SFE Plus C18 Lipid Cleanup Method for Selective Extraction and GC

Quantitative Determination of Polycyclic Aromatic Hydrocarbons in Barbecued Meat Sausages by Gas Chromatography Coupled to Mass Spectrometry...
0 downloads 0 Views 135KB Size
Anal. Chem. 1998, 70, 3242-3248

SFE Plus C18 Lipid Cleanup Method for Selective Extraction and GC/MS Quantitation of Polycyclic Aromatic Hydrocarbons in Biological Tissues Md. Yeakub Ali and Richard B. Cole*

Department of Chemistry, University of New Orleans, Lakefront, New Orleans, Louisiana 70148

Lipid material represents a potential interference for determination of nonpolar compounds (e.g., polycyclic aromatic hydrocarbons) in biological tissue samples. This study reports the development of a selective extraction method using supercritical CO2 that allows the GC/MS quantitation of PAHs in the presence of a substantial lipid background. Selective extraction of PAHs relies upon addition of C18 adsorbent beads to the initial sample slurry. The dried mixture, including C18 adsorbent, is placed in the supercritical fluid extraction (SFE) chamber. During the SFE process, lipids are preferentially retained on the C18 beads. This “SFE plus C18” procedure was developed by first optimizing SFE conditions (100 °C, 350 bar) for recovery of PAH standards. PAHs containing added model lipid compounds (stearic acid and cholesterol) were then subjected to SFE plus C18 treatment followed by GC/MS analysis. Using this approach, a recovery of 94-100% of PAHs was obtained while only 9-17% of the lipid material present was coextracted from the same test sample. The developed method is demonstrated to permit efficient recovery and detection of PAHs spiked into crab tissue, a matrix with a high lipid content. Determinations of polycyclic aromatic hydrocarbons (PAHs) in seafood or animal meat have important bearing because a significant number of these compounds are known or suspect carcinogens.1-4 Nonpolar, environmentally persistent pollutants such as PAHs have a strong tendency to accumulate in the fat of ingesting organisms. Magnification of accumulation levels can occur in higher members of the food chain, including humans. This has prompted much interest in the continued refinement of methodologies for accurately measuring levels of PAHs in the environment, especially in water and in biological tissues. The conventional approach to analysis of PAHs in seafood,5-7 meat,8 and chicken eggs9 usually involves a liquid solvent (1) Freudenthal, R. I.; Jones, P. W. Carcinog.sCompr. Surv. 1976, 1, 418419. (2) Marsch, G. A.; Jankowiak, R.; Farhat, J. H.; Small, G. J. Anal. Chem. 1992, 64, 3038-3044. (3) Miller, E. C.; Miller, J. A. Cancer 1981, 47, 2327-2345. (4) Miller, J. A. Cancer Res. 1970, 30, 559-576. (5) Akpan, V.; Lodovici, M.; Dolara, P. Bull. Environ. Contam. Toxicol. 1994, 53, 246-253. (6) Saxton, W. L.; Newton, R. T.; Roberg, J.; Sutton J.; Johnson, L. E. Bull. Environ. Contam. Toxicol. 1993, 51, 515-522. (7) Cocchieri, R. A.; Prete, U. D.; Arnese, A.; Giuliano, M.; Roncioni, A. Bull. Environ. Contam. Toxicol. 1993, 50, 618-625.

3242 Analytical Chemistry, Vol. 70, No. 15, August 1, 1998

extraction step (e.g., Soxhlet extraction) followed by a multistep separation procedure. This approach is labor intensive, and it consumes large volumes of costly and toxic solvents (e.g., dichloromethane, benzene) to the detriment of the environment. In the present decade, supercritical fluid extraction (SFE) has risen in popularity as a viable approach to environmental sample analyses.10-18 SFE employing CO2 has several distinct advantages over traditional solvent extraction including faster speed, controllable solvent power, zero toxicity, ease in running small samples, and easy automation.19 In recent years, the use of SFE for investigations of low-polarity constituents (especially polychlorinated biphenyls) present in lipidcontaining matrixes has been reported.16,20-25 In these studies, lipids were always prone to coextraction with the low-polarity analytes of interest. Accurate quantitation with low detection limits (e.g., by gas chromatography with FID, ECD, or MS detection) often relied upon subjecting the extract to a cleanup step to remove lipids prior to analysis. Conventional approaches to lipid removal that add an extra labor-intensive step to the overall analytical protocol include gel permeation chromatography, nor(8) Sitarska, E.; Klucinski, W.; Faundez, R.; Duszewska, A. M.; Winnicka, A.; Goralczyk, K. Bull. Environ. Contam. Toxicol. 1995, 55, 865-869. (9) Hothem, R. L.; Zador, S. G. Bull. Environ. Contam. Toxicol. 1995, 55, 658665. (10) Barnabas, I. J.; Dean, J. R.; Tomlinson, W. R.; Owen, S. P. Anal. Chem. 1995, 67, 2064-2069. (11) Langenfeld, J. J.; Hawthorne, S. B.; Miller, D. J.; Pawliszyn, J. Anal. Chem. 1993, 65, 338-344. (12) Burford, M. D.; Hawthorne, S. B.; Miller, D. J. Anal. Chem. 1993, 65, 14971505. (13) Dankers, J.; Groenenboom, M.; Scholtis, L. H. A.; ven der Heiden, C. J. Chromatogr. 1993, 641, 357-362. (14) Paschke, T.; Hawthorne, S. B.; Miller, D. J.; Wenclawiak, B. J. Chromatogr. 1992, 609, 333-340. (15) Wenclawiak, B.; Rathmann, C.; Teuber, A. Fresenius J. Anal. Chem. 1992, 344, 497-500. (16) David, F.; Verschuere, M.; Sandra, P. Fresenius J. Anal. Chem. 1992, 344, 479-485. (17) Robbat, A.; Liu, T. Y.; Abraham, B. M. Anal. Chem. 1992, 64, 1477-1483. (18) Hawthorne, S. B.; Miller, D. J. Anal. Chem. 1987, 59, 1705-1708. (19) Hawthorne, S. B. Anal. Chem. 1990, 62, 633A-642A. (20) Johansen, H. R.; Becher, G.; Greibrokk, T. Fresnius J. Anal. Chem. 1992, 344, 486-491. (21) Bowadt, S.; Johansson, B.; Fruekilde, P.; Hansen, M.; Zilli, D.; Larsen, B.; de Boer, J. J. Chromatogr. 1994, 675, 189-204. (22) Djordjevic, M. V.; Haffmann, D.; Fan, J.; Prokopezyk, B.; Citron, M. L.; Stellman, S. D. Carcinogenesis 1994, 15, 2581-2585. (23) Hale, R. C.; Gaylor, M. O. Environ. Sci. Technol. 1995, 29, 1043-1047. (24) Lee, H. B.; Peart, T. E.; Niimi, A. J.; Knipe, C. R. J. AOAC Int. 1995, 78, 437-444. (25) Alley, E. G.; Lu, G. J. AOAC Int. 1995, 78, 1051-1054. S0003-2700(98)00201-7 CCC: $15.00

© 1998 American Chemical Society Published on Web 06/26/1998

mal- or reversed-phase liquid chromatography, liquid-liquid partitioning, and Florisil column treatment.21,23 Barker and coworkers26-29 have used C18 (octadecyl siloxane) as a nonpolar adsorbent in a variety of biological matrixes. In these clinical and environmental analyses, C18 beads were used to adsorb nonpolar to slightly polar analytes in milk27 and animal tissues.26,28,29 In other studies, alumina has been used in conjunction with SFE to retain fats coextracted with either chlorinated pesticides from poultry fat30 or polychlorinated biphenyls from fish20,23 and crab.20 However, other than a conference proceedings report,31 to our knowledge no publication has documented the use of a solid trapping material inside the SFE chamber for selective extraction of PAHs in lipid-containing matrixes. The purpose of the current study is to develop a new extraction and sample preparation methodology where nonpolar analytes present in matrixes of high lipid content can be quantified by GC, employing a suitable detector, without further cleanup. Ten carcinogenic or suspect carcinogenic PAHs were selected as nonpolar test analytes. A SFE-GC/MS method was developed and tested using pure PAH standards in the presence of two model interfering lipids (i.e., stearic acid and cholesterol) present in increasing concentrations. The method was then applied to the determination of PAHs spiked into crab tissue containing high levels of indigenous lipids. EXPERIMENTAL SECTION Sample Handling. (1) Preparation of PAH Standards and Lipids. Ten chromatographic grade solid PAHs were purchased from Sigma Chemical Co. (St. Louis, MO). Stock solutions of the individual PAHs were prepared by serial dilutions in analytical grade methylene chloride (EM Science, Gibbstown, NJ) and then stored at -5 °C. Two solid chromatographic grade lipids (stearic acid, cholesterol) were purchased from Sigma. Stock solutions of each lipid were prepared by serial dilutions in pure methanol (EM Science) and stored at 0 °C. (2) Pretreatment of Sample Matrix. (a) C18 Beads. Analytical grade C18 nonpolar adsorbent (35-75 µm size, 60-Å porosity) purchased from Alltech Associates, Inc. (Deerfield, IL) was washed sequentially with at least two bed volumes of hexane, methanol, and methylene chloride and was then dried and stored at room temperature. (b) Diatomaceous Earth. Analytical grade diatomaceous earth (J. T. Baker Inc., Phillipsburg, NJ) was washed sequentially with hexane, methanol, and methylene chloride and was then dried and stored at room temperature. (c) Crab Tissue. Live whole crabs were collected from a local New Orleans market. Meat from five to six crabs was removed and was then pureed in a blender until homogeneous and smooth liquid mixtures were obtained. The pureed meat was stored at -30 °C until needed. (26) Barker, S. A.; Long, A. R.; Short, C. R. J. Chromatogr. 1989, 475, 353-361. (27) Long, A. R.; Hsieh, L. C.; Malbrough, M. S.; Short, C. R.; Barker, S. A. J. Assoc. Off. Anal. Chem. 1990, 73, 379-384. (28) Long, A. R.; Hsieh, L. C.; Bello, A. C.; Malbrough, M. S.; Short, C. R.; Barker, S. A. J. Agric. Food Chem. 1990, 38, 427-429. (29) Long, A. R.; Soliman, M. M.; Barker, S. A. J. Assoc. Off. Anal. Chem. 1991, 74, 493-496. (30) France, J. E.; King, J. W.; Snyder, J. M. J. Agric. Food Chem. 1991, 39, 1871-1874. (31) Ali, Md. Y.; Cole, R. B. Proceedings of the 45th Annual Conference on Mass Spectrometry and Allied Topics, Palm Springs, CA, June 1-5, 1997; p 911.

(3) Spiking of Sample Matrixes. (a) PAHs in C18 Adsorbent and in Other Matrixes. In a porcelain mortar, 2.0 g of pretreated C18 adsorbent was mixed with 50.0 µg each of 10 PAH standards and was allowed to stand for 15 min to evaporate excess solvent. The mixture was stirred gently until homogeneous and transferred either to a SFE extraction thimble or a Soxhlet extraction thimble as required. Spiked samples were prepared (in triplicate) for each extraction along with a blank. The above steps were followed exactly for the spiking of PAH mixtures in other sample matrixes, e.g., diatomaceous earth, considered to be an inert matrix, and for filter paper. (b) PAHs and Lipids in C18 Adsorbent and in Filter Paper. A lipid mixture in 100 µL of methanol containing equal amounts of stearic acid and cholesterol in progressively higher concentrations (50, 250, 500, and 2500 µg) was added to separate containers of 2.0 g of C18 matrix. Immediately afterward, a mixture of eight PAH standards (50 µg each) was added to each C18 adsorbent sample containing the lipids. Each sample was then homogenized by gentle stirring for 15 min. Finally, each sample was transferred to a separate extraction thimble, and a blank was prepared. In another series, after the above steps were performed, the PAHs and the lipids were added onto filter paper. (c) PAHs in Crab Tissue. Pureed crab tissue (0.5 g) was placed into a porcelain mortar. A mixture of 10 PAHs (50 µg each) was spiked into crab tissue and allowed to stand for 15 min. Two grams of C18 beads was added to the mixture, which was then homogenized by gentle pestle stirring. The homogenized sample was transferred into an extraction thimble. In another set of experiments, an identical mixture of PAHs spiked into crab tissue was prepared, which was homogenized with 2.0 g of filter paper and subsequently transferred to an extraction thimble. Soxhlet Extraction. For purposes of comparison of the developed method with an established method, Soxhlet extraction was used for the extraction of PAHs in an inert matrix. A mixture of the same 10 PAH compounds, each present at 50 µg, was spiked into 2.0 g of pretreated diatomaceous earth. The sample was then wrapped in filter paper and placed into an extraction thimble. Constant heating was maintained for 48 h, enabling extraction at 4-6 cycles/h. Supercritical Fluid Extraction. A Hewlett-Packard model HP 7680A (Palo Alto, CA), supercritical fluid extractor was used for all SFE work in this study. Two-gram samples were subjected to supercritical CO2 for 5 min of static equilibration time at 100 °C and 350 bar. Afterward, a CO2 stream flowed through the extraction thimble at the rate of 1.5 mL/min for a 25-min extraction. Three 1.0-mL methylene chloride rinses of the stainless steel trapping beads (held at 30 °C during extractions) allowed quantitative transfer of the extracted compounds. Gas Chromatography/Mass Spectrometry. (1) Acquisition. A HP-5985 GC/MS system (Hewlett-Packard) was used for all determinations in this study, except for the crab tissue work that was performed on a Fisons 8000 GC coupled to a Quattro II MS (Micromass, Inc., Manchester, U.K.). A 30 m × 0.32 mm id × 0.17 µm film thickness, Ultra-2 cross-linked 5% phenyl-methyl silicone (Hewlett-Packard) fused-silica capillary was used throughout as the GC column. The GC column was temperature programmed as follows: 1 min isothermal at 100 °C, and then at 8 °C/min to 300 °C and held isothermal at 300 °C for 2 min. Analytical Chemistry, Vol. 70, No. 15, August 1, 1998

3243

Helium gas served as the carrier gas at a flow rate of ∼3-4 mL/ min. The sample injection port was heated to 325 °C. The quadrupole mass spectrometer was operated in the electron ionization (EI) mode (electron energy 70 eV, source temperature 150 °C) scanning from m/z 50 to 450 at 1.6 s/scan with unit resolution. The MS data were acquired on a Vector-1 (Teknivent Inc., formerly of Maryland Heights, MO) data system except for crab tissue data that were acquired on a MassLynx (Micromass, Inc.) data system. (2) Quantitation. Standard working curves were generated each day that samples were run. All sample extractions were performed in triplicate, and each extract was injected into the GC/ MS instrument three times; thus, a total of nine measurements were made for each sample to determine the standard deviations reported in Tables 1-4. Quantitation for data presented in Tables 1-3 was performed on a Vector-2 (Teknivent) data system. These recoveries were calculated employing a one-point measurement of the peak area of a standard, followed by comparison (ratio) of the peak area of each sample extract to the standard. Quantitation for data presented in Table 4 was performed using a MassLynx 3.0 (Micromass) data system. Here, a three-point response (peak area) vs concentration calibration curve was generated for each PAH analyte (three standard concentrations). RESULTS AND DISCUSSION The goal of this study is to develop a rapid and efficient SFE method for the analysis of nonpolar compounds, such as PAHs, that may be contained in a matrix containing potential nonpolar interferences such as lipids. Not only do lipid materials mask analyte peaks, but they may also irreversibly degrade GC columns. The approach used here is to add C18 adsorbent directly to the slurry that is prepared for SFE analysis. The entire sample mixture, including C18 adsorbent, is placed in the extraction chamber and subjected to SFE conditions. Ten carcinogenic or suspect carcenogenic PAHs were selected as analytes of interest in this study (Figure 1) and they are given in order of chromatographic elution on the cross-linked 5% phenylmethyl silicone capillary column: phenanthrene (m/z 178), anthracene (m/z 178), fluoranthene (m/z 202), pyrene (m/z 202), benz[a]anthracene (m/z 228), chrysene (m/z 228), benzofluoranthene (m/z 252), benzo[e]pyrene (m/z 252), benzo[a]pyrene (m/z 252), and perylene (m/z 252). Two lipids were selected as model interference compounds, and they are stearic acid (m/z 284) and cholesterol (m/z 386). (1) Optimization of SFE Conditions. SFE conditions employing supercritical CO2 were initially optimized to get a maximum recovery of PAHs spiked onto the C18 matrix. As temperature and pressure are the most influential variable parameters in SFE, the mixture containing equivalent weights (50 µg) of 10 PAHs mixed with C18 beads was extracted at 350 bar (maximum workable supercritical CO2 pressure of the instrument) while the extraction chamber temperature was raised incrementally (50, 75, 100, and 120 °C) in separate experiments, keeping all other SFE parameters constant. Figure 2 is the plot of percent recovery vs temperature for each of the 10 PAHs extracted from this group of samples. The plot shows that the percent recovery of all ten PAHs increased with temperature, reaching a plateau at 100 °C or above. The maximum recovery of the first six PAHs, i.e., phenanthrene, anthracene, fluoranthene, pyrene, benz[a]3244 Analytical Chemistry, Vol. 70, No. 15, August 1, 1998

Figure 1. Structures of employed PAHs.

Figure 2. SFE optimization plot of percent recovery vs temperature at constant pressure (350 bar).

anthracene, chrysene (named in order of elution), was ∼100%, while the recoveries of the remaining four (benzofluoranthene, benzo[e]pyrene, benzo[a]pyrene, perylene) were ∼95% at 100 °C or above. Similarly, in separate experiments, an identical mixture of PAHs was extracted while the supercritical CO2 pressure incrementally changed from 100 to 190 to 350 bar while keeping other parameters constant including temperature (100 °C). All extracts were run in triplicate directly by GC/MS with temperature programming. Figure 3, a plot of percent recovery vs pressure at 100 °C chamber temperature for each of the 10 extracted PAHs, indicates that the recoveries of all PAHs increased with increasing supercritical CO2 pressure. The recoveries of all PAHs were highest at a pressure of ∼350 bar (maximum workable pressure of instrument); thus, 100 °C chamber temperature and 350 bar

Table 2. Comparison of SFE Recoveries of PAHs from Filter Paper and C18 Matrixes % recovery of PAHs (mean ( SD)

Figure 3. SFE optimization plot of percent recovery vs pressure at constant temperature (100 °C). Table 1. Comparison of PAH Recoveries from Inert Matrix Using SFE and Soxhlet Extraction % recovery of PAHs (mean ( SD) compounds

Soxhlet

SFE

phenanthrene anthracene fluoranthene pyrene benz[a]anthracene chrysene benzofluoranthene benzo[e]pyrene benzo[a]pyrene perylene

101.3 ( 2.8 101.0 ( 3.6 100.0 ( 2.2 101.8 ( 2.5 99.0 ( 2.6 101.5 ( 2.9 99.0 ( 1.4 100.5 ( 1.7 99.8 ( 3.3 101.0 ( 3.7

100.7 ( 3.1 100.3 ( 2.3 100.7 ( 3.5 101.3 ( 2.1 100.0 ( 2.6 100.6 ( 2.1 100.3 ( 2.1 99.7 ( 4.0 99.3 ( 2.1 99.7 ( 2.1

supercritical CO2 pressure were considered to be optimal SFE conditions for PAH determinations. (2) SFE and Soxhlet Extraction Recovery Comparison. The conventional approach to PAH extraction utilizes Soxhlet extraction, a method offering the possibility for quantitative recoveries. The disadvantages of the method, however, are that it is time-consuming and labor intensive and it consumes high levels of toxic solvents. A comparison of SFE and Soxhlet extraction was made for the recoveries of PAHs that were spiked onto a chemically inert matrix, i.e., diatomaceous earth. A mixture of the same 10 PAH standards (Figure 1) was spiked onto this inert matrix; then SFE extraction was carried out for three replicate samples. All extractions were done at previously optimized SFE conditions: extraction chamber temperature 100 °C; supercritical CO2 pressure 350 bar. Four replicate 48-h Soxhlet extractions were performed with methylene chloride solvent on an identically prepared PAH mixture that had likewise been spiked onto diatomaceous earth. Obtained recovery data for SFE and Soxhlet extractions of the 10 PAH standards spiked onto diatomaceous earth are given in Table 1. The data show that both Soxhlet and SFE exhibit similar results with 100% recoveries obtained for all 10 PAHs compounds on the inert matrix. This indicates that SFE has the same excellent extraction efficiency as Soxhlet extraction for recovery of PAHs from the inert matrix diatomaceous earth. The detection limit of the HP5985 GC/MS instrument was 1.0 ng/µL employing 1.0-µL sample injections.

compounds

filter paper

C18 adsorbent

phenanthrene anthracene fluoranthene pyrene benz[a]anthracene chrysene benzofluoranthene benzo[e]pyrene benzo[a]pyrene perylene

100.3 ( 1.5 100.0 ( 2.9 100.0 ( 1.6 100.5 ( 1.9 99.8 ( 2.4 100.8 ( 2.5 100.0 ( 1.4 99.3 ( 1.9 99.5 ( 2.4 100.3 ( 2.4

100.0 ( 2.9 100.3 ( 2.4 99.5 ( 1.3 100.3 ( 2.4 100.0 ( 2.2 100.3 ( 2.1 97.0 ( 1.8 95.5 ( 1.9 95.0 ( 2.6 95.0 ( 2.0

(3) Matrix Comparison. (a) Inert Matrixes. In separate experiments, recoveries of PAHs from filter paper were compared to recoveries from diatomaceous earth under optimized conditions. Recovery studies of a mixture of the same 10 PAHs performed in triplicate revealed that all 10 were quantitatively recovered from filter paper, thus establishing that filter paper does not irreversibly bind (or otherwise alter) PAH compounds. Filter paper can thus be used as an alternative inert SFE matrix in place of diatomaceous earth for PAH extraction. This circumvents the problem of entraining a very fine powder of diatomaceous earth in the SF extractor plumbing line, nozzle, trap etc., necessitating frequent cleaning of the SFE instrument. (b) Inert Matrix vs Nonpolar Adsorbent. Under optimized SFE conditions, a comparison of SFE recoveries of PAHs on filter paper (inert matrix) vs C18 (nonpolar adsorbent) was made to characterize the adsorption behavior of PAHs on the nonpolar adsorbent. SFE recovery data for PAHs on filter paper vs C18 nonpolar adsorbent are shown in Table 2. The data indicate that the first six PAHs, i.e., phenanthrene, anthracene, fluoranthene, pyrene, benz[a]anthracene, and chrysene, were extracted with 100% recoveries from both filter paper and C18 matrixes. But the last four compounds, i.e., benzofluoranthene, benzo[e]pyrene, benzo[a]pyrene, and perylene, were recovered at 95-97% levels using the C18 adsorbent, whereas all PAHs in filter paper were extracted with 100% recoveries (in agreement with previous results). The large molecular size of the last four compounds may be the main cause for the slightly lowered recoveries in C18 adsorbent, as it may be presumed that a nonpolar molecule of larger size has more adsorptive interaction with the nonpolar C18 beads than one of smaller size. (4) Recoveries in the Presence of Lipid Interferences. To test the recovery of PAHs in the presence of interfering lipid materials, increasing concentrations of two model lipids, stearic acid and cholesterol (at 50, 250, 500, or 2500 µg each), were added to a mixture containing equivalent weights (50 µg) of eight PAHs (phenanthrene, anthracene, fluoranthene, pyrene, benz[a]anthracene, chrysene, benzo[a]pyrene, and perylene; 50 µg each). The homogenized mixtures were placed either onto filter paper or onto 2.0 g of C18 adsorbent. All extractions were performed under optimum SFE conditions and the extracts were then directly run by GC/MS. From the inert filter paper, Figure 4a (top) shows a substantial coextraction of lipids (peaks 5 and 10) along with the PAHs (peaks 1, 2, 3, 4, 6, 7, 8, and 9). By contrast, Figure 4a (bottom) shows that the SFE plus C18 method significantly Analytical Chemistry, Vol. 70, No. 15, August 1, 1998

3245

Figure 4. GC/MS total ion chromatograms of extracts from mixtures containing PAHs and model lipids: (top) conventional SFE using inert sorbent; (bottom) SFE plus C18. Each sample contains 50 µg of eight PAHs (phenanthrene, anthracene, fluoranthene, pyrene, benz[a]anthracene, chrysene, benzo[a]pyrene, and perylene, peaks 1-4 and 6-9, respectively). Peaks 5 and 10 correspond to the model lipids stearic acid and cholesterol, respectively. Samples used to obtain panels a-d contain 50, 250, 500, and 2500 µg of each lipid, respectively.

reduced the coextraction of interfering lipid materials (peaks 5 and 10), while all spiked PAHs were recovered in high yields. In Figure 4b-d, progressively higher quantities of lipid materials were present. Upon supercritical CO2 extraction from the C18 matrix (lower traces), it was found that the PAHs (peaks 1, 2, 3, 4, 6, 7, 8, and 9) remained readily quantifiable due largely to a reduced coextraction of lipids (peaks 5 and 10). By contrast, in the filter paper extracts (upper traces), the lipid peaks truly dominate. The difference is most striking when the upper and lower total ion chromatograms of Figure 4d are compared. 3246 Analytical Chemistry, Vol. 70, No. 15, August 1, 1998

In treating the Figure 4 data, the two model lipids were found to be extracted from filter paper (conventional SFE) in 99-100% efficiencies regardless of their initial concentrations. For simultaneously increasing concentrations of the two model lipids, i.e., 50, 250, 500, and 2500 µg, the SFE plus C18 method yielded stearic acid detection at 13, 14, 16, and 17% of the initial added quantity, respectively, while cholesterol was detected at 9, 10, 12, and 11%, respectively. Calculated uncertainties for all of these obtained values were between (1 and (3%. Thus, at any concentration level employed in Figure 4, only 9-17% of the model lipids present

Table 3. SFE plus C18 Recovery of PAHs in the Presence of Model Lipids % recovery of PAHs in C18a (mean ( SD)

a

compounds

no lipids

50 µg SA 50 µg Chol

250 µg SA 250 µg Chol

500 µg SA 500 µg Chol

2500 µg SA 2500 µg Chol

phenanthrene anthracene fluoranthene pyrene benz[a]anthracene chrysene benzo[a]pyrene perylene

100.0 ( 2.9 100.3 ( 2.4 99.5 ( 1.3 100.3 ( 2.4 100.0 ( 2.2 100.3 ( 2.1 95.0 ( 2.6 95.0 ( 1.8

99.8 ( 3.0 100.0 ( 2.7 100.2 ( 1.9 99.8 ( 2.4 99.5 ( 2.8 100.2 ( 2.5 94.2 ( 2.0 94.7 ( 2.8

100.2 ( 2.5 99.5 ( 3.1 99.7 ( 2.4 100.0 ( 1.9 99.8 ( 2.1 99.6 ( 2.7 94.4 ( 2.7 94.0 ( 2.1

99.4 ( 2.6 100.0 ( 2.1 99.5 ( 2.1 99.7 ( 2.8 99.3 ( 2.5 99.8 ( 2.0 94.0 ( 2.4 93.7 ( 2.6

99.2 ( 2.1 99.7 ( 2.7 99.0 ( 2.3 99.4 ( 2.5 99.8 ( 2.0 99.0 ( 2.6 93.7 ( 2.8 93.5 ( 3.3

SA ) stearic acid; Chol ) cholesterol.

were coextracted from C18 beads as compared to filter paper. In other words, 83-91% of the lipid material was selectively retained using SFE plus C18. A comparison of the calculated percentage recovery of PAHs in the presence and in the absence of lipids is given in Table 3. The data show that the percentage recovery of PAHs using SFE plus C18 in the presence of low to high concentration of lipids is the same as the percentage recovery in the absence of lipids. This indicates that the presence of lipids does not interfere with the recovery of PAHs by the SFE plus C18 method. It is noted that, in the GC/MS traces (Figure 4), the peaks of the two selected “model lipids” do not overlap with the peaks of PAHs. Of course, coeluting peaks present an added obstacle to accurate quantitation, as will be more precisely demonstrated below. The above studies lead to the conclusion that the SFE plus C18 method can selectively retain lipid materials (83-91% retention) that represent potential interferences, while allowing highly adequate (94-100%) extraction of PAHs from lipid-containing matrixes. (5) SFE plus C18 for PAH Determination in a Seafood Matrix. The SFE plus C18 sample preparation method was tested using a real sample matrix, crab meat. No PAHs were detectable in the initial commercially obtained product. A mixture of 10 PAHs (phenanthrene, anthracene, fluoranthene, pyrene, benz[a]anthracene, chrysene, benzofluoranthene, benzo[e]pyrene, benzo[a]pyrene, and perylene, each at 50 µg) was spiked in 0.5 g of pureed crab tissue and was then homogenized with 2.0 g of C18 adsorbent. The sample was transferred to an extraction thimble. Similarly, an identical preparation of PAH-spiked crab tissue was also homogenized with 2.0 g of filter paper and was then transferred to an extraction thimble. Triplicate samples were extracted by SFE under optimized conditions for each sample set. Extracts were then directly run by GC/MS with temperature programming as previously detailed. Figure 5a shows the total ion chromatogram (TIC) of the homogenate containing crab tissue, PAHs, and filter paper. The peaks labeled with letters (a-j) correspond to coextracted lipids, while those labeled with numbers (1-10) are those of the spiked PAHs. The indigenous lipids dominate the top chromatogram, and four of the PAHs were inadequately separated from coeluting background lipids. By contrast, the comparable TIC trace appearing in Figure 5b obtained from the SFE plus C18 extract of crab tissue and PAHs reveals clearly distinct PAHs (major peaks 1-10) present on a minor background of lipid material (peaks b, c, f, g, and h). Notably, lipids peaks a, d, e, i, and j that were

Figure 5. GC/MS total ion chromatograms of extract from pureed crab tissue spiked with 10 PAHs: (a) conventional SFE using inert sorbent; (b) SFE plus C18. Peaks labeled with numbers correspond to PAHs: 1, phenanthrene; 2, anthracene; 3, fluoranthene; 4, pyrene; 5, benz[a]anthracene; 6, chrysene; 7, benzofluoranthene; 8, benzo[e]pyrene; 9, benzo[a]pyrene; 10, perylene. Peaks labeled with letters correspond to various indigenous lipids. Table 4. Recovery of PAHs Spiked in Crab Tissue % recovery of PAHs (mean ( SD) compounds

conventional SFE

SFE plus C18

phenanthrene anthracene fluoranthene pyrene benz[a]anthracene chrysene benzofluoranthene benzo[e]pyrene benzo[a]pyrene perylene

80.1 ( 3.8 55.1 ( 5.5 49.6 ( 7.0 80.1 ( 6.8 88.9 ( 5.0 90.8 ( 4.7 93.4 ( 4.2 91.5 ( 5.9 91.5 ( 5.8 90.4 ( 5.4

97.8 ( 2.9 98.1 ( 2.5 99.6 ( 3.9 100.0 ( 2.5 98.3 ( 3.6 99.2 ( 3.6 96.6 ( 3.6 95.3 ( 3.5 95.5 ( 3.1 96.3 ( 3.6

readily detected in Figure 5a are not observable in Figure 5b. This comparison indicates that SFE plus C18 is quite effective at retaining indigenous lipids present in the crab tissue extract. SFE plus C18 recovery data for PAHs in crab tissue and analogous recoveries from filter paper are given in Table 4. This table shows that the recovery of all PAHs in spiked crab tissue using SFE plus C18 is between 95 and 100%. By contrast, the recovery of PAHs in spiked crab tissue from filter paper is only 50-93%. The overlap of PAH peaks 1, 2, 3, and 4 (i.e, phenanthrene, anthracene, fluoranthene, and pyrene, respectively) by larger quantities of coeluting lipids (peaks a, d, and e) caused the lowering of recoveries of these PAHs in the filter paper extract Analytical Chemistry, Vol. 70, No. 15, August 1, 1998

3247

(Figure 5a). Specifically, the recoveries of phenanthrene (peak 1) and anthracene (peak 2) were reduced from 98 to 80%, and from 98 to 55%, respectively, due to peak masking by a lipid (labeled “a”) whose molecular weight is 228. The recovery of fluoranthene (peak 3) was reduced by the largest extent, from 100 to 50%, due to significant overlap of lipid d whose molecular weight is 270. The recovery of pyrene (peak 4) was also reduced (from 100 to 80%) due to partial overlap of background lipid e whose molecular weight is 270. The recoveries of the other six PAHs were 4-9% lower from filter paper as compared to the respective recoveries obtained from SFE plus C18. These data indicate that the SFE plus C18 method can significantly improve recovery and quantitation of PAHs extracted in the presence of lipid materials; the improvement is most pronounced when lipid materials coelute with analyte peaks. While SFE plus C18 has been shown to selectively retain lipids indigenous to crab tissue, in no case were PAH recoveries below 95%, closely mimicking Table 2 results obtained in the absence of lipids. CONCLUSION The SFE plus C18 method developed in this study, employing supercritical CO2 is capable of virtually quantitative extraction (95100%) of PAHs. Compared to Soxhlet extraction, the SFE-based approach allowed much faster extractions and required ∼2 orders of magnitude less solvent. Addition of the nonpolar adsorbent

3248 Analytical Chemistry, Vol. 70, No. 15, August 1, 1998

C18 (octadecylsiloxane) to the SFE vessel allowed selective retention of interfering model lipids, reducing their presence to 9-17% of the quantity measured in the absence of C18 adsorbent. While 83-91% of the lipids remained adhered to C18 beads, PAHs were recovered in 94-100% efficiencies. The SFE plus C18 method allowed the efficient, selective extraction of PAHs from a model biological matrix, crab tissue, while largely inhibiting the coextraction of interfering indigenous lipid materials. For 1.0-µL GC/ MS injections, the detection limit was 1.0 ng. Future work will examine further application of the method to the analysis of nonpolar compounds in the presence of high levels of lipids in other types of seafood, meat, and milk. ACKNOWLEDGMENT Financial support for this project was provided by the Louisiana Education Quality Support Fund (Louisiana Board of Regents Fund) through Grant LEQSF(RF/1995-97)-RD-A-44. Mass spectrometry support was provided by the National Science Foundation through Grant CHE-9512155 and by the W. M. Keck Foundation.

Received for review February 23, 1998. Accepted May 5, 1998. AC980201+