ANALYST, JANUARY 1987, VOL. 112 37 Analytical Reference Materials Part VII.* Development and Certification of a Sediment Reference Material for Total Polychlorinated Biphenylst Hing-Biu Lee and Alfred S. Y. Chau Quality Assurance and Methods Section, Analytical Methods Division, National Water Research Institute, Environment Canada, Burlington, Ontario, 17R 4A6, Canada A lake sediment reference material naturally contaminated with PCBs was prepared. Subsamples of this material were subjected t o repetitive in-house analyses for total PCBs. The quantitative recovery of PCBs was demonstrated by performing ultrasonic and Soxhlet extractions under various conditions. Sample extracts were usually cleaned-up with Florisil and the cleaned extracts were shown to be free from other major interferences, except for sulphur, which was later removed by mercury. The presence of PCBs in the samples was confirmed by the perchlorination of the sample extract and by GC - MS techniques.Most of the sample extracts were quantitated against a 1 + 1 + I mixture of Aroclors 1242, 1254 and 1260 on a 3% OV-I packed column using the technique described by Webb and McCall. These results were further confirmed by the quantitative GC - M S analyses of EC-1 and by interlaboratory results provided by independent laboratories. The reference material showed no signs of degradation of its PCB content over a six-year storage period a t -20 "C in the dark. Keywords: Polychlorinated biphenyl determination; certified reference materials; quality assurance; lake sediment samples Aroclors or mixtures of polychlorinated biphenyls (PCBs) with various percentages of chlorine by weight were manufac- tured in the United States by Monsanto Chemical Company.Because of their general inertness, PCBs were widely used as transformer dielectric fluids, plasticisers and flame retardants, etc., in industry. Although the production of Aroclors has been curtailed in recent years owing to government regula- tions, the disposal, dump leaching and accidental spillage of the millions of pounds of these persistent Aroclors manufac- tured in the preceding years still cause environmental pollu- tion problems in air, water and biota samples. Aroclors are classified as priority pollutants by the US Environmental Protection Agency and environmental moni- toring of the compounds was begun in the 1960s because of their toxicity and persistence. Although PCB determinations are considered routine, the results from many naturally contaminated samples are often unsatisfactory, as indicated by the many interlaboratory round-robin studies organised by our section192 and by other parties.3.4 Typically, the interlabor- atory relative standard deviation of PCB results in naturally contaminated sediment samples at low pg g-1 levels is between 25 and 50%.Hence, there is a need to develop a real-life reference material in order to monitor the perfor- mance of laboratories involved in PCB determinations. Several years ago, our section initiated the research and development of a number of sediment certified reference materials (CRMs) for toxic organics such as PCBs,5 chloro- benzenes6 and polynuclear aromatic hydrocarbons,7 in order to fulfil the requirements of our on-going quality assurance programmes. Two marine sediment CRMs were recently available commercially, however, no details were given on how these CRMs were characterised.In this paper, we report our rigorous approach to analyse and certify the total PCB contents in our sediment CRM coded EC-1. * For Part VI of this series, see p. 31. t This material is currently not for sale and not available for general distribution. Experimental Preparation of Sediment Reference Material Detailed procedures for sample collection, preliminary dry- ing, freeze-drying, blending, bottling and homogenity testing before and after subsampling have been described in previous publications .5,6 Extraction and Clean-up of Sediment Samples The Soxhlet and ultrasonic extractions of PCBs in sediment samples and the partitioning and evaporation steps were identical to those employed for the chlorobenzene determina- tions.6 The clean-up of sediment extracts was carried out in a 500 x 19 mm i.d.glass column filled with 20.0 g of activated Florisil and 10 mm of anhydrous sodium sulphate at the top. The PCBs in the extract were eluted by 200 ml of hexane. After the addition of 3 ml of iso-octane and a few boiling chips, this hexane fraction was evaporated down to ca. 3-5 ml using a three-stage macro Snyder column and a heating mantle. The concentrated extract was diluted to 10.0 ml with iso-octane. Sulphur and sulphur compounds in the cleaned-up extract were removed by vigorous agitation with mercury until the metal remained shiny.Gas Chromatography with Electron-capture Detection (GC - ECD) A Hewlett Packard 5700 series gas chromatograph equipped with a Ni-63 electron-capture detector, a Model 7671A autosampler and a Model 3390 reporting integrator was used. The column was a 1.8 m x 2 mm i.d. glass column packed with 3% OV-1 on 100-120 mesh Gas Chrom Q. The temperatures were: injection port, 250 "C; detector, 300 "C; and column, 185 "C. The flow-rate was 25 mi min-1 and the carrier gas argon - methane (95 + 5). Aroclors 1242,1254 and 1260 were supplied by the US Environmental Protection Agency. The working standard was a mixture of 1 + 1 + 1 Aroclors 1242, 1254 and 1260 in iso-octane with a total concentration of 600 pg p1-1.The quantitation of PCBs was carried out by the peak matching technique described by Webb and McCall.8ANALYST, JANUARY 1987, VOL. 112 38 Table 1. Ions, congeners and concentrations of calibration standards used in the capillary column GC - MSD analysis of EC-1 mtt Calibration congener Concentration/ Homologue Quantitate ion Confirm ion Congener no. * pg 4-1 c11 . . . . . . . . Cl;! . . . . . . . . c13 . . . . . . . . Cl'$ . . . . . . . . c15 . . . . . . . . C16 . . . . . . . . c1, . . . . . . . . C18 . . . . . . . . c19 . . . . . . . . CllO . . . . . . . . 188 222 256 292 326 360 394 430 464 498 190 224 258 290 328 362 396 432 466 500 1 5 29 50 87 154 181 200 209 209 50 50 50 100 100 100 150 150 250 250 250 - - Chrysene-d,, .. . . 240 * Adopted from reference 9. Table 2. Summary of results (pg g-1 total PCBs) from EC-1 sediment reference material by packed column Webb - McCall quantitation method Extraction Ultrasonic Soxhlet No. ofanalyses . . . . 97 72 R a n g e k g-1 . . . . 1.85-2.15 1.88-2.17 Mean k SD . . . . . . 2.02 k 0.07 1.97 k 0.08 Gas Chromatography with Mass-selective Detection (GC - MSD) A Hewlett Packard 5880A gas chromatograph equipped with a split - splitless injection port, a Level I1 terminal, a Model 7671A autosampler, a Model 5970B mass-selective detector (MSD), a Model 9816s computer and a Model 9133XV 15 megabytesdiscdrivewereused, togetherwitha30m x 0.25mm i.d. DB-5 fused-silica capillary column, which was directly interfaced to the electron impact ion source for maximum sensitivity. The GC conditions were: injection port, 275 "C; interface, 280 "C; column initial temperature, 70 "C (held for 1.5 min); programming rate, 30 "C min-1 (70- 170 "C), 2.5 "C min-1 (170-260 "C); and oven temperature held at 260 "C for 15 min.The splitless valve was on for 1.5 min and the column head pressure was 4 lb in-2. Fully automated sample injection, data acquisition, data editing and report generation was made possible with the existing "Sequencing" software and a keystroke program on the GC terminal. The detector was operated in the selected ion monitoring (SIM) mode for both confirmation and quantitation. In both instances, two characteristic ions (one for quantitation and the other for confirmation) were monitored for each PCB homologous series (Table 1).For quantitative work, the procedure described by Budde and co-workers10J1 was used, except that the 2,2' ,3,4,4' ,5,6-heptachlorobiphenyl (congener 181) instead of 2,2',3,4',5,6,6'-heptachlorobiphenyl (congener 188) was used owing to availability. Chrysene-d12 was used as an internal standard and the dwell time for each ion was 100 ms. Sediment extracts were quantitated against an iso-octane mixture of nine congeners of various concentrations (Table 1). At each level of chlorination, one PCB congener in the calibration mixture was used as the concentration standard for all isomers in that group, e.g., congener 29 was used to quantitate all trichlorobiphenyls in the sample. The only exception was that decachlorobiphenyl was used as the concentration calibration standard for both nona- and deca- chlorobiphenyls in the sample.The total PCB concentration was obtained by the summation of all concentrations in each homologous series. All PCB congeners were obtained from Ultra Scientific, Hope, RI, USA. Chrysene-d12 was obtained from Aldrich Chemical, Milwaukee, WI, USA. Results and Discussion Efforts were made to ensure that the PCBs in EC-1 were quantitatively recovered. The ultrasonic and Soxhlet extrac- tion techniques that are routinely used for the extraction of organics in sediments were employed and compared. In order to achieve a valid comparison between the extraction methods, identical clean-up (Florisil column) and quantitation (packed column Webb - McCall) procedures (see later discussions) were used in these samples.A total of 97 PCB determinations were carried out on EC-1 by the ultrasonic extraction technique using a 1 + 1 mixture of acetone and hexane. A total of 72 determinations were also carried out on the same certified reference material by Soxhlet extraction using 59 + 41 acetone - hexane. All of these results are summarised in Table 2. It is obvious from these data that nearly identical results were obtained from both techniques. As the presence of moisture has been reported to provide better recoveries of some organochlorines in soil samples,12 EC-1 samples with 0 or 30% moisture content were Soxhlet extracted simultaneously for comparison. A t-test was applied to the means of these results and no difference was found in these means at the 95% significance level, indicating that a moisture content of 0 or 30% in the sediment samples had no effect on the recovery of PCBs.Another t-test was applied to the sample results obtained by ultrasonic extraction under similar conditions and again no difference in PCB recovery was observed. As similar PCB results were obtained from samples Soxhlet extracted for 3,7, 20 and 72 h, it was implied that PCBs were exhaustively removed from this sediment after 3 h of Soxhlet extraction. Different solvents were also employed to see if the recovery of PCBs was solvent dependent. The results indi- cated that both acetone and methylene chloride gave similar recoveries of PCBs to 59+41 acetone - hexane, whereas non-polar solvents such as hexane and benzene gave slightly lower (ca.90%) recovery. Florisil has been used by many workers13 to remove co-extractives in sediment samples before organochlorine and PCB determinations. Although some workers have preferred to use Florisil deactivated with a few percent. of water for the column clean-up7 activated Florisil was used in this study because it gave a better separation of some organochlorines and polynuclear aromatic hydrocarbons from PCBs. In order to ensure that the Florisil clean-up was effective, some of the samples were subjected to additional clean-up steps. In triplicate runs, the PCB fraction of the Florisil- cleaned EC-1 extract was further cleaned-up on a Celite and 3% deactivated silica-gel column14 and, in another instance, on an activated neutral alumina coIumn,15 according to published methods.In both instances, no change in the PCB profile and amount could be observed in the EC-1 extractsANALYST, JANUARY 1987, VOL. 112 39 0 10 20 30 40 50 Tim eim i n Fig. 1. GC - ECD chromatogram of the PCB fraction in EC-1. A 1.8 m X 2 mm i.d. 3% OV-1 column was used Table 3. Composition of PCBs in EC-1 as determined by quantitative GC - MS analyses (six replicates). Uncertainty is one standard deviation EC-1 Homologue c11 . . . . . . . . Clz . . . . . . . . c13 . . . . . . . . Clq . . . . . . . . c1s . . . . . . . . C16 . . . . . . . . c1, . . . . . . . . Clfj . . . . . . . . c19 . . . . . . . . CllO . . . . . . . . Concentration/ ni3g-l Nd * Nd 184 k 12 (9.9)t 453 k 24 (24.3) 688 k 33 (36.9) 280 f 21 (15.0) 161 k 6.7 (8.6) 98 k 7.6 (5.3) Nd Nd Total concentration .. 1864 * Nd = None detected. t Figures in parentheses show % m/m of each chlorobiphenyl in EC-1. after additional clean-up. The Florisil-cleaned EC-1 extract was also subjected to an ethanolic KOH treatment at 80-90 "C for 30 min. Again, no change in the PCB components could be observed before and after the additional clean-up. These experiments indicated that Florisil-cleaned EC-1 extracts were free from any major interference from other organochlorines. It should be noted that a few chlorobenzenes, p,p'-DDE and Mirex present in EC-1 could not be separated from the PCBs by column chromatography. Their presence, however, would not affect the PCB results as their concentrations were relatively low compared to those of the PCBs.Most of the EC-1 extracts were chromatographed on a 6-ft 3% OV-1 column operated at 185 "C. The PCBs were quantitated by the established peak-matching technique developed by Webb and McCall.8 This method was used as it has been demonstrated to be better than other packed column techniques in a collaborative study.16 It is still the official method for the quantitation of total PCBs in our Water Quality laboratories and is also approved by the US EPA (Method 608).17 Samples were quantitated against a 1 + 1 + 1 mixture of Aroclors 1242, 1254 and 1260; preliminary runs of sample extracts had indicated that the PCB components in EC-1 were very similar to this mixture (Fig. 1). Extracts of EC-1 were subjected to the perchlorination procedure described by Armour18 using antimony penta- chloride.The formation of decachlorobiphenyl in these reactions confirmed the presence of PCBs in the EC-1 extracts. The perchlorination results were not used quantita- Table 4. Summary of interlaboratory results for total PCBs in EC-1 Studynumber . . . . N-27 DQC-3 No. of laboratories . . 15 14 No. of results used* . . 25 12 Range of results/pg g-1 0.96-3.41 1.11-3.26 Mediadpgg-1 . . . . 1.96 1.75 Mean f s.d./vgg-1 . . 2.05 k 0.61 1.98 2 0.69 * After rejection of outliers. tively to determine the PCB concentration in EC-1 as the latter was a complicated mixture of several Aroclors. A concentrated extract of EC-1 (containing approximately 3.0 pg ml-1 of total PCBs) was analysed on a 30-m DB-5 column interfaced to a mass-selective detector.Data were acquired in the selected ion monitoring mode for the detection of the ten chlorobiphenyl homologous series, i. e., from mono- to decachlorobiphenyl. Two characteristic ions were used for each homologous series: one for quantitation and the other for confirmation, as shown in Table 1. Although the mono-, di-, nona- and decachlorobiphenyls were not present in EC-1 in detectable amounts, the presence of tri-, tetra-, penta-, hexa-, hepta- and octachlorobiphenyls in this reference material was confirmed by the presence of both characteristic ions at the right retention times and in the expected ratios for each of the six homologous series listed above. PCBs in EC-1 extracts were also quantitatively determined by GC - MSD using the method described by Budde and co-workers, 10911 As PCB homologues have overlapping reten- tion time windows, special precautions were taken to avoid interferences by PCB congeners containing more chlorines. Under the electron ionisation mode, a PCB molecule under- goes fragm5ntation by the loss of two chlorines, and to a lesser extent by the loss of HC1 and Cl,19,2O thus causing interference in the determination of PCBs with one or two less chlorine atoms.In this work, the level of chlorination in each PCB peak was previously determined by a full scan run of a concentrated Aroclor mixture. The level of chlorination in a sample PCB peak was first assigned by the observed relative abundance of the two corresponding characteristic ions. This, together with the information obtained in the full scan run, was generally sufficient to eliminate interference generated by fragmenta- tion ions produced by co-eluting PCBs with more chlorines.The results of the quantitative GC - MSD determination of PCBs in EC-1 are shown in Table 3. In this instance, the concentration of each PCB homologous series and the total PCB concentration were obtained. As indicated, the GC - MS results further confirmed the GC - ECD results as the total PCB concentrations obtained by these two different quantita- tive methods (Tables 2 and 3) varied by less than 10%. The over-all lower sensitivity of the MSD to most PCBs, especially the hepta- and higher chlorobiphenyls, rendered some PCB peaks undetected by this detector at low concentrations. This could be the reason why, in the example of EC-1, that the total PCB results obtained by mass-selective detection were slightly lower than those obtained by electron-capture detection.Reference material EC-1 was used in two interlaboratory round-robin studies in two different years. In both instances, the participants were requested to analyse the material for total PCBs by using their own in-house methods and stan- dards. The interlaboratory results (Table 4) were diversified because of the different extraction and clean-up method- ologies and the calibration standards and quantitation tech- niques employed by various participants. 1~ Despite all these variations, the interlaboratory medians and means of the PCB results in both studies were in excellent agreement with the in-house results as summarised in Tables 2 and 4.The stability of PCBs in EC-1 under cold storage conditions (-20 "C in the dark) was monitored twice annually. The40 ANALYST, JANUARY 1987, VOL. 112 results give no evidence of degradation during storage over the last six years.21 In conclusion, we have successfully prepared and certified a naturally contaminated lake sediment reference material (EC-1) for total PCB contents on the basis of 169 in-house determinations. The reference value, 2.00 k 0.05 pg g-1 (uncertainty is one standard deviation), generated by com- bining all the GC - ECD results (Table 2), was further supported by GC - MS results and interlaboratory results in two round-robin studies. This certified reference material is currently being used in many of our intralaboratory and interlaboratory quality assurance programs for PCB deter- minations.1. 2. 3. 4. 5. 6. References Lee, H. B., and Chau, A. S. Y., “National Interlaboratory Quality Control Study No. 27-PCBs in Naturally Contami- nated Dry Sediments,” Inland Waters Directorate Report Series No. 72, Environment Canada, 1981. Lee, H. B., Dookhran, G., and Chau, A. S . Y., “Dredging Quality Control Study No. 3 (DQC3)-Analysis of PCBs in Naturally Contaminated Dry Sediments and Standard Solu- tions,” National Water Research Institute Contribution, Burl- ington, Ontario, 1985 Alford-Stevens, A. L., Budde, W. L., and Bellar, T. A., Anal. Chem., 1985, 57, 2452. Musial, C. J., and Uthe, J. F., J . Assoc. Off. Anal. Chem., 1983, 66, 22. Chau, A. S. Y., and Lee, H. B., J. Assoc. Off. Anal. Chem., 1980,63,947. Lee, H. B., Hong-You, R. L., and Chau, A. S. Y., Analyst, 1986, 111, 81. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. Lee, H. B., Dookhran, G., and Chau, A. S. Y., Analyst, 1987, 112, 31. Webb, R. G., and McCall, A. C., J . Chromatogr. Sci., 1973, 11, 366. Ballschmiter, K . , and Zell, M., Fresenius 2. Anal. Chem., 1980,302, 20. Gebhart, J. E., Hayes, T. L., Alford-Stevens, A. L., and Budde, W. L., Anal. Chem., 1985,57, 2456. Silvon, L. E., Gebhart, J. E., Hayes, T. L., Alford-Stevens, A. L., and Budde, W. L., Anal. Chem., 1985,57,2464. Williams, I. H., J. Assoc. Off. Anal. Chem., 1968, 51, 715. Lee, H. B., Chau, A. S. Y., and Kawahara, F. K., in Chau, A. S. Y., and Afghan, B. K., Editors “Analysis of Pesticides in Water,” Volume 11, CRC Press, Boca Raton, FL, 1982, pp. 1-40. Armour, J. A., and Burke, J. A., J. Assoc. Off. Anal. Chem., 1970,53,761. Telling, G. M., Sissions, D. J., and Brinkman, H. W., J. Chromatogr., 1977, 137, 405. Sawyer, L. D., J . Assoc. Off. Anal. Chem., 1978, 61, 282. U.S. Environmental Protection Agency, Fed. Regist., 1984,49, No. 209, 89. Armour, J. A., J. Assoc. Off. Anal. Chem., 1973, 56, 987. Safe, S., and Hutzinger, O., J. Chem. SOC. Perkin Trans. I . 1972, 686. Tindall, G. W., and Wininger, P. E., J. Chromatogr., 1980, 196, 109. Lee, H. B., and Chau, A. S. Y.. unpublished results. No~~-Reference~7 is to Part VI of this sehes. Paper A61220 Received July 9th, 1986 Accepted August 26th, 1986