United States      Office of Research and    EPA/600/R-01/052
         Environmental Protection   Development       August 2001
         Agency        Washington, D.C. 20460
&EPA   Environmental Technology
        Verification Report

        PCB Detection Technology

        Hybrizyme
        DELFIA™ PCB Assay
             Oak Ridge National Laboratory
ET V ETY ET

-------
TECHNICAL REPORT DATA
1. Report No.
EPA/600/R-01/052
4. Title and Subtitle
Environmental Technology Verification Report - PCB De
DELFIA TM PCB Assay
2.
jtection Technology, Hybrizyme
7. Author(s)
Dindal, A.B., Banyne, C.K., and Jenkins, R.A., Koglin, E.N.
9. Performing Organization Name & Address
Oak Ridge National Laboratory, Oak Ridge, TN. 37831-6120


12. Sponsoring Agency Name & Address
U.S. EPA, Office of Research & Development, National Exposure Research Laboratory
Environmental Sciences Division, Environmental Chemistry Branch
944 E. Harmon Ave.
Las Vegas, NV89119
3. Recipient's Accession No.
5. Report Date
07/00
6. Performing Organization Code
8. Performing Organization Report
No. (TIPS)
NERL-LV0 1-086
10. Program Element Number
11. Contract/Grant No.
13. Type of Report & Period Covered
ETV/FY-01
14. Sponsoring Agency Code
EPA 600/07
15. Supplemental Notes or Complete Citation^
.Dindal, A.B., Banyne, C.K., and Jenkins, R.A., Koglin, EPA/600/R-01/052 E.N. Environmental Technology
Verification Report - PCB Detection Technology, Hybrizyme DELFIA TM PCB Assay
Las Vegas, NV:U.S. Environmental Protection Agency
IG. Abstract The DELFIA PCB Assay is a solid-phase time-resolved fluoroimmunoassay based on the sequential addition
of sample extract and europium-labeled PCB tracer to a monoclonal antibody reagent specific for PCBS. In this assay,
the antibody reagent and sample extract are added to a strip of Microtiter plate wells and allowed to react. The strips
have been specially treated to trap the antibody reagent or antibody-PCB complexes that may have formed. A wash
step removes sample matrix from the captured antibody. This step significantly reduces any potential matrix
interferences before the addition of the PCB tracer, resulting in an unusually robust assay system. The PCB tracer is
then added and allowed to bind to the antibodies that are not complexed with sample PCBS. A wash step is used to
separate antibody-bound tracer from the tracer free in solution. The addition of an enhancement solution forms
highly fluorescent chelates with the bound europium ions. The amount of fluorescence measured is inversely
proportional to the concentration of PCBs in the sample. The lowest reporting level is typically 0.5 ppm.
17. Key Words and Document Analysis
a. Descriptors:
Characterization, Monitoring, PCB'S Immunoassay
18. Distribution Statement
UELEASE TO THE PUBLIC
b. Identifiers/Open-ended Terms
19. Security Class (This Report)
UNCLASSIFIED
20. Security Class (This Page)
UNCLASSIFIED
c. COSATI
21. No. of Pages
51
22. Price
PA 2220-1

-------
         United States      Office of Research and    EPA/600/R-01/052
         Environmental Protection   Development       August 2001
         Agency        Washington, D.C. 20460
&EPA   Environmental Technology
        Verification Report

        PCB Detection Technology

        Hybrizyme
        DELFIA™ PCB Assay
             Oak Ridge National Laboratory
ET V ETY ET

-------
                 THE ENVIRONMENTAL TECHNOLOGY VERIFICATION
                                       PROGRAM,
                                      ET
ElAMA
V
                                                              Oak Ridge National Laboratory
                        Joint Verification  Statement
     TECHNOLOGY TYPE:     EMMUNOASSAY

     APPLICATION:            MEASUREMENT OF PCBs IN CONTAMINATED SOIL
                                AND SOLVENT EXTRACTS

     TECHNOLOGY NAME:    DELFIA™ PCB Assay

     COMPANY:               Hybrizyme

     ADDRESS:                2801 Blue Ridge Rd          PHONE: (919) 783-9595
                                Raleigh, NC 27607           FAX: (919) 782-9585

     WEB SITE:                www.hybrizyme.com
     EMAIL:                   rallen@hybriyzme.com
 The U.S. Environmental Protection Agency (EPA) has created the Environmental Technology
 Verification Program (ETV) to facilitate the deployment of innovative or improved environmental
 technologies through performance verification and dissemination of information. The goal of the ETV
 Program is to further environmental protection by substantially accelerating the acceptance and use of
 unproved and cost-effective-technologies. ETV seeks to achieve this goal by providing high-quality,
 peer-reviewed data on technology performance to those involved in the design, distribution, financing,
 permitting, purchase, and use of environmental technologies.

 ETV works in partnership with recognized standards and testing organizations and stakeholder groups
 consisting of regulators, buyers, and vendor organizations, with the full participation of individual
 technology developers. The program evaluates the performance of innovative technologies by developing
 test plans that are responsive to the needs of stakeholders, conducting field or laboratory tests (as
 appropriate), collecting and analyzing data, and preparing peer-reviewed reports. All evaluations are
 conducted in accordance with rigorous quality assurance protocols to ensure that data of known and
 adequate quality are generated and that the results are defensible.

 Oak Ridge National Laboratory (ORNL) is one of the verification organizations operating under the Site
 Characterization and Monitoring Technologies (SCMT) program. SCMT, which is administered by
 EPA's National Exposure Research Laboratory, is one of six technology centers under ETV. In this
 verification test, ORNL evaluated the performance of polychlorinated biphenyl (PCB) detection
 technologies. This verification statement provides a  summary of the test results for Hybrizyme's
 DELFIA™ PCB Assay.
EPA-VS-SCM-47            The accompanying notice is an integral part of this verification statement.               August 2001

-------
 VERIFICATION TEST DESCRIPTION

 This verification test was designed to evaluate technologies that detect and measure PCBs in soil and
 solvent extracts. The test was conducted at ORNL in Oak Ridge, Tennessee, from August 21 through 24,
 2000. Spiked samples of known concentration were used to assess the accuracy of the technology.
 Environmentally contaminated soil samples collected from U.S. Department of Energy sites in Ohio,
 Kentucky, and Tennessee and ranging in concentration from 0 to approximately 700 parts per million
 (ppm) were used to assess several performance characteristics. Tests were 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. Solutions of PCBs were also analyzed to
 simulate extracted surface wipe samples. The extracts were not analyzed by the reference laboratory.
 The results of the soil analyses conducted  by the technology were compared with results from analyses of
 homogeneous replicate samples conducted by conventional EPA SW-846 methodology in a reference
 laboratory. Details of the test, including a  data summary and discussion of results, maybe found in the
 report entitled Environmental Technology Verification Report: PCS Detection Technology—Hybrizyme,
 DELFIA™PCS Assay, EPA/600/R-01/052.

 TECHNOLOGY DESCRIPTION

 The DELFIA PCB Assay is a solid-phase  time-resolved fluoroimmunoassay based on the sequential
 addition  of sample extract and europium-labeled PCB tracer to a monoclonal antibody reagent specific
 for PCBs. In this assay, the antibody reagent and sample extract are added to a strip of microtiter plate
 wells and allowed to react. The strips have been specially treated to trap the antibody reagent or
 antibody-PCB complexes that may have formed. A wash step removes sample matrix from the captured
 antibody. This step significantly reduces any potential matrix interferences before the addition of the
 PCB tracer, resulting in an unusually robust assay system. The PCB tracer is then added and allowed to
 bind to the antibodies that are not complexed with sample PCBs. A wash step is used to separate
 antibody-bound tracer from the tracer free in solution. The addition of an enhancement solution forms
 highly fluorescent chelates with the bound europium ions. The amount of fluorescence measured is
 inversely proportional to the concentration of PCBs in the sample. The lowest reporting level is typically
 0.5 ppm.

 VERIFICATION OF PERFORMANCE

 The following performance characteristics of the DELFIA PCB Assay were observed:

 Precision: The mean relative standard deviations (RSDs) for the soil and extract  samples were 20°/o and
  15%, respectively, indicating that the analyses for both matrices were precise.

 Accuracy: Accuracy was assessed using the nominal concentrations of the spiked soils. The percentages
 of recovery were significantly different for data generated under the outdoor and  the chamber conditions.
 The results were biased slightly high under the outdoor conditions (mean % recovery = 124%), and
 biased slightly low under the chamber conditions (mean % recovery = 72%).  Additional testing of the
 data demonstrated that the results generated under the outdoor and the chamber conditions were
 statistically different, indicating that the DELFIA PCB Assay performed differently under different
 environmental conditions. For the extracts, all samples were biased high, with larger bias observed under
 the outdoor conditions.

 False positive/false negative results: No false positives were reported for the soil and extract blanks.  In
  addition, false positive and false negative  results were determined by comparing the DELFIA PCB Assay
 result to the reference laboratory result for the environmental and the spiked samples. None of the results
 were reported as false positives, but 2% (4 of 192 samples) were false negatives relative to *h? rrfcicricc
  laboratory.
EPA-VS-SCM-47             The accompanying notice is an integral part of this verification statement.                August 2001

-------
  Completeness: The DELFIA PCS Assay generated results for all 208 soil samples and 24 extract
  samples, for a completeness of 100%.

  Comparability: A one-to-one sample comparison of the DELFIA PCB Assay results and the reference
  laboratory results was performed for all samples (spiked and environmental) that were reported as
  detections. The correlation coefficient (r) for the comparison of the entire soil data set was 0.50 [slope
  (m) = 0.20]. If six justifiably suspect values are excluded from the data set, the r value improves to 0.89,
  with a slope of 0.78. As stated in the Accuracy section, the DELFIA PCB Assay's performance was
  different under the outdoor and the chamber conditions. When the performance of the field technology is
  compared with the results from the reference laboratory (rather than with the nominal concentrations, as
  was used in the accuracy assessment), there is no statistical difference between the data sets generated
  outdoors and in the chamber. The comparison with the reference  laboratory results did not show
  statistical differences because of the uncertainty (i.e., variability) in the two data sets.

  Sample Throughput: Operating both in the field and in the chamber, the Hybrizyme team accomplished
  a sample throughput rate of approximately six samples per hour for the soil and extract analyses. Two
  operators were used for the PCB analyses, but the technology can be run by a single trained operator.

  Regulatory Decision-Making: One objective of this verification  test was to assess the technology's
  ability to  perform at regulatory decision-making levels for PCBs—specifically, 50 ppm for soils,
  including both performance evaluation and environmental samples. The performance of the DELFIA
  PCB Assay for this concentration range was precise (mean RSD = 14%), unbiased (mean % recovery =
  94%), and comparable to the reference laboratory (mean % difference = 27%).

  Overall Evaluation: The verification team found that the DELFIA PCB  Assay was relatively simple for
  the trained analyst to operate in the field, requiring less than an hour for initial setup. The overall
  performance of the DELFIA PCB Assay for the analysis of PCBs in soil and extract samples was
  characterized as biased (dependent on environmental conditions) but precise. 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
W.Frank Harris, Ph.D.
Associate Laboratory Director
Biological and Environmental Sciences
Oak Ridge National Laboratory
     NOTICE: EPA verifications are based on evaluations of technology performance under specific, predetermined criteria
     and appropriate quality assurance procedures. EPA and ORNL make no expressed or implied warranties as to the
     performance of the technology and do not certify that a technology will always operate as verified. The end user is solely
     responsible for complying with any and all applicable federal, state, and local requirements. Mention of commercial
     product names does not imply endorsement or recommendation.
EPA-VS-SCM-47
                          The accompanying notice is an integral part of this verification statement.
                                       August 2001

-------
                                       EPA/600/R-01/052
                                         August 2001
Environmental Technology
Verification  Report

PCB Detection Technology

Hybrizyme
DELFIA™  PCB Assay
                       By
                    Amy B. Dindal
                  Charles K. Bayne, Ph.D.
                  Roger A. Jenkins, Ph.D.
                 Oak Ridge National Laboratory
                Oak Ridge, Tennessee 37831-6120
                    Eric N. Koglin
               U.S. Environmental Protection Agency
                Environmental Sciences Division
               National Exposure Research Laboratory
                 Las Vegas, Nevada 89193-3478

-------
                                          Notice


The U.S. Environmental Protection Agency (EPA), through its Office of Research and Development (ORD),
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.

-------
                                 Table of Contents


List of Figures 	:	    v
List of Tables	   vii
Acknowledgments	   ix
Abbreviations and Acronyms 	   xi

1. INTRODUCTION	    1

2. TECHNOLOGY DESCRIPTION	    2
   Principle of the Assay	    2
   Calculation of Results	    2
   Sensitivity and Quality Control	    2
   Test Kit Components	    2
   Soil Sampling Processing 	    3
   Quantitative Assay Procedure	    3

3. VERIFICATION TEST DESIGN	    5
   Objective	    5
   Testing Location and Conditions 	    5
   What Are PCBs?	    5
   Soil Sample Description 	    5
       Sources of Samples	    5
          Oak Ridge, Tennessee  	    5
          Portsmouth,  Ohio	    5
          Paducah, Kentucky	    6
       Performance Evaluation Samples	    6
       Soil Sample Collection	    6
       Soil Sample Preparation	    6
   Extract Sample Description	    7
   Sample Randomization	    7
   Summary of Experimental Design  	    7
   Description of Performance Factors	    8
       Precision	    8
       Accuracy	    8
       False Positive/False Negative Results 	    8
       Completeness 	    9
       Comparability	    9
       Sample Throughput	    9
       Applicability to Regulatory Decision-Making	    9
       Ease of Use	    9
       Cost	   10
       Miscellaneous Factors 	   10

4. REFERENCE LABORATORY ANALYSES	   11
   Reference Laboratory Selection 	   11
   Reference Laboratory Method  	   11
   Reference Laboratory Performance	   11

5. TECHNOLOGY EVALUATION	   13
   Objective and Approach 	   13
   Precision  	   13
   Accuracy	   13
   False Positive/False  Negative Results	   1 -
   Completeness	         ....   li
   Comparability	  	   15
   Comparison of Performance under Different Environmental Conditions  	     ..   o

                                            iii

-------
   Application to Regulatory Decision-Making  	   17
   Sample Throughput	   17
   Ease of Use 	   17
   Cost Assessment	   17
      DELFIA PCB Assay Costs	   17
          Labor 	   17
          Equipment  	   18
          Waste Disposal	   18
      Reference Laboratory Costs	   18
          Sample Shipment	   18
          Labor, Equipment, and Waste Disposal	   19
      Cost Assessment Summary	   19
      Miscellaneous Factors  	   19
   Summary of Performance 	   19

6. TECHNOLOGY UPDATE AND REPRESENTATIVE APPLICATIONS	   21
   Temperature Control	   21
   Food Test Validation	   21

7. REFERENCES	   22

Appendix A — Hybrizyme's DELFIA PCB Assay Results Compared with Reference
   Laboratory Results  	   23
Appendix B — Data Quality Objective (DQO) Example	   31
                                           IV

-------
                            List of Figures


Comparison of Hybrizyme and reference laboratory PCB results, excluding
nondetects and suspect vales (N = 164) 	   16
Range of percent difference values	   16

-------
                                  List of Tables


 1   Summary of DELFIA PCB Assay's Cross-Reactivity	    2
 2   Test Kit Components 	    3
 3   Recommended Sequence for Well Use	    4
 4   Summary Protocol Sheet	    4
 5   Summary of PCB Verification Test Design	    7
 6   Range of Characterization Values by Sample Source 	    8
 7   Summary of the Reference Laboratory Performance	   12
 8   Summary of the DELFIA PCB Assay Precision	   13
 9   Summary of the DELFIA PCB Assay Accuracy for Soils	   13
10   Number of DELFIA PCB Assay Results within Acceptance Ranges for Spiked Soils	   14
11   Summary of DELFIA PCB Assay Accuracy for Extracts 	   14
12   Summary of DELFIA PCB Assay False Positive Performance on Blank Samples	   14
13   Summary of the DELFIA PCB Assay Detect/Nondetect Performance Relative
     to the Reference Laboratory Results for Soil Samples (N = 192)	   14
14   DELFIA PCB Assay Correlation with Reference Data	   15
15   Performance of DELFIA PCB Assay on Regulatory Sample PCB
     Concentrations (40-60 ppm)  	   17
16   Estimated Analytical Costs for PCB-Contaminated Samples  	   18
17   Performance Summary for the DELFIA PCB Assay	   20
                                          Vll

-------
                                 Acknowledgments


The authors wish to acknowledge the support of all those who helped plan and conduct the verification test,
analyze the data, and prepare this report. In particular, we recognize the technical expertise of Viorica
Lopez-Avila (Midwest Research Institute), who was the peer reviewer of this report. For sample collection
support, we thank Wade Hollinger, Charlotte Schaefer, and Arlin Yeager [Lockheed Martin Energy Systems
(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 evaluation samples, Michael Wilson (EPA's Office of Solid Waste and Emergency Response's
Analytical Operations and Data Quality Center); and for technical guidance and project management of the
verification test, 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 and Gary Robertson (EPA's National Exposure Research Laboratory, Las Vegas, Nevada). The
authors also acknowledge the participation of Hybrizyme, hi particular, Randy Allen and Tom Stewart, who
performed the analyses during the test.

For more information on  the PCB Detection Technology Verification contact

Eric N. Koglin                                    Roger A. Jenkins
Project Technical Leader                          Program Manager
Environmental Protection Agency                   Oak Ridge National Laboratory
Environmental Sciences Division                    Chemical and Analytical Sciences Division
National Exposure Research Laboratory              P.O. Box 2008
P.O. Box 93478                                  Oak Ridge, TN 37831- 6120
Las Vegas, NV 89193-3478                        (865)574-4871
(702) 798-2432                                   jenkinsra@ornl.gov
koglin.eric@epa.gov

For more information on Hybrizyme's DELFIA PCB Assay contact

Randy Allen
Hybrizyme
2801 Blue Ridge Road
Raleigh, NC 27607
(919) 783-9595
rallen@hybrizyme.com
www.hybrizyme.com
                                             IX

-------
                          Abbreviations and Acronyms


AL              action level
BHC             benzenehexachloride
DOE             U. S. Department of Energy
DQO             data quality objective
EPA             U.S. Environmental Protection Agency
ERA             Environmental Resource Associates
ETTP            East Tennessee Technology Park
ETV             Environmental Technology Verification (Program, EPA)
FA              false acceptance decision error rate
fh               false negative result
fp               false positive result
FR              false rejection decision error rate
HEPA            high-efficiency particulate air
ID               inner diameter
N                number of samples
NERL            National Exposure Research Laboratory (EPA)
ORD             Office of Research and Development (EPA)
ORNL            Oak Ridge National Laboratory
PCB             polychlorinated biphenyl
PE              performance evaluation
ppb              parts per billion
ppm             parts per million (equivalent units: mg/kg for soils and Jig/mL for extracts)
Pr               probability
QA              quality assurance
QC              quality control
RH              relative humidity
RSD             relative standard deviation (percentage)
RT              regulatory threshold
SCMT           Site Characterization and Monitoring Technologies  .
SD              standard deviation
SSM             synthetic soil matrix
TSCA            Toxic Substances Control Act
%D              percent difference

-------
                              Section  1 — Introduction
The U.S. Environmental Protection Agency (EPA)
created the Environmental Technology Verification
Program (ETV) to facilitate the deployment of
innovative or improved environmental technologies
through performance verification and dissemination
of information. The goal of the ETV Program is to
further environmental protection by substantially
accelerating the acceptance and use of improved and
cost-effective technologies. ETV seeks to achieve
this goal by providing high-quality, peer-reviewed
data on technology performance to those involved in
the design, distribution, financing, permitting,
purchase, and use of environmental technologies.

ETV works in partnership with recognized standards
and testing organizations and stakeholder groups
consisting of regulators, buyers, and vendor
organizations, with the full participation of
individual technology developers. The program
evaluates the performance of innovative tech-
nologies by developing verification test plans that
are responsive to the needs of stakeholders, con-
ducting field or laboratory tests (as appropriate),
collecting and analyzing data, and preparing peer-
reviewed reports. All evaluations are conducted in
accordance with rigorous quality assurance (QA)
protocols to ensure that data of known and adequate
quality are generated and that the results are
defensible.

ETV is a voluntary program that seeks  to provide
objective performance information to all of the
participants in the environmental marketplace and to
assist them in making informed technology
decisions. ETV does not rank technologies or
compare their performance, label or list technologies
as acceptable or unacceptable, seek to determine
"best available technology," or approve or
disapprove technologies. The program  does not
evaluate technologies at the bench or pilot scale and
does not conduct or support research. Rather, it
conducts and reports on testing designed to describe
the performance of technologies under  a range of
environmental conditions and matrices.
The program now operates six centers covering a
broad range of environmental areas. ETV began
with a 5-year pilot phase (1995-2000) to test a wide
range of partner and procedural alternatives in
various technology areas, as well as the true market
demand for and response to such a program. In these
centers, EPA utilizes the expertise of partner
"verification organizations" to design efficient
processes for conducting performance tests of
innovative technologies. These expert partners are
both public and private organizations, including
federal laboratories, states, industry consortia, and
private sector entities. Verification organizations
oversee and report verification activities based on
testing and QA protocols developed with input from
all major stakeholder/customer groups associated
with the technology area. The verification described
in this report was administered by the Site Charac-
terization and Monitoring Technologies (SCMT)
Center, with Oak Ridge National Laboratory
(ORNL) serving as the verification organization.
(To learn more about ETV, visit ETV's Web site at
http://www.epa.gov/etv.) The SCMT Center is
administered by EPA's National Exposure Research
Laboratory (NERL), Environmental Sciences
Division, in Las Vegas, Nevada.

The verification of a field analytical technology for
polychlorinated biphenyls (PCBs) detection is
described in this report. The verification test was
conducted at ORNL in Oak Ridge, Tennessee, from
August 21 through August 24,2000. The perfor-
mance of Hybrizyme's DELFIA™ PCB Assay was
determined under both field and controlled
atmosphere (i.e., chamber) conditions. The
technology was evaluated by comparing its results
with those obtained using an approved reference
method, EPA SW-846 Method 8081. The
verification was designed to evaluate the field
technology's ability to detect and measure PCBs in
soil and solvent extracts.

-------
                      Section 2 — Technology Description
In this section, the vendor (with minimal editorial changes by ORNL) provides a description of the
technology and the analytical procedure used during the verification testing activities.
Principle of the Assay
The Hybrizyme DELFIA PCB immunoassay system
has been designed for the quantitative or qualitative
detection of PCBs in sample extracts. The DELFIA
technology is based on time-resolved fluorometry of
lanthanide compounds, such as europium. Lan-
thanide ions exhibit a unique fluorescence that is
characterized by narrow band emission lines, long
decay times, and large Stake's shifts. The specific
fluorescence of the lanthanide label is measured
after a certain time delay following an activation
pulse. The delay eliminates essentially all of the
nonspecific background, resulting in an ultra-
sensitive assay system. Hybrizyme's DELFIA
products incorporate many components and
instrumentation manufactured by Perkin Elmer® that
are used in hospitals worldwide for clinical analysis.

The DELFIA PCB assay is a solid-phase time-
resolved fluoroimmunoassay based on the sequential
addition of sample extract and europium-labeled
PCB tracer to a monoclonal antibody reagent
specific for PCBs. In this assay, the antibody reagent
and sample extract are added to a strip of microtiter
plate wells and allowed to react. The strips have
been specially treated to trap the antibody reagent or
antibody-PCB complexes that may have formed. A
wash step removes the remaining sample from the
captured antibody. This step significantly reduces
any potential matrix interferences prior to the
addition of the PCB tracer, resulting in an unusually
robust assay system. The PCB tracer is then added
and allowed to bind to the antibodies that are not
complexed with sample PCBs. Another wash step is
used to separate antibody-bound tracer from the
tracer free hi solution. The addition of an
enhancement solution forms highly fluorescent
chelates with the bound europium ions. The amount
of fluorescence measured is inversely proportional
to the concentration of PCBs in the sample.

Calculation of Results
The DELFIA PCB assay system was developed for
use in fixed or mobile laboratories for high-
throughput PCB analysis. Normal batch sizes range
from 5 to 20 samples per run. Results are generated
from stored calibration curves, eliminating the need
to run calibrators with each assay. For characterized
sites, the data-reduction package automatically
generates a spreadsheet of results for Aroclors 1260,
1254,1248, and 1242. The user can easily add
custom calibration curves for any mixture of PCB
congener to the instrumentation at any tune. For
uncharacterized sites, the cross-reactivity of the
DELFIA PCB assay to various Aroclors can be  used
to develop qualitative screening strategies.

Sensitivity and Quality Control
Hybrizyme reports that the immunoassay can detect
<100 parts per billion (ppb) PCBs hi methanol.  The
sensitivity of the assay can be adjusted to higher
detection levels by altering sample dilution
protocols. Values that lie outside the detection range
of the assay are automatically flagged as low or
high. Results are calculated from die duplicate
analysis of each extract. If the values between the
duplicates are outside the acceptable range of
variation, the result will automatically be flagged for
review. A PCB standard is available from
Hybrizyme for verification purposes. The ability of
the assay to detect various Aroclors is shown in
Table 1. If the Aroclor is known, the sample results
can be adjusted based on cross-reactively.

Test Kit Components
Each Hybrizyme DELFIA PCB Test Kit (see
Table 2) contains reagents for testing a maximum of
40 samples in duplicate. The reagents must be stored

   Table 1.  Summary of DELFIA PCB
             Assay's  Cross-Reactivitya
Aroclor
1262
1260
1254
1248
1242
1016
1232
% Reactivity
110
130
160
100
40
25
~\j |
    a Cross-reactivity represents the amount of response
    to the various Aroclors.

-------
  Table 2. Test Kit Components
Component
Europium-labeled PCB
tracer
PCB monoclonal
antibody
Wash concentrate
Assay buffer
Enhancement solution
Microtitration strips
Description
The tracer is lyophilized in a Tris-buffered salt solution with
bovine serum albumin, glycine, and <0. 1 % sodium azide. It is
reconstituted with 0.6 mL of deionized water and should be used
within 2 weeks after reconstitution
The antibody is in a Tris-buffered salt solution with casein and
<0.1 % sodium azide
A 25-fold concentration of Tris-buffered (pH 7.8) salt solution
with Tween 20 and <0. 1 % sodium azide. It is prepared for use by
mixing entire contents with 960 mL of deionized water and placing
in platewasher WASH bottle
Ready-to-use Tris-buffered (pH 7.8) salt solution with casein and
<0. 1 % sodium azide
Ready-to-use reagent with Triton X-100, acetic acid, and chelators
Unused strips must be kept sealed and in the plastic tray
Quantity
1 vial
1 vial (0.6 mL)
1 bottle (40 mL)
1 bottle (50 mL)
bottle (50 mL)
plate (8 x 12 wells)
between 2°C and 8°C when not in use. The
expiration date of an unopened test kit is stated on
the outer label. All analyses must be conducted
within 2 weeks of tracer reconstitution.

Soil Sample Processing
The following is an example of the extraction
procedure if the user is interested in a 1-ppm PCB
detection level; this is the procedure that was used
in the verification test.

 1.  Place 5.0 g of soil sample in a 40-mL glass
     vial.
 2.  Add 25 mL of methanol.
 3.  Cap vial and vortex (or shake) for 3 min.
 4.  Remove vial from vortex and allow soil to
     settle for 10 min.
 5.  Transfer a 4-jJ.L aliquot of the extract to the
     PCB test.

The detection level of the test can be varied by
changing the amount of soil, the volume of
methanol, and the volume of extract added to the
PCB test. The lowest reported concentration hi the
verification test was 0.5 ppm.

Quantitative  Assay Procedure
The quantitative detection of PCBs in sample
extracts is performed by comparing the test response
of sample extracts to the test response of a control.

Research-grade methanol is used as the control.
Each determination is performed in duplicate for the
both the control and samples. All sample extracts
must be in methanol for analysis. All reagents and
samples must be brought to room temperature prior
to use.

 1.  Prepare the PCB tracer solution by diluting
    50 JlL of PCB tracer stock solution in 1.5 mL
    of PCB assay buffer for each strip of wells
    used. For example, if three strips of wells will
    be used, dilute 150 \lL of tracer stock solution
    into 4.5 mL of PCB assay buffer. Use within
    one hour of preparation.
 2.  Prepare the PCB antibody solution by diluting
    50 |lL of PCB antibody stock solution in
    1.5 mL  of PCB assay buffer per strip of wells
    used. Use within one hour of preparation.
 3.  Place the required number of microtitration
    strips in a strip frame. Wash the strips using the
    "PREWASH" program of the plate washer.
    Tap the strips upside-down gently on a paper
    towel to blot away any excess wash solution
    that may remain in the wells.
 4.  Pipet 100 \iL of the diluted PCB antibody
    solution into each well.
 5.  Pipet 4  JJ.L of each control or sample into a
    well using the sequence shown in Table 3. It is
    recommended that columns 1  and 2 on each
    strip of wells be used for controls.

-------
  Table 3. Recommended Sequence for Well Use
   Row
                                                  Well
                                                                8
                                                                   10
                                                                  11
                                                               12
          Control
         Control
          1st
         Unk
        1st
       Unk
       2nd
       Unk
       2nd
       Unk
       3rd
       Unk
3rd
Unk
4th
Unk
4th
Unk
 5th
Unk
 5th
Unk
     B
Control
Control
6th
Unk
6th
Unk
 7th
Unk
7th
Unk
   Unk = unknown sample
   " The plate is a 12 by 8 well configuration. Each of the 8 rows holds one strip that can contain two controls and five samples
   run in duplicate. The user can run one to eight strips at a time, for a maximum of 40 samples.
 6.  Shake the wells for 15 min using an automated
    shaker.
 7.  Wash the strips using the "3 WASHES"
    program on the plate washer. Tap the strips
    upside-down gently on a paper towel to blot
    away any excess wash solution that may
    remain in the wells.
 8.  Pipet 100 \iL of the diluted PCB tracer solution
    into each well.
 9.  Shake the wells for 5 min.
10.  Repeat step 8.
11.  Add 150 [O.L of enhancement solution to each
    well.
                                         12.  Select "PCB Quant" from the list of protocols
                                             in the time-resolved fluorometer and measure
                                             the fluorescence in each well. The protocol will
                                             automatically shake the wells for 1 min and
                                             calculate the concentration of PCB in the
                                             extracts. The amount of PCB hi the sample
                                             must be correlated using the sample processing
                                             concentration factor or dilution factor.

                                         A summary protocol sheet is presented in Table 4.
               Table 4.  Summary Protocol Sheet
Task
1
2
3
4
5
6
7
8
9
10
11
12
Prepare PCB tracer solution
Prepare PCB antibody solution
Prewash strips
Add antibody solution
Add control and samples
Incubate
Wash
Add tracer solution
Incubate
Wash
Enhance
Incubate and count
Action
50 p,L tracer per 1 .5 mL assay buffer per
microtitration strip
50 U-L antibody per 1 .5 mL assay buffer
per microtitration strip
"PREWASH" program
100 M.L
4 flL
Shake for 15 min
"3 WASHES" program
100 (J.L
Shake for 5 min
"3 WASHES" program
150 U.L
Use a "PCB Quant" protocol to shake for
2 min and measure fluorescence

-------
                     Section 3 — Verification Test Design
Objective
The purpose of this section is to describe the
verification test design. It is a summary of the test
plan (ORNL 2000).

Testing Location and Conditions
The verification of field analytical technologies for
PCBs was conducted at ORNL's Building 5507, in
Oak Ridge, Tennessee. Testing activities occurred at
two sites: a natural outdoor environment (the
outdoor site) and inside a controlled environmental
atmosphere chamber (the chamber site). The
temperature and relative humidity (RH) were
monitored during testing. Over the two days of
outdoor testing, the average temperature was 86°F
and ranged from 63 to 98°F. The average relative
humidity was 50% and ranged from 27 to 85%.

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. 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. During the two days of testing in the
controlled atmosphere, the chamber conditions were
set to 55°F and 50% RH and were maintained at
those conditions with little variation.

What Are PCBs?
PCBs (CizHn^Cy 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 hi
transformers, capacitors, paints, pesticides,  and inks
(Erickson  1997). 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%).
Soil Sample Descriptions
The samples used in this study were shipped to the
testing location for evaluation by the vendor. PCB-
contaminated soils from Kentucky, Ohio, and
Tennessee were used in this verification. Because
samples were obtained from multiple U.S.
Department of Energy (DOE) sites, the samples
represented a reasonable cross section of the
population of PCB-contaminated matrices, such that
the versatility of the field technology could be
evaluated. 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 verification test. More specific
details about the samples are presented below.

Sources of Samples
Oak Ridge, Tennessee
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 National Security Complex (formerly
known as the Oak Ridge Y-12 Plant), and East
Tennessee Technology Park (ETTP). Chemical
processing and warhead component production have
occurred at Y-12, 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. Excavation activities occurred
between 1991 and 1995. The Oak Ridge samples
were composed 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). The PCB concentrations in these
samples ranged from approximately 0.5 to 300 ppm.

Portsmouth, Ohio
A population of over 50CC  anuiis 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 hi these soils. Therefore, no EPA hazardous
waste codes are assigned to this waste. The PCB
concentrations in these samples ranged from
approximately 1 to 700 ppm.

Paducah, Kentucky
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. The PCB
concentrations in these samples ranged from
approximately 1 to 500 ppm.

Performance Evaluation Samples
Samples of Tennessee reference soil (Maskarinec
1992) served as the blanks. Preprepared certified
performance evaluation (PE) samples were obtained
from Environmental Resource Associates (ERA) of
Arvada, Colorado, and from 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 EPA samples were also spiked with target
pesticides [benzenehexachloride (BHC),
methoxychlor, and endrin ketone] to introduce some
compounds that were likely to be present in an
actual environmental soil. The hydrocarbon
background from the original sample and the spiked
pesticides produced a challenging matrix.

The PE soils required no additional preparation by
ORNL and were split for the vendor and reference
laboratory analyses as received. The PCB
concentrations in PE soils ranged from 2 to 50 ppm.

Soil 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. 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 and then poured
into a plastic-lined 5-gal container. All samples were
subjected to radiological screening and were
determined to be nonradioactive. Soil samples were
collected from 55-gal drums stored at Paducah in a
similar fashion and were shipped to ORNL in  lined
5-gal containers.

Soil Sample Preparation
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 10-mesh screen (2-mm particle
size). Last, the soil was thoroughly mixed with a
spatula. A comparison of dried and undried soils
showed that a minimal amount of PCBs '<>C<"r  , ~
lost during sample drying, iriaiaiig mis procedure
suitable for use in the preparation of the soil

-------
samples. The Paducah samples, because of their
sandy characteristics, required only the sieving and
mixing preparation steps.

To provide the vendors with soils contaminated at
higher PCB concentrations, some of the
environmental soils were spiked with additional
PCBs. Spiked soil 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 10-mesh screen (2-mm particle
size). Approximately 1500 g of the sieved soil was
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
OKNL prior to the verification test. Soil sample
homogeneity was confirmed by extracting 3-5 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
EPA's SW-846 dual-column Method 8081 (EPA
1994).
Extract Sample Description
Extract samples were prepared by making solutions
of PCBs in methanol at two concentration levels
(10 and 100 |Ig/mL). Aroclor 1242 was used to
prepare the 10-[J.g/mL samples, and Aroclor 1254
was used for the 100-|ig/mL samples. Multiple
aliquots of each sample were analyzed using the
Method 8081 to confirm the accurate preparation of
the samples with respect to PCB concentration.

Sample Randomization
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 vendor. The samples were
randomized in two stages. First, the order in which
the filled jars were distributed was randomized so
that the same vendor did not always receive the first
jar filled for a given sample set. Second, the order of
analysis was randomized so that each participant
analyzed the same set of samples, but hi a different
order. Each jar was labeled with a sample number.
Replicate samples were assigned unique (but not
sequential) sample numbers. Spiked materials and
blanks were labeled in the same manner, such that
these quality control (QC) samples were
indistinguishable from other samples. All samples
were analyzed blindly by both the vendor and the
reference laboratory.

Summary of Experimental Design
The distribution of samples from the various sites is
shown hi Table 5. A total of 208 soil samples were
analyzed, with approximately 70% of the samples
                     Table 5. Summary of PCB Verification Test Design
Sample source
Oak Ridge soil
Portsmouth soil
Paducah soil
Spiked soil
Blank soil
Spiked extract
Blank extract
Total
Number of samples
Outdoor site
48
0
20
32
4
8
4
116
Chamber site
0
48
20
32
4
8
4
116

-------
being naturally contaminated environmental soils
and the remaining 30% being spikes and blanks.
Twenty-four extract samples were also analyzed, for
a grand total of 232 samples in the verification test,
with 116 samples analyzed at each of the two sites.
Four replicates were analyzed for each sample type.
For example, 48 samples were analyzed from the
Oak Ridge site, indicating that 12 different original
samples were used in the study. As Table 5
indicates, the Paducah, PE, and extract samples
were analyzed at both the outdoor and chamber sites
so that performance under different environmental
conditions could be evaluated. Table 6 contains a
characterization summary of the environmental
samples.

Description of Performance Factors
In Section 5, technology performance is described in
terms of precision, accuracy, completeness, and
comparability, which are indicators of data quality
(EPA 1996). False positive and negative results,
sample throughput, and ease of use are also
described. Each of these performance characteristics
is defined in this section.

Precision
Precision is the reproducibility of measurements
under a given set of conditions. Standard deviation
(SD) and relative standard deviation (RSD) for
replicate results are used to assess precision, using
the following equation:

RSD = (SD/average concentration) x 100% .
                                        (Eq.l)

The overall RSD is characterized by three summary
values:

•   mean — i.e., average;
    median — i.e.,  50th percentile value, at which
    50% of all individual RSD values are below and
    50% are above; and
    range — i.e., the highest and lowest RSD values
    that were reported.

The average RSD may not be the best representation
of precision, but it is reported for convenient
reference. RSDs greater than 100% should be
viewed as indicators of large variability and possibly
non-normal distributions.

Accuracy
Accuracy represents the closeness of the tech-
nology's measured concentrations to known (in this
case, PE) values. Accuracy is assessed in terms of
percent recovery, calculated by the following
equation:
% recovery = (measured concentration/
    known concentration)  x 100%  .
                                        (Eq.2)
As with precision, the overall percentage of
recovery is characterized by three summary values:
mean, median, and range.

False Positive/False Negative Results
A false positive (fp) result is one in which the
technology detects PCBs in the sample when there
actually are none (Berger, McCarty, and Smith
1996). A false negative (fh) result is one in which
the technology indicates that no PCBs are present in
the sample when there actually are (Berger,
McCarty, and Smith 1996). The evaluation of fp and
fh results is influenced by the actual concentration
in the sample and includes an assessment of the
reporting limits of the technology.

False positive results are assessed in two ways.
First, the results are assessed relative to the blanks
(i.e., the technology reports a detected value when
the sample is a blank). Second, the results are
assessed on environmental and spiked samples
where the analyte was not detected by the reference
laboratory (i.e., the reference laboratory reports a
    Table 6. Range of Characterization Values by Sample Source
Sample source
Oak Ridge
Paducah
Portsmouth
Composition (%)
Gravel
0-2.3
0-0.4
0-1.3
Sand
85.6-99.3
83.6-93.7
65.8-87.1
Silt + clay
0.2-14.4
5.8-16.3
12.9-34.2
Total organic carbon
(mg/kg)
5,384-38,907
1,296-6.09?
1,328-10,687
pH
7.1-7.7
7.4-7.7 1
7.6-7.9 j

-------
nondetect and the field technology reports a
detection).

False negative results, also assessed for
environmental and spiked samples, indicate the
frequency with which the technology reported a
nondetect (i.e., less than reporting limits) and the
reference laboratory reported a detection.

The reference laboratory results were validated by
ORNL so that fp/fh assessment would not be
influenced by faulty laboratory data. The reporting
limit is considered in the evaluation. For example, if
the reference laboratory reported a result as
0.9 ppm, and the technology's paired result was
reported as below reporting limits (<1 ppm), the
technology's result was considered correct and not a
false negative result.

Completeness
Completeness is defined as the percentage of
measurements that are judged to be usable (i.e., the
result is not rejected). The acceptable completeness
is 95% or greater.

Comparability
Comparability refers to how well the field
technology and reference laboratory data agree. The
difference between accuracy and comparability is
that accuracy is judged relative to a known value,
and comparability is judged relative to the results of
a standard or reference procedure, which may or
may not report the results accurately. The reference
laboratory result is not assumed to be the "correct"
result. This evaluation is performed to compare the
result from the field analytical technology with what
a typical fixed analytical laboratory might report for
the same sample. A one-to-one sample comparison
of the technology results and the reference
laboratory results is performed in Section 5.

A correlation coefficient quantifies the linear
relationship between two measurements (Draper and
Smith 1981). The correlation coefficient, denoted by
the letter r, ranges in value from -1 to +1, where 0
indicates the absence of any linear relationship. The
value r — -1 indicates a perfect negative linear
relation (one measurement decreases as the second
measurement increases); the value r = +1 indicates a
perfect positive linear relation (one measurement
increases as the second measurement increases).
The slope of the linear regression line, denoted by
the letter m, is related to /-. Whereas r represents the
linear association between the vendor and reference
laboratory concentrations, m quantifies the amount
of change in the vendor's measurements relative to
the reference laboratory's measurements. A value of
+1 for the slope indicates perfect agreement. (It
should be noted that the intercept of the line must be
close to zero [i.e., not statistically different from
zero], in order for the slope value of+1 to indicate
perfect agreement.) Values greater than 1 indicate
that the  vendor results are generally higher than
those of the reference laboratory, while values less
than 1 indicate that the vendor results are usually
lower than the values from the reference laboratory.

In addition, a direct comparison between the field
technology and reference laboratory data is
performed by evaluating the percent difference
(%D) between the measured concentrations,
defined as

%D = (\field technology}- [reflab\)l(reflati)
    x 100% .                              (Eq. 3)

The range of %D values is summarized and reported
in Section 5.

Sample Throughput
Sample throughput is a measure of the number of
samples that can be processed and reported by a
technology in a given period of time. This is
reported in Section 5 as number of samples per hour
or day times the number of analysts.

Applicability to Regulatory
Decision-Making
The concentration level of regulatory concern for
PCBs is 50 ppm. When the level of contamination is
above 50 ppm, the material must be managed
according to Toxic Substances Control Act (TSCA)
regulations. To address this issue, the performance
of the technology for samples that fall in the range
of 40 to 60 ppm is independently evaluated.
Precision, accuracy, and comparability to the
reference laboratory are assessed specifically for
this concentration range in Section 5.

Ease of Use
A significant factor in purchasing ?r_ :..iL.Uiiient or a
test kit is how easy the iecnnoiogy is to use. Several
factors are evaluated and reported on in Section 5:

-------
•   What is the required operator skill level (e.g.,
    technician or advanced degree)?
•   How many operators were used during the test?
    Could the technology be run by a single person?
•   How much training would be required in order
    to run this technology?
•   How much subjective decision-making is
    required?

Cost
Another important factor in the consideration of
whether to purchase a technology is cost. Costs
involved with operating the technology and the
standard reference analyses are estimated in
Section 5. To account for the variability in cost data
and assumptions, the economic analysis is presented
as a list of cost elements and a range of costs for
sample analysis. Several factors affect the cost of
analysis. Where possible, these factors are addressed
so that decision makers can independently complete
a site-specific economic analysis to suit their needs.

Miscellaneous Factors
Any other information that might be useful to a
person who is considering purchasing the
technology is documented in Section 5. Examples of
information that might be useful to a prospective
purchaser are the amount of hazardous waste
generated during the analyses, the ruggedness of the
technology, the amount of electrical or battery
power necessary to operate the technology, and
aspects of the technology or method that make it
user-friendly or user-unfriendly.
                                                 10

-------
                 Section 4 — Reference Laboratory Analyses
Reference Laboratory Selection
The verification process is based on the presence of
a statistically validated data set against which the
performance of the technology may be compared.
The choice of an appropriate reference method and
reference laboratory are critical to the success of the
verification test. To assess the performance of the
PCB field analytical technology, the data obtained
from verification test participants were compared
with data obtained using conventional analytical
methods.

The first evaluation of PCB detection technologies
under the ETV program occurred in 1997. LAS
Laboratories, of Las Vegas, Nevada, was selected as
the reference laboratory for that study. A readiness
review conducted by ORNL confirmed the selection
of LAS as the reference laboratory. Acceptance of
the reference laboratory was finalized by
satisfactory performance in a predemonstration
study. ORNL contracted LAS  to provide full data
packages for the verification study sample analyses
within 30 days  of sample shipment. An on-site audit
of LAS occurred August 11-12, 1997, during the
analysis of the verification samples. This
surveillance focused specifically on the procedures
that were currently in use for the analysis of the
verification samples. The audit 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.

A sample holding time study performed by ORNL in
April 2000 indicated that the concentration of PCBs
in the samples had not changed significantly.
Therefore, archived soil samples and the reference
laboratory data generated in 1997 were used for
comparison with the vendor results for the 2000
verification test.

Reference Laboratory Method
The reference laboratory's analytical method,
presented in the technology test plan, followed the
guidelines established in EPA SW-846 Method 8081
(EPA 1994). (Note that since the time of the original
PCB analyses,  Method 8081 was updated to Method
8082 for PCB analyses.) According to LAS
procedures, 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 *
0.53 mm ID x 0.5 |im; confirmatory column: RTX-
5, 30 m x 0.53 mm ID x 0.5 (im). PCBs were
identified when peak patterns from a sample extract
matched the patterns of standards for both columns.
PCBs were quantified on the basis of the initial
calibration of the primary column.

Reference Laboratory Performance
ORNL validated all of the reference laboratory data
according to the procedure described in the test plan
(ORNL 2000). During the validation, the following
aspects of the data were reviewed: completeness of
the data package, adherence to holding time
requirements, correctness of the data, correlation
between replicate sample results, evaluation of QC
sample results, and evaluation of spiked sample
results. Each of these categories  is described in
detail in the test plan. 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.
Approximately 8% of the data had correctable errors
(e.g., transcription, calculation, and interpretation
errors). A small portion of the sample results (5%)
were considered suspect because the reference
laboratory did not report a quantitative result or
because the result was significantly different from
replicate results. The reference laboratory's
performance was evaluated with and without the
suspect values to represent, respectively, the worst-
and best-case scenarios.

The performance of the reference laboratory was
evaluated by statistical analysis of the data
provides a summary of the performance of the
                                               11

-------
   Table 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 no. 110
Sample no. 112
PE
Environmental
<125 ppm
>125ppm
All samples
PE
Environmental
<125 ppm
> 125 ppm
All samples
10 ppm of Aroclor 1242
100 ppm of Aroclor 1254
All samples
Number of
samples
8
16
4
4
63
107
17
187
64
108
20
192
16
16
32
Precision
(av % RSD)
n/a"
n/a"
18
23
• 19
21
21
26
56
28
19
8
14
Accuracy
(av % recovery)
All samples were
reported as nondetects.
AH samples were
reported as nondetects.
101
n/a6
n/a*
101
105
n/a*
n/a*
n/a*
104
64
84
    Because the results were reported as nondetects, precision assessment is not applicable.
   * n/a = not applicable; accuracy assessment calculated for samples of known concentration only.
reference laboratory for the analysis of all sample
types used in the technology verification study.

As shown in Table 7, the precision for the PE soils
was comparable to that for the environmental soils.
A weighted average, based on the number of
samples, gave a best-case precision (i.e., excluding
suspect values) of 21% and a worst-case precision
(i.e., including suspect values) 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 (which ranged from 0 to 50 ppm PCBs)
was  101% for the best case (which excluded the
suspect value) and 105% for the worst case (which
included the suspect value). These results indicate
that the reference laboratory results were unbiased
estimates of the certified PE concentrations.

The 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 nondetects but had difficulty with two
soil samples that contained chemical interferences
(Oak Ridge 2, samples 4 and 6, see Appendix A).
Overall, it was concluded that the reference
laboratory results were acceptable for comparison
with the field analytical technology.
                                                 12

-------
                       Section 5 — Technology Evaluation
Objective and Approach
The purpose of this section is to present a statistical
evaluation of the DELFIA PCB Assay data and
determine the technology's ability to measure PCBs
in contaminated soil and extract samples. This
section includes an evaluation of comparability
through a one-to-one comparison with the reference
laboratory data. Other aspects of the technology
(such as cost, sample throughput, hazardous waste
generation, and logistical operation) are also
evaluated in this section. Appendix A contains the
raw data provided by the vendor during the
verification test that were used to assess the
performance of the DELFIA PCB Assay. During the
verification test, Hybrizyme was provided with
information as to which Aroclor or Aroclors were
present in the sample based on what was reported by
the reference laboratory. Hybrizyme used this
information to determine the final sample results. Li
Appendix B, a data quality objective (DQO)
example of how the data in this report might be used
in a real-world application is presented.

Precision
Precision is the reproducibility of measurements
under a given set of conditions. Precision was
determined by examining the results of blind
analyses for four replicate samples.  Data were
evaluated only for those samples where all four
replicates were reported as a detection. For example,
NR = 43 (43 sets of four replicates) represents a total
of 172 individual sample analyses. A summary of
the overall precision of the DELFIA PCB Assay for
the soil and extract sample results is presented in
Table 8. The mean RSDs for the soil and extract
 Table 8.  Summary of the DELFIA PCB
           Assay Precision
Statistic
Mean
Median
Range
RSD (%)"
Soil samples
(NR = 43*)
20
14
3-99
Extract samples
(NH = 4*)
15
12
8-26
samples were comparable at 20% and 15%,
respectively. The technology's precision was
statistically the same for both outdoor and chamber
conditions.

Accuracy
Accuracy represents the closeness of the DELFIA
PCB Assay's measured concentrations to the known
content of spiked samples. A summary of the
assay's overall accuracy for the soil results is
presented in Table 9. The percent recoveries were
significantly different for data generated under the
outdoor and chamber conditions. The results were
biased high (mean % recovery = 124%) under the
outdoor conditions and biased low (mean %
recovery = 72%) under the chamber conditions.
Based on the performance acceptance ranges shown
in Table 10, which are the guidelines established by
the provider of the spiked materials to gauge
acceptable analytical results, 78% of the results (25
of 32) met the acceptance criteria under the outdoor
conditions, while 88% (28 of 32 of the results) met
the criteria under the chamber conditions. The
accuracy of the extract samples is shown in Table
11. Most of the extract results were biased high,
with larger bias observed under the outdoor
conditions.

False Positive/False Negative Results
Table 12 shows the DELFIA PCB Assay
performance for false positive results for blank
samples. No  fp results were reported for the soil and
extract samples. Table 13 summarizes the assay's fp
and fh results relative to the reference laboratory
results. (See  Section 3 for a more detailed
discussion of this evaluation.) For the environmental
                                                 Table 9.  Summary of the DELFIA PCB Assay
                                                           Accuracy for Soils
 " Calculated only from those samples where all four
 replicates were reported as a detect.
 * NR = number of replicate sets.
Statistic
Mean
Median
Range of results
% recovery
Outdoor
conditions
(N = 32)
124
109
81-387
Chamber
conditions
(N = 32)
72
58
36-188
All data
(N = 64)
98
87
36-387
                                               13

-------
Table 10.  Number of DELFIA PCB Assay Results within Acceptance Ranges for Spiked Soils
Spike concentration
(ppm)
2
20
5
50
10.9
50
2
49.8
Total
Outdoor conditions
Acceptance range
(ppm)
0.7-2.2
11.4-32.4
2.1-6.2
19.7-63.0
4.0-12.8
11.9-75.9
0.9-2.5
23.0-60.8

No. of results
within range
3 of 4
4 of 4
Iof4
4 of 4
Iof4
4 of 4
4 of 4
4 of 4
25 of 32 results
Chamber conditions
Acceptance range
(ppm)
0.7-2.2
11.4-32.4
2.1-6.2
19.7-63.0
4.0-12.8
11.9-75.9
0.9-2.5
23.0-60.8

No. of results
within range
4 of 4
Oof 4
4 of 4
4 of 4
4 of 4
4 of 4
4 of 4
4 of 4
28 of 32 results
     Table 11.  Summary of DELFIA PCB Assay Accuracy for Extracts
Statistic
Mean
Median
Range of results
% recovery
Outdoor conditions
(N = 8)
300
284
267-359
Chamber conditions
(N = 8)
145
153
76-208
All data
(N = 16)
222
238
76-359
                 Table 12. Summary of DELFIA PCB Assay False Positive
                            Performance on Blank Samples
Statistic
Number of data points
Number of fp results
% of fp results
Soil samples
8
0
0
Extract samples
8
0
0
              Table 13. Summary of the DELFIA PCB Assay Detect/
                         Nondetect Performance Relative to the Reference
                         Laboratory Results for Soil Samples (N = 192)
Statistic
False positive (fp) results
False negative (fh) results
No.
0
4
%
0
2
              Note: The reference laboratory did not analyze the extract samples, so fp/fh relative to
              the reference laboratory results could not be evaluated.
                 Of 208 samples, this evaluation excludes the 8 blanks and 8 reference laboratory
              results for which a results could not be generated. ( See Section 4 for more ^.ionviauou
              on these suspect samples.) All remaining 192 samples were reported as detects.
                                             14

-------
and spiked soils, none of the PCS results were
reported as false positives .relative to the reference
laboratory results because the laboratory did not
report any of the 192 samples as a nondetect. Four of
192 samples—2% of the results—were false
negatives, where the laboratory reported a detection
but Hybrizyme reported a nondetect. For those four
samples, Hybrizyme reported each as <0.6 ppm,
while the reference laboratory reported values
between 1.0 and 1.6 ppm. The fp/fh evaluation could
not be performed for the extract samples because the
reference laboratory did not analyze these samples.

Completeness
Completeness is defined as the percentage of
measurements that are judged to be usable (i.e., the
result was not rejected). The DELFIA PCB  Assay
obtained valid results for all 208 soil samples and
24 extract samples. Therefore, completeness was
100%.

Comparability
Comparability refers to how well the DELFIA PCB
Assay and reference laboratory data agreed. In this
evaluation, the laboratory results are not presumed to
be the "correct" answers. Rather, these results
represent what a typical fixed laboratory would
report for these types of samples. A one-to-one
sample comparison of the DELFIA PCB Assay
results and the reference laboratory results was
performed for all environmental and spiked samples
that were reported as a detection (N =  170).  (See
Appendix A to review the raw data and Section 4 for
a complete evaluation of the reference laboratory
results.) Table 14 presents the comparability of the
results in terms of correlation coefficients (r) and
slopes (m). As shown in Table 14, a few suspect
values (two for the reference laboratory and four for
Hybrizyme) influence both the correlation
coefficient (0.50 vs 0.89) and the slope (0.20 vs
0.78). Figure 1 is a plot of the DELFIA PCB Assay
results versus those for the reference laboratory for
all results (N = 164), excluding the Hybrizyme and
reference laboratory suspect values. As this figure
illustrates, Hybrizyme's results generally agreed with
those of the reference laboratory.

Another metric of comparability is the percent
difference (%D) between the reference laboratory
and the DELFIA PCB Assay results (see Section 3).
The ranges of %D values for the PCB results are
presented in Figure 2. Acceptable %D values would
be between -25% and 25%, or near the middle of the
x-axis of the plots. Approximately 45% of the results
are between -25% and 25%.

Comparison of Performance  under
Different Environmental Conditions
The Paducah and PE soil samples were analyzed
under both the outdoor and the chamber  conditions
so that the performance of the DELFIA PCB Assay
could be assessed under different environmental
conditions. When the performance of the DELFIA
PCB Assay is compared with that of the  reference
laboratory for these samples, there is no  statistical
difference between the data set that was  generated
outdoors and that generated in the chamber. The data
sets overlap and are statistically indistinguishable.
However, as shown in Tables 9 and 10, when
DELFIA's results are compared with the nominal
concentrations of the spiked PE samples, there is a
statistical difference between the results  generated
outdoors and those generated in the chamber. The
comparison with the reference laboratory results did
not show statistical differences because of more
uncertainty (i.e., variability) in these two data sets.
       Table 14. DELFIA PCB Assay Correlation with Reference Data
Description of sample set
All values where a detection was reported
Excluding reference suspect values
Excluding Hybrizyme suspect values
Excluding reference and Hybrizyme
suspect values
Number of
samples
170
168
166
164
Correlation coefficient
(r)
0.50
0.50
0.81
0.89
Slope
(m)
0.20
0.20
0.61
0.78
I
                                                15

-------
       120 i
                     20       40        60        80        100       120
                        Reference laboratory PCB concentration (ppm)
140
Figure 1.  Comparison of Hybrizyme and reference laboratory PCB results, excluding nondetects and
          suspect values (N = 164). The slope of the linear regression line is 0.78 and the intercept is 2.6
          ppm.
                               Range of percent difference values
Figure 2. Range of percent difference values.
                                           16

-------
Application to Regulatory
Decision-Making
One of the objectives of this verification test was to
assess the technology's ability to perform at
regulatory decision-making levels for
PCBs—specifically, to detect PCBs at a level
>50 ppm in soils. The technology's performance in
detecting PCBs ranging in concentration from 40 to
60 ppm in PE and environmental soil samples were
used to assess this ability. The performance of the
DELFIA PCB Assay for this concentration range, as
shown in Table 15, was precise (mean RSD = 14%),
unbiased (mean % recovery = 94%), and comparable
to the performance of the reference laboratory (mean
of the absolute value of %D = 27%).
 Table 15. Performance of DELFIA PCB
           Assay on Regulatory Sample PCB
           Concentrations (40-60 ppm)
Statistic
Mean
Median
%
RSD
14
13
%
recovery
94
99
%D
(absolute value)
27
24
Sample Throughput
Sample throughput is representative of the estimated
amount of time required to prepare and analyze the
sample and perform the data analysis. Operating in
both the field and the chamber, the two-person
Hybrizyme team accomplished a sample throughput
rate of approximately six samples per hour for the
208 soil and 24 extract samples.

Ease of Use
Two operators were used for the test because of the
number of samples and working conditions, but the
technology can be operated by a single person. Users
unfamiliar with immunoassay techniques may need
approximately one-half day of additional training to
operate the instrument. No particular level of
educational training is required for the operator.

Cost Assessment
The purpose of this economic analysis is to estimate
the range of costs for analysis of PCB-contaminated
soil samples using the DELFIA PCB Assay and a
conventional analytical reference laboratory method.
The analysis was based on the results and experience
gained from this verification test, costs provided by
Hybrizyme, and representative costs provided by the
reference analytical laboratories that offered to
analyze these samples. To account for the variability
in cost data and assumptions, the economic analysis
is presented as a list of cost elements and a range of
costs for sample analysis by the DELFIA PCB Assay
instrument and by the reference laboratory.

Several factors affected the cost of analysis. Where
possible, these factors were addressed so that
decision makers can complete a site-specific
economic analysis to suit their needs. The following
categories are considered in the estimate:

    sample shipment costs,
•   labor costs, and
•   equipment costs.

Each of these cost factors is defined and discussed
and serves as the basis for the estimated cost ranges
presented in Table 16. This analysis assumed that  the
individuals performing the analyses  were fully
trained to operate the technology. Costs for sample
acquisition and pre-analytical sample preparation,
which are tasks common to both methods, were not
included in this assessment.

DELFIA PCB Assay Costs
The costs associated with using the DELFIA PCB
Assay instrument included labor, equipment, and
waste disposal costs.  No sample shipment charges
were associated with the cost of operating the
instrument because the samples were analyzed on-
site.

Labor
Labor costs included mobilization and
demobilization, travel, per diem expenses, and on-
site labor.

•   Mobilization and demobilization. This cost
    element included the time for one person to
    prepare for and travel to each site. This estimate
    ranged from zero (if the analyst is located on
    site) to 5 h, at a rate of $50/h.
•   Travel. This element was the cost for the
    analyst(s) to travel to the site. If the analyst is
    located at the site, the cost of commuting to the
    site would be zero. The estimated cost of °r.
                                               17

-------
    Table 16. Estimated Analytical Costs for PCB-Contaminated Samples
   Analysis method:
   Analyst/manufacturer:
DELFIA PCB Assay
Hybrizyme
    Sample throughput:    6 samples/h
Analysis method:
Analyst/manufacturer:
EPA SW-486 Method 8081
Reference laboratory
                         Typical turnaround:    14-30 working days
    Cost category
      Cost (S)
Cost category
      Cost (S)
    Sample shipment
    Labor
      Mobilization/demobilization
      Travel
      Per diem expenses
      Rate
    Equipment
      Mobilization/demobilization
      Instrument purchase price
      Instrument lease price
      Reagents/supplies	
      0
      0-250
      0-1,000 per analyst
      0-150/day per analyst
      30-75/h per analyst

      0-150
      30,000
      500 per week
      22.50 per sample
Sample shipment
   Labor
   Overnight shipping
Labor
   Mobilization/demobilization
   Travel
   Per diem expenses
   Rate
Equipment
      100-200
      50-150

      Included"
      Included
      Included
      44-239 per sample
      Included
   " "Included" indicates that the cost is included in the labor rate.
   analyst traveling to the site for this verification
   test ($1000) included the cost of airline travel
   and rental car fees.
•  Per diem expenses. This cost element included
   food, lodging, and incidental expenses. The
   estimate ranged from zero (for a local site) to
   $150/day for each analyst.
•  Rate. The cost of the on-site labor was estimated
   at a rate  of $30-75/h, depending on the required
   expertise level of the analyst. This cost element
   included the labor involved with the entire
   analytical process, comprising sample
   preparation, sample management, analysis, and
   reporting.

Equipment
Equipment costs included mobilization and
demobilization, rental fees or purchase of equipment,
and the reagents and other consumable supplies
necessary to complete the analysis.

•  Mobilization and 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  verification test, the cost of shipping
   equipment and supplies was estimated at $150.
•  Instrument purchase/lease. The time-resolved
   fluorometer can be purchased for $30,000. The
   instrument can also be leased on a weekly basis
   for $500 per week.
                          •   Reagents and supplies. Hybrizyme PCB
                             DELFIA Reagent Kit provides 40 sample
                             analysis. The retail price is $22.50 per sample
                             (which includes duplicates and controls).

                          Reference Laboratory Costs
                          Sample Shipment
                          Sample shipment costs to the reference laboratory
                          included overnight shipping charges, as well as labor
                          charges associated with the various organizations
                          involved in the shipping process.

                          •   Labor. This cost element included all of the
                             tasks associated with the shipment of the
                             samples  to the reference laboratory. Tasks
                             included packing the shipping coolers,
                             completing the chain-of-custody documentation,
                             and completing the shipping forms. The estimate
                             to complete this task ranged from 2 to 4 h at
                             $50/h.
                          •   Overnight shipping. The overnight express
                             shipping service cost was estimated to be $50 for
                             one 50-lb cooler of samples.

                          Labor, Equipment, and Waste Disposal
                          The labor bids from commercial analytical reference
                          laboratories  that offered to perform the reference
                          analysis for this verification test rangea rrom $44 to
                          $239 per sample. The bid was dependent on many
                          factors, including the perceived difficulty of the
                                                 18

-------
sample matrix, the current workload of the
laboratory, and the competitiveness of the market.
This rate was a fully loaded analytical cost that
included equipment, labor, waste disposal, and
report preparation.

Cost Assessment Summary
An overall cost estimate for use of the DELFIA PCB
Assay instrument versus use of the reference
laboratory was not made because of the extent of
variation in the different cost factors, as outlined in
Table 16. The overall costs for the application of any
technology would 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's
providing immediate answers versus the reference
laboratory's provision of reporting data within
30 days of receipt of samples.

Miscellaneous Factors
The following are general observations regarding the
field operation and performance of the DELFIA PCB
Assay instrument:

•   The system included a time-resolved fluorometer
    that was transportable by one person; however, it
    is rather large instrument (41.5 kg) that requires
    110 V of electrical power.
•   During outdoor tests, the Hybrizyme team used a
    portable air conditioner to cool their tent setup.
    Because the tent was not air-tight, the
    temperature inside the tent was not much cooler
    than the .outdoor temperature.
•   The Hybrizyme technology allowed the
    processing of 40 samples at one time.
•   All 208 soil samples and 24 extracts were
    initially analyzed using a protocol to detect 1
    ppm PCBs (a range of 0.5 to 3.2 ppm). Sample
    dilution and additional analyses were required to
    detect PCB concentrations from 3.2 ppm to > 150
    ppm. In all, the Hybrizyme team performed 436
    analyses over the  four days of testing.
•   Hybrizyme used information on which Aroclors
    were in the samples to determine the final
    sample result (based on instrumental response
    for each Aroclor). If the Aroclor had been
    unknown, Hybrizyme would have used the
    calibration curve for Aroclor 1248.
•   Tests with the Hybrizyme assay generated the
    following wastes: 13 L of soil/methanol mixture
    (classified as RCRA/TSCA waste), 95 L of
    TSCA-regulated solids (glass, paper, plastic,
    etc.), and 6.8 L of PCB-detectable, non-TSCA
    aqueous waste.

Summary of Performance
A summary of the performance of DELFIA PCB
Assay is presented in Table 17. Precision, defined as
the mean RSD, was 20% for soils and 15% for
extracts. Accuracy, defined as the mean percent
recovery relative to the spiked concentration, was
124% under the  outdoor conditions (biased high) and
72% under the chamber conditions (biased low).
There was a statistical difference between the data
generated under the outdoor and chamber conditions.
For the extracts, most of the sample results were
biased high. No  false positives were reported for the
soil and extract blanks. Additionally,  false positive
and false negative results were determined by
comparing the DELFIA PCB Assay result to the
reference laboratory result for the environmental and
spiked samples.  None of the results were reported as
false positives, but 2% were false negatives. A
subset of the data was evaluated to assess the
technology's ability to detect PCB contamination at
levels that are of regulatory concern (i.e., >50 ppm).
The technology  was precise (14% RSD), accurate
(94% recovery), and comparable to the reference
laboratory (27% absolute value of %D) for this soil
concentration range.

The verification test found that the DELFIA PCB
Assay instrument was relatively simple for a trained
analyst to operate in the field, requiring less than an
hour for initial setup. The sample throughput of the
DELFIA PCB Assay was six samples per hour. Two
operators analyzed samples during the verification
test, but the technology can be run by a single trained
operator. The overall performance of the DELFIA
PCB Assay for the analysis of PCBs  in soil and
solvent extracts  was characterized as  biased
(dependent on environmental conditions) but precise.
                                               19

-------
Table 17. Performance Summary for the DELFIA PCB Assay
Feature/parameter
Precision
Accuracy
False positive results on blank
samples
False positive results relative to
reference laboratory results
False negative results relative to
reference laboratory results
Comparison with reference laboratory
results (all data, excluding suspect
values)
Regulatory decision-making
(40 to 60 ppm soil)
Completeness
Weight of time-resolved fluorimeter
Sample throughput (2 operators)
Power requirements
Training requirements
Cost
Waste generated
Overall evaluation
Performance summary
Mean RSD
Soil: 20%
Extract: 15%
Mean recovery (significantly different for the two conditions)
Soil
Outdoor: 124%
Chamber: 72%
Extract
Outdoor: 300%
Chamber: 145%
Soil: none
Extract: none
None
2% (4 of 192 samples)
Median absolute
r m % D
All values: 0.50 0.20 29%
Excluding suspect values: 0.89 0.78 29%
RSD: 14%
% recovery: 94%
Abs %D: 27%
100% of 208 soil samples and 24 extract samples
41.5kg
6 samples/h
110V
One-half day technology-specific training
Instrument purchase: $30,000
Instrument lease: $500 per week
Reagents/supplies: $22.50 per sample
13 L of soil/methanol mixture (classified as RCRA/TSCA)
95 L of TSCA-regulated solids (glass, paper, plastic, etc.)
6.8 L of PCB-detectable, non-TSCA aqueous waste
(Total number of samples analyzed: 232)
Precise
Biased high for outdoor conditions
Biased low for chamber conditions
                                     20

-------
                      Section 6 — Technology Update and
                           Representative Applications
In this section, the vendor (with minimal editorial changes by ORNL) provides information regarding new
developments with its technology since the verification activities. In addition, the vendor provides a list of
representative applications in which its technology has been used.
Temperature Control
The Hybrizyme assay system is designed for
laboratory or mobile laboratory use. For applications
beyond the normal temperature variations that occur
indoors, the Victor™ Time-Resolved Fluorometer
can be equipped with temperature control. In
addition, calibrators included within each sample
batch can be used to automatically compensate for
extreme temperature conditions. The data contained
within Uiis ETV report was obtained without
controlling for temperature fluctuations.
Food Test Validation
Hybrizyme's DELFIA PCB assay has been validated
for testing food products by the European
Commission's Joint Research Centre, Institute for
Health and Consumer Protection, Food Products
Unit, Ispra, Italy. A report on the validation results,
entitled "Use of an immunoassay as a means to
detect polychlorinated biphenyls in animal fat," by
S. Jaborek-Hugo et al., has been accepted for
publication in Food Additives & Contaminants, ed.
John Gilbert (Taylor & Francis Ltd., London).
                                             21

-------
                               Section 7 — References
American Society for Testing and Materials (ASTM). 1997a. Standard Practice for Generation of
Environmental Data Related to Waste Management Activities: Quality Assurance and Quality Control
Planning and Implementation. D5283-92.

American Society for Testing and Materials (ASTM). 1997b. Standard Practice for Generation of
Environmental Data Related to Waste Management Activities: Development of Data Quality Objectives.
D5792-95.

Berger, W., H. McCarty, and R-K. Smith. 1996. Environmental Laboratory Data Evaluation. Genium
Publishing Corp., Schenectady, N.Y.

Draper, N. R., andH. Smith. 1981. Applied Regression Analysis. 2nded. John Wiley & Sons, New York.

EPA (U.S. Environmental Protection Agency). 1994. "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 EL Office of Solid Waste and Emergency
Response, Washington, D.C., September.

EPA (U.S. Environmental Protection Agency). 1996. Guidance for Data Quality Assessment. EPA QA/G-9;
EPA/600/R-96/084. EPA, Washington, D.C., July.

Erickson, M. D. 1997. Analytical Chemistry of PCBs. 2nd ed. CRC Press/Lewis Publishers, Boca Raton, Fla.

Maskarinec, M. P., et al. 1992. Stability of Volatile Organics in Environmental Soil Samples.  ORNL/TM-
12128. Oak Ridge National Laboratory, Oak Ridge, Term., November.

ORNL (Oak Ridge National Laboratory). 2000. Technology Verification Test Plan: Evaluation of
Poly chlorinated Biphenyl (PCB) Field Analytical Techniques. Chemical and Analytical Sciences Division,
Oak Ridge National Laboratory, Oak Ridge, Term., August.
                                              22

-------
                  Appendix A

Hybrizyme's DELFIA PCB Assay Results Compared
       with Reference Laboratory Results
Location
Outside
Outside
Outside
Outside
Outside
Outside
Outside
Outside
Outside
Outside
Outside
Outside
Outside
Outside
Outside
Outside
Outside
Outside
Outside
Outside
Outside
Outside
Outside
Outside
Outside
Outside
Outside
Outside
Sample site
or type
Oak Ridge 1
Oak Ridge 1
Oak Ridge 1
Oak Ridge 1
Oak Ridge 1
Oak Ridge 1
Oak Ridge 1
Oak Ridge 1
Oak Ridge 1
Oak Ridge 1
Oak Ridge 1
Oak Ridge 1
Oak Ridge 1
Oak Ridge 1
Oak Ridge 1
Oak Ridge 1
Oak Ridge 1
Oak Ridge 1
Oak Ridge 1
Oak Ridge 1
Oak Ridge 1
Oak Ridge 1
Oak Ridge 1
Oak Ridge 1
Oak Ridge 2
Oak Ridge 2
Oak Ridge 2
Oak Ridge 2
Sample
no.
1
1
1
1
2
2
2
2
3
3
3
3
4
4
4
4
5
5
5
5
6
6
6
6
1
1
1
1
Sample
replicate
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
Total PCB
DELFIA
1.0
0.6
0.8
0.5
2.5
3.6
2.5
3.2
3.9
3.7
6.0
5.7
10.6
11.2
12.3
10.5
56.6
61.3
61.5
51.2
>150
>150
>150
>150
1.2
1.3
2.8
1.1
cone, (ppm)
Reference
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
Hybrizyme
analysis
order"
1091
1025
1063
1056
1001
1062
1085
1059
1094
1015
1020
1027
1058
1070
1082
1054
1098
1013
1017
1076
1030
1009
1053
1103
1022
1074
1100
1101
                      23

-------
Location
Outside
Outside
Outside
Outside
Outside
Outside
Outside
Outside
Outside
Outside
Outside
Outside
Outside
Outside
Outside
Outside
Outside
Outside
Outside
Outside
Outside
Outside
Outside
Outside
Outside
Outside
Outside
Outside
Outside
Outside
Outside
Outside
Sample site
or type
Oak Ridge 2
Oak Ridge 2
Oak Ridge 2
Oak Ridge 2
Oak Ridge 2
Oak Ridge 2
Oak Ridge 2
Oak Ridge 2
Oak Ridge 2
Oak Ridge 2
Oak Ridge 2
Oak Ridge 2
Oak Ridge 2
Oak Ridge 2
Oak Ridge 2
Oak Ridge 2
Oak Ridge 2
Oak Ridge 2
Oak Ridge 2
Oak Ridge 2
Paducah
Paducah
Paducah
Paducah
Paducah
Paducah
Paducah
Paducah
Paducah
Paducah
Paducah
Paducah
Sample
no.
2
2
2
2
3
3
3
3
4
4
4
4
5
5
5
5
6
6
6
6
1
1
1
1
2
2
2
• 2
3
3
3
3
Sample
replicate
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
Total PCB
DELFIA
1.6
1.5
1.5
1.8
2.2
1.9
2.2
1.9
30.9
20.9
32.2
37.2
58.3
45.2
52.9
36.3
139.0
129.0
128.0
158.0
1.1
1.0
1.0
1.0
1.1
1.1
0.8
1.1
17.2
16.0
17.3
13.7
cone, (ppm)
Reference
1.7
2.0
1.7
1.9
1.5
2.1
1.8
2.4
<490
<99
<66
<98
44.5
36.0
39.3
35.1
<66
<200
<130
<200
0.7
1.1
0.6
1.9
1.1
1.2
1.3
1.7
14.9
12.4
15.0
16.9
Hybrizyme
analysis
order "
1057
1023
1081
1061
1031
1087
1044
1084
1037 *
1093 *
1008 *
1032 *
1099
1066
1014
1072
1086 b
1083 *
1007 *
1034 *
1065
1041
1090
1067
1026
1010
1052
1033
1028
1080
1073
1006
24

-------
Location
Outside
Outside
Outside
Outside
Outside
Outside
Outside
Outside
Outside
Outside
Outside
Outside
Outside
Outside
Outside
Outside
Outside
Outside
Outside
Outside
Outside
Outside
Outside
Outside
Outside
Outside
Outside
Outside
Outside
Outside
Outside
Outside
Sample site
or type
Paducah
Paducah
Paducah
Paducah
Paducah
Paducah
Paducah
Paducah
Spike/PE
Spike/PE
Spike/PE
Spike/PE
Spike/PE
Spike/PE
Spike/PE
Spike/PE
Spike/PE
Spike/PE
Spike/PE
Spike/PE
Spike/PE
Spike/PE
Spike/PE
Spike/PE
Spike/PE
Spike/PE
Spike/PE
Spike/PE
Spike/PE
Spike/PE
Spike/PE
Spike/PE
Sample
no.
4
4
4
4
5
5
5
5
1
1
1
1
2
2
2
2
3
3
3
3
4
4
4 .
4
5
5
5
5
6
6
6
6
Sample
replicate
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
Total PCB
DELFIA
42.2
37.0
22.8
47.1
>150
>150
>150
>150
1.8
2.3
1.7
1.9
17.9
16.1
17.1
16.6
8.7
9.6
7.5
5.7
52.1
62.6
53.6
52.6
14.0
42.2
21.6
11.8
67.1
63.5
65.4
48.5
cone, (ppm)
Reference
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
58.7
55.7
53.2
50.9
12.2
10.9
11.3
10.0
59.2
56.9
66.8
57.5
Hybrizyme
analysis
order °
1078
1075
1029
1002
1024
1102
1096
1092
1046
1097
1051
1048
1047
1060
1036
1055
1071
1035
1079
1050
1064
1089
1043
1003
1077
1040
1016
1069
1012
1049
1039
1095
25

-------
Location
Outside
Outside
Outside
Outside
Outside
Outside
Outside
Outside
Outside
Outside
Outside
Outside
Outside
Outside
Outside
Outside
Outside
Outside
Outside
Outside
Outside
Outside
Outside
Outside
Chamber
Chamber
Chamber
Chamber
Chamber
Chamber
Chamber
Chamber
Sample site
or type
Spike/PE
Spike/PE
Spike/PE
Spike/PE
Spike/PE
Spike/PE
Spike/PE
Spike/PE
Soil Blank
Soil Blank
Soil Blank
Soil Blank
Extract Blank
Extract Blank
Extract Blank
Extract Blank
Extract
Extract
Extract
Extract
Extract
Extract
Extract
Extract
Paducah
Paducah
Paducah
Paducah
Paducah
Paducah
Paducah
Paducah
Sample
no.
7
7
7
7
8
8
8
8
1
1
1
1
1
1
1
1
1
1
1
1
2
2
2
2
1
1
1
1
2
2
2
2
Sample
replicate
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
Total PCB
DELFIA
2.4
2.0
2.2
2.1
54.0
55.0
56.0
52.1
<0.5
<0.5
<0.5
<0.5
<0.5
<0.5
<0.5
<0.5
28.4
28.4
27.9
35.9
267.0
271.0
342.0
311.0
0.8
<0.6
0.7
0.7
0.8
<0.6
<0.6
<0.6
cone, (ppm)
Reference
1.8
1.4
1.9
1.8
32.0
41.3
46.0
32.2
<0.1
<0.1
<0.2
<1.3
n/a'
n/ac
n/a c
n/a c
n/ac
n/ac
n/ac
n/ac
n/ac
n/ac
n/ac
n/ac
1.0
1.0
1.1
0.6
1.4
1.6
1.2
1.5
Hybrizyme
analysis
order "
1045
1005
1042
1038
1018
1068
1088
1004
1011
1021
1019
1104
1116
1106
1111
1108
1113
1112
1105
1115
1109
1110
1107
1114
2020
2052
2059
2048
2079
2066
2099
2017
26

-------
Location
Chamber
Chamber
Chamber
Chamber
Chamber
Chamber
Chamber
Chamber
Chamber
Chamber
Chamber
Chamber
Chamber
Chamber
Chamber
Chamber
Chamber
Chamber
Chamber
Chamber
Chamber
Chamber
Chamber
Chamber
Chamber
Chamber
Chamber
Chamber
Chamber
Chamber
Chamber
Chamber
Sample site
or type
Paducah
Paducah
Paducah
Paducah
Paducah
Paducah
Paducah
Paducah
Paducah
Paducah
Paducah
Paducah
Portsmouth 1
Portsmouth 1
Portsmouth 1
Portsmouth 1
Portsmouth 1
Portsmouth 1
Portsmouth 1
Portsmouth 1
Portsmouth 1
Portsmouth 1
Portsmouth 1
Portsmouth 1
Portsmouth 1
Portsmouth 1
Portsmouth 1
Portsmouth 1
Portsmouth 1
Portsmouth 1
Portsmouth 1
Portsmouth 1
Sample
no.
3
3
3
3
4
4
4
4
5
5
5
5
1
1
1
1
2
2
2
2
3
3
3
3
4
4
4
4
5
5
5
5
Sample
replicate
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
Total PCS
DELFIA
9.0
8.1
9.2
7.5
42.7
58.0
51.0
46.4
>150
>150
>150
>150
0.9
0.8
1.0
1.1
13.5
12.5
16.3
12.0
19.8
28.7
22.9
33.5
35.1
29.6
25.6
28.7
61.0
64.1
67.1
50.2
cone, (ppm)
Reference
14.0
12.8
16.2
12.4
43.1
45.3
41.0
47.7
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
Hybrizyme
analysis
order "
2096
2053
2102
2022
2057
2103
2067
2031
2098
2078
2080
2011
2076
2028
2047
2004
2039
2007
2026
2005
2033
2100
2070
2063
2032
2094
2069
2025
2101
2071
2006
2081
27

-------
Location
Chamber
Chamber
Chamber
Chamber
Chamber
Chamber
Chamber
Chamber
Chamber
Chamber
Chamber
Chamber
Chamber
Chamber
Chamber
Chamber
Chamber
Chamber
Chamber
Chamber
Chamber
Chamber
Chamber
Chamber
Chamber
Chamber
Chamber
Chamber
Chamber
Chamber
Chamber
Chamber
Sample site
or type
Portsmouth 1
Portsmouth 1
Portsmouth 1
Portsmouth 1
Portsmouth 2
Portsmouth 2
Portsmouth 2
Portsmouth 2
Portsmouth 2
Portsmouth 2
Portsmouth 2
Portsmouth 2
Portsmouth 2
Portsmouth 2
Portsmouth 2
Portsmouth 2
Portsmouth 2
Portsmouth 2
Portsmouth 2
Portsmouth 2
Portsmouth 2
Portsmouth 2
Portsmouth 2
Portsmouth 2
Portsmouth 2
Portsmouth 2
Portsmouth 2
Portsmouth 2
Spike/PE
Spike/PE
Spike/PE
Spike/PE
Sample
no.
6
6
6
6
1
1
1
1
2
2
2
2
3
3
3
3
4
4
4
4
5
5
5
5
6
6
6
6
1
1
1
1
Sample
replicate
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
Total PCB
DELFIA
124.0
>150
>150
>150
2.9
2.6
3.0
6.0
4.9
3.2
2.7
5.5
30.3
21.9
17.4
18.8
22.9
17.9
3.1
24.8
35.5
31.1
31.3
40.5
>150
>150
3.0
>150
1.4
1.3
2.0
1.4
cone, (ppm)
Reference
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
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
Hybrizyme
analysis
order "
2015
2092
2073
2012
2087
2010
2008
2002
2058
2061
2049
2104
2097
2093
.2019
2077
2036
2035
2050
2060
2030
2056
2091
2089
2074
2014
2045
2021
2038
2084
2040
2023
28

-------
Location
Chamber
Chamber
Chamber
Chamber
Chamber
Chamber
Chamber
Chamber
Chamber
Chamber
Chamber
Chamber
Chamber
Chamber
Chamber
Chamber
Chamber
Chamber
Chamber
Chamber
Chamber
Chamber
Chamber
Chamber
Chamber
Chamber
Chamber
Chamber
Chamber
Chamber
Chamber
Chamber
Sample site
or type
Spike/PE
Spike/PE
Spike/PE
Spike/PE
Spike/PE
Spike/PE
Spike/PE
Spike/PE
Spike/PE
Spike/PE
Spike/PE
Spike/PE
Spike/PE
Spike/PE
Spike/PE
Spike/PE
Spike/PE
Spike/PE
Spike/PE
Spike/PE
Spike/PE
Spike/PE
Spike/PE
Spike/PE
Spike/PE
Spike/PE
Spike/PE
Spike/PE
Soil Blank
Soil Blank
Soil Blank
Soil Blank
Sample
no.
2
2
2
2
3
3
3
3
4
4
4
4
5
5
5
5
6
6
6
6
7
7
7
7
8
8
8
8
1
1
1
1
Sample
replicate
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
Total PCB
DELFIA
7.5
7.1
37.6
8.4
3.1
2.6
2.8
3.0
22.8
33.3
29.7
38.9
6.7
8.8
6.9
7.3
34.2
34.1
33.8
40.8
1.3
1.5
1.6
1.7
44.2
40.4
50.2
37.4
<0.5
<0.5
<0.6
<0.6
cone, (ppm)
Reference
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 A
12.7
12.7
56.6
50.3
49.9
66.4
2.2
1.2
1.4
2.1
56.4
36.5
32.1
146.0
<0.1
<0.8
<0.1
<0.1
Hybrizyme
analysis
order "
2024
2042
2034
2027
2018
2016
2088
2083
2062
2085
2090
2003
2082
2001
2051
2043
2013
2046
2075
2064
2037
2065
2041
2068
2072
2086
2029
2095
2009
2044
2054
2055
29

-------
Location
Chamber
Chamber
Chamber
Chamber
Chamber
Chamber
Chamber
Chamber
Chamber
Chamber
Chamber
Chamber
Sample site
or type
Extract Blank
Extract Blank
Extract Blank
Extract Blank
Extract
Extract
Extract
Extract
Extract
Extract
Extract
Extract
Sample
no.
1
1
1
1
1
1
1
1
2
2
2
2
Sample
replicate
1
2
3
4
1
2
3
4
1
2
3
4
Total PCB
DELFIA
<0.5
<0.5
<0.5
<0.5
20.8
17.2
18.5
19.0
83.8
76.3
111.0
133.7
cone, (ppm)
Reference
n/a c
n/a c
n/a c
n/a *
n/ac
n/a<
n/ac
n/ac
n/ac
n/ac
n/ac
n/ac
Hybrizyme
analysis
order "
2111
2113
2112
2114
2106
2115
2105
2109
2107
2116
2108
2110
" Indicates order of analysis by Hybrizyme; for example, 1001 was analyzed first, then 1002, etc.
* Reference laboratory had trouble analyzing these samples. See Section 4 for more details.
c Reference laboratory did not analyze these extract samples.
                                                    30

-------
                                       Appendix B

                    Data Quality Objective (DQO) Example
Disclaimer
The following hypothetical example demonstrates how the information provided in this report may be used in
the data quality objective (DQO) process. While this example illustrates the application of quantitative DQOs
to a decision process, it cannot attempt to provide a thorough education in this topic. Please refer to other
educational or technical resources for further details (e.g., ASTM 1997a, b; EPA 1996). In addition, because
the focus of this report is on the analytical technology, this example makes simplifying assumptions (such as
that the sample is homogeneous and that the reference laboratory results represent the true concentration) that
may not be valid in the real world.

Background and Problem Statement
An industrial company discovered a land area contaminated with PCBs from an unknown source. The
contaminated soil was excavated into waste drums. Preliminary characterization determined that the PCB
concentration in a single drum was homogenous, but PCB concentrations varied greatly from drum to drum.
The company's DQO team was considering the use of Hybrizyme's DELFIA PCB Assay to measure the PCB
concentration in each drum. The DQO team decided that drums will be disposed of by incineration if the PCB
concentration is k50 ppm ("hot"). A concentration of 50 ppm is the TSCA regulatory threshold (RT) for this
environmental problem. Those drums with PCB concentrations <50 ppm will be put into a landfill because
incineration of soil is very expensive. With regulator agreement, the DQO team determined that a decision
rule for disposal would be based on the average concentration of PCBs in each drum.
                                     General Decision Rule

          If average PCB concentration < action level, then send the soil drum to the landfill.

          If average PCB concentration > action level, then send the soil drum to the
          incinerator.
DQO Goals
The DQO team's primary goal was to calculate how many samples would need to be analyzed by the DELFIA
PCB Assay in order to confidently make a decision about remediating the processed soil, given the
uncertainties of the technology's results. The worst possible mistake would be to send a drum to the landfill
with PCB concentrations exceeding 50 ppm. The error rate of this false-rejection decision would serve as the
primary determinant for the number of samples measured. A secondary decision error would be to
unnecessarily send an excessive number of drums to the incinerator if the average PCB concentration was <50
ppm. This decision error would be a false-acceptance decision error. Both the false-rejection decision error
and the false-acceptance decision error were taken into account to determine the final sampling plan.

EPA required that a sufficient number of samples be measured from each drum so that the false-rejection error
rate (FR) for the decision  rule was 0.05 or less if the true drum concentration was >50 ppm. This DQO goal
represents a 5% chance of sending to a landfill those drums with PCB concentrations >50 ppm.

The DQO team did not want to send an excessive number of drums to the incinerator if the average PCB
concentration was <50 ppm because of the expense. In this situation, a false-acceptance decision is made
when it is concluded that a drum is "hot" when,  in actuality, the drum contains soil with PCB contamination
                                               31

-------
<50 ppm. Therefore, the DQO team recommended that the false-acceptance decision error rate (FA) be 0.10 if
the true PCB concentration is 40 ppm. That is, there would be a 10% probability of sending a drum to the
incinerator (denoted as Pr[Take Drum to Incinerator]) if the true PCB concentration for a drum is 40 ppm.
                  Permissible FR and FA Error Rates and Critical Decision Points

            FR: PrfTake Drum to Landfill] ^0.05 when true PCB concentration = 50 ppm

            FA: Pr[Take Drum to Incinerator] ^0.10 when true PCB concentration = 40 ppm
Use of Technology Performance Information to Implement the Decision Rule
Technology performance information is used to evaluate whether a particular analytical technology can
produce data of sufficient quality to support the site decision. Because the DQO team was considering the use
of the Hybrizyme's DELFIA PCB Assay, the performance of this technology (as reported in this ETV report)
was used to assess its applicability to this project. Two questions arise:

1.  How many samples are needed from a single drum to permit a valid estimation of the true average
   concentration of PCBs in the drum to the specified certainty? Recall that the simplifying assumption was
   made that the PCB distribution throughout the soil within a single drum is homogeneous, and thus, matrix
   heterogeneity will not contribute to overall variability. The only variability, then, to be considered hi this
   example is the variability in the DELFIA PCB Assay's analytical method, which is determined by
   precision studies.

2.   What is the appropriate action level (AL) for using the Hybrizyme's DELFIA PCB Assay to make
   decisions in the field? After the required number of samples have been collected from a drum and
   analyzed, the results are averaged together to get an estimate of the "true" PCB concentration of the drum.
   When using the DELFIA PCB Assay, what is the value (here called "the action level for the decision
   rule") to which that average is compared to decide if the drum is "hot" or not? This method-specific or
   site-specific action level is derived from evaluations of the method's accuracy using an appropriate quality
   control regimen.

Hybrizyme's DELFIA PCB Assay Accuracy
The ETV verification test results indicated that the DELFIA PCB Assay's accuracy for soil samples showed a
statistically significant difference between data generated  under the outdoor and chamber conditions. The
results were biased slightly high (mean % recovery = 124%) under the outdoor conditions, and biased slightly
low (mean % recovery = 72%) under the chamber conditions. For this example, the testing will occur during
warm temperatures similar to the outdoor test runs. Colder temperatures would be similar to the chamber
conditions. Average replicate PCB concentrations determined by the DELFIA PCB Assay in outdoor
conditions showed a strong linear correlation (R2 = 0.96) with the certified values for the performance
evaluation samples. This correlation is represented by a line fitted to the data that predicts the expected
DELFIA PCB Assay's concentration from the certified PE value. Figure B-l shows this linear relationship
with the PCB concentrations plotted against the certified PCB values for the PE samples, which included the
concentration range of 0 to 50 ppm. The arrow on the plot in Figure B-l demonstrates a method to quickly
estimate a corrected PCB concentration from a DELFIA PCB Assay measurement. For example, a DELFIA
PCB Assay concentration of 50 ppm would correspond to a certified PCB concentration of 44  ppm.  The
equation for the PCB prediction line is

Delfia Result = 1.65 + J.JO x (CertifiedPE Value)                                          (Eq. B-l)


                                              32

-------
The critical decision points, 40 ppm and 50 ppm,
correspond to DELFIA PCB Assay results of
45.7 ppm and 56.7 ppm, respectively. The DQO team
knew that if they selected the DELFIA PCB Assay
for this project, they would have to compensate for
the bias. Compensation may be performed either by a
graphical method using a calibration line such as
Figure B-l or by a calibration equation such as B-l.

Determining the Number of Samples
With the critical decision points selected, the DQO
team could then determine the number of samples
needed from each drum to calculate the drum's "true"
average PCB concentration. For a homogeneous
matrix, the number of samples required depends on
the precision of the analytical method.

The DELFIA PCB Assay's replicate results for each
sample from the ETV verification test established
that the standard deviation for PE samples could be
approximated by a linear model within the
concentration range of 0 to 50 ppm (see Figure B-2).
The equation for the line is

DELFIA SD = 2.80 + 0.05 x  (CertifiedPE Value)
                                       (Eq.B-2)

This estimate of analytical variability (precision) is
used to calculate the number of soil samples required
to be analyzed from each drum to achieve the DQO
goals for FR and FA error rates. A formula is
provided in EPA's Guidance for Data Quality
Assessment (EPA 1996, pp. 3.2-3, Box 3.2-1) that can
be adapted to this example for calculating the number
of samples required to meet the FR and FA
requirements:
N =
            (Zl-FR  * Zl-FAf
             (XT - c
                                       (Eq.B-3)
where
    N
    5"
    RT
    CFA
    FR
                                                       Certified PCB Concentration (ppm)
                                              Figure B-l.  A linear model for predicting DELFIA PCB
                                                         Assay concentrations from certified PCB
                                                         concentrations with 95% confidence
                                                         intervals (dashed lines).
                                                          10
                                                                20
                                                                       30
                                                                                    a>
                                                        Certified PCB Cm nitration (ppr)
                                              Figure B-2.  A linear model fitted to DELFIA PCB Assay
                                                         standard deviation versus certified PCB
                                                         concentration.
       number of samples from a drum to be measured
       variance for the measurement [e.g., 5° = (2.80 + 0.05 * Certified PE Value)2 ]
       regulatory threshold (e.g., RT= 50 ppm)
       concentration at which the FA is specified (e.g., CFA = 40 ppm)
       false-rejection decision error rate (e.g., FR = 0.05)
                                               33

-------
   FA   = false-acceptance decision error rate (e.g., FA = 0. 10)
   Z,_p  =  the (1 - p)"1 percentile of the standard normal distribution (see EPA 1996, Appendix A, Table A-
           1) (e.g., Z(1_ra) = Z0.95 = 1.645).
Incorporating the appropriate values for the DELFIA PCB Assay into Eq. B-3 gives
            N __  (2.80 + 0.05 x 50)* (1.645 +  1.282)* +  (0.5)(L645)2 = 3J6 m ,
                               (50  - 40)2
Therefore, four samples from each drum would be analyzed by Hybrizyme's DELFIA PCB Assay to meet the
criteria established by the DQO process. Note that, to be conservative, one would evaluate the standard
deviation at 50 ppm and round the sample size up to the next integer. These four samples are averaged (by
taking the arithmetic mean) to produce an DELFIA PCB Assay value for a drum's PCB concentration. As
discussed earlier, this DELFIA PCB Assay value can then be converted to a corrected average drum
concentration by using a graph such as Figure B-l or an equation for the PCB prediction line such as Eq. B-2.

Determining the Action Level
Now that the number of samples that need to be analyzed from each drum to meet the DQO goals has been
determined, the action level (AL) can be calculated. The AL is the decision criterion (or "cut-off value) that
will be compared with the unbiased average PCB concentration determined for each drum. The AL for the
decision rule is calculated on the basis of regulation-driven requirements (the TSCA regulatory threshold of
50 ppm) and on the basis of controlling the FR established in the DQO process. Recall that the team set the
permissible FR error rate at 5%.

The formula to compute the action level (EPA 1996) is
                                   AL = RT -
Computing the AL in this instance, we find the following:
                    AL =  50 ppm - (1.645) x 2.80 + 0.05 x 50 = 456ppm
To summarize, four random samples from each drum are analyzed, and the biased results are corrected. The
four corrected results are averaged to produce the average PCB concentration for the drum, which is then
compared to the AL for the decision rule (45.6 ppm). Therefore, the decision rule using the DELFIA PCB
Assay to satisfy a 5% FR and a 10% FA (after correcting the results for bias) is as shown in the box below.
                                               34

-------
                              Decision Rule for 5% FR and 10% FA

          If the corrected average PCB concentration of four random soil samples from a drum
          < 45.6 ppm, then send the drum to the landfill.

          If the corrected average PCB concentration of four random soil samples from a drum
          £ 45.6 ppm, then send the drum to the incinerator.
The decision performance curve (see EPA 1996,
pp. 34—36) calculates the probability of sending a drum
to the incinerator for different values of true PCB
concentration in a drum. Figure B-3 shows that the
decision performance curve has the value of PrfTake
Drum to Incinerator] = 0.965 for True = 50 ppm. This
indicates that the decision rule meets the DQO team's
FR percentage of 5%. The Pr[Take Drum to
Incinerator] = 0.009 for True = 40 ppm, which is better
(at 0.9%) than the FA percentage of 10% that the DQO
team had originally specified. This improved
performance is due to rounding up the number of
samples to the next integer in the calculation of
number of samples required.
                                                   o  1'°
                                                   a  0.9
                                                   0)
                                                   C  O.B
                                                   'o
                                                   _C  0.7

                                                   5  0.6
                                                   E
                                                   3
                                                   Q  0.4
                                                   0>
o.s -
                                                   m
                                                      0.3 -
                                                   o  0.1
                                                   £  0.0
        False Positive = 3.5%
                                                        40  41  42  43  44  45  46  47  48  49  50
                                                          True PCB Concentration (ppm)
                                                  Figure B-3. Decision performance curve for PCB drum
                                                            example.
Alternative FR Parameter
Because of random sampling and analysis error, there
is always some chance that analytical results will not
accurately reflect the true nature of a decision unit
(such as a drum, in this example). Often, 95% certainty (a 5% FR) is customary and sufficient to meet
stakeholder comfort. But suppose that the DQO team wanted to be even more cautious about limiting the
possibility that a drum might be sent to a landfill when its true value is 50 ppm. If the team wanted to be 99%
certain that a drum was correctly sent to a landfill, the following describes how changing the FR requirement
from 5% to 1% would affect the decision rule.

Using FR = 0.01, the sample size is calculated to be seven and the action level is calculated to be 45.3 ppm.
The decision performance curve has the value of Pr[Take Drum to Incinerator] = 0.995 for True = 50 ppm.
This indicates that the decision rule meets the DQO team's FR of 1%. The Pr[Take Drum to Incinerator] =
0.002 for True = 40 ppm is better than the FA percentage of 10% that the DQO team had specified. This
improved performance is due to rounding up the number of samples to the next integer in the calculation of
number of samples required. The decision rule for the lower FR would be as shown below.
                             Decision Rule for FR = 1% and FA = 10%

       If the corrected average PCB concentration of seven random soil samples from a drum  <
       45.3 ppm, then send the drum to the landfill.

       If the corrected average PCB concentration of seven random soil samples from a drum £
       45.3 ppm, then send the drum to the incinerator.
                                               35

-------
Comparison with Reference Laboratory
A statistical analysis of the results from the reference laboratory over the range 0 to 60 ppm gave a linear
approximation to the standard deviation of Sn{= 0.14 + 0.134 x (Certified PE Value). Decision rules can be
calculated on the basis of this standard deviation. Table B-l compares the decision rules for Hybrizyme's
DELFIA PCB Assay with those of the reference laboratory.
 Table B-l.  Comparison of Decision Rules for DELFIA PCB Assay Measurements and Reference
             Laboratory Measurements

Analysis
Method

DELFIA
Reference Lab
FR = 5%andFA = 10%

N

4
6

AL
(ppm)
45.6
45.4
FR = l%andFA = 10%

N

7
9

AL
(ppm)
45.3
44.7

Cost per
sample

$22.50 °
$144

Turnaround
time

6 samples/hr
14-30 working days
   Plus instrument purchase or rental cost (see Table 16).
                                              36

-------