V>EPA Method 1668A Interlaboratory
        Validation Study Report
        March 2010

-------
                                            Method 1668A Interlaboratory Validation Study
                  U.S. Environmental Protection Agency
                            Office of Water
                    Office of Science and Technology
                Engineering and Analysis Division (4303T)
                    1200 Pennsylvania Avenue, NW
                        Washington,  DC  20460
                          EPA-820-R-10-004
March 2010

-------
                                                          Method 1668A Interlaboratory Validation Study
                                    Executive Summary

       This report presents the results of EPA's interlaboratory validation of EPA Method 1668A
Chlorinated Biphenyl Congeners in Water, Soil, Sediment, and Tissue by HRGC/HRMS. This study was
conducted in 2003-2004 to validate the performance of EPA Method 1668A in municipal wastewater, fish
tissue, and biosolids matrices.

       EPA used the results of the study to evaluate and revise Method 1668A quality control (QC)
acceptance criteria for initial precision and recovery, ongoing precision and recovery, and labeled
compound recovery from real world samples. These interlaboratory criteria (Table 5-1) replace the
single-laboratory criteria, and are published in Table 6 of the revised version of this PCB-congener
method, EPA Method 1668B.
                                     Acknowledgments

       This report was written under contract for EPA by CSC Systems & Solutions, LLC, and Interface,
Inc.  EPA acknowledges the volunteer laboratories that participated in the study and, in particular, those
laboratories that took the extra effort to comment on EPA Method 1668A and to provide suggestions for
improvements.
                                         Disclaimer

       Mention of trade names or commercial products does not constitute endorsement or
recommendation for use.
                                          Contacts

Richard Reding, Ph.D., Chief
Engineering & Analytical Support Branch
Engineering and Analysis Division (4303T)
Office of Science and Technology, Office of Water
U.S. Environmental Protection Agency
1200 Pennsylvania Avenue NW
Washington, DC 20460
http://www.epa.gov/waterscience
ostcwamethods@epa.gov
March 2010

-------
                                                           Method 1668A Interlaboratory Validation Study
                                   Table of Contents
Executive Summary	ii
Acknowledgments	ii
Disclaimer       	ii
Contacts         	ii
Section 1    Introduction And Background	1
            1.1   Introduction	1
            1.2   Background	2
            1.3   2003 Revision to Method 1668A	2
Section 2    Study Management, Objectives, Design, and Implementation	3
            2.1   Study Management	3
            2.2   Study Objectives and Design	3
            2.3   Laboratory Selection	3
            2.4   Sample Selection	4
            2.5   Preparation of Study Samples	5
                 2.5.1 Biosolids and Tissues	5
                 2.5.2 Wastewater	6
                 2.5.3 Labeling and Shipping	7
            2.6   Sample Analysis and Data Reporting	7
            2.7   Deviations from the Method or Study Design	8
                 2.7.1 Instrument Calibration	8
                 2.7.3 Tissue	10
                 2.7.4 Wastewater	10
Section 3    Data Review and Validation	11
Section 4    Results and Discussion	13
            4.1   Background and Homogeneity Testing	13
                 4.1.1 Wastewater Sample Homogeneity	13
                 4.1.2 Tissue and Biosolids Sample Homogeneity	13
            4.2   Congener Concentrations in Samples	13
            4.3   Congener Concentrations in Blanks	15
            4.4   Wastewater Sample Recovery and Precision	15
            4.5   Variability as a Function of Concentration	17
                 4.5.1 Variability vs. Concentration for Wastewater	17
                 4.5.2 Variability vs. Concentration for Tissue	19
                 4.5.3 Variability vs. Concentration for Biosolids	20
            4.6   Labeled Compound Recovery and Precision	21
Section 5    Revision of Quality Control Acceptance Criteria	23
            5.1   Calibration	23
            5.2   Calibration Verification	23
            5.3   Initial Precision and Recovery	23
            5.4   Ongoing Precision and Recovery	24
            5.5   Labeled Compound Recovery from Samples, Blanks, and IPR and OPR Standards.... 24
Section 6    Conclusions	27

Appendix A Statistical Procedures Used to Develop QC Acceptance Criteria	A-1
Appendix B Study Plan for Interlaboratory Validation of EPA Method  1668A for Determination of
            Chlorinated Biphenyl Congeners in Water, Biosolids, and Tissue by HRGC/HRMS	B-l
March 2010

-------
                                                           Method 1668A Interlaboratory Validation Study
                                      List of Tables


Table 2-1.  Laboratories Participating in the Method 1668A Validation Study
Table 2-2.  Congener Concentrations in Wastewater Samples by Level of Chlorination
Table 2-3.  Sample Pairs for Distribution to 14 Participant Laboratories
Table 3 -1.  Summary of Data Received from Participant Laboratories
Table 4-1.  Congener Concentrations in Study Samples by Level of Chlorination
Table 4-2.  Congener Concentrations in Blanks by Level of Chlorination
Table 4-3.  Wastewater Sample Recovery and Precision by Level of Chlorination
Table 4-4.  Labeled Compound Recovery and Precision by Level of Chlorination
Table 5-1.  Revised QC Acceptance Criteria for IPR, OPR, and Labeled Compounds in Samples
                                     List of Figures

Figure 4-1. Mean Recovery vs. Spike Concentration, PCB Congeners in Wastewater
Figure 4-2. Concentration Standard Deviation vs. Spike Concentration, PCB Congeners in Wastewater
Figure 4-3. Relative Standard Deviation vs. Spike Concentration, PCB Congeners in Wastewater
Figure 4-4. Mean vs. Standard Deviation of Measured Tissue Results
Figure 4-5. Mean vs. Relative Standard Deviation of Measured Tissue Results
Figure 4-6. Mean vs. Standard Deviation of Measured Biosolids Results
Figure 4-7. Mean vs. Relative Standard Deviation of Measured Biosolids Results
March 2010

-------
                                                           Method 1668A Interlaboratory Validation Study
                                         Section  1
                           Introduction and Background
 1.1   Introduction
       This report describes the interlaboratory validation study of EPA Method 1668A that EPA
conducted in 2003 - 2004 on municipal wastewater, biosolids and fish tissue matrices.  The study was
conducted according to the Study Plan for Interlaboratory Validation of EPA Method 1668A for
Determination of Chlorinated Biphenyl Congeners in Water, Biosolids, and Tissue by HRGC/HRMS,
November 2003, which is an appendix to this report. A draft of this report was peer reviewed, and the
following changes were made as a result of this review:

       •   Rounded all numbers to 3 significant figures maximum.

       •   Expanded  discussion of how QC acceptance criteria were generated

       •   Moved the definition of "Youden pair" from Section 2.5.1 to its first use in this paragraph.

       •   Table 4-4:  Truncated numbers at the decimal point in the "# pairs" column.

       •   Section 2.4 was expanded to give greater detail about the nature of the fish and biosolids
           samples.

       •   Section 2.5.2 was expanded to give greater detail about how the wastewater sample was
           prepared.

       •   A paragraph was inserted into Section 2.7.2 stating that the participating labs were required to
           determine  the solids content of the biosolids sample and report the result in units of dry
           weight.

       •   A result in Table 4-1 was corrected.

       •   Footnote 1 to Table 4-1 was revised to make clear that results for biosolids are in units of dry
           weight and results for fish are in units of wet weight.

       •   Footnote 2 to Table 4-1 was revised to state that the mean, median, and maximum
           concentrations at each LOC are based on any detected congeners in that LOC and when
           coelution of two or more congeners occurred, the combined value of those co-eluted
           congeners  was used.

       •   The same Footnote 2 was applied to Table 4-2  because it is called out in the header row. The
           existing "2" applied to the cell with the number of sand/oil blanks was changed to a "3" in
           that cell and the existing Footnote 2 was renumbered as Footnote 3.

       •   Section 4.3 was expanded to clarify that blanks were to be analyzed in the same way as
           samples.

       •   The header to column 3  in Table 4-2 was changed from "# labs" to "# blanks" to indicate the
           number of blanks analyzed in the study.
March 2010

-------
                                                           Method 1668A Interlaboratory Validation Study
        •   Footnote 1 to Table 4-2 was expanded to indicate that results for the sand/oil blank were to be
           reported as wet weight.

        •   Section 4.4 was expanded to explain that, even though the native congeners weren't recovered
           within the range expected, the labeled congeners were, thus indicating that the native
           congeners most likely were lost in transit.

        •   Table 4-3 was expanded to include recovery and precision for the labeled compounds by
           level of chlorination

        •   Section 4.6 and Table 4-4 were expanded to present results for the 27 individual labeled
           rather than by level of chlorination.

        •   A fifth footnote was called out in the  original Table 5-1, for labeled congeners 156L and
           157L, but did not appear below the table.  The missing Footnote 5 was added in this version.

        EPA used the results of the study to revise Method 1668A and publish, in 2008, Method 1668B.
Quality control (QC) acceptance criteria for initial precision and recovery, ongoing precision and
recovery, and labeled compound recovery from real world samples are in Table 5-1 of this report. These
interlaboratory criteria replace the single-laboratory criteria, and are published in Table 6 of EPA Method
1668B.

1.2     Background

        Method 1668A is for determination of chlorinated biphenyl congeners (PCBs) in water, soil,
sediment, biosolids, and tissue by high resolution (capillary column) gas chromatography combined with
high resolution mass spectrometry (HRGC/HRMS). These 209 PCB-congeners are the individual
chemicals that comprise a class of pollutants known as Aroclors. Since publication in 1999, Method
1668A has been used to measure PCBs in biosolids in EPA's 2001 National Sewage Sludge Survey, and
fish tissue in EPA's four-year National Study of Chemical Residues in Lake Fish Tissue. Additional
background on the nature and determination of PCBs and on the history of development, validation, and
peer-review of EPA Method 1668A is in the  study plan.

1.3     2003 Revision to Method 1668A

        Minor revisions to Method 1668A were made in August 2003 for use in this interlab study.  The
changes corrected technical and typographical errors and reflected practice of the method by laboratories
based on comments received.
March 2010

-------
                                                            Method 1668A Interlaboratory Validation Study
                                         Section 2
       Study Management,  Objectives,  Design,  and Implementation


2.1     Study Management

        This study was designed and managed by the Engineering and Analytical Support Branch (EASE,
formerly the Statistics and Analytical Support Branch) of the Engineering and Analysis Division in the
Office of Science and Technology within EPA's Office of Water. Day-to-day coordination of study
activities was performed by the contractor-operated Sample Control  Center (SCC).1

        Preliminary results of this study were presented at the 2004 National Environmental Monitoring
Conference in Washington, DC, July 20, 2004. Since that presentation, the results have been further
evaluated and presented in this report. Therefore, this report supersedes any material previously
presented or published.

2.2     Study Objectives and Design

        Objectives of this study were to 1) characterize the performance of Method 1668A in multiple
laboratories and matrices, and 2) evaluate and, if appropriate, revise the QC acceptance criteria in the
method.

        EASE designed the study in accordance with guidelines published by EPA and ASTM
International (ASTM).2'3  These guidelines recommend a minimum of six complete data sets for
evaluation of a method. To allow for some loss of data due to error,  lost samples, outlier removal,  or
other unforeseen causes, EPA included 14 participant laboratories in the study.  The study design is
detailed in an appendix to this report.

2.3     Laboratory Selection

        EPA used volunteer laboratories for participation in the study. Each interested laboratory was
asked to demonstrate that it had recent experience in using HRGC/HRMS to determine chlorinated
pollutants in environmental samples and confirm that it would determine all 209 congeners using an SPB-
Octyl column, as described in Method 1668A.  The intent was to ensure that study participants already
possessed the facilities, equipment, and trained staff necessary to implement the method.

        Fourteen (14) volunteer laboratories were selected to participate in this  study. The laboratories
were notified of their selection at  least two weeks before the study began, so that they would have time to
review the method and study-specific instructions. Of the 14 laboratories selected, 11 were commercial
laboratories and 3 were EPA Regional laboratories. Laboratories were not required to validate the
method  in all three matrices; as a  result, the number of participant laboratories varied, depending on the
matrix tested.  As discussed in section 3 of this report, because of scheduling problems 3 of these 14
volunteer labs did not submit data.

        To offset costs to the laboratories, EPA provided each laboratory with a set of analytical
standards necessary to identify and measure the 209 PCB congeners  targeted by Method 1668A. EPA
1 The Sample Control Center (SCC) is operated by CSC Systems & Solutions, LLC under contract to EPA.
2 Guidelines for Selection and Validation of US EPA's Measurement Methods, U.S. EPA Office of Acid Deposition,
Environmental Monitoring and Quality Assurance (OADEMQA), Office of Research and Development, U.S. Environmental
Protection Agency, August 1987 Draft.
3 ASTM Standard D2777-98, "Standard Practice for Determination of Precision and Bias of Methods of Committee D-19 on
Water," Annual Book of ASTM Standards, Vol. 11.01, ASTM International, West Conshohocken, PA 19428.

March 2010

-------
                                                            Method 1668A Interlaboratory Validation Study
also provided the laboratories sets of standard solutions containing native and carbon-13 labeled
compounds necessary to calibrate their instruments and to conduct all analyses.  The packaged sets of
standards were purchased from Cambridge Isotope Laboratories (CIL) and AccuStandard, Inc.
Laboratories were provided with detailed instructions for combining and diluting standards to preclude
injudicious use of standards.  The instructions were based on procedures given in Method  1668A.

        In addition to the 14 volunteer participant laboratories, a sample processing laboratory was
contracted to perform all activities necessary to ensure that the participant laboratories received
homogenized, spiked, and aliquoted samples. Homogenization of bulk sample volume was necessary to
prepare replicate samples for analysis by participant laboratories.  The participant and sample processing
laboratories are listed in Table 2-1.
Table 2-1. Laboratories Participating in the Method 1668A Validation Study
Alta Analytical Laboratory Inc.
1104WindfieldWay
El Dorado Hills, CA 95762
Phone: 916-933-1640
EPA Region 7
300 Minnesota Ave.
Kansas City, KS66101
Phone: 913-551-5120
AXYS Analytical Services, Ltd.
2045 Mills Road West
Sidney, BC V8L 3S8 Canada
Phone: 250-655-5800
Pace Analytical Services
1700 West Albany
Broken Arrow, OK 74012
Phone: 918-251-2858
Pacific Analytical, Inc.
6056 Corte del Cedro
Carlsbad, CA 92009
Phone: 760-496-2200
Battelle-Columbus Laboratories
505 King Avenue
Columbus, OH 43201
Phone: 614-424-7884
Columbia Analytical Services
10655 Richmond Avenue, Suite 130A
Houston, TX 77042
Phone: 713-266-1599
Severn Trent Laboratories - Knoxville
5815 MiddlebrookPike
Knoxville, TN 37921
Phone: 865-291-3000
Paradigm Analytical Laboratories, Inc.
5500 Business Drive
Wilmington, NC 28405
Phone: 910-350-1903
Data Analysis Technologies, Inc.
7715 Corporate Blvd.
Plain City, OH 43064
Phone: 800-733-8644
Philip Analytical Services Corporation
5555 North Service Road
Burlington, ON L7L 5H5 CANADA
Phone: 800-668-0639
Enviro-Test Laboratories
9936-67* Avenue
Edmonton, AB T6E OPS CANADA
Phone: 780-413-6481
EPA Region 3
701 Mapes Road
Fort Meade, MD 20755-5350
Phone: 410-305-2606
Severn Trent Laboratories -
Sacramento
880 Riverside Parkway
West Sacramento, CA 95605
Phone: 916-374-4433
EPA Region 4
980 College Station Rd.
Athens, GA 30605-2720
Phone: 706-355-8807
Note:   The primary purpose of this study was to evaluate the performance of Method 1668A.
While results obtained by individual laboratories were used relative to this purpose, no attempt
was made to assess performance of individual laboratories. No endorsement of these
laboratories is implied, nor should any be inferred. To preserve confidentiality, laboratories that
volunteered for this study, including three that did not submit lab data, were assigned numbers
randomly from 1 to 14.  The lab identities and that of the sample processing laboratory are not
revealed in the data or lists in this report.

2.4     Sample Selection

        To minimize burden on volunteer laboratories, the study was designed so that no more than two
samples of each matrix type would be analyzed, with each sample containing varying concentrations of
the target PCB congeners.  EPA provided existing (archived)  fish tissue and biosolids samples to the
sample processing laboratory to prepare study samples representing these matrices. No archived sample
volume was available for wastewater, therefore, the sample processing laboratory prepared the
wastewater samples.  In preparing study samples, EPA's objective was to ensure that the congeners
present in each sample matrix would span the anticipated measurement range of Method 1668A, from the
upper end of the calibration range down to "not detected."
March 2010

-------
                                                           Method 1668A Interlaboratory Validation Study
       Tissue and biosolids samples were generated from excess samples collected during EPA's 1999-
2000 National Lake Fish Tissue Study (NLFTS) and EPA's 2001 National Sewage Sludge Survey,
respectively.  These samples had been stored in freezers at an EPA sample repository. All of the tissue
samples used in the validation study were from bottom-dwelling fish species and were originally prepared
as whole-fish composite samples. The samples for the NLFTS were prepared as finely ground tissue by a
single laboratory.  Excess sample beyond that shipped to the laboratories during the NLFTS was archived
in 500-mL jars and stored frozen. The NLFTS also collected samples of predator species, from which
fillets were taken and composited for analysis. The bottom-dweller samples provided more tissue than
the predator samples, and thus a greater excess that was available for other purposes such as a Method
1668A validation study.  The tissue samples used for the validation study were prepared from the 500-mL
archive jars.

       The original biosolids samples used to prepare the study samples were collected as solid sewage
sludges, as  opposed to the pourable liquid sewage sludges that may be produced at some wastewater
treatment facilities. Each laboratory that analyzed the biosolids samples determined the percent solid
contents. The reported values were in the range of 30 to 40% for the two  samples, amounts that are
typical for many biosolids produced in the U.S.

       So  that a sufficient amount of each sample was available to support the study, EPA identified
several samples of each matrix type that could be combined to produce large volumes of Youden pairs
with the desired congener distribution.  (Youden pairs are defined as two samples of the same matrix
containing similar, but not exact, concentrations of the analytes of interest.)  Once these stored samples
were identified, they were forwarded on ice to the sample processing laboratory. Although PCBs are
stable and do not require preservation, ice was used to prevent decomposition offish tissue and to retard
gas production in the biosolids.  For wastewater, amounts of effluent grab samples were collected from a
publicly owned treatment works (POTW) that were sufficient to provide enough samples for all of the
participant  laboratories, and excess sample in case of breakage, spillage, or other problems. Bulk
wastewater was collected in polyethylene carboys and shipped overnight to the  sample processing
laboratory for spiking and distribution.

2.5    Preparation of Study Samples

       The sample processing laboratory was provided with a detailed set of instructions for:

       •   Combining and homogenizing the biosolids samples
       •   Combining and homogenizing fish tissues
       •   The number of aliquots to be prepared from each combined/homogenized matrix
       •   Aliquoting and spiking the wastewater samples
       •   Labeling and shipping the prepared sample aliquots.

2.5.1  Biosolids and Tissues

       Because the biosolids and tissue samples used in this study were already known to contain PCBs
at levels sufficient to cover the analytical range of Method 1668A, the sample processing laboratory did
not have to spike these matrices with PCBs.  This eliminated concerns about how well spiked  constituents
would be incorporated into  these matrices and whether spiked samples were representative of real-world
samples.  The goal of the mixing and aliquoting scheme for biosolids and tissues was to obtain Youden
pairs for each matrix of interest (i.e., composites A and B). As described in ASTM Practice D2777, the
concentrations of Youden pairs should differ by no more than 20%. Because the available "excess"
volumes of the biosolids  and fish tissues were limited and the number of laboratory participants was
relatively large, the Youden pairs were prepared in a multi-step process.  For the biosolids, the first  step
was to combine and homogenize five biosolids samples to form a composite. This composite  was then
divided approximately in half. One half of the composite was designated as biosolids sample  "A" while

March 2010                                                                                    5~

-------
                                                           Method 1668A Interlaboratory Validation Study
the other half was used to prepare biosolids sample "B." Biosolids sample "B" was prepared by adding
material from a sixth existing biosolids sample, plus some clean sand, to produce a composite with PCB
congener concentrations that were approximately 20 % different from those in sample "A." For the tissue
samples, two existing tissue samples were homogenized.  The composite was then divided approximately
in half, with one half being designated as tissue sample "A." Tissue sample "B" was prepared by adding
tissue from a third existing sample to the remaining half of the initial composite.

       The sample processing laboratory was required to perform background and homogeneity analyses
of both the biosolids and tissue matrices. The laboratory was instructed to analyze one 10-g dry weight
aliquot of sample  "A" as the background analysis, and two 10-g dry weight aliquots of sample "B" as the
homogeneity aliquots for both the biosolids and tissue matrices. Because of the mixing scheme for both
of these matrices,  it was assumed that if the homogeneity for sample "B" is found to be acceptable, the
homogeneity of sample "A" would also be acceptable.  This approach was used to preserve sample mass.
Results of tissue and biosolids background and homogeneity analyses are discussed in Section 4.1.

2.5.2   Wastewater

       Based on  previous experience, municipal wastewater discharges would be unlikely to contain
PCB congeners at concentrations sufficient to adequately test the capabilities of the method. Thus, the
sample processing laboratory was instructed to  first analyze an aliquot of wastewater from a publicly
owned treatment works (POTW) to determine background PCB congener levels.  Following a review of
the background results by SCC, EPA defined the spiking levels, and provided the sample processing
laboratory with detailed instructions to divide the unspiked POTW matrix into the required number of
aliquots and spike each aliquot separately (rather than spiking a bulk volume of wastewater and then
subdividing the spiked sample into replicate aliquots) to the appropriate concentrations. Spiking each
aliquot separately avoids problems with "wall effects," whereby organic pollutants spiked into a bulk
sample tend to adhere to the walls of the container, making it difficult to divide a bulk sample into
multiple aliquots containing the same analyte concentrations.

       In addition, the study-specific instructions provided to each participating laboratory required that
the laboratory filter the wastewater sample prior to extraction and treat both the filtrate and any  solids on
the filter in the manner described in Method 1688A. This instruction was included to prevent problems in
which some laboratories followed the method as written, and others deciding to skip the filtration step if
the wastewater did not appear turbid.

       The unspiked wastewater sample was analyzed by the sample preparation laboratory using
Method 1668A. Out of the 209 PCB congeners, 39 congeners were detected in the sample. All of those
congeners were between PCB 001 and PCB 168, and all of the concentrations were between 17 and 247
pg/L well below the concentrations of the spikes of the congeners into the wastewater (see Table 2-2 of
the Report for the  spiking levels). Results for the blanks were all below the calibration range of Method
1668A as practiced by the sample preparation laboratory. The sample preparation lab thus flagged results
for the blanks as estimates. In preparing the actual study samples, EPA decided that these background
levels were low enough that adjustments need not be made to the amount of each analyte spiked into the
samples. The solvent used for spiking the congeners into wastewater was acetone.  The volume of
acetone used to spike the Youden pair samples was either 0.5 or 0.6 mL per 1-L volume of wastewater.

       Because of the difficulty that would be encountered in preparing custom spiking solutions,
wastewater samples were spiked with varying amounts of "individual native CB congener solutions" A2
through E2 listed in Table 4 of EPA Method  1668A.  Concentrations of the congeners in the wastewater
samples, by level of chlorination, are given in Table 2-2.
March 2010

-------
                                                            Method 1668A Interlaboratory Validation Study
Table 2-2. Spiked Congener Concentrations in Wastewater Samples (by Level of Chlorination)
Congeners
24 Mono- through Trichlorinated biphenyl congeners
6 Mono- through Dichlorinated biphenyl congeners
9 Mono- through Trichlorinated biphenyl congeners
74 Tetra- through Heptachlorinated biphenyl congeners
38 Tetra- through Heptachlorinated biphenyl congeners
42 Tetra- through Heptachlorinated biphenyl congeners
13 Octa- through Decachlorinated biphenyl congeners
3 Octachlorinated biphenyl congeners
Concentration (pg/L)
Youden Pair#1
900
1,200
1,500
1,800
2,400
3,000
2,700
4,500
Youden Pair #2
750
1,000
1,250
1,500
2,000
2,500
2,250
3,750
       The sample processing laboratory analyzed two random aliquots of one concentration level for
homogeneity determination.  Results of the homogeneity analyses are discussed in Section 4.1.

2.5.3  Labeling and Shipping

       SCC provided the sample processing laboratory with a unique 5-digit sample number for each
sample. After the aliquots were prepared, the sample processing laboratory labeled each sample container
and cap with the corresponding unique sample number.  The sample processing laboratory then shipped
the prepared, numbered samples to the participant laboratories via air courier.  Although PCBs are
persistent, and thus do not require preservation, biosolids and tissue samples were shipped on ice to
hinder decomposition of the tissues and gas formation in the biosolids. The sample processing laboratory
notified SCC of the shipping date, and SCC notified participant laboratories of the shipping and
scheduled arrival dates. Table 2-3 lists the numbers of wastewater, biosolids, and tissue samples that
were prepared for distribution to the  14 participant laboratories.
Table 2-3. Sample Pairs for Distribution to 14 Participant Laboratories
Matrix
Wastewater
Biosolids
Tissue
All Three Matrices
Samples per Laboratory
2 (1 Youden Pair)
2 (1 Youden Pair)
2 (1 Youden Pair)
6
Number of Aliquots Distributed
28
28
28
84
2.6     Sample Analysis and Data Reporting

        Participant laboratories did not know the concentration of PCBs in the samples received, and
were instructed to prepare and analyze the samples according to Method 1668A procedures, except where
stated otherwise in the participant's scope of work. In addition to the analysis of study samples,
laboratories also were required to prepare and analyze two ongoing precision and recovery (OPR)
samples in reagent water, one reagent water blank, and one solids/tissue blank (playground sand mixed
with corn oil).

        Because study results were to be used to evaluate or further develop QC acceptance criteria,
laboratories were prohibited from performing multiple analyses to improve results. Laboratories were,
however, allowed to implement corrective action and reanalyses for QC failures attributable to analyst
error, instrument failure, or identified contamination. The laboratories also were instructed that any
deviations from the method and Statement of Work (SOW) must be pre-approved by EPA.

        Laboratories were required to submit electronic and hard copies of summary sample results, and
hard copies of all supporting  raw data, run chronologies, chromatograms, example equations, and case
narratives to SCC for review and data validation. Additionally, laboratories were asked to provide a
March 2010

-------
                                                             Method 1668A Interlaboratory Validation Study
detailed narrative describing any problems or recommendations and a description of any modifications to
procedures specified in the method.  All submitted data were reviewed against the study and method
requirements prior to use for evaluation of method performance. Laboratories were asked to adhere to the
following rules in reporting results:

        •  Report results to the lowest level possible, using a signal-to-noise ratio of 3 as the sample-
           specific detection limit.
        •  For congeners that are not detected, report as "
-------
                                                            Method 1668A Interlaboratory Validation Study
        Some laboratories submitted results for the analysis of biosolids samples that: used a smaller
sample size than suggested in the method, analyzing more dilute extracts than suggested in the method, or
both.

        •   Laboratory 2 used a 15-g (wet weight) sample as opposed to the 30-g sample suggested by
           the method, resulting in a two-fold dilution.
        •   Laboratory 12 used a 6-g (wet weight) sample as opposed to the 30-g sample, resulting in a
           five-fold dilution.
        •   Laboratory 7 used a 10-g (wet weight) sample as opposed to the 30-g sample and
           concentrated the extract to a final volume of 100 (iL, as opposed to 20 (iL, resulting in a 15-
           fold dilution.
        •   Laboratory 8 used the full sample  size, but concentrated the extract to a final volume of 200
           \\L as opposed to 20 \\L, resulting in a ten-fold dilution.
        •   Laboratory 13 used a 1-g (wet weight) sample as opposed to the 30-g sample, and
           concentrated the extract to a final volume of 100 (iL as opposed to 20 (iL, resulting in a 150-
           fold dilution. Discussions with this laboratory revealed no attempt to analyze a 30-g sample.
           Instead, based on past experience with GC/HRMS analyses, the laboratory used a 1-g sample,
           and concentrated the extract to 100 (iL. Their general experience has been that using a 30-g
           sample results in difficulties during instrumental analysis (lock-mass problems). Based on
           their GC/HRMS experience, Laboratory  13 also did not use the prescribed sample amounts
           for the fish tissue and wastewater samples.

        Two  laboratories (4 and 10) did not submit biosolids data because of difficulties encountered with
clean-up and  analysis. Both  of these laboratories attempted analyses on 30-g samples, as suggested in the
method.

        Laboratory 4 reported difficulties with the cleanup of both the biosolids samples.  In one of the
biosolids samples, upon the first acid wash, the sample appeared black in color and the phases could not
be distinguished. The laboratory proceeded with the  addition of sodium chloride in an attempt to mitigate
the problem.  During the subsequent acid wash steps  (second, third and fourth) no color appeared in the
aqueous layer.  The extract layer contained suspended particles and had a tar-like appearance and
viscosity.  The sample was then put through an acid/base silica column before the gel permeation
chromatography (GPC) step  in hopes that the extract would then not plug the filter used in the GPC.  In
the case of the second biosolids sample, an emulsion resulted during back-extraction with base (Section
12.5 in the method).  The laboratory unsuccessfully attempted to break the emulsion by adding sodium
chloride and cooling, and tried diluting the extract with sodium chloride solution and hexane, followed by
hexane rinses, and addition of sulfuric acid.  The extract was drained into a round bottom flask and
concentrated  by heating mantle. The sample was then washed with the maximum number (4) of acid
washes suggested in the method.

        Laboratory 10 reported difficulties with the cleanup and extraction of both biosolids samples and
reported that, despite having  made two separate attempts to cleanup and extract the biosolids samples,
they were not able to obtain reportable results. The samples were initially extracted using approximately
22 grams of each sample (dry weight basis). A total of six cleanup steps were applied to each sample.
According to the laboratory narrative, even after these measures, the final extracts contained significant
amounts of white crystals. The remaining liquid portions of the extracts were separated from the crystals
and injected.  These extracts  did not yield reportable results.  The laboratory attempted to extract the
samples a second time, this time using 2 grams each. Two cleanup steps were applied to these samples.
No crystals were present in the final extracts; however the laboratory was still unable to obtain reportable
results.

        Laboratories 7 and 12 reported biosolids results on a wet-weight basis, whereas laboratories 8 and
13 reported biosolids results  on a dry-weight basis. The dry-weight data for laboratory 8 were corrected

March 2010                                                                                      J

-------
                                                             Method 1668A Interlaboratory Validation Study
to wet weight based on percent solids data provided by laboratory 8 (33.3% solids for Youden 1 and
39.3% for Youden 2). Because laboratory 13 did not provide percent solids data, the laboratory 13 dry-
weight data were corrected to wet weight, based on the mean of the percent solids data provided by the:
sample preparation laboratory, laboratory 2, and laboratory 8. These three laboratories were the only labs
that provided percent solids data (33.3% solids for Youden 1 and 35.9% for Youden 2).

        The laboratory narratives suggest that many laboratories lacked experience extracting and
cleaning up a biosolids matrix. The resulting deviations from the method and study-specific instructions
for analysis of biosolids samples by different laboratories resulted in some unusable and inconsistent data.
Thus, EPA excluded some biosolids results, as described in Section 3 of this report.

2.7.3   Tissue

        Two of the seven laboratories that submitted usable tissue data used a smaller sample size than
that suggested in the method, or analyzed a more dilute extract than suggested in the method.

        •   Laboratory 7 used a 5-g (wet weight) sample as opposed to the 10-g sample suggested in the
           method, resulting in a two-fold dilution.
        •   For reasons similar to their deviation in biosolid sample volume (i.e., previous experience
           with GC/HRMS analyses), Laboratory 13 concentrated the extract to a final volume of 100
           (iL as opposed to the 20-(iL volume suggested in the method, resulting in a 5-fold dilution.

    Laboratory 6 did not submit tissue data, and reported difficulties with the analysis of this matrix  due
to interferences from lipids.  The  laboratory reported unsuccessful use of an acid-base wash extraction,
and two rounds of silica gel cleanup.

2.7.4   Waste water

        One of the eight laboratories that submitted usable wastewater data analyzed a more dilute extract
than suggested in the method. Specifically Laboratory 13, for reasons explained previously (GC/HRMS
experience), concentrated the extract to a final volume of 100 (iL as opposed to the 20-(iL volume
suggested in the method, resulting in a 5-fold dilution.
March 2010                                                                                      10

-------
                                                            Method 1668A Interlaboratory Validation Study
                                          Section 3
                              Data  Review and Validation
        Three of the 14 volunteer laboratories that were selected to participate in this study failed to
submit data despite repeated requests and offers to extend the submission deadlines. In all three cases, the
laboratories cited scheduling conflicts as the reason for their inability to complete the study.

        Data from the 11 laboratories that submitted results were reviewed and validated by SCC as soon
as possible after receipt. Data packages included sample tracking logs, summary results, QC summaries,
raw data, sample calculations, laboratory narratives (including descriptions of any problems encountered,
corrective actions taken, and comments on method procedures), and electronic data reporting
spreadsheets. Data were reviewed against requirements in the study plan and the method to ensure that
results from each laboratory were complete (i.e., that all required data were present, including results of
all required tests, sample lists, run chronologies, summaries of analytical results, raw data, example
questions). This included verification that: all samples were analyzed properly; appropriate spike levels
were used; the analytical systems were properly calibrated; results calculation procedures were followed
correctly;  and that raw data supported the results.  A fundamental objective of this review was to
maximize data use, and every attempt was made to resolve data discrepancies with laboratories. This
review disclosed the following facts:

        •   Data from Laboratories 3 and 11 failed to meet one or more of the chromatographic
           resolution requirements in Section 6.9.1.1.2 of Method 1668A. This section specifies that the
           SPB-Octyl GC column must resolve congener pairs 34/23 and 187/182, and that congener
           pair 156/157 must coelute.

        •   Laboratory 3 data showed coelutions across several chlorination levels, inability to  detect
           many of the congeners in the low (CS-1) calibration standard, and high baseline noise that
           made integration difficult.  Laboratory 3 also reported loss of sensitivity, column
           deterioration, and expressed general dissatisfaction with the method. Laboratory 3  reported
           results for only 1 wastewater sample,  1 blank sample, and no other QC or sample results.

        •   Laboratory 11 data indicated an inability to recover 25 of the 34 labeled compounds spiked
           into the biosolids samples without an acknowledgment or explanation of the difficulties, and
           incomplete raw supporting data.  For example, selected ion current profiles for samples in
           which the laboratory reported very high recoveries of some labeled  compounds (e.g.,  572%)
           were not provided.  SCC contacted the laboratory repeatedly, but did not receive an
           explanation.

        •   Laboratory 2 submitted only summary level sample results, and provided little or no
           calibration data. During attempts to obtain details and raw supporting data, SCC learned that
           the laboratory manager was no longer with the company and that the laboratory was closing.
           Without sufficient information to support the summary level results submitted, it was not
           possible to investigate potential causes of the observed low recoveries. Sample results for
           this laboratory were consistently below those for all other laboratories, and the labeled
           compound recoveries were generally low in both study and QC samples.

        •   Laboratory 6 did not submit tissue sample results and did not provide OPR results associated
           with the wastewater samples.  In e-mail correspondence, the laboratory indicated that the
           lighter PCB congeners were lost during final transfer, and therefore, results were not
           submitted.  This laboratory provided some raw data (e.g.,  selected ion current profiles) for
           some QC samples, but only summary level data for the results of calibration, calibration

March 2010                                                                                     77

-------
                                                            Method 1668A Interlaboratory Validation Study
           verification, and field samples. SCC was unable to obtain additional information or
           supporting data.  The laboratory reported results for 167 peaks containing one or more
           congeners. This is more than most other laboratories, and more than the 159 peaks described
           in the method, making a side-by-side comparison with data from other laboratories difficult.

        •   Mean relative response (RR) and response factor (RF) values reported by Laboratory 14 were
           reported inconsistently across the laboratory's report forms.  For example, page 318 of the
           laboratory's data package lists the  RR for 13C-labeled PCB congener 81 as 2.7956 and page
           319 has RR values ranging from 2.01 to 2.28 (with a mean of 2.09).  Many of the congeners
           have only 5 RR values, while many others appear to have 6 RR values. Conversely, for
           congener 77L, SCC could reproduce the mean RR value of 2.16 reported on page 319, but
           this value does not match the value of 2.8026 on report Form 3B. Results differ most for the
           early-eluting labeled congeners. SCC examined the calibration data for these congeners and
           compared them to calibration data from other laboratories in the study. Although some
           differences in the responses are expected between different GC/MS instruments, results from
           Laboratory 14 were inconsistent with results from the other laboratories.

        Of the remaining six laboratories:

        •   Four laboratories (7, 8, 11, and 13) provided usable data for all three  of the matrices used in
           this study, and
        •   Two laboratories (4 and 10) provided usable data for wastewater and tissue matrices only.

        Thus, this validation study using volunteer labs yielded six usable data sets for wastewater and
tissue matrices, and  four for biosolids.  Data obtained from these laboratories followed the requirements
of the study plan and the method and included  results for the required accompanying QC analyses; i.e.,
calibration, calibration verification (where submitted), OPR, reagent water blank, and solids/tissue blank
(playground sand mixed with corn oil).  Table  3-1 summarizes the status of results from the laboratories.
Table 3-1. Summary of Data Received from Participant Laboratories
Laboratory
1
2
3
4
5
6
7
8
9
10
11
12
13
14
Total usable data packages
Submitted data?
Wastewater
No
Yes, but unusable
Yes, but unusable
Yes
No
Yes, but unusable
Yes
Yes
No
Yes
Yes, but unusable
Yes
Yes
Yes, but unusable
6
Tissue
No
Yes, but unusable
No
Yes
No
No
Yes
Yes
No
Yes
Yes, but unusable
Yes
Yes
Yes, but unusable
6
Biosolids
No
Yes, but unusable
No
No
No
Yes, but unusable
Yes
Yes
No
No
Yes, but unusable
Yes
Yes
Yes, but unusable
4
        Study samples were assessed for outlying results using Grubbs' outlier test, performed in
accordance with Standard Practice for Determination of Precision and Bias of Applicable Test Methods
of Committee D-19 on Water (ASTMD2 777-98). Details on the outlier analyses are presented in
Appendix A.
March 2010
12

-------
                                                          Method 1668A Interlaboratory Validation Study
                                        Section 4
                               Results and Discussion
4.1    Background and Homogeneity testing

       As described in Section 2.3 of this report, the sample processing laboratory was required to
perform background analyses of the wastewater matrix, and homogeneity analyses of the tissue and
biosolids matrices.

4.1.1  Wastewater Sample Homogeneity

       Wastewater samples were prepared: by determining the background concentration to select
appropriate spike levels, by spiking and aliquoting the samples as described in Section 2.3, and analyzing
two random aliquots for homogeneity verification. Relative percent differences (RPDs) for all congener
concentrations between wastewater homogeneity test aliquots Bl and B2 were less than 16%, and all but
five were less than 10%, confirming the adequacy of the homogenization and aliquoting process.

4.1.2  Tissue and Biosolids Sample Homogeneity

       For tissue and biosolids, the sample processing laboratory was instructed to analyze one 10-g dry
weight aliquot of sample "A" as the background analysis, and two 10-g dry weight aliquots of sample "B"
as the homogeneity aliquots for both the biosolids and tissue matrices. Because of the mixing scheme for
the tissue matrix it was assumed, if the homogeneity for sample "B" was found acceptable, that the
homogeneity of sample "A" would be acceptable. This approach was used to preserve mass of sample for
the study itself by taking the two homogeneity aliquots from the larger aliquot (sample "B"). Relative
percent differences (RPDs) between tissue homogeneity test aliquots Bl and B2 were calculated to verify
the homogenization and aliquoting scheme.  All but five RPD values were 20%; the remaining five were
associated with sample concentrations below the sample-specific minimum level of quantitation (ML; see
Table 2 of Method 1668 for MLs), where greater uncertainty  is expected.

4.2    Congener Concentrations in Samples

       The frequency of detection and the mean, median, and maximum concentrations of the congeners
found in tissue, wastewater, and biosolids samples by level of chlorination (LOG) are in Table 4-1.  The
total number of congeners analyzed reflects the total number  of congeners or coeluted congener groups
analyzed by all labs in both samples for the given chlorination level. For example, 12 congeners were
analyzed in water for LOG 10.  This equates to one congener reported by six labs in two samples (12 = 1
x 6 x 2). Although the same six labs provided usable data for tissue and water, there are differences
between these matrices in the number of congeners analyzed  for a given LOG. For example for LOG 4, a
total of 356 tetrachlorinated congeners (or co-eluting congeners) were analyzed in the wastewater Youden
pairs, but only 352 tetrachlorinated congeners were analyzed  in  the in tissue Youden pair.  The difference
is attributable to the removal of outliers.  The next two columns in the table provide information on the
number of detected congeners in each LOG, and the percentage  of analyzed congeners that were detected.
Finally, the mean, median, and  maximum concentrations in each LOG represent all congeners within that
level; when coelutions of two or more congeners occurred, the combined value of those co-eluted
congeners was used.

       In wastewater samples, all congeners at LOCs 4 and higher were detected by all laboratories.
Only LOG 1 had a rate of detection below 90%.  The rate of congener detection across laboratories was
generally consistent for the different LOCs in biosolids and tissue samples, ranging between 69% and
100% for tissue and between 70% and 100% for biosolids.  With the exception of LOCs 9 and 10 (which

March 2010                                                                                 7J

-------
                                                              Method 1668A Interlaboratory Validation Study
include only congeners 3 and 1, respectively), at least one congener was not detected in the solids
matrices by at least one laboratory for each LOG.  The reason that all laboratories do not detect the same
congeners in each sample is likely due to differences in coelutions and because some laboratories
concentrated extracts to 100 or 50 \\L instead of 20 \\L as required by EPA Method 1668A.  Those
laboratories that did not concentrate extracts to 20 p.L would not measure to as low a level as laboratories
that did, and low concentration congeners would, therefore, not be detected by these laboratories.
Table 4-1. Congener Detection Rates and Concentrations in Study Samples (by Matrix and Level
of Chlorination)
Matrix
Biosolids
Tissue
Water
LOG
1
2
3
4
5
6
7
8
9
10
1
2
3
4
5
6
7
8
9
10
1
2
3
4
5
6
7
8
9
10
#
Labs
4
6
6
# Congeners
Analyzed
24
88
160
240
237
254
169
81
24
8
36
131
232
352
347
362
240
114
35
12
36
128
233
356
344
362
235
116
35
12
# Congeners
Detected
23
64
134
195
166
196
129
72
23
8
26
90
181
288
258
270
182
105
35
12
25
118
223
356
344
362
235
116
35
12
% Congeners
Detected
96
73
84
81
70
77
76
89
96
100
72
69
78
82
74
75
76
92
100
100
69
92
96
100
100
100
100
100
100
100
Concentration (Detects Only)1'2
Mean
142
494
972
1270
2070
1120
665
377
280
313
4
47
267
402
418
429
276
157
162
200
27
533
1100
2850
2660
2190
1750
2410
1760
1740
Median
159
265
482
372
742
407
344
259
230
299
3
27
150
130
128
108
120
108
137
201
20
505
946
2170
1750
1660
1420
1740
1520
1510
Maximum
281
2780
7130
12400
13400
12300
4810
1750
821
493
12
188
1610
3330
15700
10700
3560
709
390
236
106
1460
3430
15300
21800
11800
7370
9560
3350
3170
1  Biosolids (dry weight) and tissue (wet weight) concentrations in ng/kg (pg/g); water concentration in pg/L
2  Mean, median, and maximum concentrations at each LOG are based on any detected congeners in that LOG.
  When coelution of two or more congeners occurred, the combined value of those co-eluted congeners was used.
March 2010
14

-------
                                                            Method 1668A Interlaboratory Validation Study
4.3     Congener Concentrations in Blanks
        Table 4-2 gives mean, medium, and maximum congener concentrations found in the water and
sand/corn oil blanks, by level of chlorination. PCBs can be ubiquitous in the laboratory environment.
Congener detection rates in blank samples ranged from 8-33%, with most of the detected congeners being
reported at very low concentrations relative to the concentrations reported in samples. The relatively low
frequency of detection of congeners in blanks by all laboratories is thought to be attributable to the failure
by some laboratories to concentrate extracts to 20 \\L and to lesser PCB backgrounds in some
laboratories.

        Method 1668A requires that blanks be analyzed in the same way as environmental and IPR/OPR
samples, including use of a 1-L aliquot for water or 10 g of an appropriate reference matrix for solids (see
Section 7.6 of Method 1668A), and including concentration of extracts of blanks to 20 (iL.
Table 4-2. Congener Detection Rates and Concentrations in Blanks (by Matrix and Level of
Chlorination)
Matrix
Sand/oil
Water
LOG
1
2
3
4
5
6
7
8
9
10
1
2
3
4
5
6
7
8
9
10
# Blank
samples
103
6
# Congeners
Analyzed
50
130
227
328
370
355
245
122
50
20
30
79
135
197
220
213
146
74
30
12
# Congeners
Detected
14
20
61
77
70
100
65
30
4
3
10
9
33
43
29
52
35
15
3
2
% Congeners
Detected
28
15
27
23
19
28
27
25
8
15
33
11
24
22
13
24
24
20
10
17
Concentration (Detects Only)1'2
Mean
4.0
5.3
2.5
4.4
6.7
5.7
3.3
1.3
2.0
0.7
25.8
34.8
17.3
79.5
23.9
12.2
6.7
9.4
25.4
13.2
Median
3.5
5.1
1.4
2.0
3.3
0.4
0.6
0.2
2.1
0.1
15.1
21.3
11.0
10.0
20.2
2.2
2.5
9.9
32.0
13.2
Maximum
9.6
12.9
12.3
29.0
37.7
60.8
20.0
5.6
3.0
1.9
82.1
113
57.7
2280
74.2
85.0
39.2
28.4
33.2
26.1
  Sand/oil concentration in ng/kg (pg/g) (wet weight); water concentration in pg/L
2 Mean, median, and maximum concentrations at each LOG are based on any detected congeners in that LOG.
  When coelution of two or more congeners occurred, the combined value of those co-eluted congeners was used..
3 Six labs provided usable data for sand/oil blanks. Four of the six labs (Labs 7, 8, 10, and 13) analyzed two
  sand/oil blanks, yielding a total often sand/oil blanks.

4.4     Wastewater Sample Recovery and Precision

        Table 4-3 summarizes the laboratories' ability to recover congeners from the wastewater samples,
presenting the recovery and precision of congener determination by level of chlorination.
March 2010
15

-------
                                                            Method 1668A Interlaboratory Validation Study
        The mean and median recoveries of nearly all native congeners were in the 60 - 110 percent
range, typical for recovery of organic compounds extracted from wastewater.  Excluding data at LOCs 1
and 2, the median recovery across all native congeners and all labs is approximately 75%, and the median
RSD is approximately 10%.  Low recoveries of the native congeners at LOCs 1 and 2 (Table 4-3) may be
due to loss during transport from the sample preparation laboratory to the participant laboratories. (Data
reported by the sample prep laboratory indicate that the congeners were present immediately after
shipment.)  Even though the native congeners were not recovered within the range expected, the labeled
congeners were recovered from spikes into the wastewater samples by the laboratories.  Recoveries of the
labeled compounds are shown by LOC in Table 4-3 for comparison.  Recovery of the labeled congeners
indicates that the loss of the native congeners could not be by evaporation during the solvent evaporation
step because the labeled congeners would have been lost also.

The exact reason for loss of the native congeners at LOCs 1  and 2 is not known. Possible causes may be
loss by evaporation into the headspace of the sample container during shipment, with subsequent release
to the atmosphere when the container is opened, or to biological or other degradation during transit,
although selective degradation of congeners at LOCs 1 and 2 only would  be unusual.  In Figure 4-1, low
recoveries for congeners in the 750 - 900 pg/L range are, almost exclusively, attributable to these partial
losses of the mono- and dichloro- congeners.
Table 4-3. Wastewater Sample Recovery and Precision by Level of Chlorination
LOG
Percent Recovery (%)
# Results
Mean
Median
Min.
Max.
Within-pair Relative Standard Deviations (%)
# Pairs
Pooled*
Median
Min.
Max.
Native congeners spiked by sample prep lab
1
2
3
4
5
6
7
8
9
10
25
118
223
356
344
362
235
116
35
12
3.15
54.2
89.5
95.6
81.4
75.3
72.3
68
70.8
70
2.71
44.6
82.8
91.4
72.2
68.8
64.4
59.3
57.8
59.1
0.49
2.63
34.2
38.1
30.6
8.14
10.4
18.1
44.4
49.3
11.8
162
164
201
182
196
155
135
126
118
11
57
111
178
170
178
114
56
17
6
29.7
12.2
7.62
7.36
10.8
12.1
9.63
11.3
8.91
5.67
17.5
4.42
5.06
6.16
7.99
8.64
5.24
7.26
6.66
4.23
5.8
0.17
0.02
0
0.06
0.16
0.14
0.12
0.57
0.37
80.8
62.4
24
20.6
40.5
46.8
39.3
32.3
18.5
9.05
Labeled congeners spiked by participant labs
1
2
3
4
5
6
7
8
9
10
24
24
36
33
83
50
36
23
24
12
51.4
58.5
67.4
60.5
77.2
75.6
76.9
76.6
71.2
73.1
48.9
55.1
62.6
57.5
81.0
74.3
77.0
79.5
70.0
74.5
21.0
25.0
26.0
35.0
41.0
38.6
5.00
38.4
49.1
52.8
84
90
108
101
110
106
123
94
98
98.2
12
12
18
15
41
25
18
11
12
6
19.7
15.7
12.5
8.24
11.0
9.83
30.4
6.20
6.33
6.81
13.8
11.9
5.32
3.34
2.61
7.13
3.72
5.12
4.24
5.26
3.45
0
0.516
0.873
0
1.32
0
2.29
0
4.07
42.9
29.2
33.3
17.7
28.7
21.8
126
12.4
9.92
12.0
* Pooled RSD calculated as the square root of the mean of the squared within-pair RSDs

       Recovery, as a function of concentration, is plotted in Figure 4-1. (Plots of absolute and relative
precision as a function of concentration are addressed with precision for biosolids and tissue samples in
March 2010
16

-------
                                                             Method 1668A Interlaboratory Validation Study
Section 4.5.) The spike concentrations displayed in Figure 4-1 do not match exactly the concentrations
that were spiked (see Table 2-2) because coelutions result in combined concentrations.
               120
               100
              £ 80
              o
              u
              £
              I 60

              1
              " ,
                20
!f •• •*  *!
          •i
 .-   4   •
.   .Ğ;Ğ'!;
                       tt  *'
                              2000         4000         6000
                                      Spike Concentration (pg/L)
                                                                    8000
                                                                                10000
            Figure 4-1. Mean Recovery vs. Spike Concentration, PCB Congeners in Wastewater
4.5     Variability as a Function of Concentration

        Because true congener concentrations in the tissue and biosolids samples were not known, it was
not possible to calculate recoveries of congeners from tissue and biosolids. Variability (precision) vs.
concentration was determined for wastewater, soil, and, tissue matrices.  The following subsections
present plots of absolute precision (as standard deviation of the determined concentrations) and relative
precision (as relative standard deviation) as functions of congener concentration for each of these
matrices.  For the three matrices, standard deviation increased approximately linearly with increasing
concentration. It was expected that, at the very low concentrations in the tissue and biosolids samples,
standard deviation would become constant and the plots would resemble a "hockey stick." (The
wastewater sample was not spiked at low enough concentrations to demonstrate this effect.) The lack of a
hockey stick appearance for the tissue and biosolids; i.e., the lack of constant standard deviation at low
concentrations, is good because this indicates that measurements are being made in the quantitative range
for the congeners. This is not surprising because the rigorous congener identification criteria in Method
1668A are that the signal-to-noise ratio must be greater than 3 and ratio of the peak heights or areas for
the 2 exact m/z's must be within 15% of theoretical, in addition to the requirement that the relative
retention time of the congener must be within a specified window based on a calibration or calibration
verification  standard. Thus, the identification criteria raise the lowest level of congeners that are
determined to levels above the region of constant standard deviation.

4.5.1   Variability vs. Concentration for Wastewater

4.5.1.1   Absolute variability vs. concentration for wastewater

        Figure 4-2 is a plot of the standard deviation as a function of concentration for the congeners
spiked into wastewater.  The congener concentrations are defined by the spiking solutions, as described in
Section 2.5.2.  Results appear slightly skewed to lower standard deviation at low concentration. The
skewed appearance is likely due to the higher concentrations of the coeluted congeners.
March 2010
                                                                        17

-------
                                                            Method 1668A Interlaboratory Validation Study
               4000 -,
               3500
              rsooo
               2500
              i2000
             §1500
             ?;1000
                500
                              2000
                                                                  8000
                                                                              10000
                                          4000         6000
                                       Spike Concentration (pg/L)
            Figure 4-2. Concentration Standard Deviation vs. Spike Concentration, PCB
                       Congeners in Wastewater
4.5.1.2  Relative variability vs. concentration for wastewater

       Figure 4-3 is a plot of RSDs as a function of concentration for the congeners spiked into
wastewater. The congener concentrations are defined by the spiking solutions, as described in Section
2.5.2. The  variability is somewhat higher than expected at the higher concentrations, with RSDs of
approximately 40%. The reason for these higher than expected RSDs is not known.
               120
               100
              I80
                60
              TJ
              C
              rc
              g
              > 40
                20
•••   S   !
44

                             2000        4000         6000
                                       Spike Concentration (pg/L)
                                                                  8000
                                                                              10000
            Figure 4-3. Relative Standard Deviation vs. Spike Concentration, PCB Congeners in
                       Wastewater
March 2010
                                                                   18

-------
                                                             Method 1668A Interlaboratory Validation Study
4.5.2   Variability vs. Concentration for Tissue

4.5.2.1  Absolute Variability vs. Concentration for Tissue

        Figure 4-4 is a plot of standard deviation as a function of concentration for congeners detected in
tissue. Congeners were detected in tissue from as low as a few parts-per-trillion (ppt; pg/g) to well into
the part-per-billion (ppb; ng/g) range.
               1000 -i
               1 100
              i
              t>
              Q
              E
              i
                 0.1
                           ir *• •
                        • *v*?* •
                       ЕР .tĞğğ
*.  %*-.
                                      *..•
                                   10              100
                                        MeanResult(pg/g)
                                                                 1000
                                                                                10000
           Figure 4-4. Mean vs. Standard Deviation of Measured Tissue Results

4.5.2.2   Relative Variability vs. Concentration for Tissue

        Figure 4-5 is a plot of RSD as a function of concentration for the congeners detected in tissue.
RSDs are mostly between 10 and 30 percent, as expected, with a few outlying high values. The unusually
high RSDs occurred in congeners that are only rarely detected (2-3 laboratories.)
March 2010
                                                                 19

-------
                                                             Method 1668A Interlaboratory Validation Study
               140
               120
              rioo
              B
              >
              ? 80
              0
              •D
              1
                60
                40
                20
                                  10
                                                 100
                                         Mean Result (pg/g)
                                                                1000
                                                                                10000
            Figure 4-5.  Mean vs. Relative Standard Deviation of Measured Tissue Results

4.5.3   Variability vs. Concentration for Biosolids

4.5.3.1  Absolute Variability vs. Concentration for Biosolids

        Figure 4-6 is a plot of standard deviation as a function of concentration for congeners detected in
biosolids. Congeners were detected in biosolids from as low as a few ppt to well into the ppb range.
               10000
               11000
              |  100
              6
                  10
                 O.I
~*     *~iĞ*." * /  *"
  V  *Ğ•*•.   •  Ğ
                    i
                                                                 1000
                                                                                I 0000
                                   10             100
                                        Mean Result (pg/g)
            Figure 4-6. Mean vs. Standard Deviation of Measured Biosolids Results
4.5.3.2   Relative Variability vs. Concentration for Biosolids

        Figure 4-7 is a plot of RSDs as a function of concentration for the congeners detected in
biosolids. Unlike the plots of relative variability for tissue and wastewater samples, this plot does not
March 2010
                                                           20

-------
                                                           Method 1668A Interlaboratory Validation Study
suggest a strong relationship between variability and concentration. RSDs are mostly between the
expected ranges of 10 to 30 percent, with a few outlying high values.
               100 -i
                so
              §
                60
             Q
             •2
             1C
             •o
                40
              0)
              ct
                  1
                                   X  •
                                                               1000
                                                                              10000
                                 10             100
                                     Mean Concentration (pg/g)
            Figure 4-7. Mean vs. Relative Standard Deviation of Measured Biosolids Results
4.6     Labeled Compound Recovery and Precision

        Table 4-4 lists labeled compound recovery and precision for the 27 labeled congeners spiked into
wastewater, biosolids, and tissue samples.  Except for congener 1L, median recoveries ranged from 56 to
94 percent. Except for congener 178, pooled RSDs ranged between 5 and 22 percent. The low recovery
of congener 1L and, to some extent other congeners at low chlorination levels, is thought to be caused by
loss in the solvent evaporation step.  The reason for the high RSD for congener 178 is not known.
Table 4-4. Recovery and Precision of Labeled Compounds Spiked into Samples1
LOG2
1
1
2
2
3
3
3
4
4
4
5
5
5
5
5
5
Labeled
Congener3
1L
3L
4L
15L
19L
28L
37L
54L
77L
81L
104L
105L
111L
114L
118L
123L
Recovery (%)
# Results
32
32
32
32
32
32
32
31
31
31
32
31
29
32
32
32
Mean
52.9
60.4
60.8
64.8
59.0
78.8
77.2
61.6
72.7
74.9
76.5
81.9
85.2
83.0
83.9
84.7
Median
48.5
61.1
56.1
69.0
59.0
87.8
77.6
59.1
67.3
71.0
80.0
86.3
86.3
86.3
87.5
88.8
Min.
13.0
15.9
32.0
25.0
4.5
25.0
35.0
25.5
43.0
26.0
41.0
52.7
63.0
48.0
53.8
53.8
Max.
95.0
107
120
96.3
112
118
110
109
106
129
102
103
110
119
120
116
Within-pair RSD (%)
# Pairs
16
16
16
16
16
16
16
15
15
15
16
15
13
16
16
16
Pooled4
22.0
18.9
16.7
13.7
12.0
11.7
10.0
10.3
9.1
9.0
10.1
10.9
5.4
10.8
10.8
11.4
Median
17.4
10.9
11.9
2.8
5.4
7.8
2.4
5.4
5.7
6.6
4.8
5.2
3.7
5.2
6.4
6.5
Min.
0.81
2.70
1.30
0
0.69
1.40
0.52
0.87
1.10
1.30
0
0.50
0.29
1.00
0.36
0.48
Max.
43.4
47.6
31.6
31.4
26.5
33.3
23.6
22.0
19.8
18.8
24.4
25.2
12.7
28.3
26.6
23.4
March 2010
21

-------
                                                                  Method 1668A Interlaboratory Validation Study
Table 4-4. Recovery and Precision of Labeled Compounds Spiked into Samples1
LOG2
5
6
6
6
6
7
7
7
8
8
9
9
10
Labeled
Congener3
126L
155L
156L+157L
167L
169L
178L
188L
189L
202L
205L
206L
208L
209L
Recovery (%)
# Results
32
32
32
31
32
30
32
32
31
32
32
32
32
Mean
81.0
77.3
87.8
84.3
78.3
89.1
78.0
82.8
84.9
82.3
80.8
80.4
78.7
Median
80.8
81.3
76.5
82.0
75.3
93.8
85.2
82.0
87.8
81.5
81.3
84.7
79.8
Min.
59.0
33.0
44.0
57.3
30.5
5.0
32.0
54.0
38.4
51.0
50.0
49.0
47.7
Max.
123
106
216
110
112
126
136
118
145
118
115
129
110
Within-pair RSD (%)
# Pairs
16
16
16
15
16
14
16
16
15
16
16
16
16
Pooled4
11.8
7.8
12.9
9.9
14.8
34.6
8.5
8.1
9.2
7.9
9.2
7.6
9.2
Median
4.5
3.6
8.0
8.6
8.0
4.7
4.6
4.8
5.4
4.8
7.7
4.9
6.3
Min.
0
0
1.10
1.30
2.40
0.69
0.22
0
1.30
0.25
0
0
2.20
Max.
28.7
21.8
30.3
22.6
37.5
126
17.4
18.2
19.9
17.3
19.8
17.2
18.9
1  Wastewater, biosolids, and tissue

2 Level of chlorination
3
  Labeled analog of World Health Organization dioxin-like (Toxic) congener shown in bold
  Pooled RSD calculated as the square root of the mean of the squared within-pair RSDs
March 2010
22

-------
                                                           Method 1668A Interlaboratory Validation Study
                                         Section 5
                Revision of Quality Control Acceptance Criteria
       Interlaboratory quality control (QC) acceptance criteria were developed for initial precision and
recovery (IPR), on-going recovery (OPR; laboratory control sample, LCS), and for recovery of labeled
compounds from samples.  These revised criteria are in Table 5-1 of this report, and Table 6 of the
successor method, 1668B.  The statistical details for development of these criteria are in an appendix to
this report.  The tests to which these criteria are applied are discussed in this Section of this report.

5.1    Calibration

       The study plan and study-specific instructions suggested a 5- or 6- point calibration. Laboratories
did not provide enough calibration data to permit revision of the QC acceptance criterion for calibration
linearity. Therefore, the criterion for which an average relative response may be used for a given
congener remains at 20%, as stated in Section 10.4.4 of EPA Method 1668A; otherwise, a calibration
curve must be used for that congener.  This calibration linearity criterion applies to congeners determined
by isotope dilution only (i.e., the "toxics," "level of chlorination," and "GC window-defining" congeners)
because all other congeners are calibrated at a single point.

5.2    Calibration Verification

       The study plan and study-specific instructions suggest single calibration verification after
calibration. Because only two laboratories submitted calibration verification data, EPA did not revise the
calibration verification QC acceptance criteria.  The calibration verification QC acceptance criteria in
Table 5-1 remain unchanged from previous revisions of EPA Method 1668A. If EPA receives calibration
verification data from enough laboratories, EPA may revise these criteria in future versions of 1668.

5.3    Initial Precision and Recovery

       To minimize resource burden on volunteer participants, laboratories were not required to prepare
and analyze IPR samples. Instead, EPA used the OPR data gathered in the study to develop revised IPR
and OPR QC acceptance criteria.  In addition, results from the aqueous and solids (sand/corn oil) OPRs
were combined to yield a single set of OPR QC acceptance criteria that would be applicable to aqueous,
solids, and tissue samples.  Two laboratories resolved labeled and native congeners 156 and 157, while
the other laboratories reported these congeners as coeluting pairs.  Similarly,  one  laboratory reported
coelution of congeners 4 and 10, one laboratory reported coelution of congeners 114 and 122, and two
laboratories reported congener 106 coeluting with either congener 107 or 109. Because calculations of
IPR/OPR QC acceptance criteria were based on recoveries, coelution was ignored when generating the
revised criteria.

       Data from Laboratories 2, 3, 6, 11, and 14 were excluded for the reasons described in Section 3 of
this report.  The remaining dataset yielded a total of 15 usable reagent water and solid matrix OPR
samples.  After performing Grubbs' outlier tests on these OPRs, a total of 13 individual data points were
identified as outliers and removed from the dataset prior to development of revised IPR/OPR QC
acceptance criteria. Table 5-1 presents revised IPR QC acceptance criteria. When compared to  QC
acceptance criteria in Method 1668A, recoveries windows are generally narrower than those in the
method.  Recoveries for low molecular weight congeners are centered lower than for the other congeners
and for recoveries of low molecular weight congeners in Method 1668A.  These lower recovery windows
reflect that these congeners are partially lost in the  solvent evaporation step(s).
March 2010                                                                                    23

-------
                                                         Method 1668A Interlaboratory Validation Study
       QC acceptance criteria for IPR precision, as relative standard deviation (RSD) of recoveries, are
also presented in Table 5-1.  The RSDs generally are higher than those in Method 1668A for some of the
low molecular weight congeners, and lower for some of the other congeners. The higher RSDs for the
low molecular weight congeners reflect partial loss of these congeners in the solvent evaporation step(s)
in Method 1668A, resulting  in greater variability in results for these congeners.

5.4    Ongoing Precision and Recovery

       Each participating laboratory was required to spike and analyze two reagent water OPR samples.
These samples were used to evaluate laboratory and method performance and to update IPR and OPR QC
acceptance criteria.  Although not required by the study design, four laboratories analyzed at least one
solids matrix OPR sample. In some cases, laboratories provided one solids matrix OPR and one reagent
water OPR instead of two reagent water OPRs.  In other cases, laboratories supplemented the two reagent
water OPRs with one or more solids matrix OPRs.

       Revised OPR QC acceptance criteria are in Table 5-1. As with the IPR QC acceptance criteria,
OPR recovery windows are, generally, narrower than those in Method 1668A, and centered lower for
some of the low molecular weight congeners.

5.5    Labeled Compound Recovery from Samples, Blanks, and IPR and OPR standards

       Labeled compound recovery data from samples were used to construct revised QC acceptance
criteria for labeled compound recoveries.  Results from a total of 24 analyses were used to develop the
labeled compound recovery  QC acceptance criteria (Table 5-1.)  The IPR and OPR recovery windows are
centered lower, for the low molecular weight congeners.
March 2010                                                                                 24

-------
                                                                                                                   Method 1668A Interlaboratory Validation Study
Table 5-1. Revised QC Acceptance Criteria for IPR, OPR, and Labeled Compounds in Samples
Congener
2-MoCB
4-MoCB
2,2'-DiCB
4,4'-DiCB
2,2',6-TrCB
3,4,4'-TrCB
2,2',6,6'TeCB
3,3',4,4'-TeCB
3,4,4',5-TeCB
2,2',4,6,6'-PeCB
2,3,3',4,4'-PeCB
2,3,4,4',5-PeCB
2,3',4,4',5-PeCB
2',3,4,4',5-PeCB
3,3',4,4',5-PeCB
2,2',4,4',6,6'-HxCB
2,3,3',4,4',5-HxCB4
2,3,3',4,4',5'-HxCB4
2,3',4,4',5,5'-HxCB
3,3',4,4',5,5'-HxCB
2,2',3,4',5,6,6'-HpCB
2,3,3',4,4',5,5'-HpCB
2,2',3,3',5,51,6,61-OcCB
2,3,3',4,4',5,5',6-OcCB
2,2',3,3',4,41,5,51,6-NoCB
2,2',3,3,'4,5,51,6,61-NoCB
DeCB
Congener
number1
1
3
4
15
19
37
54
77
81
104
105
114
118
123
126
155
156
157
167
169
188
189
202
205
206
208
209
Test cone.
(ng/mL)2
50
50
50
50
50
50
50
50
50
50
50
50
50
50
50
50
50
50
50
50
50
50
50
50
50
50
50
IPR
RSD (%)
25
22
18
17
13
26
17
20
20
19
19
18
13
16
17
15
16
16
13
16
14
16
17
15
17
17
20
Recovery (%)3
84-119
83-112
82-105
85-107
86-103
77-109
84-106
81 -106
81 -106
83-107
83-107
83-105
88-105
82-102
82-104
86-105
87-108
87-108
85-101
80-100
88-106
85-106
82-104
87-107
85-106
86-108
81 -106
OPR
Recovery (%)3
71 -132
72-123
73-114
76-116
79-109
64-122
76-114
71 -116
70-116
74-117
73-117
74-113
81 -112
74-109
74-113
79-112
78-117
78-117
79-107
73-108
81 -113
77-114
74-112
79-115
76-115
77-116
71 -116
Labeled Compound Recovery in
Samples and Blanks (%)3
NA
Labeled Compounds
l3C12-2-MoCB
13C12-4-MoCB
13C12-2,2'-DiCB
l3C12-4,4'-DiCB
13C12-2,2',6-TrCB
1L
3L
4L
15L
19L
100
100
100
100
100
78
63
56
70
68
21 -100
31 -100
35-100
34-100
32-100
2-100
13-100
18-100
10-118
10-106
4-100
11 -106
14-107
19-107
1 -108
March 2010
25

-------
                                                                                                        Method 1668A Interlaboratory Validation Study
Table 5-1. Revised QC Acceptance Criteria for IPR, OPR, and Labeled Compounds in Samples
Congener
l3C12-3,4,4'-TrCB
l3C12-2,2',6,6'-TeCB
13C12-3,3',4,4'-TeCB
l3C12-3,4,4',5-TeCB
13C12-2,2',4,6,6'-PeCB
13C12-2,3,3',4,4'-PeCB
l3C12-2,3,4,4',5-PeCB
13C12-2,3',4,4',5-PeCB
l3C12-2',3,4,4',5-PeCB
l3C12-3,3',4,4',5-PeCB
13C12-2,2',4,4',6,6'-HxCB
l3C12-2,3,3',4,4',5-HxCBb
13C12-2,3,3',4,4',5'-HxCB5
13C12-2,3',4,4',5,5'-HxCB
l3C12-3,3',4,4',5,5'-HxCB
13C12-2,2',3,4',5,6,6'-HpCB
l3C12-2',3,3',4,4',5,5'-HpCB
l3C12-2,2',3,3',5,5',6,6'-OcCB
13C12-2,3,3',4,4',5,5',6-OcCB
l3C12-2,2',3,3',4,4',5,5',6-NoCB
13C12-2,2', 3,3', 4,5,5', 6,6'-NoCB
13C12-2,2',3,3',4,4',5,5',6,6'-DeCB
Cleanup standards
13C12-2,4,4'-TrCB
13C12-2,3,3',5,5'-PeCB
l3C12-2,2',3,3',5,5',6-HpCB
Congener
number
37L
54L
77L
81L
104L
105L
114L
118L
123L
126L
155L
156L
157L
167L
169L
188L
189L
202L
205L
206L
208L
209L

28L
111L
178L
Test cone.
(ng/mL)2
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100

100
100
100
IPR
RSD (%)
57
62
35
33
48
31
41
33
32
29
42
35
35
24
33
47
28
50
21
29
32
30

63
23
30
Recovery (%)3
47-104
37-100
57-100
57-100
49-100
66-101
57-100
65-102
66-103
67-100
58-103
61 -100
61 -100
74-103
66-103
53-102
68-100
56-113
70-100
64-100
62-100
65-100

43-106
75-102
78-117
OPR
Recovery (%)3
24-128
16-111
43-105
44-102
30-115
52-116
39-117
51 -117
52-118
54-113
40-121
46-115
46-115
63-115
51 -117
33-121
55-112
33-136
61 -103
51 -107
48-111
52 - 1 1 1

18-131
64-113
62-133
Labeled Compound Recovery in
Samples and Blanks (%)
25-123
13-105
31 -109
14-127
36-115
50-111
41 -121
49-111
49-116
50-106
25-124
40-120
40-120
45-118
37-117
23-125
47-116
31 -134
46-115
38-122
31 -126
43-115

14-131
57-112
57-125
1  Suffix "L" indicates labeled compound.
2 See Table 5 in EPA Method 1668A.
3 Where necessary, the limit was increased to include 100% recovery.
4 PCBs 156 and 157 are tested as the sum of the two concentrations.
5 Labeled PCBs 156L and 157L are tested as the sum of the two concentrations.
  NA = Not applicable
March 2010
26

-------
                                                           Method 1668A Interlaboratory Validation Study
                                         Section 6
                                       Conclusions
       This study demonstrated that PCB congeners can be measured in water, biosolids, and tissue in
multiple laboratories using EPA Method 1668A. Results show that recovery is nearly constant as a
function of concentration, and that precision is proportional to concentration.  Of significance with this
method is the benefit that measured concentrations are corrected by the isotope dilution technique, even
when the recovery of the labeled compounds is low.

       The results of this interlaboratory study met our objectives to characterize the performance of
Method 1668A in several laboratories and matrices, and use the results to replace the single-laboratory
QC acceptance criteria in 1668A with interlaboratory criteria. These new interlaboratory QC criteria are
in Table 6 of the successor EPA Method,  1668B.
March 2010                                                                                    27

-------
                                                         Method 1668A Interlaboratory Validation Study
                                      Appendix A
     Statistical  Procedures Used to Develop QC Acceptance Criteria


1.0    Initial Precision and Recovery (IPR) and Ongoing Precision and Recovery (OPR)

       IPR and OPR QC acceptance criteria were calculated using OPR results for all matrix types for
each given congener. The acceptance criteria were calculated as prediction limits for mean and individual
recoveries, set at the 95% confidence level.

       Prior to calculation of QC acceptance criteria, Grubbs' outliertest, as described in ASTM E178-
02, was first run on the individual OPR sample recoveries. Based on Grubbs' test, a single outlying
recovery was removed for 13 of the native or labeled congeners. These results were not included in the
subsequent IPR and OPR QC acceptance criteria calculations.

Upper and lower limits for IPR samples were calculated as:

                                  -^ — t(0.975,n-l)  S


where: X is the overall mean of all OPR recoveries for the given congener,
       s is the standard deviation of all OPR recoveries for the given congener, and
       n is the number of OPR recoveries for the given congener.

Upper and lower limits for OPR samples were calculated as:

                                 X + t^o-,      *s*Jl + -
                                      (0.975,n-l)     -y    n

The maximum RSD for IPR samples was calculated as:
                                    — *  /F
                                    y  V (0.95,n-l)

2.0    Labeled Compound Recovery from Samples and Blanks

       QC acceptance criteria for the recovery of labeled compound from samples and blanks were
calculated using all labeled sample results for all matrix types for the given congener.  The acceptance
criteria were calculated as prediction limits for mean and individual recoveries, set at the 95% confidence
level.

       Prior to calculation of QC acceptance criteria, the Grubbs' outliertest, described in ASTM E178-
02, was first run on the individual labeled sample recoveries. Based on Grubbs' test, two outlying
recoveries were each removed for two of the native or labeled congeners. These results were not included
in the subsequent labeled sample recovery QC acceptance criteria calculations.
Upper and lower limits for IPR samples were calculated as:

                                   ^ — t(Q.975,n-l)


where: X is the overall mean of all labeled sample recoveries for the given congener,
       s is the standard deviation of all labeled sample recoveries for the given congener, and
       n is the number of labeled sample recoveries for the given congener.
March 2010                                                                                A-l

-------
                                                 Method 1668A Interlaboratory Validation Study
                                Appendix B
                                 Study Plan
                                     for
             Interlaboratory Validation of EPA Method 1668A
    for Determination of Chlorinated Biphenyl Congeners in Water,
                  Biosolids, and Tissue  by HRGC/HRMS
                                  Prepared for:

                    William A. Telliard, Director of Analytical Methods
                       Engineering and Analysis Division (4303T)
                          Office of Science and Technology
                                 Office of Water
                         U.S. Environmental Protection Agency
                            1200 Pennsylvania Avenue NW
                              Washington, DC  20460
                                  Prepared by:

                          DynCorp Systems & Solutions LLC
                              6101 Stevenson Avenue
                              Alexandria, VA 22304
                                   May 2003
March 2010                                                                     B-l

-------
                                                           Method 1668A Interlaboratory Validation Study
Acknowledgments

This study plan was prepared under the direction of William A. Telliard of the Engineering and Analysis
Division within the U.S. Environmental Protection Agency (EPA) Office of Water.
Disclaimer

This study plan has been reviewed and approved by EPA's Engineering and Analysis Division. Mention
of company names, trade names, or commercial products does not constitute endorsement or
recommendation for use.
March 2010                                                                                   B-2

-------
                                                           Method 1668A Interlaboratory Validation Study
1.0    Introduction

       This study plan is for interlaboratory validation of EPA Method 1668, Revision A: Chlorinated
Biphenyl Congeners in Water, Soil, Sediment, and Tissue by HRGC/HRMS ("Method 1668A"). Method
1668A is for determination of the 12 polychlorinated biphenyl (PCB) congeners designated as "toxic" by
the World Health Organization (WHO), and the remaining 197 chlorinated biphenyl (CB) congeners,
either as individual congeners or as congener groups.4

2.0    Background

       From the 1940s into the early 1980s, PCBs were manufactured under several trade names, most
predominantly "Aroclor" in the U.S. The Aroclor name was accompanied by a four-digit number
indicating the degree of chlorination of the commercial mixture (e.g., Aroclor  1016, Aroclor 1260, etc.).
In general, the higher the number, the higher the degree of chlorination.

       From the late 1950s through the  1970s, PCBs were determined as Aroclors by low resolution
(packed column) GC with an electron capture detector (BCD). In the late 1970s and early 1980s,
heightened interest in PCBs and ambiguities in PCB identification led several researchers to separate and
identify all 209 PCB congeners using high resolution (open tubular capillary) GC columns coupled with
low resolution mass spectrometry (LRMS).  In the early to mid-1990s, researchers began to investigate
use of high resolution mass spectrometry (HRMS) more intensely as a means to reduce or eliminate
interferences that compromise measurement of PCBs by BCD or LRMS.

       In 1995, EPA developed Method 1668, which uses high resolution GC (HRGC)  combined with
HRMS for determination of 13 dioxin-like PCBs that the World Health Organization (WHO) designated
as "toxic" in 1994. Method 1668 was based on data from  studies conducted at Pacific Analytical, Inc.,
Carlsbad, CA. In  1997, interest in additional congeners led EPA to investigate determination of as many
congeners as possible in a single HRGC/HRMS run. This led to draft Revision A of EPA Method 1668.
At about the same time that Method 1668 A was drafted, WHO modified the list of dioxin-like congeners
by adding congener 81 and deleting congeners 170 and 180, resulting in the current list of 12 PCBs that
exhibit "dioxin-like" toxicity.

       Method 1668A was validated in  a single-laboratory study at AXYS Analytical Services Ltd.,
Sidney, BC, Canada. AXYS Analytical produced a report that was subsequently published in March,
2000, in two parts titled: Development of a Full-Congener Version of EPA Method 1668 and Application
to Determination of 209 CB Congeners in Aroclors (Part I) and Development of Method  1668A (Part II).

       Draft Method 1668A was subjected to formal peer review in September-October of 1999. The
peer review was conducted in accordance with EPA's Science Policy Council Peer Review Handbook
(EPA 100-B-98-001, January 1998). Based on the peer review, EPA revised and published Method
1668A without the word "Draft" in December of 1999 (EPA 821-R-00-002). EPA also published a report
titled Peer Review of Draft EPA Method 1668, Revision A: Chlorinated Biphenyl Congeners in Water,
Soil, Sediment, and Tissue by HRGC/HRMS in February 2000.
4Although some congeners have only a single chlorine atom, the entire suite of 209 chlorinated biphenyl congeners
will be referred to as "PCBs" in the remainder of this study plan for consistency with common usage.


March 2010                                                                                  #

-------
                                                            Method 1668A Interlaboratory Validation Study
3.0     Study Objectives

        The objectives of this study are to 1) characterize the performance of EPA Method 1668A in
multiple laboratories and matrices and 2) evaluate and, if appropriate, revise the quality control (QC)
acceptance criteria in the method. The ultimate objective is to propose and promulgate Method 1668 at
40 CFR part 136 for use in EPA's Clean Water Act programs.

        To ensure that these study objectives are met, EPA will require that:

    •   Each laboratory follow all analytical and quality control procedures in EPA Method 1668A and
        study-specific instructions,

    •   Any laboratory that wishes to deviate from the procedures in Method 1668A or the study-specific
        instructions obtain prior approval of the changes and document those approved changes in detail

    •   All data produced be capable of being verified by an independent person reviewing the analytical
        data package

    •   Each laboratory has a comprehensive quality assurance (QA) program in place and operating
        throughout the study. This QA program will ensure that the data produced are of appropriate and
        documented quality.
4.0     Study Management

        The study will be managed by the Statistics and Analytical Support Branch (SASB) in the
Engineering and Analysis Division within EPA's Office of Science and Technology. Day-to-day
management and coordination of study activities will be performed by the contractor-operated Sample
Control Center (SCC) under SASB guidance.5 SCC will coordinate the purchase of standards, sample
collection, sample and data tracking, and monitor day-to-day study activities. SCC also will establish
schedules for activities given in this study plan and will keep SASB informed as to the status of the study.
SASB will draw conclusions from the study and produce a report presenting study results.  If appropriate,
SASB will revise Method 1668A as necessary to reflect study findings.

5.0     Study Design

        The design of this study is intended to provide EPA with a sufficient amount of data to evaluate
method performance in accordance with the guidelines published by EPA, AOAC-International, and
ASTM International.6'7'8  These guidelines recommend a minimum of six data sets for evaluation of a
method. In order to allow for some loss of data due to error, lost samples, outlier removal, or other
unforeseen causes, EPA plans to identify at least nine laboratories willing and able to participate in the
5The Sample Control Center (SCC) is operated by DynCorp Systems & Solutions LLC under EPA Contract No. 68-
C-01-091.  All SCC activities are performed under the direction and guidance of EPA SASB.

6 Guidelines for Selection and Validation of US EPA 's Measurement Methods, U.S. EPA Office of Acid Deposition,
Environmental Monitoring and Quality Assurance (OADEMQA), Office of Research and Development, U.S.
Environmental Protection Agency, August 1987 DRAFT.

'"Report of the Committee on Collaborative Interlaboratory Studies," J. Assoc. Office. Anal. Chem., 67, (2), 1984


March 2010                                                                                    B-4

-------
                                                            Method 1668A Interlaboratory Validation Study
study.  However, given the relatively limited number of laboratories with the equipment and experience
necessary to analyze for PCBs using HRGC/HRMS, it may not be possible to identify nine laboratories
willing to participate as volunteers or to obtain at least six usable sets of data. If it is not possible to
obtain at least six usable data sets, EPA may utilize any Method 1668A data available to assess method
performance, develop revised QC acceptance criteria, and for other purposes.

        Due to budget limitations, EPA intends to seek as much volunteer participation as possible in this
study.  To help offset study costs, EPA will provide volunteer laboratories with a set of analytical
standards necessary to implement Method 1668A. Even  so, it is not reasonable to expect laboratories to
donate tens of thousands of dollars worth of analyses. Therefore, the number of analyses will be balanced
against the need to obtain a sufficient number of participant laboratories.

        An interlaboratory study designed in accordance  with ASTM standard D-2777 would involve
spikes of all 209 congeners at multiple and replicate concentrations in multiple matrices, plus  initial and
batch QC. The total number of analyses per laboratory could be upwards of 75 if calibration,  QC, and  a
method detection limit (MDL) study are included. Given that a single HRGC/HRMS analysis costs $750
- 1200, the cost for such a study in a minimum of nine laboratories would exceed available EPA resources
and be impractical for volunteers.

        To address these cost concerns, EPA intends to include no more than two samples of each matrix
type, with each sample containing varying concentrations of the target PCB congeners.  EPA anticipates
validating the method in wastewater, biosolids, and fish tissue. To further reduce study costs, EPA plans
to use excess sample volume collected from previous studies of biosolids and fish.  Biosolid samples or
sample locations will be selected based on results of EPA's 2001 National Sewage Sludge Survey; tissue
samples or sample locations will be selected based on results of EPA's ongoing National Study of
Chemical Residues in Fish Tissue. EPA does not have a similar supply of stored wastewater sample
volume.  Therefore, EPA  plans to collect and spike wastewater samples with PCBs.

        Given the above considerations, EPA believes that the study can be conducted with a  total of 10
analyses per laboratory (in addition to 5 runs necessary to determine the absolute and relative  retention
time for each congener, and an initial 6-point instrument  calibration) as follows:

    •   2 reagent water  samples,
    •   2 biosolid samples,
    •   2 tissue samples,
    •   2 wastewater samples,
    •   1 reagent water  blank, and
    •   1 solids/tissue blank (playground sand spiked with corn oil).

        EPA believes that increasing the number of samples beyond the numbers described above would
significantly limit the number of laboratories willing to participate in the study, even with enticements
offered by the recognition gained through participation in the  study and EPA-provided standards.

6.0     Study Implementation

        The study will be conducted in four phases: (1) identifying and selecting the participant
laboratories; (2) collecting, preparing, and shipping standards and samples; (3) sample analysis and data
reporting; and (4) data review and assessment.  Details of each phase are summarized below.


8 ASTM Standard D2777-98," Standard Practice for Determination of Precision and Bias of Methods of Committee
D-19 on Water," Annual Book of ASTM Standards, Vol. 11.01, ASTM International, West Conshohocken, PA
19428.

March 2010                                                                                    #T

-------
                                                            Method 1668A Interlaboratory Validation Study
6.1    Phase 1 - Laboratory Identification and Selection

       The study will involve one sample processing laboratory and a group of participant laboratories.
The total number  of participant laboratories will depend upon laboratory capability, availability, cost, and
scheduling constraints.  Participant laboratories may include commercial laboratories, academic
laboratories, State laboratories, EPA laboratories, and/or municipal laboratories. EPA will also request
participation by international laboratories so that study results reflect worldwide application of EPA
Method 1668A. EPA recognizes international environmental concerns and abilities to implement
laboratory analytical techniques targeting PCBs, and successfully included international participation in
validating EPA Method 1613 (chlorinated dibenzo-p-dioxins and dibenzofurans), EPA Method 1622
(Cryptosporidium), EPA Method 1623 (Cryptosporidium and Giardia), and EPA  Method 1631
(mercury).  As noted  in Section 5 above, EPA  1) plans to identify at least nine laboratories willing and
able to participate in the study and 2) intends to seek as much volunteer participation as possible.

       All laboratories that participate in the study will be required to demonstrate that they have recent
experience in analyzing for chlorinated pollutants in environmental samples by HRGC/HRMS with
selected ion monitoring (SIM).  This is intended to ensure that study participants already have the
facilities, equipment, and trained staff necessary to implement Method 1668A.  Once qualified participant
laboratories have been identified, they will be provided with at least two weeks notice of their selection to
participate  in the study before the study begins. This is intended to provide study participants with a
reasonable amount of time to review any study-specific instructions.

Note:  Given the relatively limited number of laboratories with HRGC/HRMS instrumentation, and
       EPA 's desire to obtain volunteer support, it may not be possible to achieve a sufficient number of
       laboratories to meet the study design.  If a sufficient number of volunteer laboratories are not
       identified, EPA may consider issuing contracts with one or more qualified laboratories through a
       competitive bidding process.

6.2    Phase 2 - Collection, Preparation, and Shipment of Samples and Standard Solutions

6.2.1   Sample Identification and Collection

       Biosolid samples will be generated from excess sample volume collected during EPA's 2001
National  Sewage Sludge Survey. Tissue samples will be generated from excess sample volume collected
during EPA's National Study of Chemical Residues in Fish Tissues. Excess sample volume from both
studies is currently stored in freezers at an EPA sample repository.  As described in Section 5, EPA will
examine biosolids and fish tissue data from these studies to identify samples that contain PCB congeners
at concentrations of interest. In selecting the samples,  EPA's objective will be to maximize the number of
congeners represented and ensure that the congeners span the anticipated measurement range of the
method, ranging from the upper end of the calibration range down to "not detected."  In order to ensure
that a sufficient volume of each sample is available to support the needs of this study, EPA will identify
several samples of each matrix type that can be combined to produce large volumes of Youden pairs with
the desired congener distribution. Once these frozen, stored samples are identified, they will be
forwarded on ice to the Sample Processing laboratory.  (Although PCBs are stable and do not require
preservation, ice will be used to prevent decomposition of the fish and retard gas production in the
biosolids.)

       Because EPA does not have a stored supply of excess wastewater sample volume, wastewater
will be collected for this study from a publicly owned-treatment works (POTW) located near the  sampling
organization. Based on previous experience, EPA believes that municipal wastewater discharges are
unlikely to contain a sufficient number of PCB congeners at concentrations to adequately test the
capabilities of the method. Depending on available resources and the selected site location, these

March 2010                                                                                    ~B^>

-------
                                                            Method 1668A Interlaboratory Validation Study
wastewater samples may be collected by SCC, the Sample Processing Laboratory, States, or EPA
Regional staff.  Samples will be collected by individuals trained in appropriate sample collection and
handling procedures. The sampling team will collect a sufficient volume to allow for testing in all
laboratories and to provide extra volume in case of sample breakage, lost shipment, or other unforeseen
problems. Samples will be collected into pre-cleaned bottles (e.g., from a bottle-manufacturing process
that includes high-temperature annealing or that have been cleaned by a laboratory experienced in the
determination of PCBs by HRGC/HRMS). Because PCBs are known to be persistent in the environment,
wastewater samples will not be stored on ice.

        Immediately after sample shipment (i.e., as soon as samples are in the custody of the carrier), the
sample repository or sampling team will call SCC and provide information on the shipment, including
sample numbers, numbers of coolers, and courier and air bill number. SCC will notify the processing
laboratory of the scheduled shipment and confirm that samples have arrived in good condition and as
scheduled.  If necessary, SCC will implement tracking activities to locate any lost shipment(s).

6.2.2   Sample Processing at the Processing Laboratory

        Each set of tissue and biosolid samples sent to the sample processing laboratory will be
accompanied by a detailed set of instructions concerning combination and homogenization of sample
volumes, the number of aliquots to be prepared from each combined/homogenized sample, and
instructions for labeling the  prepared sample aliquots. These instructions will reflect the considerations
described in Section 6.2.1 (i.e., creating sufficient volume of samples that contain a large number of PCB
congeners at a wide  range of concentrations). The sample processing laboratory will combine and
homogenize the tissue and biosolid samples according to these instructions.

        EPA also will direct the sample processing laboratory to divide the unspiked wastewater into the
required number of aliquots and spike each aliquot separately (rather than spiking a bulk volume
wastewater and then subdividing the spiked sample into replicate aliquots).  Spiking each aliquot
separately avoids the problems  with "wall effects," whereby organic pollutants spiked into a bulk sample
in a solvent tend to adhere to the walls of the container, making it difficult, if not impossible, to divide the
bulk sample into multiple aliquots containing the same concentration. EPA will provide detailed
instructions to the sample processing laboratory regarding the number of aliquots, the PCBs to be used for
spiking, and spiking concentrations.  In developing those instructions, EPA will assume that any
background concentration of PCB congeners in the wastewater samples is minimal.

        Because PCBs are ubiquitous in the environment, including laboratories, the sample processing
laboratory must judiciously  guard against sample contamination.  To minimize  contamination, the
processing facility will homogenize the samples and divide the homogenized samples into replicate
aliquots under controlled conditions.

Note:   It is not necessary that the exact congener concentrations of each sample be known because 1)
        each sample will have been designed to ensure that a wide variety of congeners and
        concentrations are present, 2) the purpose of the study is to compare Interlaboratory
        measurements rather than to definitively characterize specific samples, and 3) spikes of labeled
        compounds into these matrices will be used to measure recovery.

6.2.3   Sample Shipment

        Once the study samples have been prepared, aliquoted, and labeled, the sample processing
laboratory will ship the samples to the participant laboratories via air courier. Because of the stability of
PCBs, the samples will not require preservation. Biosolids and tissue samples will be shipped on ice,
however, to hinder decomposition of tissues  and gas formation in  the biosolids. The processing
laboratory will notify SCC of the shipping date so that SCC can notify all participant laboratories of the

March 2010                                                                                    ~

-------
                                                             Method 1668A Interlaboratory Validation Study
shipping and scheduled arrival dates, and if necessary, implement tracking procedures for any lost
shipments.

Note:   If overseas laboratories are included in the study, biosolids and tissues may be freeze dried so
        that they can be shipped without concern that ice may melt during extended transit times.

6.2.4   Standards Acquisition, Packaging, and Shipment

        To reduce the cost to volunteer laboratories, EPA will provide each volunteer laboratory with a
single set of standards sufficient to calibrate their instrumentation and conduct all analyses.  Sets of
standards solutions will be acquired from suppliers of native and carbon-13 labeled compounds. If
possible, a single supplier will aggregate all standards solutions into a set, and package a set of standards
for shipment to each laboratory. To preclude injudicious use of standards, EPA will remind laboratories
of the instructions given in Method 1668A for combining and diluting standards.

6.3     Phase 3 - Sample Analysis and Data Reporting

6.3.1   Sample Analysis

        Participant laboratories will be required to analyze samples in a timely fashion in accordance with
the study schedule, and will follow procedures  for preparation, handling, and analysis of standards
solutions and samples provided in EPA Method 1668A.

        If analytical results appear unreasonable, laboratories will be instructed to investigate possible
causes,  first by checking for transcription and calculation mistakes, and then by reanalysis. Although
laboratories will be prohibited from performing multiple analyses to improve results, they will be allowed
to implement corrective action and reanalysis for QC failures that are attributable to instrument failure or
to analyst error (e.g., incorrect spiking levels).

6.3.2   Data Reporting

        Specific reporting requirements will  be provided in detailed instructions to the laboratories.
Gathering data from analyses of 209 congeners in IPR, OPR, blank, and the study sample(s) could
represent a formidable challenge because of the multiplicity of possible data reporting formats. To
simplify data evaluation, EPA will provide an electronic spreadsheet template and request that
laboratories submit data in this suggested format.

        Each laboratory will be asked to report the following:

        •  Summary level data in spreadsheet format;
        •  Summary level and raw data in hardcopy format;
        •  Individual results, including results for all congeners found in all blanks.
           (Note: Laboratories will not be allowed to average results or perform other data
           manipulations beyond those described in Method 1668A. When results are below the
           minimum level of quantitation but  are detected, laboratories will be required to report the
           actual calculated result, regardless  of its value);
        •  A list of the composition and concentrations of PCB congeners in the calibration, IPR,
           blank(s), OPR, samples analyzed, and a run chronology;
        •  Copies of all raw data, including chromatograms, quantitation reports, spectra, bench sheets,
           and laboratory notebooks showing  weights, volumes, and other data that will allow
           verification of the calculations performed and will allow the final results reported to be traced
           to the raw data.  Each data element must be clearly identified in the laboratory's data package;


March 2010                                                                                     ~BJ

-------
                                                            Method 1668A Interlaboratory Validation Study
        •   A written report that details any problems associated with analysis of samples or standard
           solutions.  The written report also must provide comments on the performance of any part of
           Method 1668A; and
        •   A detailed description of any modifications to the procedures specified in Method 1668A.
           Details and raw data from all runs will be reviewed for determination as to whether further
           testing is required.

        Laboratories also will be instructed to use the following rules in reporting results:

        •   Quantitative results above or at the MDL - report value;
        •   Quantitative results below the MDL - report value but flag with footnote giving the MDL;
        •   Nonquantitative results - report as less than the  MDL value and state the MDL value; and
        •   The terms "zero," "trace," or "ND (not detected)" are not to be used.

        EPA will request that laboratories submit analytical results within 45 days of receipt of samples.

6.4     Phase 4 - Data Review and Assessment

        Upon receipt of laboratory data packages, SCC will review the data to ensure all results were
generated in accordance with the method and with the requirements of this study plan and any associated
laboratory instructions. An objective of this review will be  to  maximize data use. If a discrepancy
occurs,  it will be resolved with the laboratory, where possible.  Data and laboratory comments and
recommendations will be assessed in the context of the objectives of the study and the ultimate uses of
EPA Method 1668A under the Clean Water Act.

        The objective of this assessment will be to evaluate the precision, recovery, and comparability of
results obtained by multiple laboratories employing the method, and to determine if the QC acceptance
criteria  in Method 1668A should be revised based on study  results. Results of this assessment will be
published in a study report.

        EPA plans to perform a statistical analysis  of the data to determine  acceptability and suitability
for use.  This statistical analysis will be  performed  in accordance with Standard Practice for
Determination of Precision  and Bias of Applicable Test Methods of Committee D-19 on Water (ASTM
D2777) or other accepted statistical practice.

7.0     LIMITATIONS

        The study design does not include a requirement that each laboratory perform an MDL or IPR
study in each reference matrix. In order to ensure that the MDL and IPR specifications published in the
final method can be achieved in these matrices by multiple laboratories, EPA intends to supplement this
study with MDL and IPR data gathered  from at least three sources. One of these sources will include
existing MDL data generated in reference tissue, solids, and aqueous matrices. EPA already has such
data from AXYS Analytical Services Ltd., and will contact  other laboratories to determine if such
existing data are available.

        Given the cost of Method 1668A analyses, EPA believes it is neither feasible nor necessary to
validate the method in each and every possible matrix or to  validate each and every congener at low,
medium and high concentrations. This study design focuses on representative matrices and
concentrations. EPA believes that application of a method to one or more matrices in multiple
laboratories usually can reflect the performance of the method across multiple matrices.  EPA also
believes that PCBs are extremely stable and are not subject to  adsorption and other processes that cause
percent  recovery to vary as  a function of concentration for some analytes (e.g., nitrophenols). If EPA is
able to gather data from other matrices and concentrations not tested in this study, EPA will make such

March 2010                                                                                   ~B4

-------
                                                               Method 1668A Interlaboratory Validation Study
data available to interested parties, either upon request or as an addendum to the final study report.  EPA
also is willing to consider expanding the study if the additional analyses can be justified in terms of the
additional information that they will provide, and if external funding can be found to support the
additional analyses.
March 2010                                                                                      B-10

-------