Separation of Toxic Coplanar PCB Congeners from Other Congeners in Marine Sediments and Marine Fish Tissues Using Florisil and GC ECD

Transcription

1 992 LCGC NORTH AMERICA VOLUME 21 NUMBER 10 OCTOBER Separation of Toxic Coplanar PCB Congeners from Other Congeners in Marine Sediments and Marine Fish Tissues Using Florisil and GC ECD This article describes a simple, low-cost sample extraction and cleanup procedure in which the more toxic coplanar polychlorinated biphenyl (PCB) congeners Ballschmitter and Zell (BZ) numbers 77, 81, 126, and 169 are separated from other congeners in highly polluted marine and estuarine sediment samples and fish tissue and the resulting extracts are analyzed by gas chromatography with electron-capture detection. These four coplanar congeners found in the second Florisil fraction represent approximately 98% of the total toxic equivalency factors, assuming an equal distribution for the 14 known congeners on the World Health Organization list. The analytical method U.S. Environmental Protection Agency Method 8082A normally is used for analyzing chlorinated pesticide, Aroclor, and major PCB congener analyses, not minor constituents, but the method can be adapted to target these four coplanar congeners in the presence of other major congeners. R.H. Rieck U.S. Environmental Protection Agency, Region 10 Laboratory, 7411 Beach Drive East, Port Orchard, Washington Polychlorinated biphenyls (PCBs) typically have been detected and quantified as the Aroclors by matching the multipeak pattern in environmental extracts against the patterns for a series of Aroclor standards (1). The usual method of analysis by gas chromatography (GC) in conjunction with an electron-capture detection (ECD) is used widely and is relatively inexpensive. This technique has limitations in that the resulting PCB pattern can differ from the Aroclor pattern because of loss by weathering processes such as volatilization, degradation, partitioning, and metabolism as well as positive interferences that present identification and quantization difficulties. Because the technique targets the Aroclor formulation and not necessarily the individual congeners, analysts currently have elevated interest in individual congener analysis due to the difference in toxicities especially for risk assessment and toxicity studies (2). Chemists hold particular concern for those PCB congeners that have no ortho-chlorine substitution. These coplanar congeners Ballschmitter and Zell (BZ) numbers 77, 81, 126, and 169 assume a planar configuration that is similar to 2,3,7,8-tetrachloro-p-dibenzodioxin and exhibit dioxinlike toxicities (see Table I) (3). PCB congeners that have only one chlorine in the ortho position such as BZ 170 and 180 also are important and are included in the table because of their predominant contribution to Aroclor The analyses of individual PCB congeners traditionally were performed according to U.S. Environmental Protections Agency (EPA) Method 1668, which uses high-resolution mass spectrometry (MS) (4). However, laboratory workers need an inexpensive, sensitive method that uses the relatively inexpensive instrumentation already present in the laboratory. Other researchers observed that the coplanar PCB congeners can be separated from the noncoplanar congeners by chromatography using Florisil (5 8). My intention was

2 994 LCGC NORTH AMERICA VOLUME 21 NUMBER 10 OCTOBER to use this technique within the guidance of commonly used methods in SW-846 to develop a sensitive, inexpensive alternative for the analysis of these four toxic coplanar PCB congeners BZ 77, 81, 126, and 169, which are minor constituents in Aroclors 1242, 1254, and 1260 in the presence of elevated concentrations of the major congeners in marine sediment. I observed that these four congeners could be separated from the remaining congeners using a macro chromatography column with 20 g of Florisil (9) and that I could achieve low reporting limits by using relatively inexpensive GC ECD instrumentation (EPA Method 8082A). My secondary objective was to decrease the volume of solvent and other reagents used. umn and a 30 m 0.25 mm, m d f Rtx-CLPesticide2 GC column (both from Restek Corp., Bellefonte, Pennsylvania). The columns were connected in parallel to an Agilent model 6890 gas chromatograph (Agilent Technologies, Wilmington, Delaware) with dual micro electron-capture detectors, a high-volume injection programmed temperature vaporizing inlet, and a Gerstel CIS4 cooled injection system (Baltimore, Maryland). I made 30- L splitless injections. The GC temperature program was 80 C for 1.80 min; 50 C/min to 180 C and hold for 0 min; 2 C/min to 210 C and hold for 0 min; 8 C/min to 320 C and hold for 0.45 min. The total run time was 33 min. The carrier gas was helium at a 1.3-mL/min constant flow with a linear velocity of 33 cm/s. Chemicals: Solvents: I obtained hexane, diethyl ether (preserved with 2% ethanol), acetone, and isooctane (distilled in glass grade) from Burdick & Jackson (Muskegon, Michigan). Standards: The EPA PCB congener calibration check solution from Ultra Scientific (North Kingstown, Rhode Island) contained 20 congeners at 0.2 g/ml in isooctane. Nine PCB congener calibration mixtures numbers C-CS-01, C-CS-02, C-CS-03, C- CS-04, C-CS-05, C-CS-06, C-CS-07, C- CS-08, and C-CS-09 purchased from AccuStandard, Inc. (New Haven, Connecticut), provided all 209 congeners. The standard concentrations were 10 g/ml in isooctane. Individual standard solutions at 35- g/ml and 100- g/ml concentrations were Table I: PCB toxic equivalencies for coplanar and mono-ortho and di-ortho coplanar PCBs the WHO list IUPAC* Number Toxic Equivalency Type Congener (BZ Number) Factor Non-ortho 3,3,4,4,5-Pentachlorobiphenyl ,3,4,4,5,5 -Hexachlorobiphenyl ,3,4,4 -Tetrachlorobiphenyl ,4,4,5-Tetrachlorobiphenyl Mono-ortho 2,3,3,4,4 -Pentachlorobiphenyl ,3,4,4,5-Pentachlorobiphenyl ,3,3,4,4,5-Hexachlorobiphenyl ,3,3,4,4,5 -Hexachlorobiphenyl ,3,4,4,5-Pentachlorobiphenyl ,3,4,4,5-Pentachlorobiphenyl ,3,4,4,5,5 -Hexachlorobiphenyl ,3,3,4,4,5,5 -Heptachlorobiphenyl Di-ortho 2,2,3,3,4,4,5-Heptachlorobiphenyl ,2,3,4,4,5,5 -Heptachlorobiphenyl * IUPAC International Union of Pure and Applied Chemistry Experimental Supplies and instruments: I used mesh Florisil (J.T. Baker, Inc., Phillipsburg, New Jersey) that was suitable for use in chromatographic cleanup of pesticide residues. I found it was necessary to heat the Florisil at 430 C for 4 h in a muffle furnace within a flat stainless steel pan to remove interferences and obtain consistent separation performance. I cooled the Florisil, transferred it to large Erlenmeyer flask, capped the flask loosely with aluminum foil, and stored it at least overnight in a drying oven at 130 C before use. I used a 250 mm 11 mm chromatography column (Kimble/Kontes, Vineland, New Jersey) with a 200-mL reservoir capacity and a polytetrafluoroethylene (PTFE) stopcock plug. I also used a 30 m 0.25 mm, m d f Rtx-CLPesticide GC colpurchased from AccuStandard and Ultra Scientific, respectively. The final working standards were in isooctane. Experimental procedures: Preparation of the chromatography column (semi-macro): I placed a small glass wool plug at the bottom of the chromatography column. Approximately 0.25 in. of anhydrous sodium sulfate was placed on top of the glass wool. I added 10 g of prepared Florisil on top of the sodium sulfate and packed the column tightly by vertically tapping it. I put a in. layer of anhydrous sodium sulfate on top. I rinsed the column with 40 ml of hexane, which was discarded. I could not allow the column to go dry, because the separation characteristics would have been compromised and the column would need to be prepared again (9). First experiment: The objective of the first experiment was to determine the elution volumes needed to totally separate BZ 77, 81, 126, and 169 (which I will henceforth call the four coplanar congeners) on the World Health Organization (WHO) list from the rest of the congeners. I added a mixture that contained the 14 congeners on the WHO list to the chromatography column. I used two 1-mL portions of hexane to rinse the inside of the column, added 98 ml of hexane to the column, and collected the effluent in 10-mL increments in individual centrifuge tubes. (These increments were numbered 1 through 10 and represented 10 analytical samples.) Then, I added 100 ml of 6% diethyl ether in hexane to the column. Again, I collected 10-mL increments individually in centrifuge tubes and numbered them 11 through 20. These represented an additional 10 samples for analysis. The individual samples were analyzed using a single-point calibration standard by GC ECD. The results are plotted in Figure 1 as percent recovery versus 10-mL increment and reveal that the four coplanar congeners were eluted totally in the 6% Florisil fraction. This result suggests that the volume of 6% diethyl ether could be reduced to 50 or 60 ml. The remaining 10 congeners were eluted totally in the 0% Florisil fraction. As a note of caution, if the polarity of the Florisil is incorrect, BZ 105 could spill over into the second fraction (10). I repeated this experiment using Supelco LC-Florisil packing cartridges (Bellefonte, Pennsylvania) with a 60-mL reservoir, 10-g capacity, lot number SP02141B, as received. I was unable to duplicate the separation of the coplanar congeners from the

3 996 LCGC NORTH AMERICA VOLUME 21 NUMBER 10 OCTOBER rest of the congeners and did not use them for this project. I repeated this experiment several times with the same outcome. However, the variability data are unavailable. Second experiment: My objective was to determine how many of the 209 PCB congeners would be eluted partially or totally in the second Florisil fraction (6%) using the nine congener mixes. I prepared 10 Florisil chromatography columns and added 25 L at 10 ng/ L of AccuStandard congener mix 1 (C-CS-01) to the first column with 1 ml of hexane. I also added a 500- L aliquot of surrogate spike mix that contained tetrachloro-m-xylene; 4,4 -dibromooctafluorobiphenyl; and 2,2, 4,4,5,5 -hexabromobiphenyl at 100 pg/ L each to the head of the column. The eluent was collected in a 250-mL Kuderna-Danish concentrator assembly fitted with a 10-mL concentrator ampoule. The chromatography column was rinsed twice with approximately 1 ml of hexane, which allowed the solvent to just reach the surface of the sodium sulfate and not go dry between each hexane rinsing. The column was eluted with 100 ml of hexane. This eluent was labeled the 0% Florisil fraction. I placed a second Kuderna-Danish assembly under the column, and the column was eluted with 100 ml of 6% diethyl ether in hexane. The resulting eluent was labeled 6% Florisil fraction. Each fraction was concentrated to 10 ml using first a steam bath and then an N-Evap nitrogen evaporator (Organomation Associates, Inc., Berlin, Massachusetts). I prepared a single-point standard at the same time and used it for percent recovery determinations. The concentration of this standard, compared with a five-point calibration curve, was 25 pg/ L. I repeated this procedure for the remaining eight AccuStandard PCB congener mixes and a method blank. Each sample was analyzed using GC ECD and quantitated against the corresponding standard. The resulting recoveries are listed in an appendix, which appears on the LCGC web site ( com) along with this article. Only those congeners that were eluted partially or totally in the second fraction are listed in Table II. The remainder of the congeners were eluted in the 0% Florisil fraction with recoveries ranging from approximately 90% to 100%. Based upon this study and previous experience, this separation scheme enabled the analysis of PCBs as the Aroclors and the generally considered noncoplanar congeners in the 0% fraction. The second fraction 6% Florisil was analyzed for the four coplanar congeners. With slight adjustment, such as decreasing the initial hexane volume used for the 0% fraction, it was possible to separate additional coplanar congeners such as BZ 35, 37, 78, 79, and 127, which are minor constituents in Aroclors, from the major constituents. During this process, I determined the retention times for all 209 congeners on both analytical columns and calculated the relative retention times, compared with 2,2,4,4,5,5 -hexabromobiphenyl, for both columns. I also calculated the recoveries for 209 congeners found in the 0% and 6% Florisil fractions. These data also are listed in the on-line appendix. One of the major issues raised by other investigators when isolating planar PCB congeners, is the problem of separating congener BZ 110 (2,3,3,4,6-pentachlorobiphenyl), a major Aroclor constituent, from congener BZ 77 (3,3,4,4 -tetrachlorobiphenyl), a minor constituent (11). Both congeners normally are coeluted from most analytical columns. In this analytical scheme, BZ 110 is eluted totally in the first Florisil fraction, and BZ 77 is eluted totally in the second fraction. With the analytical column pair used, the retention times also differ substantially: using the Rtx-CLPesticide column the retention times for BZ 110 Recovery (%) Figure 1: Plots of WHO congener elution volumes. and 77 are and min, respectively; using the Rtx-CLPesticide2 column the retention times for BZ 110 and 77 are and min, respectively. Third experiment: A slight adjustment such as decreasing the hexane volume of the 0% fraction enabled the separation of additional coplanar congeners such as BZ 35, 37, 78, 79, 127, and others, all of which are minor constituents in Aroclors, from the major constituents. I prepared 36 Florisil columns as described above. Four of the columns had the same amount of AccuStandard congener mix number 1 (C-CS-01) as per the second experiment. The surrogates were not added. The first column was eluted first with 100 ml of hexane followed by 100 ml 6% diethyl ether (preserved). Each fraction was collected separately as before. The second column was eluted with 90 ml of hexane and then with 100 ml of 6% diethyl ether. The third column was eluted with 80 ml of hexane and then with 100 ml of 6% diethyl ether. The fourth column was eluted with 70 ml of hexane and then with 100 ml of 6% diethyl ether. The volume of each fraction was concentrated to 10 ml and analyzed as before and compared with a single-point standard. A single-point standard was prepared at the same time and used for percent recovery determinations. The concentration of this standard as compared with a five-point calibration curve was 25 pg/ L. I repeated this procedure for Elution volume (10-mL increments)

4 998 LCGC NORTH AMERICA VOLUME 21 NUMBER 10 OCTOBER each of the eight remaining AccuStandard congener standard mixes. Of the 209 PCB congeners, 32 were eluted either partially or totally in the 6% Florisil fraction (Table II). Essentially, nine non-ortho-substituted (coplanar) PCB congeners were separated in the 6% Florisil fraction when 100 ml of hexane was used. My results show an additional five non-ortho-substituted congeners could be separated in the 6% Florisil fraction using 70 ml of hexane for the first fraction without interference from other congeners (Table III). This adjustment enables analysts to target other coplanar congeners. Congener 38 results do not totally follow the elution trend as the rest of the congeners. The reason for this deviation is unknown. It would seem reasonable that some mix-up could have occurred in the two labels for the 80-mL pair for this sample. The reason for the empty spaces in Table III is that no detectable residues were present in the 0% fraction. No variability data were available. Extraction and analysis procedure for marine sediments: I extracted 21 contaminated marine sediments in duplicate (total of 42 real, distinct samples) and cleaned up and analyzed the samples along with associated method blanks and matrix fortified samples using this procedure. The matrixfortified samples were spiked at three levels 11 replicate spikes at 10 ng/kg (20 L at 10 pg/ L added to 20-g dry-weight basis), 4 replicate spikes at 50 ng/kg (100 L at 10 pg/ L added to 20-g dry-weight basis), and 7 replicate spikes at 250 ng/kg (500 L at 10 pg/ L added to 20-g dry-weight basis) on a trial basis to determine the estimated method detection limit. I determined that the 10-ng/kg concentration was appropriate. The two samples chosen for matrix spikes were prescreened. I determined that both samples had low concentrations of native PCBs and were suitable for this purpose. The matrix spike contained the 20 congeners from the EPA congener mix, eight additional congeners from the WHO list, and 17 additional major congeners present in Aroclors 1242, 1254, and A total of 45 congeners at 10 pg/ L were included in the composite matrix spike. The congeners are listed below: EPA congener mix: BZ 8, 18, 28, 44, 52, 66, 77, 101, 105, 118, 126, 128, 138, 153, 170, 180, 187, 195, 206, and 209 (20 congeners). WHO list: BZ 81, 169, 114, 123, 156, 157, 167, and 189 (8 congeners). (BZ 77, 105, 118, and 126 on the WHO list were included in the EPA congener mix. Table II: PCB congeners from mixes 1 9 eluted in the second Florisil fraction Presence Recovery Recovery Recovery Recovery Recovery Recovery Recovery Recovery Congener in from from from from from from from from BZ Aroclors 1242, Ortho 0% Florisil 6% Florisil 0% Florisil 6% Florisil 0% Florisil 6% Florisil 0% Florisil 6% Florisil Number 1254, and 1260* Substitution (%) (%) (%) (%) (%) (%) (%)# (%)# 1 M NA** NA N NA NA D NA NA M NA NA D NA NA D NA NA M NA NA D NA NA M NA NA M NA NA M NA NA M NA NA D NA NA D NA NA N NA NA N NA NA D NA NA N NA NA N N NA NA N M NA NA M NA NA N NA NA N NA NA N N N NA NA N NA NA * Congener at 1.0 wt%; congener at 1.0 wt%; congener not present at 0.05 wt%. Ortho substitution: N Non-ortho, M Mono-ortho, D Di-ortho or more. Analysis performed 5 March Analysis performed 12 February Analysis performed 11 October # Analysis performed in ** NA Not analyzed. Identification from Restek s congener retention time table.

5 1000 LCGC NORTH AMERICA VOLUME 21 NUMBER 10 OCTOBER Additional congeners: BZ 31, 33, 64, 70, 87, 95, 97, 110, 132, 141, 149, 151, 158, 163, 174, 194, and 196 (17 congeners). Approximately g of wet sediment was placed in an extraction thimble. A separate subsample was used for percent moisture determination, so the results were reported on a dry-weight basis. Acetone solutions of the surrogate spikes and, where appropriate, matrix spikes were added to the sample in the thimble. The samples were extracted overnight with 200 ml of acetone in a Soxhlet extractor connected to a 250-mL Erlenmeyer flask (12). After the extraction was completed, the flask was allowed to cool. A three-ball Snyder column (Kimble/Kontes) was attached to the flask, and the acetone was evaporated on a steam bath to approximately 50 ml. I poured 100 ml of hexane through the top of the Snyder column and cooled the extract. I added a sufficient amount of anhydrous sodium sulfate to the Erlenmeyer flask and mixed it to remove the water. After allowing the hexane fraction sufficient time to cool and dry, I quantitatively transferred the resulting extract with hexane rinses to a 250-mL Kuderna-Danish concentrator assembly connected to a 10-mL ampoule and concentrated to a 10- ml volume of hexane using first a steam bath and then a nitrogen evaporator. Method blanks were performed through the entire process, including the addition of surrogate spikes, extraction, cleanup, and analysis. It is imperative that all extracts be exchanged totally to hexane before initiating the following treatments. The resulting extract was transferred to a disposable 15-mL centrifuge tube that previously had been calibrated to 10 ml. (The meniscus line had been marked with a waterproof pen.) The volume then was adjusted to 10 ml. The marine sediment extracts contained substantial elemental sulfur, an analytical interference for electron-capture detectors. I removed the sulfur by adding elemental mercury to the centrifuge tube and vigorously mixing the sample for 2 min on a Vortex-Genie mixer (Scientific Industries, Bohemia, New York). The extracts were allowed to settle overnight, which allowed the reaction to reach completion. The centrifuge tube containing the extract and mercury was mixed vigorously by hand with 2 ml of concentrated sulfuric acid and allowed to settle or centrifuge as needed (13,14). (Note: The mercury was reclaimed using an in-laboratory procedure for reuse. Since the project was completed, tetrabutylammonium hydrogen sulfate has become the recommended choice for sulfur removal [15].) The sample extract was transferred quantitatively with hexane to another disposable centrifuge tube and concentrated to approximately 1 ml. The Florisil column was prepared as described above and charged with the entire 1-mL extract. The centrifuge tube was rinsed twice with approximately 1 ml of hexane and added to the column. The eluent was collected into a 250-mL Kuderna-Danish concentrator assembly with a 10-mL concentrator ampoule. The chromatography column was rinsed twice with approximately 2 ml of hexane, which allowed the solvent to reach the surface of the sodium sulfate and not go dry between each hexane rinsing. The sample was eluted with the remaining portion of the 100 ml of hexane. This fraction was labeled the 0% Florisil fraction. A second Kuderna- Danish assembly was placed under the chromatography column. The column was washed with 100 ml of 6% diethyl ether in hexane. The resulting eluent was labeled 6% Florisil fraction. The Florisil fractions were concentrated and solvent exchanged to 1.0 ml isooctane using first a steam bath and then a nitrogen evaporator (9). The PCBs as Aroclors and as congeners were determined in the cleaned sediment extracts by GC with micro-electron-capture detectors and Rtx-CLPesticide and Rtx- CLPesticide2 narrow-bore capillary columns connected in parallel for dual, dissimilar column confirmation. I injected 30 L of extract into high-volume injector with a programmed temperature vaporizing inlet and a cooling injection system, which initially was set at 50 C and ramped rapidly to 280 C after the solvent was vented. The sample was split between both analytical columns. The GC program is listed above. The total run time was 33 min, following EPA Method 8082A. Method detection limit determination: I fortified two sediment samples, which had the lowest PCB concentrations present, at 10-ng/kg dry-weight basis. The results were corrected for the native concentrations for the four coplanar congeners. The results are reported as percentage recovered in Table IV. The average percentage recoveries were 97.9, 46.4, 76.1, and 80.1% for Table III: Florisil elution fractions for non-ortho-substituted PCB congeners* Recovery Recovery Recovery Recovery Recovery Recovery Recovery Recovery Congener from 100 ml from 100 ml from 90 ml from 100 ml from 80 ml from 100 ml from 70 ml from 100 ml BZ 0% Florisil 6% Florisil 0% Florisil 6% Florisil 0% Florisil 6% Florisil 0% Florisil 0% Florisil Number (%) (%) (%) (%) (%) (%) (%) (%) * Semi-macro procedure using 10 g of Florisil and 25 L of PCB congeners at 10 ng/ L; the final volume was 10 ml, 30 L was injected.

6 1002 LCGC NORTH AMERICA VOLUME 21 NUMBER 10 OCTOBER BZ 77, 81, 126, and 169, respectively. Taking into account the respective standard deviations and converting to nanograms per kilogram (parts per trillion [ppt]), the respective detection limits were 6.7, 3.2, 3.9, and 4.5 ng/kg (ppt). I observed an inexplicable interference problem for BZ 77 and 81 for replicate number 7. The average recoveries for BZ 77, 126, and 169 are higher than that for BZ 81. Apparently, the sediment matrix polarity affected the elution so that a portion of BZ 81 partially and consistently was eluted in the 0% Florisil fraction. I could not determine the amount of BZ 81 in this fraction because it was a very minor component compared with the other congeners; however, the precision in the 6% Florisil fraction was consistent. Results and Discussion Many of the sediment samples, especially those in close proximity to a known contaminated industrialized site, were highly contaminated with PCBs. Others more remote from this area had lower PCB concentrations. All results are expressed on a dry-weight basis. A seven-point calibration curve was generated for Aroclors 1242 and A six-point calibration curve was generated for Aroclors 1221 and An eight-point calibration curve was generated for the 45 PCB congeners. In each case, I used either a linear or a quadratic fit and had reproducibility (r 2 ) of or better. Each calibration level was recalculated against its respective calibration curve, and its difference from its expected value was less than 20%. The Aroclor 1242 concentrations were g/kg. The Aroclor 1254 concentrations were g/kg. The Aroclor 1260 concentrations were g/kg. BZ 77 concentrations were g/kg and were detected in 20 of the 21 samples. BZ 81 concentrations were detected in 7 of the 21 samples and were g/kg. BZ 126 was detected in only one sample with a concentration of 0.51 g/kg; the duplicate of the sample had a similar concentration. BZ 169 was not detected in any of the samples at the reporting limit. I used 3 6 ng/kg (3 6 ppt) as the reporting limit for the four toxic PCB congeners when I used 20 g of sediment (dry-weight basis). For those samples in which correspondingly larger samples were taken, I used a lower reporting limit. Fish tissue experiment: The object of this experiment was to determine the potential of this technique for analyzing realworld, highly contaminated fish tissue. This study was completed before the sediment studies. I extracted, cleaned up, and analyzed five starry flounder samples in duplicate (10 real samples) using the same procedure as that for the sediment samples. The fortified samples received 250 L of a mixture of 20 congeners at 600 pg/ L. The final concentration in the extract was 15 pg/ L. The final extract volume was 10 ml and only 5 L was injected. I observed good agreement and precision between the Aroclor and coplanar congener results for the duplicate samples and for the recoveries of the spiked samples. The BZ 126 recoveries for the spiked samples, although lower than those for the BZ 77, could be considered acceptable. The results are summarized in Table V. Florisil characterization and standardization: During the course of the experiments, I determined that each new batch of Florisil must be muffled and stored as described above and characterized before use. I characterized the Florisil by processing a fortified blank spiked with the congeners of concern, including the four coplanar congeners. I found that 100 ml of hexane, including the extract volume and rinses, was optimal for separation of the four coplanar PCB congeners. Figure 1 shows an example of this characterization. The volume can change with the Florisil lot or supplier and must be determined for each batch of Florisil. If the coplanar congeners are not eluted totally in the second fraction, then the elements of experiment 3 must be repeated with different hexane volumes to optimize the separation. The laboratory temperature is a variable that can have a pronounced effect on the elution characteristics of the Florisil, and it must be taken into consideration. A change of only 5 C can change the optimum elution volume by ml. The characterization check and all samples should be processed at the temperature of characterization within 5 C. Figure 2 shows the temperature effect on the elution of BZ 81 and 105, which are the ones that seemed to be most affected by temperature. The temperature pattern was determined at three temperatures 4 C, 20 C, and 30 C. At 30 C, BZ 81 is split between the 0% Florisil fraction and the Table IV: Coplanar congener recovery at 10-ppt spike level Replicate Number BZ 77 Recovered (%) BZ 81 Recovered (%) BZ 126 Recovered (%) BZ 169 Recovered (%) * * Average recovery (%) Standard deviation Estimated detection limit as ng/kg (ppt) * Unexplainable interference problem. A portion of replicate 10 was lost due to a laboratory accident.

7 6% Florisil fraction. At 4 C, BZ 105 was eluted partially in the 6% Florisil fraction. Conclusion My laboratory needed a sensitive method using relatively inexpensive instrumentation already present to conduct these analyses. Based upon this study and previous experience, this separation scheme enabled the analysis of PCBs as the Aroclors and the generally considered noncoplanar congeners in the 0% Florisil fraction. The second fraction, 6% Florisil, was analyzed for the coplanar congeners; that is, BZ 77, 81, 126, and 169. With slight adjustment such as decreasing the hexane volume used for elution, it is conceivable that additional coplanar congeners such as BZ 3, 12, 15, 35, 37, 78, 79, and 127, which are minor constituents in Aroclors, also can be separated OCTOBER 2003 LCGC NORTH AMERICA VOLUME 21 NUMBER totally from the major constituents. The estimated reporting limits for the four coplanar PCB congeners in marine sediment were 3 6 ng/kg (ppt) on a dry-weight basis, which is approximately in the same range reported by laboratories using the more expensive EPA Method 1668 by high resolution MS. This procedure appears to be potentially useful for low-level fish tissue analyses. Table V: Aroclor and BZ concentrations from starry flounder samples Site Number Aroclor 1248 ( g/kg) Aroclor 1254 ( g/kg) Aroclor 1260 ( g/kg) BZ 77 ( g/kg) BZ 126 ( g/kg) 1 22* Duplicate 17* * Duplicate 17* * Duplicate 13* * Duplicate 13* * Duplicate 84* Matrix spike 91% 66% 2 Matrix spike duplicate 86% 54% * Estimated. Not detected at stated concentration. Matrix spike duplicate using the EPA congener standard mix from Ultra Scientific that contained 20 congeners.

8 1004 LCGC NORTH AMERICA VOLUME 21 NUMBER 10 OCTOBER Recovery (%) (4 C) 81 (20 C) 81 (30 C) 105 (4 C) 105 (20 C) (30 C) Elution volume (10-mL increments) Figure 2: Plots of elution volume versus elution temperature for congeners 81 and 105. Acknowledgments The author wants to express his appreciation to several colleagues at the EPA Region 10 Laboratory (Port Orchard, Washington). I thank Margaret Knight and Joseph Blazevich for reviewing the article and making suggestion, Randy Cummings for making suggestions, and Steven Reimer for helping with the sediment sample extractions, reviewing the article, and converting the manuscript from a WordPerfect document to a Microsoft Word document. Editors Note Please review the appendix to this article on LCGC s web site at chromatographyonline.com. References (1) SW-846, Method 8082: Polychlorinated Biphenyls (PCBs) by Gas Chromatography (U.S. Environmental Protection Agency, Washington, D.C., 1996). (2) L. Valoppi, M. Petreas, R.M. Donohoe, L. Sullivan, and C.A. Callahan, Use of PCB Congener and Homologue Analysis in Ecological Risk Assessment (U.S. Environmental Protection Agency, Region 9, Biological Technical Assistance Group, San Francisco, California, 1999). (3) M. Van den Berg, L. Birnbaum, A.T.C. Bosveld, B. Brunstrom, P. Cook, M. Feeley, J.P. Giesy, A. Hanberg, R. Hasegawa, S.W. Kennedy, T. Kubiak, J.C. Larsen, F.X.R. van Leeuwen, A.K. Djien Liem, C. Nolt, R.E. Peterson, L. Poellinger, S. Safe, D. Schrenk, D. Tillitt, M. Tysklind, M. Younes, F. Waern, and T. Zacharewski, Environ. Health Perspect. 106(12), (1998). (4) G.M. Frame, J.W. Chochran, and S. Bøwadt, J. High Resolut. Chromatogr. 19, 657 (1996). (5) E. Storr-Hansen, M. Cleeman, T. Cederberg, and B. Jansson, Chemosphere 24(3), (1992). (6) S.J. Harrad, A.S. Sewart, R. Boumphrey, R. Duarte-Davidson, and K.C. Jones, Chemosphere 24(8), (1992). (7) R. Lazar, R.C. Edwards, C.D. Metcalfe, T. Metcalfe, F.A.P.C. Gobas, and G.D. Haffner, Chemosphere 25(4), (1992). (8) C.S. Creaser, F. Krokos, and J.R. Startin, Chemosphere 25(12), (1992). (9) SW-846, Method 3620B: Florisil Cleanup (U.S. Environmental Protection Agency, Washington, D.C., 1996). (10) SOP #Or_P362A, Florisil Cleanup and Separation of Co-Planar PCB Congeners (U.S. Environmental Protection Agency, Region 10 Laboratory, Port Orchard, Washington, August 2002). (11) J.W. Anderson, Analysis of Coplanar Congeners in Aroclors Using Alumina Column Cleanup, ManTech Environmental Technology, Inc. (R.S. Kerr Environmental Research Laboratory, U.S. Environmental Protection Agency, Ada, Oklahoma, March 1992). (12) SW-846, Method 3540C: Soxhlet Extraction (U.S. Environmental Protection Agency, Washington, D.C., 1996). (13) SW-846, Method 3660: Sulfur Removal (U.S. Environmental Protection Agency, Washington, D.C., 1986). (14) SW-846, Method 3665A: Sulfuric Acid/Permanganate Cleanup (U.S. Environmental Protection Agency, Washington, D.C., 1996). (15) SOP #Or_P366A, Sulfur Removal by SW-846 Method 3660B (U.S. Environmental Protection Agency, Region 10 Laboratory, Port Orchard, Washington, August 2002).