Anal. Chem. 2001, 73, 4050-4053
Comparison of Fat Retainers in Accelerated Solvent Extraction for the Selective Extraction of PCBs from Fat-Containing Samples Erland Bjo 1 rklund,‡ Anne Mu 1 ller, and Christoph von Holst*
European Commission, Joint Research Centre, Institute for Health and Consumer Protection, Food Products and Consumer Goods Unit, I-21020 Ispra (VA), Italy
Five types of fat retainers were investigated for the lipidfree extraction of PCBs from fat-containing matrixes using accelerated solvent extraction: florisil, basic alumina, neutral alumina, acidic alumina, and sulfuric acidimpregnated silica. All of the fat retainers generated fatfree extracts when the fat/fat retainer ratio was 1:40. Sulfuric acid-impregnated silica and florisil were the only retainers that gave completely clear extracts, and the former was the only not to show any reaction when treated with sulfuric acid after the extraction. Using sulfuric acidimpregnated silica as fat retainer, on-line cleanup of fatcontaining matrixes was possible, as demonstrated for naturally contaminated fish meal as well as certified cod liver oil (CRM 349). Currently there is increasing public awareness of food quality. The recent crisis in Belgium once again put focus on dioxins and polychlorinated biphenyls as a result of the contamination of pork and chicken meat1. The analysis of the worst cases of these food items revealed large PCB contamination, sometimes exceeding the tolerance level (200 ng, sum of seven indicator PCBs/g fat, including PCBs 28, 52, 101, 118, 138, 153, and 180) set by the European Commission by a factor of 2502. Determination of PCBs in fatty matrixes is important; however, the extraction of trace compounds in the presence of extractable major sample components such as lipids offers special problems: In gas chromatography, large amounts of injected fat may cause problems in the injector and at the top of the column;3,4 when using mass spectrometry detection, the ion source might become contaminated, causing impaired analytical performance.5 Extraction procedures often rely on Soxhlet extraction6 in which the extracts contain large amounts of lipids that need to be removed. Several * Corresponding author. ‡ Present address: Department of Analytical Chemistry, Lund University, P.O. Box 124, S-221 00 Lund, Sweden. (1) European Commission. Official Journal of the European Commission 1999, L 310, 62-70. (2) Bernard, A.; Hermans, C.; Broeckaert, F.; De Poorter, G.; De Cock, A.; Housins, G. Nature 1999, 401, 62-70. (3) Grob, K., Jr. J. Chromatogr. 1984, 287, 1-15. (4) Grob, K., Jr.; Bossard, M. J. Chromatogr. 1984, 294, 65-75 (5) Mågård, M.; Berg, H. E. B.; Tagesson, V.; Ja¨remo, M. L. G.; Karlsson, L. L. H.; Mathiasson, L. J. E.; Bonneau, M.; Hansen-Møller, J. J. Agric. Food Chem. 1995, 43, 114-120 (6) EN 1528-2. Fatty Food: Determination of Pesticides and Polychlorinated Biphenyls (PCBs), Extraction of Fat, Pesticides, and PCBs, and Determination of Fat Content; Beuth Verlag: Berlin, 1997.
4050 Analytical Chemistry, Vol. 73, No. 16, August 15, 2001
cleanup procedures are described in the literature, including column chromatography on activated florisil or gel permeation chromatography (GPC).7 Unfortunately these extraction and cleanup procedures are both solvent-consuming and tedious. At present, several novel techniques are available to help analytical chemists in their efforts to produce extracts ready for analysis with a minimum amount of time spent on sample extraction and cleanup. One of the most recent sample preparation techniques is accelerated solvent extraction (ASE), which was developed by Dionex Corporations. This extraction can be complete within a few minutes.8 A short review of the major scientific investigations related to extraction of persistent organic pollutants (including PCBs) from environmental samples was recently presented by Bjo¨rklund et al.9 PCBs in sediments and similar solid environmental matrixes have primarily been determined using ASE; the numbers of scientific publications dealing with PCB extraction from food samples are still limited. A few investigations have been devoted to oyster tissue10 and freeze-dried mussel tissue.11 In the first investigation, the extracts were injected directly onto a GCECD without any further sample cleanup. This was possible because oysters contain only 1% fat. However when the matrixes are richer in lipids, the extracts have to be cleaned prior to injection. In the latter study, size exclusion chromatography was employed (using as much as 70 mL of chlorinated organic solvent), since freeze-dried mussels contain ∼8% fat. The possibility of performing on-line cleanup in ASE for the extraction of PCBs from fatty samples was presented in the extraction of PCBs from fish, in which acidic alumina was introduced into the extraction cell.12 With this ASE setup, the extracts could be analyzed on GC-ECD without further sample treatment. This approach is very similar to selective extractions performed in SFE investigations in which PCBs are selectively extracted from fat matrixes using basic alumina.13,14 Within the field of SFE, several fat retainers have been investigated,14 but no literature is available (7) EN 1528-3 Fatty Food: Determination of Pesticides and Polychlorinated Biphenyls (PCBs), Cleanup Methods; Beuth Verlag: Berlin, 1997. (8) Richter, B. LC-GC 1999, 17, 522-528. (9) Bjo ¨rklund, E.; Nilsson, T.; Bøwadt, S. TrAC 2000, 19, 434-445. (10) Richter, B. E.; Jones, B. A.; Ezzell, J. L.; Porter, N. L.; Avdalovic, N.; Pohl, C. Anal. Chem. 1996, 68, 1033-1039. (11) Schantz, M. M.; Nichols, J. J.; Wise, S. A. Anal. Chem. 1997, 69, 42104219. (12) Dionex; Application Note ASE 322; Dionex Corporation: Sunnyvale, CA, 1996. 10.1021/ac010178j CCC: $20.00
© 2001 American Chemical Society Published on Web 07/12/2001
so far on the on-line cleanup of ASE extracts using fat retainers. In this paper, we are comparing the fat-retaining capability of five different fat retainers, florisil, basic alumina, neutral alumina, acidic alumina, and sulfuric acid-impregnated silica, to increase our knowledge of their behavior during accelerated solvent extraction procedures. Additionally, the different fat retainers were tested on fish meal naturally contaminated with PCBs to check the efficiency of removing not only lipids but also other coeluting compounds in order to find the most suitable adsorbent for realworld samples and to avoid tedious cleanup procedures. Such an approach would drastically facilitate sample throughput and reduce the workload for routine laboratories, thereby cutting overall costs. MATERIALS AND METHODS Samples. The lard was bought in a local Italian supermarket. The naturally contaminated fish meal was supplied by State Official Laboratory (ROLT, Tervuren, Belgium). Many commercially available feeds contain fish meal as a major ingredient in which the PCBs have been incorporated into the matrix via the food chain. Consequently, the PCBs in fish meal are less accessible than PCBs that are added to feed during or after production, making fish meal a good choice of matrix for method performance verification purposes. Cod liver oil CRM 349 was provided by the Institute for Reference Materials and Measurements, IRMM, of the European Commission, Joint Research Centre, Geel, Belgium. Chemicals. Basic, neutral, and acidic alumina were purchased from Machery-Nagel GmbH (Du¨ren, Germany). Florisil came from BDH (Milano, Italy). All fat retainers were heated to 400 °C overnight prior to use in order to remove residues of substances that might interfere with the determination of the target analytes. Sodium sulfate (AnalaR) was from BDH (Poole, England). Sulfuric acid (95-98%), sea sand (Pro Analysis), and n-hexane (SupraSolv, organic trace analysis) came from Merck (Darmstadt, Germany). The sodium sulfate and sea sand were heated to 500 °C for 6 h before usage. Silica gel 60 was purchased from Fluka Chemie AG (Buchs, Switzerland), SFE Wet support was from Isco, Inc. (Lincoln, NB), and cellulose filters for extracting cell caps came from Dionex Corp. (Sunnyvale, CA). Impregnated silica gel was prepared by heating 600 g of silica gel 60 overnight at 200 °C and adding 400 g of sulfuric acid to the cold material. Finally, this mixture was mixed in a head-overheel mixer for 4 h. All PCB congeners were from Dr. Ehrenstorfer, GmbH (Augsburg, Germany). A PCB spiking solution (used to spike the lard) was prepared by dissolving PCBs 28, 52, 101, 118, 153, 138, and 180 in n-hexane at a concentration of each congener of 200 ng/mL. These congeners were chosen because they are indicator PCBs that, according to European legislation, should be included in the analytical program1. PCB 209 was used as an internal standard solution at a concentration of 1000 ng/mL dissolved in n-hexane. This internal standard was added to the final extracts in all of the experiments. Quantitation was based on a five-point calibration curve in the concentration interval 0.75-10 ng/mL for the fish meal and 1.5-50 ng/mL for the cod liver oil. In both cases, PCB 209 was added at 50 ng/mL. (13) Bjo¨rklund, E.; Ja¨remo, M.; Mathiasson, L. J. Liq. Chromatogr., Relat. Technol. 2000, 23, 2337-2344. (14) Ja¨remo, M.; Bjo ¨rklund, E.; Nilsson, N.; Karlsson, L.; Mathiasson, L. J. Chromatogr. A 2000, 877, 167-180.
Figure 1. Packing of the extraction thimble.
Equipment. Gas Chromatographic Analysis. A Hewlett-Packard GC 6890 equipped with HP MSD 5973 and HP 5 MS capillary columns (length, 30 m; column i.d., 0.25 mm; film thickness, 0.25 µm) was used to analyze the PCB congeners. The MS was run in the single ion monitoring (SIM) mode, and the following masses were measured for each chlorination level of the analyzed PCBs: M and M + 2 for PCBs 28 and 52; M + 2 and M + 4 for PCBs 101, 118, 138, 153, 180, and 209. Quantification was based on the added internal standard (PCB 209). Standard Parameters for the ASE Extraction. All of the extractions were performed on an ASE 200 System (Dionex, Sunnyvale, CA), using instrumental settings suggested by the instrument’s manufacturer.12 These included a temperature of 100 °C and a static time of 5 min (2 cycles) at a pressure of 1500 psi. The solvent used was n-hexane with a flush volume of 60% and a purge time of 90 s. The final extract was always in the range of 30-40 mL when using a thimble size of 33 mL. Directly after the extraction step, 100 µL of the internal standard solution (PCB 209) was added to the extracts. Prior to GC analysis, the extracts were concentrated to ∼1 mL using an evaporator equipped with a vacuum controller and filled with n-hexane to the 2 mL mark. The dead volume material used was sand/sodium sulfate (1/1, w/w), and ∼1 mL of SFE wet support was added on top of this for practical reasons in order not to have sand corns all the way to the top of the thimble when closing it. The extraction thimble was completely filled according to the recommendations of the instrument’s manufacturer.12 A 2-g portion of fish meal and 0.5 g of cod liver oil were mixed with a sample support consisting of sand/sodium sulfate (1/1, w/w). The minimum sample size for fish meal used in this study was set at 2 g in order to meet basic requirements for sufficient homogeneity. Grinding was performed with mortar and pestle; the presence of sand supported the mechanical breaking and grinding of the matrix. The packing of the extraction thimble can be seen in Figure 1. Fat Determination. The fat content was determined gravimetrically using a Mettler PM 400 (Mettler Strumenti, Milan, Italy) analytical balance. By transferring an aliquot of the eluate into a flask and weighing the residue in the glass after evaporation of the solvent, the fat content of the sample could be determined. Analytical Chemistry, Vol. 73, No. 16, August 15, 2001
4051
weighing under a fume removal device. Wear nonpermeable gloves.
Figure 2. Amount of fat retained for five fat retainers using different fat/fat retainer ratios. Each data point is an average of three measurements. Table 1. Determination of PCBs in Fish Meal Using Two Different Standard Extraction Methodologies According to the CEN Guidelines6 soxhlet
cold column
average of averages
congener
PCB ng/g sample
RSD % n)3
PCB ng/g sample
RSD % n)8
PCB ng/g sample
28 52 101 118 153 138 180
0.52 0.96 2.30 4.12 12.50 10.11 5.34
3.6 3.2 5.8 1.1 6.0 2.8 2.1
0.48 1.05 2.31 4.06 12.52 9.67 5.61
4.9 5.4 4.5 5.6 2.6 2.4 1.9
0.50 1.00 2.30 4.09 12.51 9.89 5.47
Soxhlet Extraction. Soxhlet extractions were performed according to the CEN guidelines.6 A 10-g portion of fish meal was mixed with 10 g of sodium sulfate and transferred to Soxhlet extraction thimbles. This extraction was performed for 6 h using n-hexane as the extraction solvent. A 250-µL portion of the internal standard was added, and the eluate was concentrated to a final volume of 3 mL using an evaporator equipped with a vacuum controller and filled with n-hexane to the 5 mL mark. A 2-mL aliquot was taken and was shaken vigorously with concentrated sulfuric acid. After separation of the two phases (which can be supported by a centrifuge), the supernatant was subjected to GC/MS analysis. Cold Column Extraction. The cold column extractions of the Belgian fish meal were performed according to the CEN guidelines.6 A 10-g portion of fish meal was mixed with 100 g of sea sand/sodium sulfate (1/1, w/w). The mixture was transferred to a glass column and was extracted using 300 mL of a solution containing n-hexane/acetone (2/1, v/v). A 250-µL portion of the internal standard was added, and the eluate was concentrated to ∼3 mL using an evaporator equipped with a vacuum controller and filled up with n-hexane to the 5 mL mark. A 2-mL aliquot was taken and was shaken vigorously with concentrated sulfuric acid. After separation of the two phases, which may be supported by a centrifuge, the supernatant was subjected to GC/MS analysis. The obtained limit of detection (LOD) for the PCBs was 0.3 ng/g for ASE and 0.1 ng/g for the conventional extraction techniques. The corresponding limit of quantitation (LOQ) was 1 ng/g for ASE and 0.4 ng/g for the conventional extraction techniques. Safety Considerations. PCBs are toxic compounds. Effective safety measures have to be taken. Do all laboratory mixing and 4052 Analytical Chemistry, Vol. 73, No. 16, August 15, 2001
RESULTS AND DISCUSSION The first experiments that were carried out were designed to evaluate the accelerated solvent extraction process itself without any fat retainer present in the extraction cell. This was done to avoid interpreting instrumental problems as fat retainer effects. Consequently, a certified reference material (cod liver oil CRM 349) was extracted without fat retainer present in the extraction cell. The raw extracts were cleaned after the extraction by adding sulfuric acid, followed by injection of the supernatant. The results from a triplicate extraction of the cod liver oil gave recoveries of 114, 100, 91, 97, 96, and 102% for PCBs 28, 52, 101, 118, 153, and 180 as compared to certified data. The RSDs never exceeded 5% for the individual congeners. It was, therefore, concluded that the method was capable of extracting PCBs from a fatty matrix when no fat retainer was present in the extraction cell. The first fat retainer experiments performed were designed to elucidate the amount of fat retainer needed to achieve a quantitative removal of the coextracted lipids. Three different combinations were chosen: 1 g of fat combined with 5 g ofretainer, 1 g of fat combined with 10 g of retainer, and 0.5 g of fat combined with 10 g of retainer. These combinations resulted in fat/fat retainer ratios of 0.2, 0.1, and 0.05. Florisil, however, has a much lower density; therefore, only half of the amount of fat and fat retainer was used to achieve the same fat/fat retainer ratio. All of the combinations were tested for all five of the retainers, and each experiment was performed in triplicate. The results from these investigations are presented in Figure 2. Prior to the experiments, the absence of the target analytes in the chemicals used in the study was confirmed by extracting 10 g of each fat retainer, 5 g of SFE support, and 5 g of sodium sulfate with ASE and subsequent treatment with sulfuric acid. The data shows that all of the retainers have very similar fatretaining properties and that when using a fat/fat retainer ratio of 0.05, the amount of coextracted fat is reduced to ∼5%. To check if a completely fat-free extract could be obtained, the fat/fat retainer ratio was decreased to 0.025. In this case, the amount of coextracted fat was below 1% and at the limit of detectability for the balance that was used. Consequently, to accomplish a complete fat removal, as much as 40 times the amount of retainer, as compared to the fat content, needs to be present in the thimble. To verify that the PCBs were quantitatively recovered when the fat retainer was present in the extraction thimble, all extracts presented in Figure 2 were analyzed for their PCB content. This demonstrated that all seven of the indicator PCBs were quantitatively recovered, with individual PCB recoveries in the range of 95-110% and RSDs between 2 and 12% (n ) 3). Because all of the fat retainers had equal performance characteristics when using PCB-spiked lard, all of the retainers were also tested on a naturally contaminated fish meal sample to determine whether they behaved differently when extracting realworld matrixes. Prior to this study, the PCB content in the fish meal sample was determined using two different standard extraction techniques in order to decrease the effect of a methodological bias on the estimation of the PCB concentration when only one extraction method was used. These involved Soxhlet extraction and cold column extraction, both used according to the CEN
Table 2. Recoveries of PCBs and Lipids Extracted from 2 g of Fish Meal Using 10 g of Various Types of Fat Retainers (1 g of Fish Meal with 5 g Florisil)a sulfuric acid
basic alumina
neutral alumina
acidic alumina
PCB
PCB recovery %
RSD %
PCB recovery %
RSD %
PCB recovery %
RSD %
PCB recovery %
RSD %
PCB recovery %
RSD %
52 101 118 153 138 180
130 117 115 107 85 90
7.7 6.4 2.4 6.9 1.2 1.2
123 110 133 109 95 95
9.4 9.1 4.2 1.5 2.7 1.9
130 112 124 112 89 91
0.1 6.0 2.3 4.9 0.7 0.1
120 116 128 110 89 97
8.3 2.2 6.7 5.7 5.9 1.1
133 116 119 117 91 96
8.7 2.2 2.4 9.5 3.4 4.0
extr fat, % color react with H2SO4 a
florisil
0.7
0.7
1.4
0.4
0.7
clear
clear
light brown
light brown
light brown
no
yes, weak
yes
yes
yes
Recoveries are based on the average of the two averages, applying soxhlet extraction and cold column extraction, as presented in Table 1.
guidelines.6 The raw extracts were treated with sulfuric acid, and the supernatant was injected into the GC/MS without any further sample cleanup. The results from these extractions are presented in Table 1. Both of the methodologies showed very similar PCB concentrations for all of the congeners, and the average of the two averages was used as an estimate of the true value in the fish meal. All of the PCB concentrations obtained from the fish meal extractions using fat retainer were compared with these values. Evaluation of the different fat retainers was performed by extracting 2 g of fish meal (containing 0.22 g of fat) with 10 g of fat retainer (5 g for the florisil with 1 g of fish meal) present in the extraction thimble, giving a fat/fat retainer ratio of 0.022. The results of the PCB recoveries, including fat recoveries, color of extracts, and reaction with sulfuric acid treatment, are presented in Table 2. The PCB recoveries that were obtained using the different fat retainers were found to be very similar, but the concentration of PCB 52 determined with ASE seemed to be enhanced, as compared to the traditional extraction techniques. However, the concentration level of this compound was close to the detection limit of 1 ng/g obtained for ASE, thereby resulting in a possible higher deviation from the target concentrations established by employing the Soxhlet and cold column extraction method. Therefore, the somewhat higher recovery of PCB 52 was interpreted as a methodological effect rather than as an increased extraction efficiency resulting from the use of ASE. PCB 28 could not be determined, because its concentration was below the LOQ for ASE, which was not the case for the Soxhlet method and the cold column extraction technique. Regarding coextracted fat, only basic alumina had a fat recovery over 1%, but in general, all of the fat retainers were capable of generating fat-free or nearly fatfree extracts (
