Polychlorinated biphenyls

Transcription

1 Congener-Specific Analysis of Polychlorinated Biphenyls by GC/ECD and GC/MS: Characterization of an Aroclor Mixture Polychlorinated biphenyls (PCBs) were manufactured and marketed in the United States under the trade name Aroclor until production was phased out in Aroclors are mixtures that contain various amounts of chlorine (e.g., Aroclor 1242 contains 42% by weight, 1248 contains 48%, etc.). There are 209 theoretically possible PCB congeners with chlorine numbers ranging from 1 to 10. Traditionally, PCBcontaminated samples have been analyzed by measuring one or several PCB congeners and subjectively matching the chromatographic pattern to one of the Aroclors. This method relies heavily on the experience of the analyst. It is important to measure PCBs on a congener-specific basis because individual congeners possess different physical and chemical properties (Pellizzari et al. 1985, Schwartz et al. 1987, Eganhouse and Gossett 1991). In the environment, PCBs are degraded and metabolized at different rates (Zell and Ballschmiter 1980, Harkness et al. 1993). Furthermore, toxicity studies have shown that some PCB congeners are more toxic than Viewing a chromatogram. would be to choose as calibration standards PCB congeners commonly encountered in the environment. Mixtures of Aroclors would fit this purpose and the question becomes how to separate and quantify Aroclors that can be used as daily calibration standards. Multidimensional gas chromatography using two columns with different stationary phases has potential for comothers (Safe 1984). Congener-specific measurements of PCBs are necessary to understand the source, fate, and transport of PCB compounds and their impact on the global ecosystem. Since PCBs have been supplied and disposed mostly in the form of Aroclors that do not contain all 209 congeners, the practical approach to the congener-specific analysis 35

2 ours is that we established relative retention times and response factors for all the PCB chromatographic peaks in the mixture so that most assignment errors arising from insufficient calibration standards have been eliminated. The Aroclor mixture quantified in this study will be used as a daily calibration standard for congener-specific analyses of PCBs in environmental samples. Los Angeles Harbor Materials and Methods plete separation and quantitation of Aroclors (Duinker et al. 1988; Schulz et al. 1989). However, single column gas chromatography has remained the most widely adopted method because of its simplicity and availability. So far, no single capillary column is able to completely resolve all chromatographic components for any Aroclor and the presence of multicomponent peaks makes accurate quantitation difficult. This problem can be partially resolved by selecting the most abundant component of a multi-component peak for calibration. The maximum calibration error for a multicomponent peak using this strategy would be less than 55%, and the error is much less for measurements of total PCBs (Eganhouse and Gossett 1991). The objective of this study was to develop a practical method for PCB congenerspecific analysis using GC/ ECD. We characterized a mixture of Aroclors 1242, 1248, 1254, and 1260 (1:1:1:1, wt%), called a secondary standard, by GC/ECD and GC/MS. Quantitation of the Aroclor mixture was made with 120 PCB congeners (primary standards) on a GC/ ECD; peaks were identified and confirmed on GC/MS. Eganhouse et al. (1989) also quantified a mixture of Aroclors 1242, 1248, 1254, and 1260 using GC/FID and GC/MS with capillary columns. They chose 27 PCB congeners as primary standards and treated each PCB isomer group with an average response factor of 1 to 6 congeners. The difference between their method and Selection of PCB Standards Schultz et al. (1989) detected 86 chromatographic peaks in a mixture of Aroclors 1016, 1242, 1254 and 1260 (1:1:1:1, wt%). We selected 85 congeners based on their results (congener 146 was not commercially available); the prominent components of multi-component peaks were selected. Eganhouse et al. (1989) further separated some of the multi-component peaks reported by Schultz et al. (1989). Hence we added 10 more congeners (congeners 1, 2, 3, 12, 13, 103, 104, 121, 165, and 166). Albro et al. (1981) characterized Aroclor 1248 and discovered two more congeners (113 and 154) that were not detectable in either of the above studies. They were included in our primary standards (although congener 113 remained unde- 36

3 tectable in the Aroclor mixture under our chromatographic conditions and congener 154 coeluted with congeners 77 and 110). Congener 81 was also included for its toxicity, although its contribution to the composition of the Aroclor mixture was minimal. The numbering of PCB congeners in this report follows the IUPAC system that is different from the system of Ballschmiter and Zell (1980; BZ numbering) in assigning congeners 199, 200, and 201. Materials Individual PCB congeners were purchased from Ultra Scientific, Inc. (North Kingstown, RI) in solution (200 mg/ml in hexane) and from AccuStandard, Inc. (New Haven, CT) in neat form. Aroclors (1000 μg/ml in hexane) used to prepare the Aroclor mixture were obtained from Supelco, Inc. (Bellefonte, PA). Reference mixtures of PCBs used as quality control standards were obtained from National Institute of Standards and Technology (Gaithersburg, MD) and Supelco, Inc. Pesticide analysis grade hexane was obtained from Baxter International, Inc. (Muskegon, MI). The primary calibration standards were prepared by diluting the individual PCB congeners with hexane to appropriate concentrations suitable for GC/ECD and GC/ MS detection. The Aroclor mixture was made by mixing equal amounts (by weight) of Aroclors 1242, 1248, 1254, and 1260 in hexane and further diluting with hexane for GC/ECD and GC/MS analyses. All the samples were analyzed using two internal standards, 4-bromobiphenyl (for peaks with retention times up to 60 min) and 2,2',4,4',5,5'-hexabromobiphenyl (for peaks with retention times greater than 60 min) purchased from Ultra Scientific, Inc. GC/ECD Analysis A Varian Model 3500 GC equipped with a 63 Ni electron capture detector and a 30m x 0.25mm I.D. (0.25μm thickness) DB-5 silica fused capillary column (J&W Scientific, Figure 1. A typical mass spectrum of PCB congeners. RELATIVE ABUNDANCE Folsom, CA) was used as the primary calibration instrument. High purity helium was the carrier gas at a flow rate of 1.6 ml/min at 80 C. The ECD detector make-up gas was ultra-high purity nitrogen at a flow rate of 30 ml/min. Both He and N 2 gases were purified with moisture, hydrocarbon, and oxygen filters before use. The column temperature was initially held at 80 C for 2 min and ramped to 140 C at 5 C/min, followed by a 1 C/min increase to 240 C; the temperature was immediately raised to 280 C with a rate of 5 C/min. The total run time of 122 min obtained 105 baseline-resolved chromatographic peaks from our Aroclor mixture. One μl sample volume was MASS NUMBER (AMU) 37

4 manually injected into a split/ splitless injector with a 1 min solvent split time. The injector temperature was maintained at 280 C and detector temperature was maintained at 300 C. Data were acquired and processed by a Perkin Elmer (PE) Nelson 2700 (Turbochrom 3) software running on a IBM compatible PC interfaced to the GC system by a PE Nelson 900 Series Interface unit. GC/MS Analysis Confirmation of PCB peak assignments was provided by a Hewlett-Packard Model 5890 II GC with a 5970 MSD under chromatographic conditions identical to the GC/ECD. Mass spectral data were acquired in the electron impact ionization mode with electron energy 70 ev. Data acquisition and processing were monitored using Hewlett-Packard ChemStation software running on an IBM compatible PC. The mass scanning range of m/e 40 to m/e 501 at a rate of 1.2 scans/s had sufficient resolution for a minimum peak width of about 10 s. The voltage supplied to the electron multiplier was 2000 volts and the ion source was maintained at 250 C. The instrument was calibrated daily with decafluorotriphenylphosphine (DFTPP) and corrections were made immediately if problems occurred. Mass spectra of PCBs were characterized by strong ion fragments of M+ and [M-2Cl]+ and much 38 Table 1. Relative retention times of PCB congeners on GC/ECD and GC/MS. GC/ECD GC/MS PCB Ave. RRT b RSD c % Ave. RRT b RSD c % Diff d % Congener a (n=14) (n=6) I.S.=4-bromobiphenyl (RT=26.24 min) a PCB congeners were numbered using the IUPAC system. b RRTs were calculated relative to 4-bromobiphenyl and hexabromobiphenyl. c RSD is the relative standard deviation in percent. d Diff is 100% x [ave.rrt(gc/ecd)-ave.rrt(gc/ms)]/ave.rrt(gc/ecd).

5 (Table 1. continued) GC/ECD GC/MS PCB Ave. RRT b RSD c % Ave. RRT b RSD c % Diff d % Congener a (n=14) (n=6) I.S.=hexabromobiphenyl (RT= min) weaker [M-Cl]+ signals (Figure 1). Results/Discussion Relative Retention Times Assignments of the PCB peaks were based on data in the literature and mass spectral analyses. One hundred twenty PCB congeners were prepared in small groups containing only the resolved peaks observed by Mullin et al. (1984) and Schulz et al. (1989). The groups were injected into GC/ECD and GC/ MS and the PCB eluting sequence was compared to those obtained by Mullin et al. (1984) and Schulz et al. (1989). Congeners with similar retention times usually had different chlorine numbers in our grouping system and could be unambiguously identified by GC/MS. Relative retention time (RRT) is the retention time of species i (RT i ) divided by the retention time of an internal standard (RT I.S. ). Average relative retention times were expressed relative to 4- bromobiphenyl for peaks with retention times up to 60 min and to hexabromobiphenyl for retention times greater than 60 min (Table 1). Congener 2 was not included because background residues interfered with detection by GC/ECD. The eluting sequence of PCBs on GC/ECD agreed well with that on GC/MS except for pairs 15 and 17, 126 and 178, and 208 and 195, which 39

6 (Table 1. continued) GC/ECD GC/MS PCB Ave. RRT b RSD c % Ave. RRT b RSD c % Diff d % have similar retention times. Although identical chromatographic conditions were adopted in GC/ECD and GC/ MS analyses, small fluctuations in carrier gas flow rate could be expected. Since closely eluting peaks were separated into different groups for analysis, small fluctuations in flow rate might cause variations in retention times from run to run. The altered relative retention times could lead to different eluting sequences for the same peaks on GC/ECD and GC/MS. The differences between RRTs on GC/ECD and GC/MS were negative for congeners using hexabromobiphenyl as the internal standard and mostly positive for those using 4-bromobiphenyl as the internal standard. This was probably due to fluctuations in the instrumental parameters used to operate the GC/ECD and GC/MS. Additional confirmation was obtained by using the method of Eganhouse et al. (1989) to compare the relative retention times in this study to those reported by Mullin et al. (1984): RRT 2 = RRT 1 + K(RRT 3 - RRT 1 ) where RRT 2 is the predicted relative retention time (PRRT) of peak 2 in this study, and RRT 1 and RRT 3 are the relative retention times of peak 1 (lower retention time) and peak 3 (higher retention time) observed by Mullin et al. (1984), and K = (RT 2 - RT 1 )/(RT 3 - RT 1 ) with RT i = retention time of peak i (i=1,2,3, with the larger number denoting higher retention time) determined in this study. The predicted relative retention times of all congener peaks, except the first and last, obtained on GC/ ECD and GC/MS were in excellent agreement with PRRTs in Mullin et al. (1984), further confirming the assignments of our PCB congener chromatograms (Table 2). Relative Response Factors Relative response factors were acquired using both GC/ ECD and GC/MS, although only the GC/ECD was used to quantify the Aroclor mixture. The GC/MS response factors were used to evaluate previous studies using GC/MS (see below). To obtain accurate 40

7 Table 2. Comparison of predicted relative retention times (PRRT). a PCB Mullin et al. b SCCWRP Congener GC/ECD GC/ECD Diff1 c % GC/MS Diff2 d % calibration data, 119 PCB congeners were divided into four groups for GC/ECD and 12 groups for GC/MS analyses based on their relative retention times. The analyses were performed on GC/ECD with five concentrations ranging from 25 to 500 ng/ml for individual PCBs and on GC/MS with three concentrations ranging from 1 to 8 μg/ml for individual PCBs. Three replicates were analyzed by GC/ ECD on different days to get average response factors; two replicates were analyzed by GC/MS. The relative response factors were calculated by fitting a linear least-square regression through the origin to the calibration data using 4- bromobiphenyl (<60 min retention time) and hexabromobiphenyl (>60 min) as the internal standards (Table 3). The relative response factors from the GC/ECD analysis scattered in a broad range, even for the same isomer group (Table 3). This was consistent with results obtained by Mullin et al. (1984) and Cooper et al. (1985). Although different internal standards were used, the relative magnitudes of RRFs agreed well with our results. The average relative response factors obtained from GC/MS analyses decreased moderately with increasing chlorine number (Tables 3 and 4). Part of this decrease can be accounted for by the numa See text for procedure for calculating PRRTs. b See Mullin et al. (1984). c Diff1 is 100% x [PRRT(GC/ECD)-PRRT (Mullin et al.)]/prrt (Mullin et al.). d Diff2 is 100% x [PRRT(GC/MS)-PRRT (Mullin et al.)]/prrt (Mullin et al.). 41

8 (Table 2. continued) PCB Mullin et al. b SCCWRP Congener GC/ECD GC/ECD Diff1 c % GC/MS Diff2 d % ber of molecules of compounds with different molecular weights in the same concentration. The number of monochlorobiphenyl molecules (congener 1, molecular weight 188) is 2.6 times the number of decachlorobiphenyl molecules (congener 209, molecular weight 494) in the same concentration. Additional losses in relative response factors can be attributed to a combination of the discrimination associated with splitless injection against, and decreasing sensitivity of the MS spectrometer toward, high molecular weight compounds. Interestingly, splitless and oncolumn injection techniques, when adjusted to the most efficient conditions, could achieve a deca-to-monochlorobiphenyl peak ratio of and (Gebhart et al. 1985), which equals the inverse ratio of their molecular weights (~0.38). Since conditions employed in this study may not have been the most efficient, some discrimination against PCB congeners with high molecular weights was expected. In a study using on-column injection, a quadrupole mass spectrometer showed higher discrimination against heavier PCB congeners than did magnetic sector mass spectrometer (Erickson et al. 1988). Previous studies with GC/ MS assumed that the response factors in an isomer group were relatively invariant so that one representative 42

9 (Table 2. continued) reference standards (Tables 5 and 6). Agreement between PCB Mullin et al. b SCCWRP the known and measured Congener GC/ECD GC/ECD Diff1 c % GC/MS Diff2 d % values was generally good, which validates the application of the RRFs in measurements of the Aroclor mixture congener could be used to calibrate the group (Gebhart et al. 1985, Slivon et al. 1985). We calculated the average relative response factors and relative standard deviations of each isomer group from the GC/MS data (Table 4). Relative standard deviations ranged from 16.4% to 31.5%, except for the isomer group with four chlorines (4- bromobiphenyl I.S.) in which the anomalously high RRF of congener 69 was probably due to an artifact. (Congener 69 had a significantly higher RRF than other congeners in the same isomer group, which may have been due to errors in preparation of standard solutions.) The discrepancy between the average RRFs and individual RRFs could be as much as 243% (congener 16), excluding congener 69. The calibration procedure using one congener to represent an isomer group is questionable. To verify the measured RRFs, we analyzed two NIST Analyses of Aroclor Mixture The concentrations of individual components in the mixture of Aroclors varied over a broad range (Figure 2). The best detection ranges on GC/ECD were obtained with 10 and 20 μg/ml of the Aroclor mixture with 1 μl injection volume. Even at these concentrations, quite a few components were out of the calibration range (25 to 500 μg/ml). In those cases, the calibration curves were assumed to be linear beyond the calibration range. The assignment of individual PCB peaks in the Aroclor mixture was done by comparing RRTs of peaks in the mixture to those of the individual PCB standards. The matching retention time windows were within 2.4% for peaks relative to 4-bromobiphenyl and 0.9% for peaks relative to hexabromobiphenyl. A total of 106 baseline-resolved PCB peaks was obtained (Table 7). We compared our Aroclor mixture with the composition of a similar mixture (different Aroclor lots) analyzed by Eganhouse et al. (1989), who used 1 to 6 congeners to represent each isomer group 43

10 Table 3. Relative response factors of PCB congeners on GC/ECD and GC/MS. GC/ECD GC/MS PCB No. Cl Ave RRF a RSD % Ave RRF a RSD % Congener (n=3) (n=2) I.S.=4-bromobiphenyl (Table 7). The differences in individual PCB contents between the two Aroclor mixtures were randomly scattered, and generally became significant when the concentrations were low. This might reflect the lot-to-lot variation in commercial Aroclors and the difference in the calibration procedures used in the two studies. Comparison of the sum of the measured individual PCB concentrations in the Aroclor mixture and the actual amount injected is an indicator of the quality of a PCB congenerspecific analytical method (Schulz et al. 1989). In our study, four replicates each of 10 and 20 μg/ml of the Aroclor mixture were analyzed by GC/ECD. The ratios of measured/actual concentrations ranged from to 1.096, with an average value of (standard deviation = 0.064). The composition of the Aroclor mixture in Table 7 was normalized to the total measured value rather than the actual amount injected, since these two values were identical within the error limits. For potential multi-component peaks, the assignment was tentative pending more conclusive experiments, such as spiking or co-injection of the Aroclor mixture with individual PCB standards. Multi-component peaks were quantified using the PCB congener with the greatest relative response factor detera RRFs were determined from 5-point calibration curves for GC/ECD and 3-point calibration curves for GC/MS. 44

11 (Table 3. continued) GC/ECD GC/MS PCB No. Cl Ave RRF a RSD % Ave RRF a RSD % Congener (n=3) (n=2) I.S.=hexabromobyphenyl mined from the measurement of individual PCB congeners. Instrument Detection Limit Determination of a method detection limit (MDL) involves analyses of at least seven replicates of the analyte(s) in a matrix of interest at a concentration close to the estimated MDL (Glaser et al. 1981). Since the individual PCB concentrations in an Aroclor or Aroclor mixture always scatter in a broad range, it is impossible to prepare a solution containing equal amounts of PCB congeners from Aroclors. Instead, it would be more realistic to estimate the instrument detection limit (IDL) that is defined as the concentration that produces a signal equal to five times the signal-to-noise ratio (American Public Health Association 1989). The IDLs were estimated for the GC/ECD by evaluating the signal-to-noise (S/N) ratios only for small peaks in a 4 μg/ ml solution. Estimated IDLs ranged from 0.05 to 1 ppb, except PCB 4 (2.5 ppb) with 1 μl injection volume. The actual quantitation limits should be higher depending on individual congeners. Conclusions A mixture of Aroclors 1242, 1248, 1254, and 1260 characterized by GC/ECD and GC/MS capillary column has paved the way for its use as a standard mixture in congener- 45

12 (Table 3. continued) GC/ECD GC/MS PCB No. Cl Ave RRF a RSD % Ave RRF a RSD % Congener (n=3) (n=2) specific analysis of PCBs in environmental samples. The relative retention times and response factors of the PCB congeners were comparable to those obtained by other researchers. However, the composition of the Aroclor mixture differed significantly from a similar mixture characterized by Eganhouse et al. (1989), which might be due to lot-to-lot variation of commercial Aroclors and different calibration procedures. The instrument detection limits of the GC/ECD ranged from 0.05 to 1 ppb depending on individual PCB congeners. References Albro, P.W., J.T. Corbett, and J.L. Schroeder Quantitative characterization of polychlorinated biphenyl mixtures (Aroclors 1248, 1254 and 1260) by gas chromatography using capillary columns. J. Chromatogr. 205: American Public Health Association Method Detection Limit. Pp to 1-20, In: Standard method for the examination of water and wastewater. 17th ed. L.S. Clesceri, A.E. Greenberg, and R.R. Trussell (eds.). Port City Press, Baltimore, MD. Ballschmiter, K. and M. Zell Analysis of polychlorinated biphenyls (PCB) by glass capillary gas chromatography. Composition of technical Aroclor- and Clophen-PCB mixtures. Fresenius Z. Anal. Chem. 302: Cooper, S.D., M.A. Moseley, and E.D. Pellizzari Surrogate standards for the determination of individual polychlorinated biphenyls using high-resolution gas chromatography with electron capture detection. Anal. Chem. 57: Duinker, J.C., D.E. Schulz, and G. Petrick Multidimensional gas chromatography with electron capture detection for the determination of toxic congeners in polychlorinated biphenyl mixtures. Anal. Chem. 60: Eganhouse, R.P. and R.W. Gossett Sources and magnitude of bias associated with determination of polychlorinated biphenyls in environmental samples. Anal. Chem. 63: Eganhouse, R.P., B.R. Gould, D.M. Olaguer, C.S. Phinney, and P.M. Sherblom Congener- 46

13 Table 4. Relative response factors of PCB congeners and isomer groups on GC/MS. PCB Number Group RSD a % Congener of chlorines Average I.S.=4-bromobiphenyl 1, ,7,68,12,13, ,18,17,24,16,34,29,26,25,31,33,22,35, ,51,45,46,69,52,49,48,44,42,41,40,67,63, b 90.1 c 70,66,56 104,96,103,100,95,121,91,92,84,101,113, I.S.=hexabromobiphenyl 81, ,83,97,87,85,110,82,107,118,114,122,132, ,154,151,135,149,134,131,165,105,141, ,138,129,166,128,167,156,157, ,176,178,175,187,183,185,174,177,171, ,180,193,191,170, ,200,197,199,198,201,203,195,194, ,207, a RSD is the relative standard deviation within an isomer group. b if the RRF of congener 69 were excluded. c 28.4% if the RRF of congener 69 were excluded. Table 5. Analyses of NIST SRM 1585 reference standards. GC/ECD GC/MS PCB Certified Measured RSD% Diff a % Certified Measured RSD% Diff a % Congener Value Value n = 4 Value Value n = 3 (ppb) (ppb) (ppm) (ppm) , a Diff is 100% x (measured value - certified value)/certified value. 47

14 Table 6. Analyses of Supelco, Inc. PCB Mix 525 standards. GC/ECD GC/MS PCB Certified Measured RSD % Diff a % Certified Measured RSD % Diff a % Congener Conc. Conc. N = 3 Conc. Conc. n = 4 (ppb) (ppb) (ppm) (ppm) a Diff is 100% x (measured value - certified value)/certified value. Figure 2. A representative chromatogram of the Aroclor mixture by GC/ECD. ABUNDANCE BROMOBIPHENYL 31(28) 52 56(60) RETENTION TIME (MIN) 194 HEXABROMOBIPHENYL specific determination of chlorobiphenyls in biological tissues using an Aroclor-based secondary calibration standard. Intern. J. Environ. Anal. Chem. 35: Erickson, M.D., J.S. Stanley, J.K. Turman, J.E. Going, D.P. Redford, and D.T. Heggem Determination of byproduct polychlorobiphenyl in commercial products and wastes by high-resolution gas chromatographiy/eclectron impact mass spectrometry. Environ. Sci. Technol. 22: Gebhart, J.E., T.L. Hayes, A.L. Alford-Stevens, and W.L. Budde Mass spectrometric determination of polychlorinated biphenyls as isomer groups. Anal. Chem. 57: Glaser, J.A., D.L. Foerst, G.D. McKee, S.A. Quave, and W.L. Budde Trace analyses for wastewaters. Environ. Sci. Technol. 15: Harkness M.R., J.B. McDermott, D.A. Abramowicz, J.J. Salvo, W.P. Flanagan, M.L. Stephens, F.J. Mondello, R.J. May, J.H. Lobos, K.M. Carroll, M.J. Brennan, 48

15 Table 7. Comparative Analyses of Aroclor Mixture (1242, 1248, 1254, in 1:1:1:1 wt%). SCCWRP Eganhouse et al. b Peak No. a PCB GC/ECD RSD % GC/FID GC/MS Diff % Diff % Congener wt %, n=8 wt %, n=8 wt % ECD-FID ECD-GC/MS nd nd (10) (9) (5) c c d d (27) (32) (28) (20) (47,75) e e (59) f f (64) a Peak numbers were adopted from Schulz et al. (1989). b See (Eganhouse et al. 1989). c-k Coeluting components. 49

16 (Table 7. continued) SCCWRP Eganhouse et al. b Peak No. a PCB GC/ECD RSD % GC/FID GC/MS Diff % Diff % Congener wt %, n=8 wt %, n=8 wt % ECD-FID ECD-GC/MS (60) (90) g g (115) (123) (146) (153) (158,160) h h

17 (Table 7. continued) SCCWRP Eganhouse et al. b Peak No. a PCB GC/ECD RSD% GC/FID GC/MS Diff % Diff % Congener wt %, n=8 wt %, n=8 wt % ECD-FID ECD-GC/MS i i i j j (190) (196) k 195 k A.A. Bracco, K.M. Fish, G.L. Warner, P.R. Wilson, D.K. Dietrich, D.T. Lin, C.B. Morgan, and W.L. Gately In situ stimulation of aerobic PCB biodegradation in Hudson River sediments. Science 259: Mullin, M.D., C.M. Pochini, S. McCrindle, M. Romkes, S.H. Safe, and L.M. Safe High resolution PCB analysis: Synthesis and chromatographic properties of all 209 PCB congeners. Environ. Sci. Technol. 18: Pellizzari, E.D., M.A. Moseley, and S.D. Cooper Recent ad vances in the analysis of polychlorinated biphenyls in environmental and biological media. J. Chromatogr. 334: Safe, S Polychlorinated biphenyls (PCBs) and polybrominated biphenyls (PBBs): biochemistry, toxicology, and mechanism of action. CRC Critical Rev. Toxicol. 13: Schulz, D.E., G. Petrick, and J.C. Duinker Complete characterization of polychlorinated biphenyl congeners in commercial Aroclor and Clophen mixtures by multidimensional gas chromatography-electron capture detection. Environ. Sci. Technol. 23: Schwartz, T.R., D.L. Stalling, and C.L. Rice Are polychlorinated biphenyl residues adequately described by Aroclor 51

18 mixture equivalents? Isomerspecific principal components analysis of such residues in fish and turtles. Environ. Sci. Technol. 21: Slivon, L.E., J.E. Gebhart, T.L., T.L. Hayes, A.L. Alford-Stevens, and W.L. Budde Auto mated procedures for mass spectrometric determination of polychlorinated biphenyls as isomer groups. Anal. Chem. 57: Zell, M. and K. Ballschmiter Baseline studies of the global pollution. III. Trace analysis of polychlorinated biphenyls (PCB) by ECD glass capillary gas chromatography in environmental samples of different trophic levels. Fresenius Z. Anal. Chem. 304: Acknowledgements Authors Eddy Zeng, Marilyn Castillo (present address: Hyperion Treatment Plant, Playa Del Rey, CA), and Azra Khan thank Dr. Robert Eganhouse (U.S. Geological Survey) for sharing his experience in PCB congener-specific analysis and providing many valuable suggestions. The authors also are indebted to the SCCWRP staff for their understanding and encouragement during this study. 52