agreement No. DW89937854 with Oak Ridge National Laboratory, the
verification effort described herein. This report has been peer and administratively reviewed and has been
approved for publication as an EPA document. Mention of trade names or commercial products does not
constitute endorsement or recommendation for use of a specific product.
-------
I
t»
UNITED STATES ENVIRONMENTAL PROTECTION AGENCY
Office of Research and Development
Washington, D.C. 20460
ENVIRONMENTAL TECHNOLOGY VERIFICATION PROGRAM
VERIFICATION STATEMENT
TECHNOLOGY TYPE:
APPLICATION:
TECHNOLOGY NAME:
COMPANY:
ADDRESS:
PHONE:
POLYCHLORINATED BIPHENYL (PCB) FIELD ANALYTICAL
TECHNIQUES
MEASUREMENT OF PCBs IN SOILS AND SOLVENT EXTRACTS
D TECH PCB TEST KIT
STRATEGIC DIAGNOSTICS INC.
Ill PENCADER DRIVE
NEWARK, DE 19702-3322
(302) 456-6789
The U.S. Environmental Protection Agency (EPA) has created a program to facilitate the deployment of innovative
technologies through performance verification and information dissemination. The goal of the Environmental Technology
Verification (ETV) Program is to further environmental protection by substantially accelerating the acceptance and use
of improved and more cost-effective technologies. The ETV Program is intended to assist and inform those involved in
the design, distribution, permitting, and purchase of environmental technologies. This document summarizes the results
of a demonstration of the Strategic Diagnostics Inc. (SDI) D TECH PCB test kit.
PROGRAM OPERATION
EPA, in partnership with recognized testing organizations, objectively and systematically evaluates the performance of
innovative technologies. Together, with the full participation of the technology developer, they develop plans, conduct
tests, collect and analyze data, and report findings. The evaluations are conducted according to a rigorous demonstration
plan and established protocols for quality assurance. EPA's National Exposure Research Laboratory, which conducts
demonstrations of field characterization and monitoring technologies, with the support of the U.S. Department of
Energy's (DOE's) Environmental Management (EM) program, selected Oak Ridge National Laboratory as the testing
organization for the performance verification of polychlorinated biphenyl (PCB) field analytical techniques.
DEMONSTRATION DESCRIPTION
In July 1997, the performance of six PCB field analytical techniques was determined under field conditions. Each
technology was independently evaluated by comparing field analysis results to those obtained using approved reference
methods. Performance evaluation (PE) samples were also used to assess independently the accuracy and comparability
of each technology.
The demonstration was designed to detect and measure PCBs in soil and solvent extracts. The demonstration was
conducted at the Oak Ridge National Laboratory (ORNL) in Oak Ridge, Tennessee, from July 22 through July 29. The
study was conducted under two environmental conditions. The first site was outdoors, with naturally fluctuating
temperatures and relative humidity conditions. The second site was inside a controlled environmental chamber, with
generally cooler temperatures and lower relative humidities. Multiple soil types, collected from sites in Ohio, Kentucky,
and Tennessee, were analyzed in this study. Solutions of PCBs were also analyzed to simulate extracted surface wipe
EPA-VS-SCM-15
The accompanying notice is an integral part of this verification statement
iii
August 1998
-------
samples. The results of the soil and extract analyses conducted under field conditions by the technology were compared
with results from analyses of homogeneous replicate samples conducted by conventional EPA SW-846 methodology in
an approved reference laboratory. Details of the demonstration, including a data summary and discussion of results, may
be found in the report entitled Environmental Technology Verification Report: Immunoassay Kit, Strategic Diagnostics,
Inc., D TECHPCB Test Kit, EPA/600/R-98/112.
TECHNOLOGY DESCRIPTION
The D TECH PCB test kit is designed to provide quick, semi-quantitative, and reliable test results for making
environmental decisions. The D TECH kit utilizes immunoassay technology to detect trace amounts of PCBs in
environmental samples. This test specifically detects Aroclors 1254, 1260, and 1262 equally; reacts well with Aroclors
1242, 1248, and 1268; reacts moderately with Aroclors 1232 and 1016; and shows little reactivity to Aroclor 1221. The
test is calibrated for Aroclor 1254 and has conversions for Aroclors 1242 and 1248. The D TECH PCB test kit uses latex
particles as the solid support component of the assay. With this immunoassay system, antibodies are immobilized on the
thousands of latex particles that are free to interact and react with the sample solution. To initiate the test, a sample
solution (soil extract) is added to a dry mixture of antibody-coated latex particles and enzyme conjugate. A competitive
reaction for binding sites on the antibodies then occurs between the analyte PCBs (in the sample solution) and the added
enzyme conjugate. When the reaction is complete, the particles are collected on the membrane surface of a collection cup
and briefly washed. The test is completed by adding a color developing solution to the surface of the collection cup. As
with the other immunoassay systems, the enzyme conjugate produces a color change reaction which can then be detected
and measured. The darker the color, the less analyte PCB is present in the sample. Measurement of the test results can
be completed instrumentally with a reflectance meter or visually with the included color card.
VERIFICATION OF PERFORMANCE
The following performance characteristics of the D TECH PCB test kit were observed:
Throughput: Throughput was 11 samples/hour in the chamber and 15 samples/hour outdoors. This rate included sample
preparation and analysis.
Ease of Use: Three operators analyzed samples during the demonstration, but the technology can be run by a single
operator. Minimal training (2 to 4 hours) is required to operate the D TECH kit, provided the user has a fundamental
understanding of basic chemical and field analytical techniques.
Completeness: The D TECH kit generated results for all 232 PCB samples, for a completeness of 100%.
Blank results: PCBs were detected above the lowest reporting interval for five of the eight blank soil samples. Therefore,
the percentage of false positive results was 62%. Two false positive results (25%) were reported for the extract samples.
The D TECH kit reported 0.5% false negative results for soils and 0% for extracts.
Precision: The overall precision, based on the percentage of combined sample sets where all four replicates were reported
as the same interval, was 44% for the PE soils, 21% for the environmental soils, and 25% for the extracts.
Accuracy: Accuracy was assessed using PE soil and extract samples. Accuracy, defined as the percentage of D TECH
results that agreed with the accepted concentrations, was 56% for PE soils and 42% for extracts. The percentage of
samples that was biased high was greater (28%) than the percentage that was biased low (17%) for the PE soil samples,
while the percentage that was biased high or low was comparable (29%) for the extract samples.
Comparability: Comparability, like accuracy, was defined as the percentage of samples that agreed with, was above (i.e.,
biased high), or was below (i.e., biased low) the reference laboratory results. The percentage of samples that agreed with
EPA-VS-SCM-15 The accompanying notice is an integral part of this verification statement August 1998
iv
-------
the reference laboratory results was 53% for all soils (PE and environmental) and 42% for extracts. The percentage of
samples that was biased high was greater (28%) than the percentage that was biased low (20%) for the soil samples,
while the percentage that was biased high or low was again comparable (29%) for the extract samples.
Regulatory decision-making: One objective of this demonstration was to assess the technology's ability to perform at
regulatory decision-making levels for PCBs, specifically 50 ppm for soils and 100 (jg/100cm2 for surface wipes. For PE
and environmental soil samples in the range of 40 to 60 ppm, 61% of the D TECH results agreed with the reference
laboratory. By contrast, 7% were biased high, while 32% were biased low. For the extract samples representing surface
wipe sample concentrations of 100 (jg/100 cm2 and 1000 (jg/100 cm2 (assuming a 100 cm2 wipe sample), 42% of the
D TECH results agreed with the extract spike concentration. In comparison, the percentage of extract samples biased
high or low was comparable (29%).
Data quality levels: The performance of the D TECH PCB test kit was characterized as biased, with approximately 50%
of the D TECH results agreeing with the accepted values (in terms of accuracy), and imprecise, with consistently less
than 50% replicate sample results reported as the same interval.
The results of the demonstration show that the D TECH PCB test kit can provide useful, cost-effective data for
environmental problem-solving and decision-making. Undoubtedly, it will be employed in a variety of applications,
ranging from serving as a complement to data generated in a fixed analytical laboratory to generating data that will stand
alone in the decision-making process. As with any technology selection, the user must determine if this technology is
appropriate for the application and the project data quality objectives. For more information on this and other verified
technologies, visit the ETV web site at http://www.epa.gov/etv.
Gary J. Foley, Ph.D.
Director
National Exposure Research Laboratory
Office of Research and Development
NOTICE: EPA verifications are based on an evaluation of technology performance under specific, predetermined criteria and the
appropriate quality assurance procedures. EPA makes no expressed or implied warranties as to the performance of the technology and
does not certify that a technology will always, under circumstances other than those tested, operate at the levels verified. The end user
is solely responsible for complying with any and all applicable Federal, State, and Local requirements.
EPA-VS-SCM-15 The accompanying notice is an integral part of this verification statement August 1998
V
-------
VI
-------
Foreword
The U.S. Environmental Protection Agency (EPA) is charged by Congress with protecting the nation's natural
resources. The National Exposure Research Laboratory (NERL) is EPA's center for the investigation of
technical and management approaches for identifying and quantifying risks to human health and the
environment. NERL's research goals are to (1) develop and evaluate technologies for the characterization and
monitoring of air, soil, and water; (2) support regulatory and policy decisions; and (3) provide the science
support needed to ensure effective implementation of environmental regulations and strategies.
EPA created the Environmental Technology Verification (ETV) Program to facilitate the deployment of
innovative technologies through performance verification and information dissemination. The goal of the ETV
Program is to further environmental protection by substantially accelerating the acceptance and use of
improved and cost-effective technologies. The ETV Program is intended to assist and inform those involved
in the design, distribution, permitting, and purchase of environmental technologies. This program is
administered by NERL's Environmental Sciences Division in Las Vegas, Nevada.
The U.S. Department of Energy's (DOE's) Environmental Management (EM) program has entered into active
partnership with EPA, providing cooperative technical management and funding support. DOE EM realizes
that its goals for rapid and cost-effective cleanup hinge on the deployment of innovative environmental
characterization and monitoring technologies. To this end, DOE EM shares the goals and objectives of the
ETV.
Candidate technologies for these programs originate from the private sector and must be commercially ready.
Through the ETV Program, developers are given the opportunity to conduct rigorous demonstrations of their
technologies under realistic field conditions. By completing the evaluation and distributing the results, EPA
establishes a baseline for acceptance and use of these technologies.
Gary J. Foley, Ph.D.
Director
National Exposure Research Laboratory
Office of Research and Development
vn
-------
-------
Abstract
In July 1997, the U.S. Environmental Protection Agency (EPA) conducted a demonstration of poly chlorinated
biphenyl (PCB) field analytical techniques. The purpose of this demonstration was to evaluate field analytical
technologies capable of detecting and quantifying PCBs in soils and solvent extracts. The fundamental
objectives of this demonstration were (1) to obtain technology performance information using environmental
and quality control samples, (2) to determine how comparable the developer field analytical results were with
conventional reference laboratory results, and (3) to report on the logistical operation of the technology. The
demonstration design was subjected to extensive review and comment by EPA's National Exposure Research
Laboratory (NERL) Environmental Sciences Division in Las Vegas, Nevada; Oak Ridge National Laboratory
(ORNL); EPA Regional Offices; the U.S. Department of Energy (DOE); and the technology developers.
The demonstration study was conducted at ORNL under two sets of environmental conditions. The first site
was outdoors, with naturally variable temperature and relative humidity conditions typical of eastern Tennessee
in the summer. A second site was located inside a controlled environmental chamber having lower, and
relatively stable, temperature and relative humidity conditions. The test samples analyzed during this
demonstration were performance evaluation soil, environmental soil, and extract samples. Actual environmental
soil samples, collected from sites in Ohio, Kentucky, and Tennessee, were analyzed, and ranged in
concentration from 0.1 to 700 parts per million (ppm). Extract samples were used to simulate surface wipe
samples, and were evaluated at concentrations ranging from 0 to 100 (jg/mL. The reference laboratory method
used to evaluate the comparability of data was EPA SW-846 Method 8081.
The field analytical technologies tested in this demonstration were the L2000 PCB/Chloride Analyzer (Dexsil
Corporation), the PCB Immunoassay Kit (Hach Company), the 4100 Vapor Detector (Electronic Sensor
Technology), and three immunoassay kits: D TECH, EnviroGard, and RaPID Assay System (Strategic
Diagnostics Inc.). The purpose of an Environmental Technology Verification Report (ETVR) is to document
the demonstration activities, present demonstration data, and verify the performance of the technology. This
ETVR presents information regarding the performance of SDI's D TECH PCB test kit. Separate ETVRs have
been published for the other technologies demonstrated.
The D TECH kit utilizes immunoassay technology to detect trace amounts of PCBs, determined as interval
threshold values, in environmental samples. This test specifically detects Aroclors 1254, 1260, and 1262
equally; reacts well with Aroclors 1242, 1248, and 1268; reacts moderately with Aroclors 1232 and 1016; and
shows little reactivity to Aroclor 1221. The D TECH PCB test kit uses latex particles, onto which antibodies
are immobilized, as the solid support component of the assay. The test is calibrated for Aroclor 1254 and has
conversions for Aroclors 1242 and 1248. As with the other immunoassay systems, the added enzyme conjugate
produces a color change reaction that can then be detected and measured. The darker the color, the less analyte
PCB is present in the sample. Measurement of the test results can be completed instrumentally with a
reflectance meter or visually with the included color card. The D TECH kit provides no information on Aroclor
identification.
ix
-------
The D TECH's quantitative results were based on the analysis of threshold standards for Aroclor 1254 that
were analyzed with every four samples. Because the D TECH kit is an interval technique, method detection
limits are not applicable. Precision, defined as the percentage of the sample sets where all four replicates were
reported as the same interval range, was 44% for the PE soils, 21% for the environmental soils, and 25% for
the extracts. Accuracy, defined as the percentage of D TECH results that agreed with the accepted
concentrations, was 56% for PE soils and 42% for extracts. In general, the percentage that was biased high
was greater (28%) than the percentage that was biased low (17%) for the PE soil samples, while the percentage
that was biased high or low was comparable (29%) for the extract samples. Comparability was defined in a
way similar to accuracy, but the D TECH results were compared to the reference laboratory results rather than
to the certified concentrations. For all soil samples (PE and environmental), the percentage of D TECH results
that agreed with the reference laboratory results was 53%, while the percentage that was biased high (28%)
was slightly greater than the percentage that was biased low (20%).
The demonstration found that the D TECH kit was simple to operate in the field, requiring about one hour for
initial setup and preparation for sample analysis. Once operational, the D TECH kit delivered a sample
throughput of 11 samples/hour under chamber conditions and 15 samples/hour under outdoor conditions. Three
operators analyzed samples during the demonstration, but the technology can be run by a single operator.
Minimal training (2 to 4 hours) is required to operate the D TECH kit, provided the user has a fundamental
understanding of basic chemical and field analytical techniques. The overall performance of the D TECH PCB
test kit was characterized as biased, with approximately 50% of the D TECH results agreeing with the accepted
values (in terms of accuracy), and imprecise, with consistently less than 50% replicate sample results reported
as the same interval.
-------
Table of Contents
Notice ii
Verification Statement iii
Foreword vii
Abstract ix
List of Figures xv
List of Tables xvii
List of Abbreviations and Acronyms xix
Acknowledgments xxiii
Section 1 Introduction 1
Technology Verification Process 2
Needs Identification and Technology Selection 2
Demonstration Planning and Implementation 3
Report Preparation 3
Information Distribution 3
Demonstration Purpose 4
Section 2 Technology Description 5
Objective 5
Principle 5
Test Kit Description 6
Procedure 6
Sampling and Extraction 6
Assay 6
Determining PCB Concentration 7
Section 3 Site Description and Demonstration Design 9
Objective 9
Demonstration Site Description 9
Site Name and Location 9
xi
-------
Site History 9
Site Characteristics 10
Experimental Design 10
Environmental Conditions during Demonstration 12
Sample Descriptions 12
Performance Evaluation Materials 12
Environmental Soil Samples 14
Extract Samples 14
Sampling Plan 14
Sample Collection 14
Sample Preparation, Labeling, and Distribution 14
Predemonstration Study 16
Predemonstration Sample Preparation 16
Predemonstration Results 17
Deviations from the Demonstration Plan 17
Section 4 Reference Laboratory Analytical Results and Evaluation 19
Objective and Approach 19
Reference Laboratory Selection 19
Reference Laboratory Method 20
Calibration 20
Sample Quantification 20
Sample Receipt, Handling, and Holding Times 21
Quality Control Results 21
Objective 21
Continuing Calibration Standard Results 21
Instrument and Method Blank Results 22
Surrogate Spike Results 22
Laboratory Control Sample Results 22
Matrix Spike Results 23
Conclusions of the Quality Control Results 23
Data Review and Validation 23
Objective 23
Corrected Results 24
Suspect Results 24
Data Assessment 25
Objective 25
Precision 25
Performance Evaluation Samples 25
Environmental Soil Samples 26
Extract Samples 28
Accuracy 28
Performance Evaluation Soil Samples 29
Extract Samples 30
xii
-------
Representativeness 30
Completeness 30
Comparability 31
Summary of Observations 31
Section 5 Technology Performance and Evaluation 33
Objective and Approach 33
Interval Reporting 33
Data Assessment 34
Objective 34
Precision 34
Performance Evaluation Samples 34
Environmental Soil Samples 35
Extract Samples 37
Precision Summary 37
Accuracy 38
Performance Evaluation Soil Samples 38
Extract Samples 40
False Positive/False Negative results 40
Representativeness 41
Completeness 41
Comparability 41
Summary of PARCC Parameters 42
Regulatory Decision-Making Applicability 43
Additional Performance Factors 43
Sample Throughput 43
Cost Assessment 44
D TECH Costs 44
Reference Laboratory Costs 46
Cost Assessment Summary 46
General Observations 47
Performance Summary 47
Section 6 Technology Update and Representative Applications 49
Objective 49
Technology Update 49
Reconfiguration of Soil Extraction (Sample Preparation) Products 49
D TECH Test Format Changes 49
Instrument Consolidation 49
Representative Applications 50
Data Quality Objective Example 50
Section 7 References 51
xin
-------
Appendix A Description of Environmental Soil Samples 53
Appendix B Characterization of Environmental Soil Samples 57
Appendix C Temperature and Relative Humidity Conditions 61
Appendix D SDFs D TECH PCB Test Kit: PCB Technology Demonstration Sample Data 67
Appendix E Data Quality Objective Example 77
Disclaimer 79
Background and Problem Statement 79
DQO Goals 79
Use of Technology Performance Information to Implement the Decision Rule 80
Determining the Number of Samples 80
Alternative FP Parameter 83
xiv
-------
List of Figures
2-1 Dispensing soil into bottle 1 6
2-2 Allowing the soil to settle 6
2-3 Diluting the extract 7
2-4 Filling the cup assembly 7
3-1 Schematic map of ORNL, indicating the demonstration area 11
C-l Summary of temperature conditions for outdoor site 63
C-2 Summary of relative humidity conditions for outdoor site 64
C-3 Summary of temperature conditions for chamber site 64
C-4 Summary of relative humidity conditions for chamber site 65
xv
-------
-------
List of Tables
2-1 Converting DTECHTOR readings to PCB concentration in soils 8
3-1 Summary of experimental design by sample type 13
3-2 Summary of D TECH predemonstration results 17
4-1 Suspect measurements within the reference laboratory data 24
4-2 Precision of the reference laboratory for PE soil samples 26
4-3 Precision of the reference laboratory for environmental soil samples 27
4-4 Precision of the reference laboratory for extract samples 28
4-5 Accuracy of the reference laboratory for PE soil samples 29
4-6 Accuracy of the reference laboratory for extract samples 30
4-7 Summary of the reference laboratory performance 32
5-1 D TECH PCB test kit reporting intervals 33
5-2 Classification of precision results 34
5-3 Precision of the D TECH PCB test kit for PE soil samples 35
5-4 Precision of the D TECH PCB test kit for environmental soil samples 36
5-5 Precision of the D TECH PCB test kit for extract samples 37
5-6 Overall precision of the D TECH PCB test kit for all sample types 38
5-7 D TECH test kit accuracy data for PE soil samples 39
5-8 Evaluation of agreement between D TECH's PE sample results and the certified
PE values as a measure of accuracy 39
5-9 Accuracy of the D TECH test kit for extract samples 40
5-10 Evaluation of agreement between D TECH's extract results and the spike concentration
as a measure of accuracy 40
5-11 Evaluation of agreement between D TECH's soil results and the reference laboratory's
results as a measure of comparability 42
5-12 Comparison of the D TECH results with the reference laboratory's suspect measurements 42
5-13 D TECH PCB test kit performance for precision, accuracy, and comparability 43
5-14 Estimated analytical costs for PCB soil samples 45
5-15 Performance summary for the D TECH PCB test kit 48
A-l Summary of soil sample descriptions 55
B-l Summary of environmental soil characterization 59
C-l Average temperature and relative humidity conditions during testing periods 63
D-l SDI's D TECH PCB test kit technology demonstration soil sample data 69
D-2 SDI's D TECH PCB test kit's technology demonstration extract sample data 75
D-3 Corrected reference laboratory data 76
xvn
-------
-------
List of Abbreviations and Acronyms
AL action level
ANOVA analysis of variance
ASTM American Society for Testing and Materials
BHC benzenehexachloride
C concentration at which the false positive error rate is specified
CASD Chemical and Analytical Sciences Division (ORNL)
CCV continuing calibration verification standard
CSCT Consortium for Site Characterization Technology
DCB decachlorobiphenyl
DOE U.S. Department of Energy
DQO data quality objective
ELISA enzyme-linked immunosorbent assay
EM Environmental Management (DOE)
EPA U.S. Environmental Protection Agency
ERA Environmental Resource Associates
EST Electronic Sensor Technology
ETTP East Tennessee Technology Park
ETV Environmental Technology Verification (Program)
ETVR Environmental Technology Verification Report
EvTEC Environmental Technology Evaluation Center
fn false negative result
FN false negative decision error rate
xix
-------
fp false positive result
FP false positive decision error rate
GC gas chromatography
HEPA high-efficiency participate air
ID identifier
LCS laboratory control sample
LMER Lockheed Martin Energy Research
LMES Lockheed Martin Energy Systems
LV Las Vegas
MDL method detection limit
MS matrix spike
MSB matrix spike duplicate
n number of samples
NERL National Exposure Research Laboratory (EPA)
NCEPI National Center for Environmental Publications and Information
NRC Nuclear Regulatory Commission
ORD Office of Research and Development (EPA)
ORNL Oak Ridge National Laboratory
ORO Oak Ridge Operations (DOE)
PARCC precision, accuracy, representativeness, completeness, comparability
PCB polychlorinated biphenyl
PE performance evaluation
ppb parts per billion
ppm parts per million; equivalent units: mg/kg for soils and (jg/mL for extracts
Pr probability
QA quality assurance
QC quality control
xx
-------
R2 coefficient of determination
RDL reporting detection limit
RH relative humidity
RFD request for disposal
RPD relative percent difference
RSD relative standard deviation (percent)
RT regulatory threshold
S2 variance for the measurement
SARA Superfund Amendments and Reauthorization Act of 1986
SD standard deviation
SDI Strategic Diagnostics Inc.
SITE Superfund Innovative Technology Evaluation
SMO sample management office
SOP standard operating procedure
SSM synthetic soil matrix
TCMX tetrachloro-m-xylene
TSCA Toxic Substance Control Act
Z]_p the (l-p)th percentile for the standard normal distribution
%D percent difference
xxi
-------
-------
Acknowledgments
The authors wish to acknowledge the support of all those who helped plan and conduct the demonstration,
analyze the data, and prepare this report. In particular, we recognize the technical expertise of Mitchell
Erickson (Environmental Measurements Laboratory), Viorica Lopez-Avila (Midwest Research Institute), and
Robert F. O'Brien (Pacific Northwest National Laboratory), who were peer reviewers of this report. For
internal peer review, we thank Stacy Barshick (ORNL); for technical support during the demonstration, Todd
Skeen and Ralph Ilgner (ORNL); for site safety and health support, Kim Thomas, Marilyn Hanner, and Fred
Smith (ORNL); for administrative support, Betty Maestas and Linda Plemmons (ORNL); for sample collection
support, Wade Hollinger, Charlotte Schaefer, and Arlin Yeager (LMES), and Mike Rudacille and W. T.
Wright (EET Corporation); for preliminary soil characterization support, Frank Gardner, John Zutman, and
Bob Schlosser (ORNL, Grand Junction, Colo.); for sample management support, Angie McGee, Suzanne
Johnson, and Mary Lane Moore (LMES); for providing performance evaluations samples, Michael Wilson
(EPA's Office of Solid Waste and Emergency Response's Analytical Operations and Data Quality Center);
and for technical guidance and project management of the demonstration, David Garden, Marty Atkins, and
Regina Chung (DOE's Oak Ridge Operations Office), David Bottrell (DOE, Headquarters), Deana Crumbling
(EPA's Technology Innovation Office), and Stephen Billets, Gary Robertson, and Eric Koglin (EPA's National
Exposure Research Laboratory, Las Vegas, Nevada). The authors also acknowledge the participation of
Strategic Diagnostics, Inc., in particular, Craig Kostyshyn, Tim Lawruk, Chris Jones, and Penny Kosinski,
who performed the analyses during the demonstration.
For more information on the PCB Field Analytical Technology Demonstration, contact
Eric N. Koglin
Project Technical Leader
Environmental Protection Agency
Characterization and Research Division
National Exposure Research Laboratory
P.O. Box 93478
Las Vegas, Nevada 89193-3478
(702) 798-2432
For more information on SDI's D TECH PCB test kit, contact
Tim Lawruk
Strategic Diagnostics Inc.
Ill Pencader Drive
Newark, DE 19702-3322
(302) 456-6789
xxni
-------
Section 1
Introduction
The performance evaluation of innovative and alternative environmental technologies is an integral part of the
U.S. Environmental Protection Agency's (EPA's) mission. Early efforts focused on evaluating technologies
that supported the implementation of the Clean Air and Clean Water Acts. In 1987, the Agency began to
evaluate the cost and performance of remediation and monitoring technologies under the Superfund Innovative
Technology Evaluation (SITE) program. This was in response to the mandate in the Superfund Amendments
and Reauthorization Act (SARA) of 1986. In 1990, the U.S. Technology Policy was announced. This policy
placed a renewed emphasis on "making the best use of technology in achieving the national goals of improved
quality of life for all Americans, continued economic growth, and national security." In the spirit of the
Technology Policy, the Agency began to direct a portion of its resources toward the promotion, recognition,
acceptance, and use of U.S.-developed innovative environmental technologies both domestically and abroad.
The Environmental Technology Verification (ETV) Program was created by the Agency to facilitate the
deployment of innovative technologies through performance verification and information dissemination. The
goal of the ETV Program is to further environmental protection by substantially accelerating the acceptance
and use of improved and cost-effective technologies. The ETV Program is intended to assist and inform those
involved in the design, distribution, permitting, and purchase of environmental technologies. The ETV Program
capitalizes upon and applies the lessons that were learned in the implementation of the SITE Program to the
verification of twelve categories of environmental technology: Drinking Water Systems, Pollution
Prevention/Waste Treatment, Pollution Prevention/ Innovative Coatings and Coatings Equipment, Indoor Air
Products, Air Pollution Control, Advanced Monitoring Systems, EvTEC (an independent, private-sector
approach), Wet Weather Flow Technologies, Pollution Prevention/Metal Finishing, Source Water Protection
Technologies, Site Characterization and Monitoring Technology [also referred to as the Consortium for Site
Characterization Technology (CSCT)], and Climate Change Technologies. The performance verification
contained in this report was based on the data collected during a demonstration of polychlorinated biphenyl
(PCB) field analytical technologies. The demonstration was administered by CSCT.
For each pilot, EPA utilizes the expertise of partner "verification organizations" to design efficient procedures
for conducting performance tests of environmental technologies. To date, EPA has partnered with federal
laboratories and state, university, and private sector entities. Verification organizations oversee and report
verification activities based on testing and quality assurance protocols developed with input from all major
stakeholder/customer groups associated with the technology area.
1
-------
In July 1997, CSCT, in cooperation with the U.S. Department of Energy's (DOE's) Environmental
Management (EM) Program, conducted a demonstration to verify the performance of six field analytical
technologies for PCBs: the L2000 PCB/Chloride Analyzer (Dexsil Corporation), the PCB Immunoassay Kit
(Hach Company), the 4100 Vapor Detector (Electronic Sensor Technology), and three immunoassay kits from
Strategic Diagnostics Inc.: D TECH, EnviroGard, and RaPID Assay System. This environmental technology
verification report (ETVR) presents the results of the demonstration study for one PCB field analytical
technology, SDFs D TECH PCB test kit. Separate ETVRs have been published for the other five technologies.
Technology Verification Process
The technology verification process is intended to serve as a template for conducting technology demonstrations
that will generate high-quality data that EPA can use to verify technology performance. Four key steps are
inherent in the process:
• Needs identification and technology selection
• Demonstration planning and implementation
• Report preparation
• Information distribution
Needs Identification and Technology Selection
The first aspect of the technology verification process is to determine technology needs of EPA and the
regulated community. EPA, DOE, the U.S. Department of Defense, industry, and state agencies are asked to
identify technology needs and interest in a technology. Once a technology need is established, a search is
conducted to identify suitable technologies that will address this need. The technology search and identification
process consists of reviewing responses to Commerce Business Daily announcements, searches of industry and
trade publications, attendance at related conferences, and leads from technology developers. Characterization
and monitoring technologies are evaluated against the following criteria:
• meets user needs;
• may be used in the field or in a mobile laboratory;
• is applicable to a variety of environmentally impacted sites;
• has high potential for resolving problems for which current methods are unsatisfactory;
• is cost competitive with current methods;
• performs better than current methods in areas such as data quality, sample preparation, or analytical
turnaround time;
-------
• uses techniques that are easier and safer than current methods; and
• is a commercially available, field-ready technology.
Demonstration Planning and Implementation
After a technology has been selected, EPA, the verification organization, and the developer agree to the
responsibilities for conducting the demonstration and evaluating the technology. The following tasks are
undertaken at this time:
• identifying demonstration sites that will provide the appropriate physical or chemical environment,
including contaminated media;
• identifying and defining the roles of demonstration participants, observers, and reviewers;
• determining logistical and support requirements (for example, field equipment, power and water
sources, mobile laboratory, communications network);
• arranging analytical and sampling support; and
• preparing and implementing a demonstration plan that addresses the experimental design, sampling
design, quality assurance/quality control (QA/QC), health and safety considerations, scheduling of
field and laboratory operations, data analysis procedures, and reporting requirements.
Report Preparation
Innovative technologies are evaluated independently and, when possible, against conventional technologies. The
field technologies are operated by the developers in the presence of independent technology observers. The
technology observers are provided by EPA or a third-party group. Demonstration data are used to evaluate the
capabilities, limitations, and field applications of each technology. Following the demonstration, all raw and
reduced data used to evaluate each technology are compiled into a technology evaluation report, which is
mandated by EPA as a record of the demonstration. A data summary and detailed evaluation of each
technology are published in an ETVR.
Information Distribution
The goal of the information distribution strategy is to ensure that ETVRs are readily available to interested
parties through traditional data distribution pathways, such as printed documents. Documents are also available
on the World Wide Web through the ETV Web site (http://www.epa.gov/etv) and through a Web site supported
by the EPA Office of Solid Waste and Emergency Response's Technology Innovation Office (http://CLU-
in.com).
-------
Demonstration Purpose
The purpose of this demonstration was to obtain performance information for PCB field analytical
technologies, to compare the results with conventional fixed-laboratory results, and to provide supplemental
information (e.g., cost, sample throughput, and training requirements) regarding the operation of the
technology. The demonstration was conducted under two climatic conditions. One set of activities was
conducted outdoors, with naturally fluctuating temperatures and relative humidity conditions. A second set was
conducted in a controlled environmental facility, with lower, relatively stable temperatures and relative
humidities. Multiple soil types, collected from sites in Ohio, Kentucky, and Tennessee, were used in this study.
PCB soil concentrations ranged from approximately 0.1 to 700 parts per million (ppm). Developers also
analyzed 24 solutions of known PCB concentration that were used to simulate extracted wipe samples. The
extract samples ranged in concentration from 0 to 100 (jg/mL.
-------
Section 2
Technology Description
Objective
This section describes the technology being demonstrated, including the operating principles underlying the
technology and the overall approach to its use. The information provided here is excerpted from that provided
by the developer. Performance characteristics described in this section are specified by the developer, which
may or may not be substantiated by the data presented in Section 5.
Principle
The D TECH PCB test kit applies the principles of enzyme-linked immunosorbent assay (ELISA) to the
determination of PCBs. In such an assay, an enzyme has been chemically linked to a PCB molecule or PCB
analog to create a labeled PCB reagent. The labeled PCB reagent (called a conjugate) is mixed with an extract
of native sample containing the PCB contaminant. A portion of the mixture is applied to a surface to which an
antibody specific for PCB has been affixed. The native PCB and PCB-enzyme conjugate compete for a limited
number of antibody sites. After a period of time, the solution is washed away, and what remain are either
PCB-antibody complexes or enzyme-PCB- antibody complexes attached to the test surface. The proportion
of the two complexes on the test surface is determined by the amount of native PCB in the original sample. The
enzyme present on the test surface is used to catalyze a color change reaction in a solution added to the test
surface. Because the amount of enzyme present is inversely proportional to the concentration of native PCB
contaminant, the amount of color development is inversely proportional to the concentration of PCB
contaminant.
The D TECH PCB test kit uses latex particles as the solid support component of the assay. With this kit,
antibodies are immobilized on thousands of latex particles that are free to interact and react with the analyte
and the enzyme conjugate sample extract. These particles are then separated on a filtration device, the color
development is induced, and the results are read visually or with a small reflectance meter. In this kit, sample
solution (i.e., soil extract) is added to a dry mixture of antibody-coated latex particles and enzyme conjugate.
The immunological reaction then occurs in this reconstituted suspension. When the reaction is complete, the
particles are collected on the membrane surface of a collection cup and washed briefly. The test is completed
by adding a color-developing solution to the surface of the collection cup. As with the other immunoassay
formats, the enzyme linked to the antibody catalyzes a color change reaction. Measurement of test results can
be completed instrumentally with a reflectance meter or visually with the included color card.
The D TECH PCB test kit is designed to provide quick, semi-quantitative, and reliable results for making
environmental decisions. The kit can screen for highly contaminated samples that require pre-dilution prior to
instrumental analysis. The D TECH PCB test kit has a working range of 0.5 to 25 ppm for soil samples and
a 10 to 250 (jg/100 cm2 for wipe samples. This test specifically detects Aroclors 1254, 1260,
-------
and 1262 equally; reacts well with Aroclors 1242, 1248, and 1268; reacts moderately with Aroclors 1232 and
1016; and shows little reactivity to Aroclor 1221.
Test Kit Description
The D TECH PCB test kit contains sufficient materials to perform four soil tests. All the materials needed to
extract PCB from soils for semi-quantitation are included. This kit has a working temperature range from 7°to
38°C (45° to 100°F). It should not be frozen or stored in direct sunlight.
Procedure
Sampling and Extraction
The D TECH soil sampling tube is used to collect field soil samples. To achieve a more
homogeneous distribution and to ensure reproducible test results, the soil sample should
be mixed thoroughly. All debris, such as sticks, stones, and leaves, should be removed
from the soil sample before using the D TECH soil sampling tube. (Note: Because the
soil demonstration samples were sandy and dry, 5 g of soil was weighed out and poured
into the soil sampling tube.)
In the D TECH procedure the soil is dispensed into bottle 1, which already contains 9 mL
of methanol liquid, by positioning the barrel into the neck of the bottle and firmly pushing
the plunger (Figure 2-1). If soil lodges in the neck of the bottle, the sampling tube is used
to push it into the bottle. The threads of the bottle neck and cap should be wiped clean of
any soil that has adhered to them before the cap is placed on the bottle. The bottle is then
capped tightly, and the soil and liquid methanol are thoroughly mixed by continuous
shaking for 3 min. The soil is then allowed to settle for approximately 1 min (Figure 2-2).
Some soils will settle more slowly than others.
Figure 2-1.
Dispensing soil
into bottle 1.
The next sequence of steps, outlined in Figure 2-3, begins the extract analysis. The
cap is removed from bottle 2 (containing diatomaceous earth). With the 2-mL
calibrated pipet, 2 mL of the liquid layer is removed from bottle 1 and dispensed
into bottle 2; this solution is then mixed well. After the filter tip is snapped onto the
neck of bottle 2, that bottle is squeezed to deliver the filtered solution into the PCB
test vial to a level between the two lines (approximately 13-14 drops). At this point,
the extract is prepared for assay.
Assay
Once the assay portion of the test is initiated, all steps must be executed sequentially
and without stopping. The contents of reagent C (white cap) are squeezed to fill the
PCB reference vial (red stopper) to a level between the 2 lines. The contents are
gently mixed by shaking the vial in a back and forth motion. After 5 min (±30 sec), the contents of the PCB
test vial are poured into the T (test) side of the cup assembly (Figure 2-4). Then the contents of the reference
vial are immediately poured into the R side of the cup assembly. The liquid is allowed to drain completely on
both sides. The reference vial concentration is equivalent to the detection limit of
Figure 2-2. Allowing
the soil to settle.
-------
*
,*~^B
r-*^ _J__ x^
Figure 2-3. Diluting the extract.
the kit (0.5 ppm). For every sample analyzed, this vial is filled using a new bottle.
The next steps use dropper-tipped bottles. When dispensing these
reagents, the analyst should not allow any dropper tip to contact any
solution(s) or surface in the device. To assure uniform color
development across the device, the drop should be dispensed onto the
sloped side of the well to lessen its impact. The analyst should not allow
the drop to fall into the middle of the well.
Ten drops (±2 drops) of reagent D solution (yellow cap) are now added
into each side of the cup assembly, and the liquid is allowed to drain
completely. Then 5 drops (±1 drop) of reagent E solution (blue cap) are
added into each side of the cup assembly. This solution must be added
to the second well immediately after addition to the first well. Allow the
wells to drain completely. Color development time is temperature-
dependent and takes approximately 10 min at 75 °F. More time is
required at lower temperatures and less time is required at higher
temperatures. The color in both wells is stable for approximately 4 h.
Figure 2-4. Filling the cup assembly.
Determining PCB Concentration
The results from the D TECH PCB test kit are interpreted using the DTECHTOR and Table 2-1. First, the
percent relative reflectance is determined by use of the DTECHTOR instrument. The DTECHTOR instrument
is a hand-held reflectometer for interpreting the results of D TECH test samples. Completely portable and
powered with a 9-V plug-in battery, the DTECHTOR uses a simple push-button procedure to provide results
in 1 min. Given the reading of this instrument, the analyst then uses the data shown in Table 2-1 to determine
the concentration range of total PCB in the sample. If the user does not know which Aroclor(s) are present in
the sample, he or she should use the values for Aroclor 1254.
-------
Table 2-1. Converting DTECHTOR readings to PCS concentration in soils
DTECHTOR Reading
(% relative reflectance)
<10
10-20
21^0
41-60
61-70
>70
Aroclor Concentration Range
(ppm)
1254
0.5
0.5-1.0
1.1^.0
4.1-15
16-25
>25
1242
<1.5
1.5-3.5
3.6-20
21-54
55-100
>100
1248
0.8
0.8-2.3
2.4-11
12-28
29-53
>53
-------
Section 3
Site Description and Demonstration Design
Objective
This section describes the demonstration site, the experimental design for the verification test, and the sampling
plan (sample types analyzed and the collection and preparation strategies). Included in this section are the
results from the predemonstration study and a description of the deviations made from the original
demonstration design.
Demonstration Site Description
Site Name and Location
The demonstration of PCB field analytical technologies was conducted at Oak Ridge National Laboratory
(ORNL) in Oak Ridge, Tennessee. PCB-contaminated soils from three DOE sites (Oak Ridge; Paducah,
Kentucky; and Piketon, Ohio) were used in this demonstration. The soil samples used in this study were
brought to the demonstration testing location for evaluation of the field analytical technologies.
Site History
Oak Ridge is located in the Tennessee River Valley, 25 miles northwest of Knoxville. Three DOE facilities are
located in Oak Ridge: ORNL, the Oak Ridge Y-12 Plant, and East Tennessee Technology Park (ETTP).
Chemical processing and warhead component production have occurred at the Y-12 Plant, and ETTP is a
former gaseous diffusion uranium enrichment plant. At both facilities, industrial processing associated with
nuclear weapons production has resulted in the production of millions of kilograms of PCB-contaminated soils.
Two other DOE facilities—the Paducah plant in Paducah, Kentucky, and the Portsmouth plant in Piketon,
Ohio—are also gaseous diffusion facilities with a history of PCB contamination. During the remediation of the
PCB-contaminated areas at the three DOE sites, soils were excavated from the ground where the PCB
contamination occurred, packaged in containers ranging in size from 55-gal to 110-gal drums, and stored as
PCB waste. Samples from these repositories —referred to as "Oak Ridge," "Portsmouth," and "Paducah"
samples in this report—were used in this demonstration.
In Oak Ridge, excavation activities occurred between 1991 and 1995. The Oak Ridge samples were comprised
of PCB-contaminated soils from both Y-12 and ETTP. Five different sources of PCB contamination resulted
in soil excavations from various dikes, drainage ditches, and catch basins. Some of the soils are EPA-listed
hazardous waste due to the presence of other contaminants (e.g., diesel fuels).
A population of over 5000 drums containing PCB-contaminated soils was generated from 1986 to 1987 during
the remediation of the East Drainage Ditch at the Portsmouth Gaseous Diffusion Plant. The ditch was reported
-------
to have three primary sources of potential contamination: (1) treated effluent from a radioactive liquid treatment
facility, (2) runoff from a biodegradation plot where waste oil and sludge were disposed of, and (3) storm sewer
discharges. In addition, waste oil was reportedly used for weed control in the ditch. Aside from PCB
contamination, no other major hazardous contaminants were detected in these soils. Therefore, no EPA
hazardous waste codes are assigned to this waste.
Twenty-nine drums of PCB-contaminated soils from the Paducah plant were generated as part of a spill
cleanup activity at an organic waste storage area (C-746-R). The waste is considered a listed hazardous waste
for spent solvents (EPA hazardous waste code F001) because it is known to contain trichloroethylene. Other
volatile organic compounds, such as xylene, dichlorobenzene, and cresol, were also detected in the preliminary
analyses of some of the Paducah samples.
Site Characteristics
PCB-contaminated environmental soil samples from Oak Ridge, Portsmouth, and Paducah were collected from
waste containers at storage repositories at ETTP and Paducah. Many of the soils contained interfering
compounds such as oils, fuels, and other chlorinated compounds (e.g., trichloroethylene). Specific sample
descriptions of the environmental soils used in this demonstration are given in Appendix A. In addition, each
sample was characterized in terms of its soil composition, pH, and total organic carbon content. Those results
are summarized in Appendix B.
Field demonstration activities occurred at two sites at ORNL: a natural outdoor environment (the outdoor site)
and inside a controlled environmental atmosphere chamber (the chamber site). Figure 3-1 shows a schematic
map of a section of ORNL indicating the demonstration area where the outdoor field activities occurred.
Generally, the average summer temperature in eastern Tennessee is 75.6° F, with July and August temperatures
averaging 79.1 °F and 76.8 °F, respectively. Average temperatures during the testing periods ranged from 79
to 85 °F, as shown in Appendix C. Studies were also conducted inside a controlled environmental atmosphere
chamber, hereafter referred to as the "chamber," located in Building 5507 at ORNL. Demonstration studies
inside the chamber were used to evaluate performance under environmental conditions that were markedly
different from the ambient outdoor conditions at the time of the test. Average temperatures in the chamber
during the testing periods ranged from 55 to 70°F. The controlled experimental atmosphere facility consists
of a room-size walk-in chamber 10 ft wide and 12 ft long with air processing equipment to control temperature
and humidity. The chamber is equipped with an environmental control system, including reverse osmosis water
purification that supplies the chamber humidity control system. High efficiency particulate air (HEPA) and
activated charcoal filters are installed for recirculation and building exhaust filtration.
Experimental Design
The analytical challenge with PCB analysis is to quantify a complex mixture that may or may not resemble the
original commercial product (i.e., Aroclor) due to environmental aging, and to report the result as a single
number [1]. The primary objective of the verification test was to compare the performance of the field
technology to laboratory-based measurements. Often, verification tests involve a direct one-to-one comparison
of results from field-acquired samples. However, because sample heterogeneity can preclude replicate field or
laboratory comparison, accuracy and precision data must often be derived from the analysis of QC and
performance evaluation (PE) samples. In this study, replicates of all three sample types (QC, PE, and
environmental soil) were analyzed. The ability to use
10
-------
Building
Figure 3-1. Schematic map of ORNL, indicating the demonstration area.
environmental soils in the verification test was made possible because the samples, collected from drums
containing PCB-contaminated soils, could be thoroughly homogenized and characterized prior to the
demonstration. This facet of the design, allowing additional precision data to be obtained on actual field-
acquired samples, provided an added performance factor in the verification test.
Another objective of this demonstration was to evaluate the field technology's capability to support regulatory
compliance decisions. For field methods to be used in these decisions, the technology must be capable of
informing the user, with known precision and accuracy, that soil concentrations are greater than or less than
50 ppm, and that wipe samples are greater than or less than 100 (jg/100 cm2 [2]. The samples selected for
analysis in the demonstration study were chosen with this objective in mind.
The experimental design is summarized in Table 3-1. This design was approved by all participants prior to the
start of the demonstration study. In total, the developers analyzed 208 soil samples, 104 each at both locations
(outdoors and chamber). The 104 soil samples comprised 68 environmental samples (17 unique environmental
samples prepared in quadruplicate) ranging in PCB concentration from 0.1 to 700 ppm and 36 PE soils (9
unique PE samples in quadruplicate) ranging in PCB concentration from 0 to 50 ppm. To determine the impact
of different environmental conditions on the technology's performance, each batch of 104 samples contained
five sets of quadruplicate soil samples from DOE's Paducah site. These were analyzed under both sets of
environmental conditions (i.e., outdoor and chamber conditions). For the developers participating in the extract
sample portion (i.e., simulated wipe samples) of the demonstration, 12 extracts, ranging in concentration from
0 to 100 pg/mL, were analyzed in each location (chamber and outdoors). All samples were analyzed without
11
-------
prior knowledge of sample type or concentration and were analyzed in a randomized order that was unique for
each developer.
Environmental Conditions during Demonstration
As mentioned above, field activities were conducted both outdoors under natural environmental conditions and
indoors in a controlled environmental atmosphere chamber to evaluate the effect of environmental conditions
on technology performance. The weather outside was relatively uncomfortable during the July demonstration,
with highs approaching 100°F and 90% relative humidity (RH). Daily average temperatures were around 85 °F
with 70% RH. While outside, the developers set up canopies to provide shade and protection from frequent late
afternoon thundershowers.
In the indoor chamber tests, conditions were initially set to 55 °F and 25% RH. An independent check of the
conditions inside the chamber revealed that the temperature was closer to 68 °F with a 38% RH on the first day
of testing. A maintenance crew was called in to address the inconsistencies between the set and actual
conditions. By the middle of the third day of testing, the chamber was operating properly at 55 °F and 50% RH.
Appendix C contains a summary of the environmental conditions (temperature and relative humidity) during
the demonstration. The SDI team analyzed samples using the D TECH test kit outdoors July 26, and in the
chamber on July 23.
Sample Descriptions
PCBs (C^HjQ.xClx) are a class of compounds that are chlorine-substituted linked benzene rings. There are 209
possible PCB compounds (also known as congeners). PCBs were commercially produced as complex mixtures
beginning in 1929 for use in transformers, capacitors, paints, pesticides, and inks [1]. Monsanto Corporation
marketed products that were mixtures of 20 to 60 PCB congeners under the trade name Aroclor. Aroclor
mixtures are identified by a number (e.g., Aroclor 1260) that represents the mixture's chlorine composition as
a percentage (e.g., 60%).
Performance Evaluation Materials
Samples of Tennessee reference soil [3] served as the blanks. Preprepared certified PE samples were obtained
from Environmental Resource Associates (ERA) of Arvada, Colorado, and the Analytical Operations and Data
Quality Center of EPA's Office of Solid Waste and Emergency Response. The soils purchased from ERA had
been prepared using ERA's semivolatile blank soil matrix. This matrix was a topsoil that had been dried,
sieved, and homogenized. Particle size was approximately 60 mesh. The soil was approximately 40% clay. The
samples acquired from EPA's Analytical Operations and Data Quality Center had been prepared using
contaminated soils from various sites around the country in the following manner: The original soils had been
homogenized and diluted with a synthetic soil matrix (SSM). The SSM had a known matrix of 6% gravel, 31%
sand, and 43% silt/clay; the remaining 20% was topsoil. The dilution of the original soils was performed by
mixing known amounts of contaminated soil with the SSM in a blender for no less than 12 h. The samples were
also spiked with target pesticides (a, P, A, and 6-BHC, methoxychlor, and endrin ketone) to introduce some
compounds that were likely to be present in an actual environmental soil. The hydrocarbon background from
the original sample and the spiked
12
-------
pesticides produced a challenging matrix. The PE soils required no additional preparation by ORNL and were
split for the developer and reference laboratory analyses as received.
Table 3-1. Summary of experimental design by sample type
Concentration
Range
Sample ID a
Outdoor Site
Chamber Site
Total #
Samples
Analyzed
PE Materials
0
2.0 ppm
2.0 ppm
5.0 ppm
10.9 ppm
20.0 ppm
49.8 ppm
50.0 ppm
50.0 ppm
126
118
124
120
122
119
125
121
123
226
218
224
220
222
219
225
221
223
8
8
8
8
8
8
8
8
8
Environmental Soils
0.1-2.0 ppm
2. 1-20.0 ppm
20. 1-50.0 ppm
50. 1-700.0 ppm
101, 107, 108, 109, 113, 114
102, 103, 104, 115
111,116
105, 106, 110, 112, 117
Extracts
0
10 |ig/mL
100 |ig/mL
Grand Total
129b/132c
127/130
128/131
116
201,202,206
203,207,212,213
204,208,209,214,215
205,210,211,216,217
229/232
227/230
228/231
116
36
32
28
40
8
8
8
232 d
a Each sample ID was analyzed in quadruplicate.
b Extract prepared in iso-octane for Dexsil and the reference laboratory.
c Extract prepared in methanol for Electronic Sensor Technology, Strategic Diagnostics Inc., and the
reference laboratory.
d All samples were analyzed in random order.
13
-------
Environmental Soil Samples
As noted in the site description above, PCB-contaminated environmental soil samples from Oak Ridge,
Portsmouth, and Paducah were used in this demonstration. The soils were contaminated with PCBs as the result
of spills and industrial processing activities at the various DOE facilities. Originally, the contaminated soils
were excavated from dikes, drainage ditches, catch basins, and organic waste storage areas. The excavated soils
were then packaged into waste containers and stored at the repositories in ETTP and Paducah in anticipation
of disposal by incineration. The environmental soil samples used in this study were collected from these waste
containers. Many of the soils contained interfering compounds such as oils, fuels, and other chlorinated
compounds, while some contained multiple Aroclors. For more information on sampling locations and sample
characteristics (soil composition, pH, and total organic carbon content), refer to Appendices A and B,
respectively.
Extract Samples
Traditionally, the amount of PCBs on a contaminated surface is determined by wiping the surface with a cotton
pad saturated with hexane. The pad is then taken to the laboratory, extracted with additional hexane, and
analyzed by gas chromatography. Unlike soil samples, which can be more readily
homogenized and divided, equivalent wipe samples (i.e., contaminated surfaces or post-wipe pads) were not
easily obtainable. Therefore, interference-free solutions of PCBs were analyzed to simulate an extracted surface
wipe pad. Extract sample analyses provided evaluation data that relied primarily on the technology's
performance rather than on elements critical to the entire method (i.e., sample collection and preparation).
Because different developers required the extract samples prepared in different solvents (e.g., methanol and iso-
octane), the reference laboratory analyzed sets of extracts in both solvents. SDI analyzed extracts prepared in
methanol. A total of 12 extracts were analyzed per site; these consisted of four replicates each of a blank and
two concentration levels (10 and 100 (jg/mL).
Sampling Plan
Sample Collection
Environmental soil samples were collected from April 17 through May 7, 1997. Portsmouth and Oak Ridge
Reservation soils were collected from either storage boxes or 55-gal drums stored at ETTP. Briefly, the
following procedure was used to collect the soil samples. Approximately 30 Ib of soil were collected from the
top of the drum or B-25 box using a scoop and placed in a plastic bag. The soil was sifted to remove rocks and
other large debris, then poured into a plastic-lined 5-gal container. All samples were subjected to radiological
screening and were determined to be nonradioactive. Similarly, soil samples were collected from 55-gal drums
stored at Paducah and shipped to ORNL in lined 5-gal containers.
Sample Preparation, Labeling, and Distribution
Aliquots of several of the environmental soils were analyzed and determined to be heterogeneous in PCB
concentration. Because this is unsatisfactory for accurately comparing the performance of the field technology
with the laboratory-based method, the environmental soils had to be homogenized prior to sample distribution.
Each Portsmouth and Oak Ridge environmental soil sample was homogenized by first placing approximately
1500 g of soil in a glass Pyrex dish. The dish was then placed in a large oven set at 35 °C, with the exhaust and
blower fans turned on to circulate the air. After drying overnight, the soil was pulverized using a conventional
blender and sieved using a 9-mesh screen (2 mm particle size).
14
-------
Last, the soil was thoroughly mixed using a spatula. A comparison of dried and undried soils showed that a
minimal amount of PCBs (< 20%) was lost due to sample drying, making this procedure suitable for use in the
preparation of the soil samples. The Paducah samples, because of their sandy characteristics, only required the
sieving and mixing preparation steps. Extract sample preparation involved making solutions of PCBs in
methanol and iso-octane at two concentration levels (10 and 100 (jg/mL). Multiple aliquots of each sample
were analyzed using the analytical procedure described below to confirm the homogeneity of the samples with
respect to PCB concentration.
To provide the developers with soils contaminated at higher concentrations of PCBs, some of the environmental
soils (those labeled with an "S" in Appendix B) were spiked with additional PCBs. Spiked soils samples were
prepared after the soil was first dried in a 35 °C oven overnight. The dry soil was ground using a conventional
blender and sieved through a 9-mesh screen (2-mm particle size). Approximately 1500 g of the sieved soil were
spiked with a diethyl ether solution of PCBs at the desired concentration. The fortified soil was agitated using
a mechanical shaker and then allowed to air-dry in a laboratory hood overnight. A minimum of four aliquots
were analyzed using the analytical procedure described below to confirm the homogeneity of the soil with
regard to the PCB concentration.
The environmental soils were characterized at ORNL prior to the demonstration study. The procedure used to
confirm the homogeneity of the soil samples entailed the extraction of 3 to 5 g of soil in a mixture of solvents
(1 mL water, 4 mL methanol, and 5 mL hexane). After the soil/solvent mixture was agitated by a mechanical
shaker, the hexane layer was removed and an aliquot was diluted for analysis. The hexane extract was analyzed
on a Hewlett Packard 6890 gas chromatograph equipped with an electron capture detector and autosampler.
The method used was a slightly modified version of EPA's SW-846 dual-column Method 8081 [4].
After analysis confirming homogeneity, the samples were split into jars for distribution. Each 4-oz sample jar
contained approximately 20 g of soil. Four replicate splits of each soil sample were prepared for each
developer. The samples were randomized in two fashions. First, the order in which the filled jars were
distributed was randomized, such that the same developer did not always receive the first jar filled for a given
sample set. Second, the order of analysis was randomized so that each developer analyzed the same set of
samples, but in a different order. The extract samples were split into 10-mL aliquots and placed into 2-oz jars.
The extracts were stored in the refrigerator (at <4°C) until released to the developers. Each sample jar had
three labels: (1) developer order number; (2) sample identifier number; and (3) a PCB warning label. The
developer order number corresponded to the order in which the developer was required to analyze the samples
(e.g., SDI 1001 through SDI 1116). The sample identifier number was in the format of "xxxyzz," where "xxx"
was the three-digit sample ID (e.g., 101) listed in Table 3-1, "y" was the replicate (e.g., 1 to 4), and "zz" was
the aliquot order of each replicate (e.g., 01 to 11). For example, sample identifier 101101 corresponded to
sample ID "101" (an Oak Ridge soil from RFD 40022, drum 02), "1" corresponded to the first replicate from
that sample, and "01" corresponded to the first jar filled in that series.
Once the samples were prepared, they were stored at a central sample distribution center. During the
demonstration study, developers were sent to the distribution center to pick up their samples. Samples were
distributed sequentially in batches of 12 to ensure that samples were analyzed in the order specified.
15
-------
Completion of chain-of-custody forms and scanning of bar code labels documented sample transfer activities.
Some of the developers received information regarding the samples prior to analysis. SDI received information
pertaining to which Aroclors were in the samples. This was provided at the request of SDI to simulate the type
of information that would be available during actual field testing. The developers returned the unused portions
of the samples with the analytical results to the distribution center when testing was completed. The sample
bar codes were scanned upon return to document sample throughput time.
Three complete sets of extra samples, called archive samples, were available for distribution in case the
integrity of a sample was compromised. Very few (<5) archive samples were utilized over the course of the
demonstration.
Predemonstration Study
Ideally, environmental soil samples are sent to the developers prior to the demonstration study to allow them
the opportunity to analyze representative samples in advance of the verification test. This gives developers the
opportunity to refine and calibrate their technologies and revise their operating procedures on the basis of the
predemonstration study results. The predemonstration study results can also be used as an indication that the
selected technologies are of the appropriate level of maturity to participate in the demonstration study.
According to ORNL regulations, however, one of two conditions must exist in order to ship environmental soils
that were once classified as mixed hazardous waste. First, the recipient—in this case, the developer's
facilities—must have proper Nuclear Regulatory Commission (NRC) licensing to receive and analyze
radiological materials. Second, the soils must be certified as entirely free of radioactivity, beyond the no-rad
certification issued from radiological screening tests based on ORNL standards. Because none of the developers
had proper NRC licensing and proving that the soils were entirely free of radioactivity was prohibitive, spiked
samples of Tennessee reference soil were used for the predemonstration study. The developers had an
opportunity to evaluate the Tennessee reference soils spiked with PCBs at concentrations similar to what would
be used in the demonstration study. The developers also analyzed two performance evaluation samples and one
solvent extract. The reference laboratory analyzed the same set of samples, which included two extracts
samples, prepared in the two solvents (methanol and iso-octane) requested by the developers.
Predemonstration Sample Preparation
Two soil samples were prepared by ORNL using Tennessee reference soil [3]. The soil was a Captina silt loam
from Roane County, Tennessee, that was slightly acidic (pH ~5) and low in organic carbons (-1.5%). The soil
composition was 7.7% sand, 29.8% clay, and 62.5% silt. To prepare a spiked sample, the soil was first ground
either using a mortar and pestle or a conventional blender. The soil was then sieved through a 16-mesh screen
(1 mm particle size). Approximately 500 g of the sieved soil was spiked with a diethyl ether solution of PCBs
at the desired concentration. The soil was agitated using a mechanical shaker, then allowed to air-dry overnight
in a laboratory hood. A minimum of five aliquots were analyzed by gas chromatography using electron capture
detection. The PCB concentration of the spiked samples was determined to be homogeneous. The remaining
two soil samples used in the predemonstration study were performance evaluation materials acquired from ERA
and EPA (see the section "Performance Evaluation Materials" above). In addition, a solvent extract was
prepared by ORNL to simulate an extracted surface wipe sample. The extracts were prepared in two different
solvents (iso-octane and methanol) to accommodate developer requests.
16
-------
Predemonstration Results
The predemonstration samples were sent to the developers and the reference laboratory on June 2, 1997.
Predemonstration results were received by June 26, 1997. Table 3-2 summarizes the test kit's results for the
predemonstration samples. Results indicated that SDFs D TECH PCB test kit was ready for field evaluation.
Table 3-2. Summary of D TECH predemonstration results
Sample Description
2 ppm of Aroclor 1260
100ppm(total)of
Aroclors 1254 and 1260
1 1 ppm of Aroclor 1260
50 ppm of Aroclor 1254
5 ppm of Aroclor 1242
Matrix
Soil
Soil
Soil
Soil
Extract
Source
ORNL
ORNL
EPA
ERA
ORNL
D TECH a
Result
(ppm)
[l,4)b
(25, »)
[15,25]
(25, co)
(3.5,6.6)
Duplicate
result (ppm)
[1,4)
(25, oo)
[15,25]
c
(3.5,6.6)
Reference Laboratory
Result
(ppm)
2.2
78.0
11.0
37.0
4.7
Duplicate
result (ppm)
2.3
89.0
9.5
c
4.9
a Results were Aroclor-adjusted (see Section 2 for more details).
b The notation [1, 4) indicates sample concentration > 1 and < 4. See Sections 2 and 5 for more information on
interval reporting.
c Replicate was not analyzed because of lack of adequate sample for second analyses.
Deviations from the Demonstration Plan
A few deviations from the demonstration plan occurred. In Appendix B of the technology demonstration plan
[5], the reference laboratory's procedure states that no more than 10 samples will be analyzed with each
analytical batch (excluding blanks, standards, QC samples, and dilutions). The analytical batch is also stated
as 10 samples in the Quality Assurance Project Plan of the demonstration plan. The reference laboratory
actually analyzed 20 samples per analytical batch. Because a 20-sample batch is recommended in SW-846
Method 8081, this deviation was deemed acceptable.
Table 5 of the demonstration plan [5] delineates the environmental soils according to concentration. The
classification was based on a preliminary analysis of the soils at ORNL. Table 3-1 of this report arranges the
concentrations as characterized by the reference laboratory. The reference laboratory determined that five
sample sets (sample IDs 102, 105, 110, 111, and 210) were in the next highest concentration range, differing
from what was originally outlined in the demonstration plan. Also, the highest concentration determined by the
reference laboratory was 700 ppm, while the preliminary analysis at ORNL found the highest concentration
to be 500 ppm.
During the demonstration study, the SDI team made one modification to the procedure described in the
technology demonstration plan [5]: the demonstration samples were weighed instead of collected volumetrically
using the D TECH soil sampling tube. This was necessary because of the dry, sandy characteristics of the soils.
17
-------
18
-------
Section 4
Reference Laboratory Analytical Results and Evaluation
Objective and Approach
The purpose of this section is to present the evaluation of the PCB data generated by the reference laboratory.
Evaluation of the results from the analysis of PE, environmental soil, and extract samples was based on
precision, accuracy, representativeness, completeness, comparability (PARCC) parameters [6]. This section
describes how the analytical data generated by the reference laboratory were used to establish a baseline
performance for PCB analysis.
Reference Laboratory Selection
The Oak Ridge Sample Management Office (SMO) has been tasked by DOE Oak Ridge Operations (DOE-
ORO) with maintaining a list of qualified laboratories to provide analytical services. The technology
demonstration plan [5] contains the SMO's standard operating procedures (SOPs) for identifying, qualifying,
and selecting analytical laboratories. Laboratories are qualified as acceptable analytical service providers for
the SMO by meeting specific requirements. These requirements include providing pertinent documentation
(such as QA and chemical hygiene plans), acceptance of the documents by the SMO, and satisfactory
performance on an on-site prequalification audit of laboratory operations. All laboratory qualifications are
approved by a laboratory selection board, composed of the SMO operations manager and appointees from all
prime contractors that conduct business with the SMO.
All of the qualified laboratories were invited to bid on the demonstration study sample analysis. The lowest-cost
bidder was LAS Laboratories, in Las Vegas, Nevada. A readiness review conducted by ORNL and the SMO
confirmed the selection of LAS as the reference laboratory. Acceptance of the reference laboratory was
finalized by satisfactory performance in the predemonstration study (see Table 3-2). The SMO contracted LAS
to provide full data packages for the demonstration study sample analyses within 30 days of sample shipment.
The SMO conducts on-site audits of LAS annually as part of the laboratory qualification program. At the time
of selection, the most recent audit of LAS had occurred in February 1997. Results from this audit indicated
that LAS was proficient in several areas, including program management, quality management, and training
programs. No findings regarding PCB analytical procedure implementation were noted. A second on-site audit
of LAS occurred August 11-12, 1997, during the analysis of the demonstration study samples. This
surveillance focused specifically on the procedures that were currently in use for the analysis of the
demonstration samples. The audit, jointly conducted by the SMO, DOE-ORO, and EPA-Las Vegas (LV),
verified that LAS was procedurally compliant. The audit team noted that LAS had excellent adherence to the
analytical protocols and that the staff were knowledgeable of the requirements of the method. No findings
impacting data quality were noted in the audit report.
19
-------
Reference Laboratory Method
The reference laboratory's analytical method, also presented in the technology demonstration plan [5], followed
the guidelines established in EPA SW-846 Method 8081 [4]. According to LAS's SOP, PCBs were extracted
from 30-g samples of soil by sonication in hexane. Each extract was then concentrated to a final volume that
was further subjected to a sulfuric acid cleanup to remove potential interferences. The analytes were identified
and quantified using a gas chromatograph equipped with dual electron-capture detectors. Each extract was
analyzed on two different chromatographic columns with slightly different separation characteristics (primary
column: RTX-1701, 30 m x 0.53 mm ID x 0.5 (jm; confirmatory column: RTX-5, 30 m x 0.53 mm ID x 0.5
(jm). PCBs were identified when peak patterns from a sample extract matched the patterns of standards for
both columns. PCBs were quantified based on the initial calibration of the primary column.
Calibration
Method 8081 states that, because Aroclors 1016 and 1260 include many of the peaks represented in the other
five Aroclor mixtures, it is only necessary to analyze two multilevel standards for these Aroclors to demonstrate
the linearity of the detector response for PCBs. However, per LAS SOPs, five-point (0.1 to 4 ppm) initial
calibration curves were generated for Aroclors 1016, 1248, 1254, and 1260 and the surrogate compounds
[decachlorobiphenyl (DCB) and tetrachloro-ra-xylene (TCMX)]. Single mid-level standards were analyzed for
the other Aroclors (1221, 1232, and 1242) to aid in pattern recognition. All of the multi-point calibration data,
fitted to quadratic models, met the QC requirement of having a coefficient of determination (R2) of 0.99 or
better over the calibration range specified. The detection limits for soil samples were 0.033 ppm (ng/g) for all
Aroclors except Aroclor 1221, which was 0.067 ppm. For extract samples, the detection limits were 0.010 ppm
((ig/mL) for all Aroclors except Aroclor 1221, which was 0.020 ppm. Reporting detection limits were
calculated based on the above detection limits, the actual sample weight, and the dilution factor.
Sample Quantification
For sample quantification, Aroclors were identified by comparing the samples' peak patterns and retention
times with those of the respective standards. Peak height ratios, peak shapes, sample weathering, and general
similarity in detector response were also considered in the identification. Aroclor quantifications were
performed by selecting three to five representative peaks, confirming that the peaks were within the established
retention time windows, integrating the selected peaks, quantifying the peaks based on the calibrations, and
averaging the results to obtain a single concentration value for the multicomponent Aroclor. If mixtures of
Aroclors were suspected to be present, the sample was typically quantified in terms of the most representative
Aroclor pattern. If the identification of multiple Aroclors was definitive, total PCBs in the sample were
calculated by summing the concentrations of both Aroclors. Aroclor concentrations were quantified within the
concentration range of the calibration curve. If PCBs were detected and the concentrations were outside of the
calibration range, the sample was diluted and reanalyzed until the concentration was within the calibration
range. If no PCBs were detected, the result was reported as a non-detect (i.e., "< reporting detection limit").
Sample Receipt, Handling, and Holding Times
The reference laboratory was scheduled to analyze a total of 256 PCB samples (208 soil samples, 24 iso-octane
extract samples, and 24 methanol extract samples). Of these same samples, the developer was scheduled to
analyze a total of 232 PCB samples (208 soil samples and 24 extract samples in solvent of choice). The
samples were shipped to LAS at the start of the technology demonstration activities (July 22). Shipment was
coordinated through the SMO. Completion of chain-of-custody forms documented sample transfer. The
20
-------
samples were shipped on ice in coolers to maintain <6°C temperatures during shipment. Samples were shipped
with custody seals to ensure sample integrity and to prevent tampering during transport.
Upon receipt of the samples, the reference laboratory checked the receipt temperature and conditions of the
sample containers, assigned each sample a unique number, and logged each into its laboratory tracking system.
All samples were received at the proper temperature and in good condition. Demonstration samples were
divided into 11 analytical batches (with no more than 20 samples per batch). The samples were analyzed in an
order specified by ORNL to ensure that the analysis of sample types was randomized. Analyses of QC samples,
supplied by the reference laboratory to indicate method performance, were performed with each analytical
batch of soils.
Prior to analysis, samples were stored in refrigerators kept at 4 to 6°C to maintain analyte integrity. The
reference laboratory was required to analyze the extract samples and to extract the soil samples within 14 days
of shipment from ORNL. Once the soils were extracted, the reference laboratory had an additional 40 days to
analyze the soil extracts. Maximum holding times were not exceeded for any of the demonstration samples. The
final reference laboratory data package for all samples was received at ORNL in 72 days, on October 1, 1997.
The contractual obligation was 30 days.
The remainder of this section is devoted to summarizing the data generated by the reference laboratory and to
assessing the analytical performance.
Quality Control Results
Objective
The purpose of this section is to provide an assessment of the data generated by the reference laboratory's QC
procedures. The QC samples included continuing calibration verification standards (CCVs), instrument blanks,
method blanks, surrogate spikes, [laboratory control samples (LCSs)], and MS/MSD samples. Each control
type is described in more detail in the following text and in the technology demonstration plan [5]. Because
extraction of these liquid samples was not required, calibration check standards and instrument blanks were
the only control samples implemented for the extract samples. The reference laboratory's implementation of
QC procedures was consistent with SW-846 guidance.
Continuing Calibration Verification Standard Results
A CCV is a single calibration standard of known concentration, usually at the midpoint of the calibration range.
This standard is evaluated as an unknown and is quantified against the initial calibration. The calculated
concentration is then compared with the nominal concentration of the standard to determine whether the initial
calibration is still valid. CCVs were analyzed with every 10 samples or at least
every 12. The requirement for acceptance was a percentage difference of less than 15% for the CCV relative
to the initial calibration. This QC requirement was met for all Aroclors and surrogates, except for one standard
that had a 16% difference for DCB. These results indicated that the reference laboratory maintained instrument
calibrations during the course of sample analysis.
21
-------
Instrument and Method Blank Results
Instrument blanks (hexane) were analyzed prior to each CCV. The QC requirement was that instrument blanks
must contain less than the reporting detection limit for any analyte. All instrument blanks were acceptable.
A method blank is an analyte-free soil matrix sample that is taken through the extraction process to verify that
there are no laboratory sources of contamination. One method blank was analyzed for each analytical batch.
The QC requirement was that method blanks must contain less than the reporting detection limit for any
Aroclor. No PCBs were detected in any of the eleven method blanks that were analyzed. These results
demonstrated that the reference laboratory was capable of maintaining sample integrity, and that it did not
introduce PCB contamination to the samples during preparation.
Surrogate Spike Results
A surrogate is a compound that is chemically similar to the analyte group but is not expected to be present in
the environmental sample. A surrogate is added to test the extraction and analysis methods to verify the ability
to isolate, identify, and quantify a compound similar to the analyte(s) of interest without interfering with the
determination. Two different surrogate compounds, DCB and TCMX, were used to bracket the retention time
window anticipated in the Aroclor chromatograms. All soil samples, including QC samples, were spiked with
surrogates at 0.030 ppm prior to extraction. Surrogate recoveries were deemed to be within QC requirements
if the measured concentration fell within the QC acceptance limits that were established by past method
performance. (For LAS this was 39 to 117% for DCB, and 66 to 128% for TCMX). The results were
calculated using the following equation:
measured amount ,nnn,
percent recovery = x 100% (4-1)
actual amount
In all undiluted samples, both of the surrogates had percentage recoveries that were inside the acceptance limits.
Surrogate recoveries in diluted samples were uninformative because the spike concentration (0.030 ppm, as
specified by the method) was diluted below the instrument detection limits. The surrogate recovery results for
undiluted samples indicated that there were no unusual matrix interferences or batch-processing errors for these
samples.
Laboratory Control Sample Results
A LCS is an aliquot of a clean soil that is spiked with known quantities of target analytes. The LCS is spiked
with the same analytes and at the same concentrations as the matrix spike (MS). (MSs are described in the next
section.) If the results of the MS analyses are questionable (i.e., indicating a potential matrix effect), the LCS
results are used to verify that the laboratory can perform the analysis in a clean, representative matrix.
Aroclors 1016 and 1260 were spiked into the clean soil matrix at approximately 0.300 ppm, according to the
reference laboratory's SOP. The QC requirements (defined as percent recovery) for the LCS analyses were
performance-based acceptance limits that ranged from 50 to 158%. In all but one of the eleven LCSs analyzed,
both Aroclor percent recoveries fell within the acceptance limits. Satisfactory recoveries for LCS verified that
the reference laboratory performed the analyses properly in a clean matrix.
22
-------
Matrix Spike Results
In contrast to a laboratory control sample (LCS), a MS sample is an actual environmental soil sample into
which target analytes are spiked at known concentrations. MS samples are used to assess the efficiency of the
extraction and analytical methods for real samples. This is accomplished by determining the amount of spiked
analyte that is quantitatively recovered from the environmental soil. A duplicate matrix spike (MSB) sample
is spiked and analyzed to provide a measure of method precision. Ideally, to evaluate the MS/MSD results, the
environmental soil is analyzed unspiked so that the background concentrations of the analyte in the sample are
considered in the recovery calculation.
For the demonstration study samples, one MS and MSB pair was analyzed with each analytical batch. The MS
samples were spiked under the same conditions and QC requirements as the LCS (50 to 158% acceptance
limits), so that MS/MSD and LCS results could be readily compared. The QC requirement for MS and MSB
samples was a relative percent difference (RPD) of less than 30% between the MS/MSD pair. RPD is defined
as:
100% (4_2)
nnr, I MS recovery - MSD recovery
KrL) —
average recovery
A total of eleven MS/MSD pairs were analyzed. Because the MS/MSD spiking technique was not always
properly applied (e.g., a sample which contained 100 ppm of Aroclor 1254 was spiked ineffectively with 0.300
ppm of Aroclor 1260), many of the MS/MSD results were uninformative. For the samples that were spiked
appropriately, all MS/MSD QC criteria were met.
Conclusions of the Quality Control Results
The reference laboratory results met performance acceptance requirements for all of the samples where proper
QC procedures were implemented. Acceptable performance on QC samples indicated that the reference
laboratory was capable of performing analyses properly.
Data Review and Validation
Objective
The purpose of validating the reference laboratory data was to ensure usability for the purposes of comparison
with the demonstration technologies. The data generated by the reference laboratory were used as a baseline
to assess the performance of the technologies for PCB analysis. The reference laboratory data were
independently validated by ORNL and SMO personnel, who conducted a thorough quality check and reviewed
all sample data for technical completeness and correctness.
Corrected Results
Approximately 8% of the results provided by the reference laboratory (20 of 256) were found to have
correctable errors. So as not to bias the assessment of the technology's performance, errors in the reference
laboratory data were corrected. These changes were made conservatively, based on the guidelines provided in
the SW-846 Method 8081 for interpreting and calculating Aroclor results. The errors (see Appendix D, Table
D-3) were categorized as transcription errors, calculation errors, and interpretation errors. The corrections
23
-------
listed in Table D-3 were made in the final data set that was used for comparison with the demonstration
technologies.
Suspect Results
Normally, one would not know if a single sample result was "suspect" unless (1) the sample was a performance
evaluation sample, where the concentration is known or (2) a result was reported and flagged as suspect for
some obvious reason (e.g., no quantitative result was determined). The experimental design implemented in this
demonstration study provided an additional indication of the abnormality of data through the inspection of the
replicate results from a homogenous soil sample set (i.e., four replicates were analyzed for each sample ID).
Data sets were considered suspect if the standard deviation (SD) of the four replicates was greater than 30 ppm
and the percent relative standard deviation (RSD) was greater than 50%. Five data sets (sample IDs 106, 205,
216,217, 225) contained measurements that were considered suspect using this criteria, and the suspect data
are summarized in Table 4-1. A number of procedural errors may have caused the suspect measurements (e.g.,
spiking heterogeneity, extraction efficiencies, dilution, etc.). In the following subsections for precision and
accuracy, the data were evaluated with and without these suspect values to represent the best and worst case
scenarios.
Table 4-1. Suspect measurements within the reference laboratory data
Criteria
SD > 30 ppm
and
RSD > 50%
Qualitative Result
Sample ID
106
205
216
217
225
110
112
PCB Concentration (ppm)
Replicate Results
(ppm)
255.9,269.9,317.6
457.0,483.3,538.7
47.0, 54.3, 64.0
542.8, 549.8, 886.7
32.1,36.5,56.4
< reporting detection
limits
Suspect Result(s)
(ppm)
649.6
3,305.0
151.6
1,913.3
146.0
< 66, < 98, < 99, < 490
< 66, < 130, < 200,< 200
Data Usability
Performed data analysis with
and without this value
Used as special case for
comparison with developer
results
Samples that did not fall into the above criteria, but were also considered suspect, were non-blank samples that
could not be quantified and were reported as "< the reporting detection limit." This was the case for
environmental soil sample IDs 110 and 112. It is believed that the reference laboratory had trouble quantifying
these soil samples because of the abundance of chemical interferences. These samples were diluted by orders
of magnitude to reduce interferences, thereby diluting the PCB concentrations to levels that were lower than
the instrument detection limits. With each dilution, the reporting detection limits values were adjusted for
sample weight and dilution, which accounts for the higher reporting detection limits (up to 490 ppm). It is
believed that these samples should have been subjected to additional pre-analytical cleanup to remove these
24
-------
interferences before quantification was attempted. Sample IDs 110 and 112 were collected from the same
cleanup site (see Appendix B), so it is not surprising that similar difficulties were encountered with both sample
sets. Because the results for sample IDs 110 and 112 were not quantitative, these data were compared with the
technology data only on a special case basis.
Data Assessment
Objective
The purpose of this section is to provide an evaluation of the performance of the reference laboratory results
through statistical analysis of the data. The reference laboratory analyzed 72 PE, 136 environmental soil, and
48 extract samples. All reference laboratory analyses were performed under the same environmental
conditions. Therefore, site differentiation was not a factor in data assessment for the reference laboratory. For
comparison with the technology data, however, the reference laboratory data are delineated into "outdoor site"
and "chamber site" in the following subsections. For consistency with the technology review, results from both
sites were also combined to determine the reference laboratory's overall performance for precision and
accuracy. This performance assessment was based on the raw data compiled in Appendix D. All statistical tests
were performed at a 5% significance level.
Precision
The term "precision" describes the reproducibility of measurements under a given set of conditions. The SD
of four replicate PCB measurements was used to quantify the precision for each sample ID. SD is an absolute
measurement of precision, regardless of the PCB concentration. To express the reproducibility relative to the
average PCB concentration, RSD is used to quantify precision, according to the following equation:
„„„ Standard Deviation ,nnn,
RSD = x 100% (4-3)
Average Concentration
Performance Evaluation Samples
The PE samples were homogenous soils containing certified concentrations of PCBs. Results for these samples
represent the best estimate of precision for soil samples analyzed in the demonstration study. Table 4-2
summarizes the precision of the reference laboratory for the analysis of PE samples. One suspect measurement
(sample ID 225, 146.0 ppm) was reported for the PE soil samples. The RSDs for the combined data ranged
from 9 to 33% when the suspect measurement was excluded, and from 9 to 79%, including the suspect
measurement. The overall precision, determined by the mean RSD for all PE
25
-------
Table 4-2. Precision of the reference laboratory for PE soil samples
Outdoor Site
Sample
ID
126 a
118
124
120
122
119
125
121
123
Average
Concentration
(ppm)
0
1.6
1.7
5.0
11.1
20.1
37.9
54.6
60.1
SD
(ppm)
n/a
0.6
0.2
1.0
0.9
3.4
6.9
3.4
4.6
RSD
(%)
n/a
39
13
20
8
17
18
6
8
Chamber Site
Sample
ID
226
218
224
220
222
219
225
221
223
Average
Concentration
(ppm)
0
2.6
1.7
5.8
12.8
23.3
41.7"
44.9
55.8
SD
(ppm)
n/a
0.2
0.5
1.8
0.3
6.1
12.9"
11.3
7.7
RSD
(%)
n/a
6
29
31
3
26
3i
25
14
Combined Sites
Average
Concentration
(ppm)
0
2.1
1.7
5.4
11.9
21.7
39. 5C
49.8
58.0
SD
(ppm)
n/a
0.7
0.4
1.4
1.1
4.9
9.2 c
9.3
6.3
RSD
(%)
n/a
33
21
26
9
23
23 c
19
11
a All PCB concentrations were reported as non-detects.
b Results excluding the suspect value (results including the suspect value: mean = 67.8 ppm,
c Results excluding the suspect value (results including the suspect value: mean =52.8 ppm,
SD = 53.2 ppm, and RSD = 79%).
SD = 38.6 ppm, and RSD = 73%).
samples, was 21% for the worst case (including the suspect result) and 18% for the best case (excluding the
suspect result).
Environmental Soil Samples
The precision of the reference laboratory for the analysis of environmental soil samples is reported in Table
4-3. In this table, results including suspect measurements are presented in parentheses. Average concentrations
were reported by the reference laboratory as ranging from 0.5 to 1,196 ppm with RSDs that ranged from 7 to
118% when the suspect results were included. Excluding the suspect results, the highest average concentration
decreased to 660 ppm, and the largest RSD decreased to 71%. Because the majority of the samples fell below
125 ppm, precision was also assessed by partitioning the results into two ranges: low concentrations (<125
ppm) and high concentrations (>125 ppm). For the low concentrations, the average RSD was 23% excluding
the suspect value and 26% including the suspect value. These average RSDs were only slightly larger than the
RSDs for the PE soils samples of comparable concentration (18% for best case and 21% for worst case). Five
soil sample sets (sample IDs: 106, 117, 205, 211 and 217) were in the high-concentration category. The
average precision for high concentrations was 56% for the worst case and 19% for the best case. The precision
estimates for the low and high concentration ranges were comparable when the suspect values were excluded.
This indicated that the reference laboratory's precision for the environmental soils was consistent
(approximately 21% RSD), and comparable to the PE soil samples when the suspect values were excluded.
26
-------
The Paducah soils (indicated as bold sample IDs in Table 4-3) were analyzed by the technologies under both
outdoor and chamber conditions to provide a measure of the effect that two different environmental conditions
had on the technology's performance. Although this was not an issue for the reference laboratory (because all
the samples were analyzed under laboratory conditions), the reference laboratory's results were delineated into
the different site categories for comparison with the technologies. Sample IDs 113 and 201, 114 and 202, 115
and 203, 116 and 204, and 117 and 205 each represent a set of eight replicate samples of the same Paducah
soil. The RSDs for four of the five Paducah pairs (excluding the suspect value for sample ID 205) ranged from
11 to 17%. The result from one pair (sample IDs 113 and 201) had an RSD of 42%, but the reported average
concentration was near the reporting limits.
Table 4-3. Precision of the reference laboratory for environmental soil samples
Outdoor Site
Sample
ID
101
102
103
104
105
106
107
108
109
110
111
112
113 c
114
115
116
117
Average
Concentration
(ppm)
0.5
2.0
2.3
9.4
59.4
281.0 (373.2) a
1.3
1.8
2.0
b
38.7
b
1.1
1.3
14.8
41.3
383.9
Standard
Deviation
(ppm)
0.1
0.3
0.6
4.0
16.5
32.4(186.2)
0.3
0.1
0.4
b
4.3
b
0.6
0.3
1.8
5.9
55.2
RSD
(%)
16
16
27
43
28
12 (50)
20
8
20
b
11
b
55
20
12
14
14
Chamber Site
Sample
ID
206
207
208
209
210
211
212
213
214
215
216
217
201
202
203
204
205
Average
Concentration
(ppm)
1.9
18.8
30.5
40.2
88.6
404.5
3.2
8.1
25.2
26.7
55.1 (79.2)
659.8(973.2)
0.9
1.4
13.9
44.3
493.0(1196.0)
Standard
Deviation
(ppm)
0.9
3.5
7.9
28.5
25.6
121.8
1.6
1.6
3.7
3.2
8.5 (48.7)
196.6(647.0)
0.2
0.2
1.7
2.9
41.7(1406.4)
RSD
(%)
49
19
26
71
29
30
50
20
15
12
15 (62)
30 (66)
24
12
12
7
8(118)
a Data in parentheses include suspect values.
b Qualitative results only were reported for this sample.
c Bold sample IDs were matching Paducah sample pairs (i.e..
113/201, 114/202, 115/203, 116/204, 117/205).
27
-------
Extract Samples
The extract samples, which were used to simulate surface wipe samples, were the simplest of all the
demonstration samples to analyze because they required no extraction and were interference-free. Three types
of extract samples were analyzed: solvent blanks, spikes of Aroclor 1242 at 10 (jg/mL, and spikes of Aroclor
1254 at 100 (jg/mL. Identical extract samples were prepared in two solvents (iso-octane and methanol) to
accommodate the developer's request. The reference laboratory analyzed both solvent sets. A Student's t-test
[7, 8] was used to compare the reference laboratory's average PCB concentrations for the two different solvents
and showed that no significant differences were observed at either concentration. Therefore, the reference
laboratory results for the two extract solvents were combined. Additionally, all blank samples were quantified
as non-detects by the reference laboratory.
Table 4-4 summarizes the reference laboratory results for the extract samples by site. RSDs for the four
replicates for each sample ID ranged from 3 to 24%. For the combined data set (16 replicate measurements),
the average RSD at the lO-pg/mL level was 19%, while the average RSD at the 100-(jg/mL level was 8%. For
the entire extract data set, an estimate of overall precision was 14%. The overall precision for the extract
samples was comparable to the best-case precision for environmental soil samples (21%) and PE soil samples
(18%).
Table 4-4. Precision of the reference laboratory for extract samples
Outdoor Site
Sampl
elD
129 a
132 a
127
130
128
131
Average
Cone
(ug/mL)
0
0
10.9
12.1
67.4
63.8
SD
(ug/mL)
n/a
n/a
0.4
2.9
2.3
5.0
RSD
(%)
n/a
n/a
4
24
3
8
Chamber Site
Sample
ID
229
232
227
230
228
231
Average
Cone
(ug/mL)
0
0
9.6
8.9
65.2
57.7
SD
(ug/mL)
n/a
n/a
0.8
1.4
5.1
3.1
RSD
(%)
n/a
n/a
8
16
8
5
Combined Sites
Average
Cone
(ug/mL)
0
10.4
63.5
SD
(ug/mL)
n/a
1.9
5.2
RSD
(%)
n/a
19
8
1 All PCB concentrations reported as non-detects by the laboratory.
Accuracy
Accuracy represents the closeness of the reference laboratory's measured PCB concentrations to the accepted
values. Accuracy was examined by comparing the measured PCB concentrations (for PE soil and extract
samples) with the certified PE values and known spiked extract concentrations. Percent recovery was used to
quantify the accuracy of the results. The optimum percent recovery value is 100%. Percent recovery values
greater than 100% indicate results that are biased high, and values less than 100% indicate results that are
biased low.
28
-------
Performance Evaluation Soil Samples
The reference laboratory's performance for the PE samples is summarized in Table 4-5. Included in this table
are the performance acceptance ranges and the certified PCB concentration values. The acceptance ranges,
based on the analytical verification data, are guidelines established by the provider of the PE materials to gauge
acceptable analytical results. As shown in Table 4-5, all of the average concentrations were within the
acceptance ranges, with the exception of sample ID 218. The average result of sample ID 225 was outside of
the acceptance range only when the suspect result was included. All of the replicate measurements in sample
ID 225 were biased slightly high. Average percent recoveries for the PE samples (excluding suspect values)
ranged from 76 to 130%. Overall accuracy was estimated as the average recovery for all PE samples. The
overall percent recovery was 105% as a worst case when the suspect value was included. Excluding the suspect
value as a best case slightly lowered the overall percent recovery to 101%. A regression analysis [9] indicated
that the reference laboratory's results overall were unbiased estimates of the PE sample concentrations.
Table 4-5. Accuracy of the reference laboratory for PE soil samples
Certified
Concentration
(ppm)
(Acceptance
Range, ppm)
Oa
(n/a)
2.0
(0.7-2.2)
2.0
(0.9-2.5)
5.0
(2.1-6.2)
10.9
(4.0-12.8)
20.0
(11.4-32.4)
49.8
(23.0-60.8)
50.0
(19.7-63.0)
50.0
(11.9-75.9)
Outdoor Site
Sample
ID
126
118
124
120
122
119
125
121
123
Average
Cone
(ppm)
0
1.6
1.7
5.0
11.1
20.1
37.9
54.6
60.1
Recovery
(%)
n/a
79
85
99
102
100
76
109
120
Chamber Site
Sample
ID
226
218
224
220
222
219
225
221
223
Average
Cone
(ppm)
0
2.6
1.7
5.8
12.8
23.3
41.7"
44.9
55.8
Recovery
(%)
n/a
130
85
117
117
116
84"
90
112
Combined Sites
Average
Cone
(ppm)
0
2.1
1.7
5.4
11.9
21.7
39. 5 c
49.8
58.0
Recovery
(%)
n/a
105
85
108
109
109
79 c
100
116
a All PCB concentrations reported as non-detects by the laboratory.
b Results excluding the suspect value (results including the suspect value: average = 67.8 ppm and recovery = 136%).
'Results excluding the suspect value (results including the suspect value: average = 52.8 ppm and Recovery = 106%).
29
-------
Extract Samples
Percent recovery results for extract samples are summarized in Table 4-6 for the reference laboratory. The
average percent recoveries for extract samples ranged from 58 to 121%. In terms of concentration levels, the
average recovery at the lO-pg/mL level (for both solvents) was 104%, compared with 64% at the 100-(jg/mL
level. The reference laboratory classified all 16 samples spiked at 10 (jg/mL as Aroclor 1016; however, these
samples were actually spiked with Aroclor 1242. Despite this misclassification, the results did not appear to
be biased. In contrast, the samples spiked at 100 (ig/mL were correctly classified as Aroclor 1254 but were
all biased low. Although these results suggested that Aroclor classification had little effect on the quantification
of the extract samples, there was an obvious, consistent error introduced into the analysis of the 100-(jg/mL
samples to cause the low bias. For the entire extract data set, the overall percent recovery was 84%.
Table 4-6. Accuracy of the reference laboratory for extract samples
Spike
Concentration
(ug/mL)
Oa
Oa
10
10
100
100
Outdoor Site
Sample
ID
129
132
127
130
128
131
Avg
Cone
(ug/mL)
0
0
10.9
12.1
67.4
63.8
Recovery
(%)
n/a
n/a
109
121
67
64
Chamber Site
Sample
ID
229
232
227
230
228
231
Avg
Cone
(ug/mL)
0
0
9.6
8.9
65.2
57.7
Recovery
(%)
n/a
n/a
96
89
65
58
Combined Sites
Avg
Cone
(ug/mL)
n
in A
f,^ s
Recovery
(%)
n/a
104
64
a All PCB concentrations reported as non-detects by the laboratory.
Representativeness
Representativeness expresses the degree to which sample data accurately and precisely represent the capability
of the method. Representativeness of the method was assessed based on the data generated for clean-QC
samples (i.e., method blanks and laboratory control samples) and PE samples. Based on the data assessment
(discussed in detail in various parts of this section), it was determined that the representativeness of the
reference laboratory data was acceptable. In addition, acceptable performance on laboratory audits
substantiated that the data set was representative of the capabilities of the method. In all cases, the performance
of the reference laboratory met all requirements for both audits and QC analyses.
Completeness
Completeness is defined as the percentage of measurements that are judged to be usable (i.e., the result was
not rejected). Usable results were obtained for 248 of the 256 samples submitted for analysis by the reference
laboratory. Eight results (for sample IDs 110 and 112) were deemed incomplete and therefore not valid because
the measurements were not quantitative. To calculate completeness, the total number of complete results were
divided by the total number of samples submitted for analysis, and then multiplied by 100 to express as a
30
-------
percentage. The completeness of the reference laboratory was 97%, where a completeness of 95% or better is
typically considered acceptable.
Comparability
Comparability refers to the confidence with which one data set can be compared with another. The
demonstration study was designed to have a one-to-one, sample-by-sample comparison of the PCB results
obtained by the reference laboratory and the PCB results obtained by the technology being evaluated. Based
on thorough examination of the data and acceptable results on the PE samples, it was concluded that the
reference laboratory's SOPs for extraction and analysis, and the data generated using these procedures, were
of acceptable quality for comparison with the field technology results. Additional information on comparability
was available because the experimental design incorporated randomized analysis of blind, replicate samples.
Evaluation of the replicate data implicated some of the individual data points as suspect (see Table D-2). The
reference laboratory's suspect data were compared with the technology data on a special-case basis, and
exceptions were noted.
Summary of Observations
Table 4-7 provides a summary of the performance of the reference laboratory for the analysis of all sample
types used in the technology demonstration study. As shown in Table 4-7, the precision of the PE soils was
comparable to the environmental soils. A weighted average, based on the number of samples, gave a best-case
precision of 21% and a worst-case precision of 28% for all the soil data (PE and environmental). The extract
samples had a smaller overall RSD of 14%. Evaluation of overall accuracy was based on samples with certified
or known spiked concentrations (i.e., PE and extract samples). The overall accuracy, based on percent
recovery, for the PE samples was 105% for the worst case (which included the suspect value) and 101% for
the best case (which excluded the suspect value). These results indicated that the reference laboratory measured
values were unbiased estimates of the certified PE concentrations (for samples that contained <50 ppm of
PCBs). Accuracy for the extract samples at 10 ppm was also unbiased, with an average percent recovery of
104%. However, the accuracy for the extract samples at 100 ppm was biased low, with an average recovery
of 64%. Overall, the average percent recovery for all extract samples was 84%. The reference laboratory
correctly reported all blank samples as non-detects, but had difficulty with two soil sample IDs (110 and 112)
that contained chemical interferences. In general, the reference laboratory's completeness would be reduced,
at the expense of an improvement in precision and accuracy, if the suspect measurements were excluded from
the data analysis. Based on this analysis, it was concluded that the reference laboratory results were acceptable
for comparison with the developer's technology.
31
-------
Table 4-7. Summary of the reference laboratory performance
Sample Matrix
Blank
Environmental soil with
interferences
Soil
Best Case
(excluding suspect data)
Soil
Worst Case
(including suspect data)
Extract
Sample Type
Soil
Extract
Sample ID 110
Sample ID 112
PE
Environmental
< 125 ppm
> 125 ppm
overall
PE
Environmental
< 125 ppm
> 125 ppm
Overall
10 ppm
100 ppm
Overall
Number of Samples
8
16
4
4
63
107
17
187
64
108
20
192
16
16
32
Precision
(Average % RSD)
n/aa
n/aa
18
23
19
21
21
26
56
28
19
8
14
Accuracy
(Average %Recovery)
All samples were
reported as non-detects.
All samples were
reported as non-detects.
101
n/ab
n/ab
101
105
n/ab
n/ab
105
104
64
84
a Because the results were reported as non-detects, precision assessment is not applicable.
b Accuracy assessment calculated for samples of known concentration only.
32
-------
Section 5
Technology Performance and Evaluation
Objective and Approach
This section presents the evaluation of the data generated by the D TECH PCB test kit. The technology's
precision and accuracy performance are presented for the data generated in the demonstration study. In
addition, an evaluation of comparability, through a one-to-one comparison with the reference laboratory data,
is presented. Evaluations of other aspects of the technology (such as cost, sample throughput, hazardous waste
generation, and logistical operation) are also presented in this section.
Interval Reporting
The D TECH results were reported as concentration ranges that were designated as intervals incorporating
parentheses/bracket notation. The parentheses indicate that the end points of the concentration range were
excluded, while brackets indicate that the end points were included. For example, in Table 5-1 the interval (1,
4] indicates that the PCB concentration range is >1 and <4.
As discussed briefly in Section 2 of this report, this technology cannot distinguish between different Aroclors.
The test kit has been calibrated to respond in a one-to-one ratio with Aroclor 1254. If site history or
information indicates that Aroclor 1242 or 1248 is present in the sample, a conversion can be applied (see
Section 2 for more information). The Aroclor-specific reporting intervals for the D TECH results are listed in
Table 5-1. For the purposes of the demonstration, SDI was provided information about the type of Aroclor
present in the samples. Dilution of samples during analysis to optimize method performance altered some of
the standard intervals shown in Table 5-1 for select samples.
Table 5-1. D TECH PCB test kit reporting intervals
Default Mode
Interval
[0,0.5)
[0.5, 1]
(1,4]
(4, 15]
(15,25]
(25, »)
Aroclor 1254
Concentration Range
0< PCBppm<0.5
0.5 25
Conversion to Secific Aroclor
Interval
[0,1.5)
[1.5,3.5]
(3.5,20]
(20, 54]
(54, 100]
(100,°°)
Aroclor 1242
Concentration Range
0< PCB ppm < 1.5
1.5 100
Interval
[0,0.8)
[0.8,2.3]
(2.3,11]
(11,28]
(28, 53]
(53, »)
Aroclor 1248
Concentration Range
0< PCB ppm < 0.8
0.8 < PCB ppm < 2.3
2. 3 < PCB ppm < 11
1 1 < PCB ppm < 28
28 < PCB ppm < 53
PCB ppm > 53
33
-------
Data Assessment
Objective
The purpose of the data assessment section is to present the evaluation of the performance of SDFs D TECH
PCB test kit through a statistical analysis of the data. PARCC parameters were used to evaluate the test kit's
ability to measure PCBs in PE, environmental soil, and extract samples. The developer analyzed splits of
replicate samples that were also analyzed by the reference laboratory (72 PE soil samples, 136 environmental
soil samples, and 24 extract samples). See Section 4 for a more detailed analysis of the reference laboratory's
results. Replicate samples were analyzed by the developer at two different sites—under outdoor conditions and
inside an environmentally controlled chamber—to evaluate the effect of environmental conditions on the test
kit's performance; see Section 3 for further details on the different sites. Evaluation of the measurements made
at each site indicated that there were no significant difference between the two data sets. Because environmental
conditions did not appear to affect the results significantly, data from both sites were also combined for each
parameter (precision and accuracy) to determine the test kit's overall performance. All statistical tests were
performed at the 5% significance level. Appendix D contains the raw data that were used to assess the
performance of the D TECH test kit.
Precision
Precision is the reproducibility of measurements under a given set of conditions. The frequency with which the
same interval was reported within a set of replicates was used to quantify precision. Examples of how the
precision was classified are presented in Table 5-2. Reporting a higher number of replicates in the same interval
for a given replicate set indicates higher precision. In other words, reporting all four replicate results as the
same interval indicates the highest possible precision.
Table 5-2. Classification of precision results
If the replicate results are . . .
[0, 1.5), [1.5, 3.5], (3.5, 20], (20, 54]
[0, 1.5), [1.5, 3.5], [1.5, 3.5], (3.5, 20]
[0, 1.5), [1.5, 3.5], [1.5, 3.5], [1.5, 3.5]
[1.5, 3.5], [1.5, 3.5], [1.5, 3.5], [1.5, 3.5]
. . . and the number
reported in identical
intervals are . . .
0
2
3
4
. . . then the precision
classification is ...
None
Low
Medium
High
Performance Evaluation Samples
Table 5-3 summarizes the precision information for the D TECH test kit's analysis of the PE samples. The D
TECH test kit reported all four replicates as the same interval (i.e., high precision) for two PE sample sets
under the outdoor conditions but achieved the highest precision category for six of eight sample sets under the
chamber conditions. Therefore, the test kit performed more precisely under the chamber conditions. Operating
under the outdoor conditions, five of eight replicate sets were classified as low precision (i.e., two of four
replicates were reported in the same interval). None of the replicate sets
34
-------
Table 5-3. Precision of the D TECH PCB test kit for PE soil samp
Certified
PE
Cone.
(ppm)
0
2.0
2.0
5.0
10.9
20.0
49.8
50.0
50.0
Outdoor Site
Sample
ID
126"
118
124
120
122
119
125
121
123
# in each precision
classification
Precision
high
# of Replicates Reported in Identical
Intervals
Oa
0
2
X
X
X
X
X
X
6
3
X
1
4
X
X
2
les
Chamber Site
Sample
ID
226"
218
224
220
222
219
225
221
223
Precision
high
# of Replicates Reported in Identical
Intervals
Oa
0
2
X
X
2
3
X
X
2
4
X
X
X
X
X
5
a Indicates that all four replicates were reported as different intervals.
b Blank data were not included in the determination of the overall precision.
were reported with the lowest precision. A more detailed analysis of the data showed that the replicates having
medium to low precision classifications were more than one interval away from the most frequently reported
interval for sample IDs 120, 121, 122, and 124.
Environmental Soil Samples
The D TECH results for the replicate environmental soil sample measurements are presented in Table 5-4. The
improved precision under the chamber conditions as described above for the PE samples was not observed for
the environmental soils. Under the outdoor conditions, 3 of 17 replicate sets achieved the highest precision
classification (i.e., the same interval was reported for all four replicates). Under the chamber conditions, 4 of
17 sample sets were classified as high precision. Under each set of conditions, 9 of 17 samples were classified
as low precision. The D TECH test kit reported four different intervals for sample ID 112. A more detailed
analysis of the data showed that, of the sample sets where precision was classified as medium to low, nine
replicate results (i.e., sample IDs 103, 105, 106, 111,112, 204, 208, 209, and 211) differed by more than one
interval range.
Because most of the measurements fell below 125 ppm, precision was also assessed by partitioning the results
into two ranges: low concentration (reference laboratory values <125 ppm) and high concentration (reference
35
-------
laboratory values >125 ppm). See Section 4 for the delineation of which sample IDs were in the low and high
categories. For the low concentrations, 20% of the sample sets were reported with all four replicates in the
same interval (i.e., highest possible precision). For the high concentration category, 40% of the sample sets
(two of five) were reported with the highest possible precision.
Table 5-4. Precision of the D TECH PCB test kit for environmental soil samples
Outdoor Site
Sample
ID
101
102
103
104
105
106
107
108
109
110
111
112
113"
114
115
116
117
# in each
precision
classification
Precision
high
# of Replicates Reported in Identical
Intervals
Oa
X
1
2
X
X
X
X
X
X
X
X
X
9
3
X
X
X
X
4
4
X
X
X
3
Chamber Site
Sample
ID
206
207
208
209
210
211
212
213
214
215
216
217
201"
202
203
204
205
Precision
# of Replicates Reported in Identical
Intervals
Oa
0
2
X
X
X
X
X
X
X
X
X
9
3
X
X
X
X
4
4
X
X
X
X
4
a Indicates that all four replicates were reported as different intervals.
b Bold sample IDs were matching Paducah sample pairs (i.e., 113/201, 114/202, 115/203, 116/204, 117/205).
36
-------
The Paducah soils (indicated by bold sample IDs in Table 5-4) were analyzed at both sites to provide an
assessment of the D TECH's performance under different environmental conditions. For these samples,
the data generated under both environmental conditions were also combined to provide an overall assessment
of precision. Sample IDs 113 and 201, 114 and 202, 115 and 203, 116 and 204, and 117 and 205 represented
replicate Paducah soil sample sets, in which the 100 series were samples analyzed under the outdoor conditions
and the 200 series were samples analyzed inside the chamber. Additional statistical analysis was used to
compare the effect of the two environmental conditions on the measurements. Results from this analysis showed
that there were no significant differences in the data generated at each site. This indicated that these different
environmental conditions did not have an impact on the performance of the test kit.
Extract Samples
The D TECH results for the replicate extract measurements are presented in Table 5-5. Under the outdoor
conditions, the 10- and 100-(jg/mL sample sets were reported with medium and high precision, respectively,
and medium and low precision under the chamber conditions. Of the sample sets reported with medium and
low precision, all differed by more than one interval range. Under both testing conditions, the blank samples
were reported with medium precision.
Table 5-5. Precision of the D TECH PCS test kit for extract samples
Outdoor Site
Sample ID
130"
131
132
# in each
precision
classification
Precision
# of Replicates Reported in Identical
Intervals
Oa
0
2
0
3
X
X
2
4
X
1
Chamber Site
Sample
ID
230"
231
232
Precision
high
# of Replicates Reported in Identical
Intervals
Oa
0
2
X
1
3
X
X
2
4
0
a Indicates that all four replicates were reported as different intervals.
b Blank data were not included in the determination of the overall precision.
Precision Summary
A summary of the overall precision of the D TECH test kit is presented by sample type (PE, environmental soil,
and extract samples) in Table 5-6. For PE and environmental soil samples, 44% and 21% of the samples,
respectively, achieved the highest possible precision (i.e., all four samples replicates were reported as the same
interval). For the extract samples, 25% of the samples achieved the highest precision.
37
-------
Table 5-6. Overall precision of the D TECH PCS test kit for all sample types
Environmental
Site
Outdoor Site
Chamber Site
Combined
Sites
Percentage of Samples Classified in Each Precision Category
PE Samples
None
0
0
0
Low
63
25
44
Med
13
13
13
High
25
63
44
Environmental Soil Samples
None
6
0
3
Low
53
53
53
Med
24
24
24
High
18
24
21
Extract Samples
None
0
0
0
Low
0
50
25
Med
50
50
50
High
50
0
25
Accuracy
Accuracy represents the closeness of the D TECH test kit's measured PCB concentrations to the certified
values. Because the D TECH test kit produced interval results, accuracy was evaluated in terms of the
percentage of samples which agreed with, were above (i.e., biased high), and were below the certified value
(i.e., biased low).
Performance Evaluation Soil Samples
Table 5-7 compares the D TECH interval results with the corresponding certified PE values. The listing of an
interval in a particular column indicates how many of the four replicates were reported as that interval. For
example, for sample ID 220, two replicates were reported as (4, 15], and two were reported as (15, 25]. For
sample ID 124, three are reported as (1,4], and one is reported as [0, 0.5). Table 5-7 also presents performance
acceptance ranges for the PE results, which are the guidelines established by the provider of the PE materials
to gauge acceptable analytical results. These ranges were not used to evaluate the D TECH results because
the acceptance ranges overlap several D TECH reporting intervals.
The data in Table 5-7 were used to derive the accuracy results presented in Table 5-8. Accuracy was based
on a comparison of the certified PE value with the interval reported by the D TECH test kit. If the interval
encompassed the certified PE value, the D TECH result "agreed" with the certified value. If the D TECH result
was above the certified value, the result was classified as "biased high." If the D TECH result was below the
certified value, the result was classified as "biased low." For example, for sample ID 219, the certified value
was 20 ppm (for Aroclor 1248). The comparison would be classified as "agreed" for the D TECH interval
result (11,28], as "biased high" for the interval result (28, 53], and as "biased low" for the interval result (2.3,
11]. Separate comparisons were made for the two environmental conditions to determine if ambient temperature
and humidity had an effect on the performance of the technology. Statistical analysis showed that there was
no significant difference between the results obtained by the test kit under the two different environmental
conditions evaluated in this demonstration. Therefore, all PE sample results were combined to determine the
overall percentage of agreement between the D TECH results and the certified PE value. The overall percentage
of agreement was 56%. Approximately 17% of the D TECH sample results were biased low. A total of 28%
of the results from the test kit were biased high.
38
-------
Table 5-7. D TECH test kit accuracy data for PE soil samples
Certified
Cone.
(ppm)
(Acceptance
Range, ppm)
0
(n/a)
2.0
(0.7-2.2)
2.0
(0.9-2.5)
5.0
(2.1-6.2)
10.9
(4.0-12.8)
20.0
(11.4-32.4)
49.8
(23.0-60.8)
50.0
(19.7-63.0)
50.0
(11.9-75.9)
Outdoor Site
Sample
ID
126
118
124
120
122
119
125
121
123
# of Replicates Reported at Each
Interval
1
[0,0.5)
0,4]
(4, 15]
[0.5, 1]
(15,25]
(4, 15]
(15,25]
2
[0,0.5)
[0.5, 1]
[0.5, 1]
[4, 15]
(2.3,11]
(11,28]
(15,25]
(25, co)
(25, oo)
3
(1,4]
4
(2.3,11]
(25, oo)
Chamber Site
Sample
ID
226
218
224
220
222
219
225
221
223
# of Replicates Reported at Each Interval
1
[0,0.5)
[0.8,2.3]
(2.3,11]
(28, 53]
2
(4, 15]
(15,25]
(11,28]
3
[0.5, 1]
(2.3,11]
4
0,4]
(25, oo)
(25, oo)
(25, oo)
(25, oo)
Table 5-8. Evaluation of agreement between D TECH's PE sample results and the certified PE values
as a measure of accuracy
Environmental
Site
Outdoor Site
Chamber Site
Combined Sites
Relative to Certified Values for Performance Evaluation
Samples
Biased
Low
31%
3%
17%
Agree
50%
61%
56%
Biased
High
19%
36%
28%
Number of Samples
36
36
72
39
-------
Extract Samples
Table 5-9 compares the D TECH interval results with the corresponding spike concentrations for the extract
samples. The D TECH test kit's percentage of agreement with the spike concentration of the extract samples
is summarized in Table 5-10. Statistical analysis showed that environmental conditions had no significant effect
upon the performance of the D TECH test kit. Therefore, the data sets generated under the outdoor and
chamber conditions were combined. Overall, 10 of 24 extract samples (42%) agreed with the spike
concentration. Approximately 29% were biased high, and 29% were biased low. All biased-high results were
from samples analyzed under the chamber conditions, and all biased-low results were from samples analyzed
under the outdoor conditions.
Table 5-9. Accuracy of the D TECH test kit for extract samples
Spike Cone.
(ug/mL)
0
10
100
Outdoor Site
Sample
ID
132
130
131
# of Replicates Reported at Each
Interval
1
(3.5,20]
[0.5, 1]
2
3
[1.5,3.5]
(15,25]
4
[0,0.5]
Chamber Site
Sample
232
230
231
# of Replicates Reported at Each
Interval
1
[0.5, 1]
(4, 15]
(54, 100]
(100,»)
2
[0,0.5)
3
(20, 54]
(25,oo)
4
Table 5-10. Evaluation of agreement between D TECH's extract results and the spike concentration
as a measure of accuracy
Site
Outdoor Site
Chamber Site
Combined Sites
Relative to Spike Concentration for Extract Samples
Biased
Low
58%
0%
29%
Agree
42%
42%
42%
Biased
High
0%
58%
29%
Number of Samples
12
12
24
False Positive/False Negative Results
A false positive (fp) result [10] is one in which the technology detects PCBs in the sample when there actually
are none. A false negative (fh) result [10] is one in which the technology indicates that there are no PCBs
present in the sample, when there actually are. Both fp and fh results are influenced by the method detection
limit of the technology. Of the eight blank soil samples, five were reported as [0.5, 1], so the fp result was 62%.
Of the 192 non-blank soil samples analyzed, D TECH reported one in the lowest reporting interval (e.g., 0 to
0.5 ppm), when the corresponding reference laboratory result was 1.8 ppm. Therefore, the fn result for the soil
40
-------
samples was 0.5%. For the eight blank extract samples, the D TECH test kit reported one blank as [0.5, 1] and
one as (4, 15]. Therefore, the fp result was 25%. All other extract samples were reported as non-blanks, so the
m result was 0%.
Representativeness
Representativeness expresses the degree to which the sample data accurately and precisely represent the
capability of the technology. The performance data was accepted as being representative of the technology
because the D TECH test kit was capable of analyzing diverse samples types (PE samples, simulated wipe
extract samples, and actual field environmental samples) under multiple environmental conditions. When this
technology is used, quality control samples should be analyzed to assess the performance of the D TECH PCB
test kit under the testing conditions.
Completeness
Completeness is defined as the percentage of measurements that are judged to be useable (i.e., the result was
not rejected). Useable results were obtained by the technology for all 232 samples. Therefore, completeness
was 100%.
Comparability
Comparability refers to the confidence with which one data set can be compared to another. A one-to-one
sample comparison of the D TECH results and the reference laboratory results was performed for all soil
samples. Accuracy was evaluated in terms of the percentage of samples that agreed with, were above (i.e.,
biased high), and were below (i.e., biased low) the certified value. For comparability, the D TECH results were
compared with the results generated by the reference laboratory, including both environmental soils and PE
samples. Sample IDs 110 and 112 were excluded because the reference laboratory did not generate quantitative
results forthese samples. The results are summarized in Table 5-11. The percentage of D TECH results that
agreed with the reference laboratory results was 53%. Approximately 28% were biased high, while
approximately 20% were biased low relative to the results reported by the reference laboratory.
For the extract samples, the comparison of the D TECH test kit's result with the reference laboratory result
was the same as the comparison with the spike concentrations (previously discussed in Table 5-10). There was
42% agreement between the laboratory method and the field technology, 29% of the D TECH results were
biased high, and 29% were biased low.
The soil data not included in previous comparability evaluations (because the replicate data for the reference
laboratory was considered suspect) are shown in Table 5-12. (Refer to Section 4, in particular Table 4-1, for
more information on the reference laboratory's suspect measurements.) The reference laboratory's suspect data
were compared with the D TECH's matching results. For sample IDs 110 and 112, the reference laboratory
obtained qualitative results only. The D TECH test kit also had some difficulty with sample IDs 110 and 112,
producing replicate results in multiple intervals. For the other five suspect values for the reference laboratory
data, three of the D TECH test kit results agreed with the replicate means of the reference laboratory; two of
the D TECH results were biased low relative to the reference laboratory replicate means. These comparisons
demonstrated that the D TECH test kit had difficulty with most of the samples that were troublesome for the
reference laboratory.
41
-------
Table 5-11. Evaluation of agreement between D TECH's soil results and the reference laboratory's
results as a measure of comparability
Environmental
Site
Outdoor Site
Chamber Site
Combined Sites
Relative to Reference Laboratory Results for Soil Samples
Biased
Low
27%
13%
20%
Agree
46%
59%
53%
Biased
High
27%
29%
28%
Number of Samples
96
104
200
Table 5-12. Comparison of the D TECH results with the reference laboratory's suspect measurements
Sample ID
110
112
106
205
216
217
225
Reference Laboratory
Suspect Measurement
(ppm)
-------
Table 5-13. D TECH PCB test kit performance for precision, accuracy, and comparability
Sample Type
PE
Environmental
Soil
Extract
Precision a
% High
Precision
44
25
21
Accuracy b
%
Biased
Low
17
n/ab
29
%in
Agreement
56
n/a
42
% Biased
High
28
n/a
29
Comparability c
% Biased
Low
19
20
29
%in
Agreement
57
50
42
% Biased
High
24
31
29
a Percentage of sample sets that achieved highest precision (i.e., all four replicates were reported as the same interval).
b D TECH result versus certified value; accuracy cannot be assessed for environmental soils.
c D TECH result versus reference laboratory result.
percentage in agreement ranged from 42 to 57%, the percentage biased high was 24 to 29%, and the percentage
biased low was 17 to 29%.
Regulatory Decision-Making Applicability
One objective of this demonstration was to assess the technology's ability to perform at regulatory decision-
making levels for PCBs, specifically 50 ppm for soils and 100 (jg/100cm2 for surface wipes. To assess this
ability, the D TECH test kit's performance for PE and environmental soil samples ranging in concentration
from 40 to 60 ppm (as determined by the paired reference laboratory analyses) can be used. For this
concentration range, the test kit's results agreed with the reference laboratory's results 61% of the time. Results
were biased high 7% of the time and biased low 32% of the time.. No false negatives were observed for this
concentration range. Of the 24 PE samples that had a nominal concentration of 50 ppm, over 80% of the
samples were reported correctly as (25, °°); the other results were biased low.
Assuming a 10 mL extract volume, extract samples (at 10 and 100 (jg/mL) represented surface wipe sample
concentrations of 100 (jg/100cm2 and 1000 (jg/100cm2. For simulated wipe extract samples, the percentage
of the D TECH's measurements that agreed with the reference laboratory results was 42%. Approximately
29% of the results were biased high and 29% were biased low. No false negative results were observed for the
extract samples.
Additional Performance Factors
Sample Throughput
Sample throughput is representative of the estimated amount of time required to extract the PCBs, to perform
appropriate reactions, and to analyze the sample. Operating under the outdoor conditions, the SDI team's
sample throughput rate was 15 samples/hour. For work in the chamber, the rate was lower, around 11
samples/hour. The higher sample throughput under the outdoor conditions may be attributed to the analysis
order; because the SDI staff analyzed samples under the chamber conditions first, they may have gained
valuable experience that was applied during the analysis of the outdoor samples. Alternatively, SDI analysts
may have had more difficulty with the sample matrices that were analyzed only under the outdoor conditions.
43
-------
Cost Assessment
The purpose of this economic analysis is to provide an estimation of the range of costs for an analysis of PCB-
contaminated soil samples using the D TECH PCB test kit and a conventional analytical reference laboratory
method. The analysis was based on the results and experience gained from this demonstration, costs provided
by SDI, and representative costs provided by the reference analytical laboratories who offered to analyze these
samples. To account for the variability in cost data and assumptions, the economic analysis was presented as
a list of cost elements and a range of costs for sample analysis by the D TECH test kit and by the reference
laboratory.
Several factors affected the cost of analysis. Where possible, these factors were addressed so that decision-
makers can independently complete a site-specific economic analysis to suit their needs. The following
categories are considered in the estimate:
• sample shipment costs,
• labor costs,
• equipment costs,
• waste disposal costs.
Each of these cost factors is defined and discussed below and serves as the basis for the estimated cost ranges
presented in Table 5-14. This analysis assumed that the individuals performing the analyses were fully-trained
to operate the technology. SDI does not offer a specific training course on the use of the D TECH kit, but does
provide free assistance, on an as-needed basis, through its technical service department. Sample acquisition
and pre-analytical sample preparation, which are tasks common to both methods, were costs that were not
included here.
D TECH Costs
Sample shipment costs. Because the samples were analyzed on-site, no sample shipment charges were
associated with the cost of operating the D TECH test kit.
Labor costs. Labor costs included mobilization/demobilization, travel, per diem, and on-site labor.
— Labor mobilization/demobilization: This cost element included the time for one person to
prepare for and travel to each site. The estimate ranged from 5 to 8 hours, at a rate of $50 per
hour.
— Travel: This element was the cost for the analyst(s) to travel to the site. If the analyst is
located near the site, the cost of commuting to the site (estimated to be 50 miles at $0.30 per
mile) would be minimal ($15). The estimated cost of an analyst traveling to the site for this
demonstration ($1,000) included the cost of airline travel and rental car fees.
— Per diem: This cost element included food, lodging, and incidental expenses, and was
estimated ranging from zero (for a local site) to $150 per day per analyst.
44
-------
Table 5-14. Estimated analytical costs for PCB soil samples
D TECH PCB test kit
Strategic Diagnostics, Inc.
Sample throughput rate: 15
11
Cost Category
Sample Shipment
Labor
Mobilization/demobilization
Travel
Per diem
Rate
Equipment
Mobilization/demobilization
Photometer purchase price
Reagents/supplies
Waste Disposal
samples per hour (outdoors)
samples per hour (chamber)
Cost (S)
0
5CMOO
15-1 000 per analyst
0-150 per day per analyst
30-75 per hour per analyst
0-150
299
36 per
sample
75-1060
EPA SW-846 Method
8080/8081/8082
Reference Laboratory
Typical turn-around time: 14-30 days
Cost Category
Sample Shipment
Labor
Overnight shipping charges
Labor
Mobilization/demobilization
Travel
Per diem
Rate
Equipment
Mobilization/demobilization
Rental/purchase of system
Reagents/supplies
Waste Disposal
Cost (S)
100-200
50-150
Included3
Included
Included
44-239 per
Included
Included
Included
sample
Included
a "Included" indicates that the cost is included in the labor rate.
— Rate: The cost of the on-site labor was estimated at a rate of $30 to $75 per hour, depending
on the required expertise level of the analyst. This cost element included the labor involved
with the entire analytical process comprising sample preparation, sample management,
analysis, and reporting.
Equipment costs. Equipment costs included mobilization/demobilization, rental fees or purchase of
equipment, and the reagents and other consumable supplies necessary to complete the analysis.
— Equipment mobilization/demobilization: This included the cost of shipping the equipment to
the test site. If the site is local, the cost would be zero. For this demonstration, the cost of
shipping equipment and supplies was estimated at $150.
— Purchase: At the time of the demonstration, the cost of purchasing the photometer was $299.
— Reagents/supplies: These items are consumable and are purchased on a per sample basis. At
the time of the demonstration, the cost of the reagents and supplies needed to prepare and
analyze PCB soil samples using the D TECH PCB test kit was $36 per sample. This cost
included the sample preparation supplies, assay supplies, and consumable reagents.
Waste disposal costs. Waste disposal costs are estimated based on the 1997 regulations for disposal
of PCB-contaminated waste. Using the D TECH test kit, SDI generated approximately 20 Ib of vials
containing soils and liquid solvents (classified as solid PCB waste suitable for disposal by incineration)
and approximately 20 Ib of other solid PCB waste (used and unused soil, gloves, paper towels,
45
-------
ampules, etc.). The cost of disposing of PCB solid waste by incineration at a commercial facility was
estimated at $1.50/lb. The cost for solid PCB waste disposal at ETTP was estimated at $18/lb. The
test kit also generated approximately 19 Ib of liquid waste. The cost for liquid PCB waste disposal at
a commercial facility was estimated at $0.25/lb, while the cost at ETTP was estimated at $11/lb.
Reference Laboratory Costs
• Sample shipment costs. Sample shipment costs to the reference laboratory included overnight
shipping charges, as well as labor charges associated with the various organizations involved in the
shipping process.
— Overnight shipping: The overnight express shipping service cost was estimated to be $50 for
one 50-lb cooler of samples.
— Labor: This cost element included all of the tasks associated with the shipment of the samples
to the reference laboratory. Tasks included packing the shipping coolers, completing the chain-
of-custody documentation, and completing the shipping forms. Because the samples contained
PCBs, the coolers were inspected by qualified personnel to ensure acceptance with the U.S.
Department of Transportation's shipping regulations for PCBs. The estimate to complete this
task ranged from 2 to 4 h at $50 per hour.
Labor, equipment, and waste disposal costs. The labor bids from commercial analytical reference
laboratories who offered to perform the PCB analysis for this demonstration ranged from $44 per
sample to $239 per sample. The bid was dependent on many factors, including the perceived difficulty
of the sample matrix, the current workload of the laboratory, and the competitiveness of the market.
In this case, the wide variation in bids may also be related to the cost of PCB waste disposal in a
particular laboratory's state. LAS Laboratories was awarded the contract to complete the analysis as
the lowest qualified bidder ($44 per sample). This rate was a fully loaded analytical cost that included
labor, equipment, waste disposal, and report preparation.
Cost Assessment Summary
An overall cost estimate for the D TECH test kit vs the reference laboratory was not made due to the extent
of variation in the difference cost factors, as outlined in Table 5-14. The overall costs for the application of
each technology will be based on the number of samples requiring analysis, the sample type, and the site
location and characteristics. Decision-making factors, such as turnaround time for results, must also be
weighed against the cost estimate to determine the value of the field technology vs the reference laboratory.
46
-------
General Observations
The following are general observations regarding the field operation and performance of the D TECH test kit:
• The system was light, easily transportable, and rugged. It took about one hour for the SDI
team to prepare to analyze samples on the first day of testing. While working at the outdoor
site, the SDI team completely disassembled their work station, bringing everything inside at
the close of each day. It took the team less than one hour each morning to prepare for sample
analyses.
• Three operators were used for the demonstration because of the number of samples and the
working conditions, but the technology can be operated by a single person. With three SDI
technologies (D TECH, EnviroGard, and RaPID Assay) being demonstrated, SDI elected to
work as a team to complete the analyses for each technology (as opposed to three SDI people
working with three different technologies).
• Operators generally require 2 to 4 h of training and should have a basic knowledge of field
analytical techniques.
Data processing and interpretation was minimal. The results were quantified relative to the
four calibration standards and reported in terms of intervals using the sample information
provided (Aroclortype and ratio).
New start and stop solutions were used with every four samples. All reagents were allowed
to come to room temperature before use. Although it is recommended that all of the reagents
in the test kit be stored under refrigerated conditions, the SDI team noted that the reagents can
be stored at ambient conditions for several hours.
• The development times for the assay were altered as outlined in the D TECH protocol. The
times, which were usually shorter as temperature increased, were determined on the basis of
the assay of the reference standard. When the color development in the reference standard read
the correct range, the assay was stopped. Temperature measurements and stop times were not
recorded.
• The D TECH test kit generated approximately 20 Ib of vials containing soils and liquid
solvents (classified as incinerable solid PCB waste) and approximately 20 Ib of other solid
PCB waste (used and unused soil, gloves, paper towels, ampules, etc.). The test kit also
generated approximately 19 Ib of liquid waste (aqueous with trace methanol).
Performance Summary
A summary of the performance characteristics of SDFs D TECH PCB test kit, presented previously in this
chapter, is shown in Table 5-15. The performance of D TECH test kit was characterized as biased, because
nearly half (44%) of the D TECH results disagreed with the certified PE values, and imprecise, because over
half (56%) of the PE replicate results were not reported as the same interval. For the soil samples, the test kit
47
-------
reported 62% of the blank sample results as false positives. One non-blank result was classified as a false
negative. For extract samples, the test kit had two false positive results (25%) and no false negative results.
Table 5-15. Performance summary for the D TECH PCS test kit
Feature/Parameter
Blank Results
Precision
Accuracy
False Positive Results
False Negative Results
Comparison with Reference Laboratory
Results
Regulatory Decision-Making Applicability
Sample Throughput
Power Requirements
Operator Requirements
Cost
Hazardous Waste Generation
Performance Summary
Soils: 3 samples reported correctly as [0, 0.5); 5 samples reported as [0.5,
1]
Extracts: 6 samples reported correctly as [0, 0.5); 1 sample reported as
[0.5, 1]; 1 sample reported as (4, 15]
Percentage of combined sample sets in which all four replicates were
reported as the same interval
PE soils: 44%
Environmental soils: 21%
Extracts: 25%
PE soils Extracts
agreed = 56% agreed = 42%
biased high =28 % biased high = 29%
biased low = 1 7% biased low = 29%
Blank soils: 62% (5 of 8 samples)
Blank extracts: 25% (2 of 8 samples)
PE and environmental soils: 0.5% (1 of 192 samples)
Spiked extracts: 0% (0 of 16 samples)
PE and environmental soils Extracts
agreed = 53% agreed = 42%
biased high = 28% biased high = 29%
biased low = 20% biased low = 29%
PE and environmental soils Extracts
(40-60 ppm) (100 ug/lOOcm2, 1000 ug/100cm2)
agreed = 61% agreed = 42%
biased high = 7% biased high = 29%
biased low = 32% biased low = 29%
1 1 samples/hour (chamber)
1 5 samples/hour (outdoors)
Battery-operated photometer
Basic knowledge of chemical techniques; 2-A h technology-specific training
Incremental: $36 per sample
Instrumental: $299 (purchase photometer)
-20 Ib of solid/liquid waste (classified as incinerable solid)
—20 Ib of solid waste (used gloves, pipettes, paper towels, etc.)
-19 Ib of liquid waste (aqueous with trace methanol)
48
-------
Section 6
Technology Update and Representative Applications
Objective
The purpose of this section is to allow SDI to describe new technology developments that have occurred since
the demonstration activities. In addition, the developer has provided a list of representative applications where
the D TECH PCB test kit has been or is currently being utilized.
Technology Update
Reconfiguration of Soil Extraction (Sample Preparation) Products
SDI is in the process of commercializing a common extraction kit for three of its four remediation
immunoassay test kit product lines. The affected product lines include the EnviroGard, EnSys (not
demonstrated here), and RaPID Assay test kit systems. The new "Universal Extraction Kit" will be used with
assay kits of these three product lines, with extraction solvents or dilution reagents specifically formulated to
match individual kits available as kit component options where required. The new test kit configuration will
provide increased user convenience and simplify the product specification and ordering process without
affecting test kit analytical performance. Commercialization of the new Universal Extraction Kit was initiated
April 1998. The new kits are not for use with the D TECH product line, which will continue to use the existing
SDI Soil Extraction Pac products.
D TECH Test Format Changes
Two modifications to SDFs D TECH product line to enhance analytical performance are presently under
development. First, the intervals between the tests' existing semi-quantitative range values will be significantly
increased by reducing the number of ranges. For example, the D TECH PCB semi-quantitative ranges will be
reduced from the current six (0 to 0.5 ppm, 0.5 to 1 ppm, >1 to 4 ppm, >4 to 15 ppm, >15 to 25 ppm, >25
ppm) to about four (0 to 1 ppm, 1 to 10 ppm, 10 to 25 ppm, and >25 ppm). This will simplify the interpretation
of results and broaden the range of results, which will make the results more accurate. Secondly, each assay
kit will include results' interpretation data based on the actual lot-specific performance (calibration curve) of
that production lot. The combination of these two modifications is expected to significantly increase the
analytical performance of the D TECH PCB test kit, while maintaining the speed, simplicity, and convenience
of this format.
Instrument Consolidation
Associated with the incorporation of several independently developed product lines into SDFs product
offerings, some consolidation of equipment and instrumentation is anticipated in the near future. This will
consist primarily of reducing the number of pipet types and photometers used to perform the assays. While
pipet types and procedures for pipeting reagents and reading and interpreting assay results may change slightly,
no effect of assay performance will result.
49
-------
Representative Applications
A 1997 report entitled Field Analytical and Site Characterization Technologies: Summary of Applications
[11] documents the use of SDI immunoassay kits at more than 30 remediation sites under state or federal
oversight. The report provides contact information for many of the immunoassay kit users at these sites. The
summary report can be obtained from the National Center for Environmental Publications and Information
(NCEPI). Hard copies of the report can be ordered, free of charge, by telephone (513-891-6561), by fax (513-
891-6685), or through the NCEPI home page on the Internet (http://www.epa.gov /ncepihom). The report is
available for viewing or download as a .pdf file from the CLU-IN Internet web site
(http: //clu-in. com/pubichar .htm).
Data Quality Objective Example
This application of SDFs D TECH PCB immunoassay kit is based on data quality objective (DQO) methods
for project planning advocated by ASTM [12, 13] and EPA [14]. ORNL derived a DQO example from the
performance results in Section 5. The example, which is presented in Appendix E, illustrates the use of SDFs
D TECH performance data from the ETV demonstration in the DQO process to select the number of samples
to characterize the decision rule's false positive and false negative error rates.
50
-------
Section 7
References
[1] Erickson, M. D. Analytical Chemistry ofPCBs, 2nd ed., CRC Press/Lewis Publishers, Boca Raton,
Fla., 1997.
[2] "Polychlorinated Biphenyls (PCBs) Manufacturing, Processing, Distribution in Commerce, and Use
Prohibitions," Code of Federal Regulations, 40 CFR, pt. 761, rev. 7, December 1994.
[3] Maskarinec, M.P., et al. Stability of Volatile Organics in Environmental Soil Samples, ORNL/TM-
12128, Oak Ridge National Laboratory, Oak Ridge, Tenn., November 1992.
[4] U.S. Environmental Protection Agency. "Method 8081: Organochlorine Pesticides and PCBs as
Aroclors by Gas Chromatography: Capillary Column Technique," in Test Methods for Evaluating
Solid Waste: Physical/Chemical Methods (SW-846), 3d ed., Final Update II, Office of Solid Waste
and Emergency Response, Washington, D.C., September 1994.
[5] Oak Ridge National Laboratory. Technology Demonstration Plan: Evaluation of Polychlorinated
Biphenyl (PCB) Field Analytical Techniques, Chemical and Analytical Sciences Division, Oak Ridge
National Laboratory, Oak Ridge, Tenn., July 1997.
[6] U.S. Environmental Protection Agency. Data Quality Objectives for Remedial Response Activities,
EPA 540/G-87/003, EPA, Washington, D.C., March 1987.
[7] Sachs, Lothar. Applied Statistics: A Handbook of Techniques, 2nd ed., Springer-Verlag, New York,
1984.
[8] Snedecor, G. W., and William G. Cochran. Statistical Methods, Iowa State University Press, Ames,
Iowa, 1967.
[9] Draper, N. R., and H. Smith. Applied Regression Analysis, 2nd ed., John Wiley & Sons, New York,
1981.
[10] Berger, Walter, Harry McCarty, and Roy-Keith Smith. Environmental Laboratory Data Evaluation,
Genium Publishing Corp., Schenectady, NY., 1996.
[11] U.S. Environmental Protection Agency. Field Analytical and Site Characterization Technologies:
Summary of Applications, EPA-542-R-97-011, Office of Solid Waste and Emergency Response,
Washington, D.C., November 1997.
51
-------
[12] American Society for Testing and Materials (ASTM). Standard Practice for Generation of
Environmental Data Related to Waste Management Activities: Quality Assurance and Quality
Control Planning and Implementation, D5283-92, 1997.
[13] American Society for Testing and Materials (ASTM). Standard Practice for Generation of
Environmental Data Related to Waste Management Activities: Development of Data Quality
Objectives, D5 792-95, 1997.
[14] U.S. Environmental Protection Agency. Guidance for Data Quality Assessment, EPA QA/G-9;
EPA/600/R-96/084, EPA, Washington, D.C., July 1996.
52
-------
Appendix A
Description of Environmental Soil Samples
53
-------
-------
Table A-l. Summary of soil sample descriptions
Location
Oak Ridge
Oak Ridge
Oak Ridge
Oak Ridge
Oak Ridge
Paducah
Portsmouth
Tennessee
Reference Soil
Request for
Disposal
(RFD)#
40022
40267
24375
43275
134555
97002
7515
n/a
Drum#
02
01
02
03
04
01
02
03
01
02
03
01
02
03
04
858
1069
1096
1898
2143
2528
3281
538
940
4096
n/a
Description
Soil from spill cleanup at the Y-12 Plant in Oak Ridge, Tennessee.
This soil is PCB-contaminated soil excavated in 1992.
Soil from the Elza Gate area, a DOE Formerly Utilized Sites Remedial
Action Program site in Oak Ridge, Tennessee. This soil is PCB-
contaminated soil that was excavated in 1992.
Catch-basin sediment from the K-71 1 area (old Powerhouse Area) at
the DOE East Tennessee Technology Park (formerly known as Oak
Ridge Gaseous Diffusion Plant) in Oak Ridge, Tennessee. This soil is
PCB-contaminated storm drain sediment that was excavated in 1991.
Soil from the K-25 Building area at the DOE East Tennessee
Technology Park (formerly known as Oak Ridge Gaseous Diffusion
Plant) in Oak Ridge, Tennessee. This soil is PCB-contaminated soil
that was excavated in 1993.
Soil from the K-707 area at the DOE East Tennessee Technology Park
(formerly known as Oak Ridge Gaseous Diffusion Plant) in Oak Ridge,
Tennessee. This soil is PCB-contaminated soil from a dike spillage that
was excavated in 1995.
Soil from the DOE Paducah Gaseous Diffusion Plant in Kentucky. This
soil is PCB-contaminated soil from a spill cleanup at the C-746-R
(Organic Waste Storage Area) that was excavated in 1989.
Soil from the DOE Portsmouth Gaseous Diffusion Plant in Ohio. This
soil is PCB-contaminated soil from a probable PCB oil spill into the
East Drainage Ditch that was excavated in 1986.
Captina silt loam from Roane County, Tennessee; used as a blank in
this study (i.e., not contaminated with PCBs)
55
-------
-------
Appendix B
Characterization of Environmental Soil Samples
57
-------
-------
Table B-l. Summary of environmental soil characterization
Location
Oak Ridge
Paducah
Portsmouth
Sample
ID
101
102
103
104
105
106
107
108
109
110
111
112
126, 226
113,201
114,202
115,203
116,204
117,205
206
207
208
209
210
216
211
217
212
213
214
215
RFD
Drum # a
40022-02
40267-03
40267-01
40267-04
40267-01 Sb
24375-03
24375-01
40267-02
24375-02
43275-01
134555-03Sb
43275-02
non-PCB soil
97002-04
97002-01
97002-03
97002-02
97002-02S b
7515-4096
7515-1898
7515-1096
7515-2143
7515-0940
7515-0538
7515-0538Sb
7515-0538Sb
7515-2528
7515-3281
7515-0858
7515-1069
Composition
% gravel
0
0.5
0.2
0.6
0.5
0.5
2.5
0.4
0.3
0
0.5
0.1
0
0
0.2
0.1
0.4
0
0.2
0.4
0
0.3
0.5
0.5
0.5
0
1.3
% sand
91.8
99.3
96.7
98.2
94.8
87.8
92.5
94.2
93.1
89.2
88.1
91.4
85.6
92.4
87.6
83.6
93.7
87.1
78.0
74.4
74.3
73.0
73.3
70.4
72.6
65.8
75.0
% silt + clay
8.2
0.2
3.1
1.2
4.7
11.7
5.0
5.4
6.6
10.8
11.4
8.5
14.4
7.6
12.2
16.3
5.8
12.9
21.8
25.2
25.7
26.7
26.3
29.1
26.8
34.2
23.7
Total Organic
Carbon
(mg/kg)
5384
13170
13503
15723
14533
19643
1196
9007
1116
14250
10422
38907
9249
1296
6097
3649
4075
3465
3721
3856
10687
7345
1328
5231
5862
6776
4875
PH
7.12
7.30
7.21
7.07
7.28
7.36
7.26
7.30
7.48
7.57
7.41
7.66
7.33
7.71
7.64
7.59
7.43
7.72
7.66
7.77
7.71
7.78
7.78
7.92
7.67
7.85
7.56
a Request for disposal drum number (see Table A-l).
b "S" indicates that the environmental soil was spiked with additional PCBs.
59
-------
-------
Appendix C
Temperature and Relative Humidity Conditions
61
-------
-------
Table C-l. Average temperature and relative humidity conditions during testing periods
Date
7/22/97
7/23/97
7/24/97
7/25/97
7/26/97
7/27/97
7/28/97
7/29/97
Outdoor Site
Average
Temperature
(°F)
85
85
85
80
85
80
79
b
Average
Relative Humidity
(%)
62
70
67
70
55
75
88
b
Chamber Site
Average
Temperature
(°F)
70 a
60 a
58
56
57
55
57
55
Average
Relative Humidity
(%)
38 a
58 a
66
54
51
49
52
50
' The chamber was not operating properly on this day. See discussion in Section 3.
' No developers were working outdoors on this day.
120
100 -
60
40
20
0 -I
7/22/97 7/23/97 7/24/97 7/25/97 7/26/97 7/27/97 7/28/97
Figure C-l. Summary of temperature conditions for outdoor site.
63
-------
120
7/22/97 7/23/97 7/24/97 7/25/97 7/26/97
7/27/97
7/28/97
Figure C-2. Summary of relative humidity conditions for the outdoor site.
7/22/97 7/23/97 7/24/97 7/25/97 7/26/97 7/27/97 7/28/97 7/29/97
Figure C-3. Summary of temperature conditions for chamber site.
64
-------
90
80
70
— 60
!5 50
E
3
HI
OL
30
20
10--
n
[In
7/22/97 7/23/97 7/24/97 7/25/97 7/26/97 7/27/97 7/28/97
7/29/97
Figure C-4. Summary of relative humidity conditions for chamber site.
65
-------
66
-------
Appendix D
SDI's D TECH PCB Test Kit
PCB Technology Demonstration Sample Data
67
-------
Legend for Appendix D Tables
Table Heading
Obs
Sample ID
Rep
D TECH Result
Ref Lab Result
Reference Aroclor
Type
Order
Definition
Observation
Sample identification
101 to 126 = outdoor site soil samples
127 to 130 = outdoor site extract samples
201 to 226 = chamber site soil samples
227 to 230 = chamber site extract samples
Replicate of sample ID (1 through 4)
Measured PCB concentration (ppm) for SDFs D TECH
LAS reference laboratory measured PCB concentration (ppm)
Values with "<" are samples that the reference
laboratory reported as "< reporting detection limit"
Aroclor(s) identified by the reference laboratory
Sample = environmental soil
1242, 1248, 1254, 1260 = Aroclor in PE samples
Blank = non-PCB-contaminated sample
Order of sample analysis by SDI (started with 2001-2116, then
1001-1116)
68
-------
Table D-l. SDI's D TECH PCB test kit technology demonstration soil sample data
Obs
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
Sample
ID
101
101
101
101
102
102
102
102
103
103
103
103
104
104
104
104
105
105
105
105
106
106
106
106
107
107
107
107
108
108
108
108
109
109
109
109
110
110
110
110
Rep
1
2
3
4
1
2
3
4
1
2
3
4
1
2
3
4
1
2
3
4
1
2
3
4
1
2
3
4
1
2
3
4
1
2
3
4
1
2
3
4
D TECH
Result
(ppm)
(2,8]
(1,2]
(2,8]
(2,8]
(4,15]
(4,15]
(4,15]
(4,15]
(25, »)
(4,15]
(1,4]
(1,4]
(4,15]
(1,4]
(4,15]
(15,25]
(1,4]
(25, »)
(15,25]
(25, »)
(25, »)
(25, »)
(25, »)
(4,15]
(1,4]
[0.5,1]
[0.5,1]
(1,4]
(1,4]
(4,15]
(4,15]
(1,4]
(1,4]
(1,4]
(1,4]
(1,4]
(15,25]
(4,15]
(4,15]
(4,15]
Ref Lab
Result
(ppm)
0.6
0.4
0.5
0.5
2.2
2.1
1.7
2.5
3.0
2.4
2.0
1.6
6.8
6.0
14.8
9.9
49.7
84.1
50.6
53.2
269.6
255.9
317.6
649.6
1.0
1.6
1.2
1.2
1.7
2.0
1.7
1.9
1.5
2.1
1.8
2.4
<490.0
<99.0
<66.0
<98.0
Reference
Aroclor
1254
1254
1254
1254
1254
1254
1260
1260
1254
1254
1260
1260
1260
1254
1254
1254
1260
1260
1260
1260
1254
1254
1254
1254
1254
1254
1254
1254
1254
1254
1254
1254
1254
1254
1254
1254
Non- Detect
Non- Detect
Non- Detect
Non- Detect
Type
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Order
1103
1097
1092
1075
1052
1084
1077
1018
1007
1004
1096
1005
1071
1100
1043
1010
1060
1062
1032
1045
1014
1041
1038
1087
1012
1030
1070
1061
1088
1020
1048
1046
1039
1050
1080
1033
1049
1028
1006
1089
69
-------
Obs
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
Sample
ID
111
111
111
111
112
112
112
112
113
113
113
113
114
114
114
114
115
115
115
115
116
116
116
116
117
117
117
117
118
118
118
118
119
119
119
119
120
120
120
120
Rep
1
2
3
4
1
2
3
4
1
2
3
4
1
2
3
4
1
2
3
4
1
2
3
4
1
2
3
4
1
2
3
4
1
2
3
4
1
2
3
4
D TECH
Result
(ppm)
(1,4]
(15,25]
(4,15]
(1,4]
(4,15]
(25, »)
(1,4]
(15,25]
[0.8,2.3]
(2.3,11]
(2.3,11]
[0.8,2.3]
(1,4]
(1,4]
(1,4]
(1,4]
(20,54]
(3.5,20]
(3.5,20]
[1.5,3.5]
(3.5,20]
(20,54]
(20,54]
(20,54]
(53, »)
(28, 53]
(28, 53]
(53, »)
(2.3,11]
(2.3,11]
(2.3,11]
(2.3,11]
(2.3,11]
(11,28]
(11,28]
(2.3,11]
(1,4]
(4,15]
[0.5,1]
[0.5,1]
Ref Lab
Result
(ppm)
44.5
36.0
39.3
35.1
<66.0
<200.0
<130.0
<200.0
0.7
1.1
0.6
1.9
1.1
1.2
1.3
1.7
14.9
12.4
15.0
16.9
41.4
41.2
48.5
34.0
431.6
406.3
304.7
392.8
2.1
1.9
0.7
1.6
21.2
17.2
17.4
24.4
4.5
4.0
6.3
5.0
Reference
Aroclor
1254
1254
1254
1254
Non- Detect
Non- Detect
Non- Detect
Non- Detect
1260
1260
1260
1248/1260
1260
1260
1260
1260
1248
1016
1248
1248
1248
1016
1248
1016
1016
1016
1016
1016
1248
1016
1248
1248
1016
1248
1248
1248
1254
1254
1254
1254
Type
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
1248
1248
1248
1248
1248
1248
1248
1248
1254
1254
1254
1254
Order
1074
1040
1008
1065
1086
1044
1058
1047
1013
1073
1017
1076
1093
1034
1099
1001
1083
1063
1094
1057
1064
1095
1027
1091
1021
1067
1069
1090
1042
1056
1035
1016
1026
1003
1079
1051
1098
1101
1085
1068
70
-------
Obs
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
Sample
ID
121
121
121
121
122
122
122
122
123
123
123
123
124
124
124
124
125
125
125
125
126
126
126
126
201
201
201
201
202
202
202
202
203
203
203
203
204
204
204
204
Rep
1
2
3
4
1
2
3
4
1
2
3
4
1
2
3
4
1
2
3
4
1
2
3
4
1
2
3
4
1
2
3
4
1
2
3
4
1
2
3
4
D TECH
Result
(ppm)
(25, »)
(25, »)
(4,15]
(15,25]
(4,15]
[0.5,1]
(15,25]
(4,15]
(25, »)
(25, »)
(25, »)
(25, »)
(1,4]
(1,4]
(1,4]
[0,0.5)
(25, »)
(15,25]
(25, »)
(15,25]
[0,0.5)
[0.5,1]
[0.5,1]
[0,0.5)
(2.3,11]
[0.8,2.3]
[0.8,2.3]
(2.3,11]
[0.5,1]
(1,4]
(4,15]
(4,15]
(54, 100]
(20,54]
(53, »)
(20,54]
(100, =0)
(100, .)
(54, 100]
(3.5,20]
Ref Lab
Result
(ppm)
58.7
55.7
53.2
50.9
12.2
10.9
11.3
10.0
59.2
56.9
66.8
57.5
1.8
1.4
1.9
1.8
32.0
41.3
46.0
32.2
<0.1
<0.1
<0.2
<1.3
1.0
1.0
1 . 1
0.6
1.4
1.6
1.2
1.5
14.0
12.8
16.2
12.4
43.1
45.3
41.0
47.7
Reference
Aroclor
1254
1254
1254
1254
1260
1260
1260
1260
1260
1260
1260
1260
1254
1260
1254
1254
1254
1254
1254
1260
Non- Detect
Non- Detect
Non- Detect
Non- Detect
1016/1260
1016/1260
1016/1260
1260
1260
1260
1260
1260
1248
1248
1248
1248
1248
1248
1248
1248
Type
1254
1254
1254
1254
1260
1260
1260
1260
1260
1260
1260
1260
1254/1260
1254/1260
1254/1260
1254/1260
1254/1260
1254/1260
1254/1260
1254/1260
Blank
Blank
Blank
Blank
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Order
1011
1029
1053
1031
1024
1066
1078
1015
1009
1036
1037
1072
1023
1102
1054
1059
1081
1025
1002
1055
1104
1082
1019
1022
2092
2044
2023
2073
2002
2059
2058
2033
2070
2019
2011
2094
2042
2055
2091
2051
71
-------
Obs
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
Sample
ID
205
205
205
205
206
206
206
206
207
207
207
207
208
208
208
208
209
209
209
209
210
210
210
210
211
211
211
211
212
212
212
212
213
213
213
213
214
214
214
214
Rep
1
2
3
4
1
2
3
4
1
2
3
4
1
2
3
4
1
2
3
4
1
2
3
4
1
2
3
4
1
2
3
4
1
2
3
4
1
2
3
4
D TECH
Result
(ppm)
(53, »)
(53, »)
(53, »)
(53, »)
(4,15]
(4,15]
(1,4]
(1,4]
(15,25]
(15,25]
(15,25]
(25, »)
(25, »)
(25, »)
(25, »)
(4,15]
(15,25]
(4,15]
(4,15]
(25, »)
(25, »)
(25, »)
(25, »)
(25, »)
(25, »)
(4,15]
(25, »)
(25, »)
(4,15]
(4,15]
(4,15]
(4,15]
(4,15]
(4,15]
(15,25]
(1,4]
(25, »)
(25, »)
(25, »)
(15,25]
Ref Lab
Result
(ppm)
3305.0
538.7
457.0
483.3
2.9
1 . 1
1.1
2.5
17.8
14.3
21.6
21.6
42.0
27.7
24.0
28.4
32.7
79.3
11.0
37.9
123.2
61.5
84.1
85.5
387.8
581.4
330.0
318.7
3.8
3.9
4.3
0.8
6.9
7.3
7.8
10.5
26.0
25.6
29.1
20.2
Reference
Aroclor
1016/1260
1016
1016
1016
1260
1260
1016/1260
1260
1260
1260
1260
1254
1260
1016/1260
1254
1260
1260
1260
1260
1260
1260
1260
1260
1260
1254
1254
1254
1254
1260
1260
1260
1260
1260
1260
1260
1260
1260
1260
1260
1260
Type
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Order
2068
2045
2066
2021
2101
2078
2077
2007
2076
2041
2084
2017
2064
2024
2087
2082
2031
2029
2050
2032
2085
2020
2072
2096
2016
2065
2053
2009
2095
2039
2010
2102
2079
2100
2038
2030
2104
2014
2093
2054
72
-------
Obs
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
Sample
ID
215
215
215
215
216
216
216
216
217
217
217
217
218
218
218
218
219
219
219
219
220
220
220
220
221
221
221
221
222
222
222
222
223
223
223
223
224
224
224
224
Rep
1
2
3
4
1
2
3
4
1
2
3
4
1
2
3
4
1
2
3
4
1
2
3
4
1
2
3
4
1
2
3
4
1
2
3
4
1
2
3
4
D TECH
Result
(ppm)
(15,25]
(25, »)
(25, »)
(15,25]
(15,25]
(25, »)
(15,25]
(25, »)
(25, »)
(25, »)
(25, »)
(25, »)
(2.3,11]
(2.3,11]
(2.3,11]
[0.8,2.3]
(28, 53]
(11,28]
(2.3,11]
(11,28]
(15,25]
(4,15]
(4,15]
(15,25]
(25, »)
(25, »)
(25, »)
(25, »)
(25, »)
(25, »)
(25, »)
(25, »)
(25, »)
(25, »)
(25, »)
(25, »)
(1,4]
(1,4]
(1,4]
(1,4]
Ref Lab
Result
(ppm)
25.1
24.1
26.2
31.2
151.6
47.0
54.3
64.0
886.7
549.8
542.8
1913.3
2.8
2.4
2.6
2.6
22.4
26.0
29.4
15.2
8.5
4.9
4.7
5.2
32.0
44.1
43.8
59.6
13.2
12.4
12.7
12.7
56.6
50.3
49.9
66.4
2.2
1.2
1.4
2.1
Reference
Aroclor
1260
1260
1260
1016/1260
1260
1260
1260
1260
1254
1254
1254
1016/1260
1248
1248
1248
1248
1248
1016
1248
1248
1254
1254
1254
1254
1016/1260
1016/1260
1254
1254
1260
1260
1260
1260
1260
1260
1260
1260
1254
1260
1260
1254
Type
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
Sample
1248
1248
1248
1248
1248
1248
1248
1248
1254
1254
1254
1254
1254
1254
1254
1254
1260
1260
1260
1260
1260
1260
1260
1260
1254/1260
1254/1260
1254/1260
1254/1260
Order
2069
2022
2056
2086
2083
2027
2048
2063
2046
2036
2047
2057
2075
2080
2008
2062
2025
2060
2001
2005
2067
2088
2028
2074
2004
2015
2103
2089
2012
2090
2006
2034
2026
2003
2035
2098
2049
2013
2037
2099
73
-------
Obs
Sample
ID
201 225
202 225
203 225
204 225
205
206
207
208
226
226
226
226
Rep
D TECH
Result
(ppm)
(25, »)
(25, »)
(25, »)
(25, »)
[0,0.5)
[0.5,1]
[0.5,1]
[0.5,1]
Ref Lab
Result
(ppm)
56.4
36.5
32.1
146.0
<0.8
Reference
Aroclor
Type
Order
1260
1016/1260
1260
1254
1254/1260
1254/1260
1254/1260
1254/1260
2018
2052
2071
2040
Non-Detect
Non-Detect
Non-Detect
Non-Detect
Blank
Blank
Blank
Blank
2061
2081
2043
2097
74
-------
Table D-2. SDI's D TECH PCB test kit's technology demonstration extract sample data
DBS
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
Sample
ID
130
130
130
130
131
131
131
131
132
132
132
132
230
230
230
230
231
231
231
231
232
232
232
232
Rep
1
2
3
4
1
2
3
4
1
2
3
4
1
2
3
4
1
2
3
4
1
2
3
4
D TECH
Result
(ppm)
(3.5,20]
[1.5,3.5]
[1.5,3.5]
[1.5,3.5]
(15,25]
(15,25]
(15,25]
[0.5,1]
[0,0.5)
[0,0.5)
[0,0.5)
[0,0.5)
(20,54]
(54, 100]
(20,54]
(20,54]
(100, =0)
(25, »)
(25, »)
(25, »)
[0,0.5)
[0.5,1]
[0,0.5)
(4,15]
Ref Lab
Result
(ppm)
16.4
10.9
10.3
10.7
67.1
57.1
62.8
68.2
<0.1
<0.1
<0.1
<0.1
9.8
10.4
7.6
7.9
55.2
55.0
61.3
59.1
<0.1
<0.1
<0.1
<0.1
Reference
Aroclor
1016
1016
1016
1016
1254
1254
1254
1254
Non- Detect
Non- Detect
Non- Detect
Non- Detect
1016
1016
1016
1016
1254
1254
1254
1254
Non- Detect
Non- Detect
Non- Detect
Non- Detect
Type
1242
1242
1242
1242
1254
1254
1254
1254
blank
blank
blank
blank
1242
1242
1242
1242
1254
1254
1254
1254
blank
blank
blank
blank
Spike '
(ppm)
10
10
10
10
100
100
100
100
0
0
0
0
10
10
10
10
100
100
100
100
0
0
0
0
Order
1107
1116
1112
1114
1105
1111
1110
1113
1108
1115
1109
1106
2113
2109
2114
2107
2116
2110
2106
2108
2115
2111
2112
2105
aNominal spike concentration of the extract sample prepared by ORNL.
75
-------
Table D-3. Corrected reference laboratory data
Error
Transcription
Calculation
Interpretation
Sample ID
106
130
205
207
210
118
119
209
214
219
101a
101a
107
109
113b
113b
119
127
201
219
Reported Result
(ppm)
<490
5.6
32,000
180
160
3.6
4.3
2.3
43.0
29.0
<0.7
<0.7
<1.3
18.0
<0.9
<1.0
18.0
7.2
< 1.0
21.0
Corrected Result
(ppm)
255.9
10.3
3,305.0
17.8
123.2
2.1
17.4
37.9
26.0
22.4
0.5
0.6
1.2
1.5
0.6
0.7
21.2
10.9
0.6
26.0
a Two of four measurements in Sample ID 101 were corrected.
b Two of four measurements in Sample ID 113 were corrected.
76
-------
Appendix E
Data Quality Objective Example
77
-------
78
-------
Disclaimer
The following hypothetical example serves to demonstrate how the information provided in this report may be
used in the data quality objectives (DQO) process. This example serves to illustrate the application of
quantitative DQOs to a decision process, but cannot attempt to provide a thorough education in this topic.
Please refer to other educational or technical resources for further details. In addition, since the focus of this
report is on the analytical technology, this example makes the simplifying assumption that the contents of these
drums will be homogeneous. In the real world, however, this assumption is seldom valid, and matrix
heterogeneity constitutes a source of considerable uncertainty which must be adequately evaluated if the overall
certainty of a site decision is to be quantified.
Background and Problem Statement
An industrial company discovered a land area contaminated with PCBs from an unknown source. The
contaminated soil was excavated into waste drums. Preliminary evaluation determined that a number of PCB
drums had to be incinerated to reduce or eliminate the PCB contamination. The incinerated soil was placed in
drums for disposal in a landfill. However, a final check of each drum was required to verify for the regulator
that the appropriate level of cleanup had been achieved. The regulator required that no drum have more than
2 ppm of PCB. The company's DQO team was considering the use of SDFs D TECH PCB kit to measure the
PCB concentration in each drum. Because the type of Aroclor was unknown, all measurements would be
reported as Aroclor 1254. The plan was to randomly select soil samples collected from each drum and test with
SDFs kit to determine if the concentration was in one of the three intervals: [0, 1], (1, 4], or (4, <=°). Recall that
this notation describes the concentration ranges 0 ppm < PCB < 1 ppm, 1 ppm < PCB < 4 ppm, and PCB >
4 ppm, as used in Section 5. The DQO team decided that a drum would be reprocessed by incineration if any
of SDI's D TECH results indicated a concentration in the intervals (1, 4], or (4, °°). In agreement with the
regulator, the DQO team determined that a decision rule for disposal would be based on the number of samples
with PCB concentrations in the intervals (1, 4], or (4, <=°).
General Decision Rule
If all of the PCB sample results show concentrations in [0, 1), then send the soil drum to the landfill.
If any of the PCB sample results are different than [0, 1), then reprocess the soil drum by incineration.
DQO Goals
EPA's Guidance for Data Quality Assessment [14] states in Secion 1.2: "The true condition that occurs with
the more severe decision error . . . should be defined as the null hypothesis." The DQO team decided that the
more severe decision error would be for a drum to be erroneously sent to a landfill if the drum's PCB
concentration actually exceeded the 2 ppm limit. Therefore, the null hypothesis is constructed to assume that
the drum's true PCB concentration exceeds the 2 ppm limit and that, as a "hot" drum, it should be sent to the
incinerator. Drums would be sent to the landfill only if the null hypothesis is rejected and it is concluded that
the "true" average PCB concentration is less than 2 ppm.
79
-------
With the null hypothesis defined in this way, a false positive decision is made when it is concluded that a drum
contains less than 2 ppm PCBs (i.e., the null hypothesis is rejected), when the drum is actually "hot" (i.e., the
null hypothesis is true). The DQO team required that the error rate for sending a "hot" drum to the landfill (i.e.,
the false positive error rate for the decision) could not be more than 5%. Therefore, a sufficient number of
samples must be taken from each drum so that the false positive decision error rate (FP) is 0.05 (or less) if the
true drum concentration is 2 ppm. This scenario represents a 5% chance of sending a drum containing 2 ppm
or more of PCBs to the landfill. The D TECH interval boundary of 1 ppm can be used as a conservative
estimate of the 2 ppm criterion.
The DQO team did not want to reprocess an excessive number of drums by incineration if the drum PCB
concentration was less than 2 ppm because of the expense. In this situation, a false negative decision is made
when it is concluded that a drum is "hot" (i.e., the null hypothesis is not rejected), when in actuality, the drum
contains soil with less than 2 ppm PCBs (i.e., the null hypothesis is actually false). After considering the
guidelines presented in Section 1.1 of EPA's Guidance for Data Quality Assessment [14], the team
recommended that the false negative decision error rate (FN) for the decision rule be 0.10 if the true drum
concentration was less than 1 ppm. That is, there would be a 10% chance of reprocessing a drum by
incineration if the true PCB concentration for a drum was less than 1 ppm.
Permissible FP and FN Error Rates and Critical Decision Point
FP: Pr[Take Drum to Landfill] < 0.05 when true PCB concentration > 1 ppm
FN: Pr[Reprocess Drum in Incinerator] < 0.10 when true PCB concentration < 1 ppm
Use of Technology Performance Information to Implement the Decision Rule
Technology performance information is used to evaluate whether a particular analytical technology can produce
data of sufficient quality to support the site decision. Because the DQO team is considering the use of the D
TECH PCB kit, the performance of this technology (as reported in this ETV report) was used to assess its
applicability to this project. The question which then arises is, How many samples are needed from a single
drum to permit a statistically valid decision at the specified certainty? Recall that a simplifying assumption was
made that the PCB distribution throughout the soil within a single drum is homogeneous and that therefore
matrix heterogeneity will not contribute to overall variability. The only variability to be considered in this
example, then, is the variability in performance of the D TECH kit's analytical method, which is determined
by precision and accuracy studies.
Determining the Number of Samples
The number of samples needed to satisfy the FP and FN requirements depends on the misclassification error
rates of the D TECH PCB kit. Two types of misclassifications have to be considered:
1 underestimating the PCB concentration—classifying a sample concentration in [0, 1] when the true
PCB concentration is greater than 1 ppm, and
2. overestimating the PCB concentration— classifying a sample concentration in (1, 4] or (4, °°) when
the PCB concentration is less than or equal to 1 ppm.
80
-------
The ETV demonstration results on performance evaluation soil samples and on environmental soil samples
indicated the error rates for the two types of misclassifications to be as follows:
PU = Pr[ Underestimating the PCB concentration ] = 0.066,
P0 = Pr[ Overestimating the PCB concentration ] = 0.412.
The probability distribution of classifying the number of soil samples in different concentration intervals
follows a binomial probability distribution [7]. This probability distribution and the requirements for FP and
FN can be used to determine the number of samples to meet the DQO goals. The FP for the decision rule is
related to PU by
FP = Pr[ All D TECH results < 1 ppm for PCB > 1 ppm ] = (P^" (E-l)
The FP error rate decreases as the sample size increases. Rearranging to solve for sample size, n, Equation E-l
becomes
Log(FP)
n = —^ (E-2)
Log(Pu}
where
n = number of samples from a drum to be measured,
FP = false positive decision error rate (e.g., FP = 0.05),
Pv = probability of underestimating the PCB concentration (e.g., Pv = 0.066).
Incorporating the appropriate values for the D TECH PCB test kit into Equation E-2 gives
n = Zog(0.05) = -1.301 = 1 1Q 2
Zog(0.066) -1.180
To be conservative, the sample size was rounded up to the next integer, which will decrease the FP. The DQO
team would have to take two samples to meet the decision rule's false positive requirement. The FN for the
decision rule is related to P0 by
FN = Pr[ Some of D TECH results > 1 ppm for PCB < 1 ppm ] = 1 - ( 1 - Po)n (E-3)
81
-------
The probability of a false negative decision (FN = sending a drum for reprocessing) actually increases with
increasing sample size because the chance of the kit overestimating a result increases with continued testing.
The sample size required to meet the FN requirement is
= Log( I - FN)
Log(l - P0) (E'4)
where
n = number of samples from a drum to be measured,
FN = false negative decision error rate (e.g., FN= 0.10),
P0 = probability of overestimating a PCB concentration.
n = Log(l - 0.10) = -0.046
Log(\ - 0.412) -0.231
The sample size must be rounded up to n = 2 because that is the number of samples required to meet the
specified FP. When n = 2 and the above equation is solved for FN, it is found that the DQO team cannot meet
their goal of 10% FN and would have to accept an FN of 65%. The DQO team would be able to meet the DQO
goals only for the FP requirement, not for the FN requirement. This situation occurs because of the 41%
overestimation error rate of the kit. If the decision about sending a drum for reprocessing is based on having
a single overestimate out of two samples, and each sample has a 41% chance of being overestimated, there is
a 65% chance that the drum will unnecessarily be sent for reprocessing through the incinerator (which is the
definition of FN). Although this amount of conservatism may be desirable in some situations, in others it may
not be. The only way to reduce the FN in this kind of scenario is to use an analytical technology with a lower
overestimation error rate.
The DQO team decided that the sampling procedure would be to randomly select two soil samples from each
drum and test the samples with SDFs D TECH PCB kit. The DQO team would send the drum to the landfill
if both of SDFs D TECH results were less than 1 ppm, and send the drum to be reprocessed by incineration
if any of SDFs D TECH results were greater than 1 ppm. To meet the FP requirement of 5%, the DQO team
would have to accept the FN of 65%.
82
-------
Decision Rule for 5% FP and 65% FN
If two randomly selected soil samples have PCB test results reported as the interval [0, 1], then send the
soil drum to the landfill.
If any of two randomly selected soil samples have PCB test results different than [0, 1], then reprocess
the soil drum by incineration.
Alternative FP Parameter
The following describes how changing the FP requirement from 5% to 0.1% would affect the decision rule.
Using FP = 0.001, the calculated sample sizes would be n = 2.54, so the sample size would be rounded up to
n = 3. The actual FP would then be 0.03% and the actual FN would be 80%. The decision rule for the lower
FP requirement would be as shown.
Decision Rule for FP = 0.1% and FN = 80%
If three randomly selected soil samples have PCB test results reported as the interval [0, 1], then send
the soil drum to the landfill.
If any of the three randomly selected soil samples have PCB test results different than [0, 1], then
reprocess the soil drum by incineration.
83
------- |