Optimisation of Pressurized Liquid Extraction for the Determination of Seven Selected Polychlorinated Biphenyls in Feed Samples 2005, 61, 391–396
C. von Holst1,&, A. Mu¨ller3, F. Serano1, S. Sporring2, E. Bjo¨rklund2 1
European Commission, DG Joint Research Centre, Institute for Reference Materials and Measurements, Retieseweg, 2440, Geel, Belgium; E-Mail:
[email protected] 2 Department of Analytical Chemistry, Lund University, P.O. Box 124, 22100, Lund, Sweden 3 European Commission, DG Joint Research Centre, Institute for Environment and Sustainability, 21020, Ispra, Italy Received: 8 October 2004 / Revised: 4 February 2005 / Accepted: 14 February 2005 Online publication: 30 March 2005
that feed was accidentally contaminated by PCB oil present in recycled fat, which regularly was used in feed production [2]. The accident caused increased levels of PCBs and Dioxins in pork and chicken meat, sometimes exceeding the tolerable level (set by the European Commission) as much as 250 times [1]. Consequently feed is also of major importance in the assurance of safe food and is today given increased attention as seen from recent research
examining the effects of dietary PCB content on animal welfare [3]. Fishmeal is a high quality ingredient used in for example poultry and pig feeds, but is also being used in the growing market of farmed fish. Typical feed for fish can contain 30 to 50% of fishmeal [4 and references therein]. Unfortunately, fishmeal frequently contains increased levels of POPs [4–7]. Another important ingredient in feed is fish oil, which often is added to feed in the order of 10–30% [4 and references therein]. The reason for this addition is the positive correlation between the total lipid content in cultivated fish and the feed lipid content, and consequently the aquaculture industry supports the addition of fish oil to promote fish growth and increase the nutritional value to humans [8–11]. However, also fish oil is known to contain rather high levels of POPs [4, 6, 7, 12], and this may lead to high levels of POPs in farmed fish [4, 7, 13, 14]. This study deals with the extraction of both fishmeal and fish oil, as these are common sources to PCB exposure from feed. Today there is a great need for new extraction techniques which can run several samples unattended with shortened extraction times and reduced organic solvent consumption [15–17] such as microwave-assisted extraction (MAE), supercritical fluid extraction (SFE) and Pressurized Liquid Extraction (PLE). The similarity between these techniques is the possibility of working at elevated temperatures and pressures, which drastically improves the speed of
Chromatographia 2005, 61, April (No. 7/8)
391
Abstract Pressurized Liquid Extraction was utilized for the extraction of seven selected polychlorinated biphenyls (28, 52, 101, 118, 153, and 180) from a naturally contaminated fishmeal and two feed samples fortified with a naturally contaminated fish oil sample. In order to assure sufficient extraction efficiencies, the extraction solvent, the extraction temperature, and the flush volume were optimised by a factorial design approach. The results of the analyses revealed that the impact of these parameters on the extraction of PCBs differed depending on which matrix that was analysed. For fishmeal, an elevated extraction temperature was important to obtain the highest values for the recovery rates whereas for the feed samples high extraction efficiencies could be obtained for all temperatures investigated in the study. In addition, the solvent had an impact on the extraction of PCBs, however, the influence was less pronounced than the impact of temperature. The final conditions, resulting in high recovery rates for all PCBs in all matrices, were found to be temperatures above 100°C using n-heptane as extraction solvent, while the flush volume had very limited effects on the extraction efficiency.
Keywords Gas chromatography Pressurized liquid extraction Polychlorinated biphenyls Feed stuffs fishmeal
Introduction Despite the observed decrease in environmental polychlorinated biphenyl (PCB) levels during the last two decades, there is still a great need for risk assessment of these compounds in various foodstuffs. The great risks associated with persistent organic pollutants (POPs) exposure through food were obvious in the Belgian dioxin crisis [1]. However, in this case it was revealed
Original DOI: 10.1365/s10337-005-0526-7 0009-5893/05/04
Ó 2005 Friedr. Vieweg & Sohn/GWV Fachverlage GmbH
the extraction process. A number of these techniques are useful also for the extraction of PCBs in food products and similar matrices [18]. One of the most recent techniques on the market is PLE which is successfully applied in various fields of POP analysis [17, 19]. Until now no investigation of suitable extraction conditions for PCBs in feed using PLE has been performed. However, Chen and co-workers [20] optimized the extraction of spiked chlorinated pesticides in pig feed, studying the effects of extraction temperature, time and solvent composition. From this study, it was concluded that a temperature of 100°C for 9 min in 2 cycles with a solvent composition of n-hexane/acetone (3:2, v/ v) was the best choice. This methodology was later compared to other extraction techniques, by the same research group, and PLE was there found to be the superior technique due to increased extraction efficiency, decreased solvent consumption, and shortened extraction times with a large degree of automation [21]. The aim of the work presented in this paper was to establish what extraction parameters of the PLE instrument that significantly influence the extraction efficiency, expressed in terms of recovery rate, when analysing PCBs in various feed matrices. In order to investigate the effects on the recovery rate a factorial design was utilized. n-Hexane has been substituted with n-heptane due to it’s lower neurotoxicity [22]. The PCBs selected are the most common measured indicator PCBs which are PCB 28, 52, 101, 118, 138, 153 and 180.
Experimental Reagents and Chemicals Sodium sulphate (AnalaRÒ) was obtained from BDH (Poole, England). Sulphuric acid (95-98%), sea sand (Pro Analysis), n-hexane and acetone (SupraSolvÒ, organic trace analysis) were all purchased from Merck (Darmstadt, Germany). The sodium sulphate and the sea sand were heated to 500°C for 6 h before usage. Isco Inc. (Lincoln, Nebraska, USA) delivered SFE Wet support, and cellulose filters for extracting cell caps came from Dionex Corp. (Sunnyvale, California, USA).
392
An internal standard solution with a concentration of 1000 ng/mL was prepared by dissolving PCB 169 in n-hexane. This standard was added (between 50 and 100 lL) directly to the PLE extracts in all experiments. All PCB congeners were from Dr. Ehrenstorfer GmbH (Augsburg, Germany). Quantitation was based on a threepoint calibration curve in the concentration interval 2.5–10 ng mL1 with PCB 169 added at 10 ng mL1 .
Samples The naturally contaminated fishmeal was supplied by State Official Laboratory (ROLT, Tervuren, Belgium). Many commercial feedingstuffs contain fishmeal as a major ingredient, in which the PCBs have been incorporated into the matrix via the food chain. Consequently the PCBs in fishmeal are less accessible than PCBs added to feeding stuff during or after production, making fishmeal a good choice of matrix for method performance verification purposes. The assigned values of the PCB concentration (Table 1) were established in a European proficiency test in which 27 laboratories from 15 Member States participated [23]. The fat content was determined to be 11.3% in the proficiency test, which was verified in-house applying PLE with n-hexane/acetone (1:1, v/v) at 100°C in 2 cycles a´ 5 min, giving a fat content of 11.4% (RSD ¼ 1.6%, n ¼ 3). In addition, an industrially produced compound feed samples fortified with cod liver oil containing the target analytes were included in the study. This sample also reflects real world conditions since in the production of compound feed fat and/or oil are added to the feed for nutritional reasons. As desribed elsewhere [1] PCBs can enter the food and feed chain via this process if the oil used in the formulation of feed has been previously contaminated with PCBs. The samples were commercially available feed for poultry ‘‘Becco Giallo’’ (Raggio di Sole Mangimi S.p.A., Italy) and vegetable cattle feed prepared in the laboratory using typical feed ingredient. Analysing these blank samples revealed that the material did not contain PCBs from other sources. Prior to extraction, the feed materials were fortified with a naturally contaminated cod liver oil which previously had been used in a proficiency test organized by Food Analysis Performance Chromatographia 2005, 61, April (No. 7/8)
Assessment Scheme (FAPAS) of the Central Science Laboratory DEFRA (Department for Environmental Food Rural Affairs) UK. Characteristics of the cod liver oil including the target values of the PCBs are described in the FAPAS report [24]. The main ingredients of ‘‘Becco Giallo’’ were maize, soybean, wheat and maize gluten, with a total fat content of 3% according to the producer of the feed. The vegetable cattle feed contained some 13 ingredients e.g. wheat, citrus pulp, molasses, minerals etc. with a fat content of 4.0%. The fat content of the two feeds were also determined inhouse applying PLE with n-hexane/acetone (1:1, v/v) at 100°C in 2 cycles a´ 5 min. It was then revealed that the fat content was 4.00% (RSD ¼ 1.6%, n ¼ 3) and 3.98% (RSD ¼ 1.3%, n ¼ 3) for the feed for poultry and vegetable cattle feed, respectively. The spiking procedure of the two feed matrices was performed by taking 20 g of cod liver oil and diluting it in n-hexane to a total volume of 25 ml. An aliquot of 0.5 ml (corresponding to 400 mg of fish oil) from this volume was added to each individual feed sample of 2 g and mixed. Before extracting the samples, the mixture was dried overnight under a hood to evaporate the n-hexane. This spiking procedure had been successfully applied in a previous study demonstrating that the procedure did not introduce a significant error [23]. The target concentrations of the feed samples and the corresponding standard error of the PCBs in the feed samples were calculated using the results of the cod liver oil obtained in the intercomparison study [24] and the dilution factor of 6 due the spiking of 2 g feed matrix with 400 mg of cod liver oil, as shown in Table 1.
Extraction and Clean-up The extractions were performed on an ASEä200 Accelerated Solvent Extraction System (Dionex, Sunnyvale, CA, USA). In this study, the influence of three factors on the recovery rate was investigated by applying a factorial design approach as described below. In all cases a 5 min static extraction was performed in 2 cycles. However, the principal outline of the experiments was as follows: The extraction cell was filled with 2 g of matrix mixed with the same amount of sand/ Original
SDffiffi Table 1. Target PCB concentrations of the investigated matrices. SE ¼ p , SD = Standard n deviation of the PCB concentration obtained in the intercomparison studies for fishmeal [23] and for the cod liver oil [24] used to spike the feed matrices
PCB
28 52 101 118 153 138 180
Fishmeal
Feed for poultry/vegetable cattle feed
Concentration (ng g)1)
Standard error (SE) (ng g)1)
Concentration (ng g)1)
Standard error (SE) (ng g)1)
0.56 1.31 2.76 4.07 13.5 9.55 4.32
0.10 0.12 0.19 0.31 0.61 0.66 0.28
1.14 3.12 7.18 6.25 13.6 13.4 4.37
0.095 0.20 0.64 0.29 0.68 0.83 0.22
sodium sulphate (1:1, w/w). The empty volume of the cells was filled with SFE support. The extraction of the samples was performed applying the various conditions according to the factorial design. After the extraction, internal standard PCB 169 was added and the extract was evaporated to about 1 mL. Prior to GC-MS analysis the extraction solution was cleaned-up by adding concentrated sulphuric acid to the extracts and shaking the samples with a Vortex, followed by centrifugation and injection of the organic solvent supernatant. Details of the method are given elsewhere [25] and also the suitability of the clean-up procedure for the analysis of feed samples has previously been demonstrated in the same study. Gas Chromatographic Analysis
The analyses of the PCB congeners were carried out on a GC 6890 (Agilent Technologies, Waldbronn, Germany) equipped with a HP MSD 5973 and a HP 5 MS capillary column (length 30 m, column ID 0.25 mm, film thickness
0.25 lm) All quantifications were based on the added internal standard (PCB 169). The mass spectrometer was operated in SIM mode and the following masses were measured for each chlorination level of the analysed PCBs: Molecule mass (M) and M + 2 for PCBs 28 and 52; M + 2 and M + 4 for PCBs 101, 118, 138, 153, 180, and 169. The oven temperature was programmed as follow: initial temperature 85°C, held for 1 min; programmed to 300°C at 8°C/min and held for a further 5 min.
Factorial Design A factorial design was utilized to investigate the impact of the following factors on the recovery rate in each matrix, respectively: Extraction temperature set at three levels: 50°C, 100°C, and 150°C. Two different extraction solvents: n-heptane (Hept) and a mixture of n-heptane and acetone (1:1, v/v), (Hept/Ac).
Flush volume set at two levels: 50% and 150%. The design required 12 experiments to be carried out reflecting different combinations of the factors. The statistical significance of the impact of the factors was evaluated by comparing the observed effects with the analytical error for each PCB congener that occurs when determining the recovery rate several times applying identical extraction conditions. The standard deviation of the analytical error was calculated from the duplicated analyses of all parameter combinations of the factorial design (12 experiments). All 24 experiments were performed in a randomised order. The statistical analysis of the results was performed using STATISTICAä software (Stat Soft Inc., USA).
Results Fishmeal The results for the fishmeal are seen in Table 2. For all PCBs, a high extraction temperature increased significantly the recovery rate. A remarkable outcome of the statistical analysis is also, that the experiments carried out at elevated temperatures often gave recoveries above 100%. This is clearly seen in Fig. 1. Evaluating the effect of the temperature on the individual PCBs revealed that recoveries above 100% were obtained for the less volatile PCBs. For instance, the recovery rate at 150°C for PCB 138 was 120% and for PCB 180 112%, whereas
Table 2. Results from the experimental design for fishmeal expressed in terms of mean values of the duplicate experiments performed on each parameter combination. PCB 28 has been excluded since it is below the limit of quantification of the analytical method. Exp. No.
Solvent
Temp. (°C)
Volume (%)
1 Hept/Ac 50 50 2 Hept/Ac 50 150 3 Hept/Ac 100 50 4 Hept/Ac 100 150 5 Hept/Ac 150 50 6 Hept/Ac 150 150 7 Hept 50 50 8 Hept 50 150 9 Hept 100 50 10 Hept 100 150 11 Hept 150 50 12 Hept 150 150 Relative standard deviation due to analytical error (%)
Original
Recovery rate (%) of each PCB congener (n = 2)
Average
52
101
118
153
138
180
(7 PCBs)
75 79 74 80 92 83 73 74 72 76 84 82 4.0
87 92 95 96 105 98 82 83 85 88 91 92 3.6
93 89 97 96 110 99 92 95 97 98 103 104 3.0
99 96 104 103 113 106 98 99 100 103 105 107 3.2
111 107 116 115 128 119 107 110 111 115 117 120 3.7
103 98 107 106 118 110 102 103 104 106 108 111 3.5
95 94 99 100 111 102 93 94 95 98 101 103
Chromatographia 2005, 61, April (No. 7/8)
393
robustness of the PLE method, since modifications of the extraction conditions did not jeopardise the extraction efficiency.
Vegetable Cattle Feed
Fig. 1. Effects of solvent and temperature on PCB recoveries for fishmeal (incurred sample). The results in each bar is the average recovery from four experiments at a constant temperature and solvent but with different flush volumes, using the average recovery for the different PCBs as an estimate of the recovery in a single experiment (error bars indicate the standard error, n=4.). For fishmeal, PCB 28 was excluded since it was below the limit of quantification
the corresponding recovery rates for PCB 101 were 96% and for PCB 118 104%. Fig. 1 also shows that the addition of a polar solvent to the extraction solution has a positive effect on the recovery rate that turned out to be significant only for PCBs 52 and 101. Therefore, it is important to note that the extraction temperature had a more pronounced effect on the recovery rate than did the composition of the extraction solvent. An interaction between the solvent and the flush volume was shown for PCB 118, 153, 138, 180 since the n-heptane values were somewhat low at a flush volume of 50%.
Feed for Poultry The results for the feed for poultry are presented in Table 3. The only factor that turned out to be significant was the solvent. For all PCBs,
pure n-heptane gave a better recovery than the mixture n-heptane/acetone, though the strength of the effect of the solvent on the recovery rate differed amongst the PCB congeners. For instance, the recovery rate of PCB 28 was 86% when using n-heptane/acetone compared to 102% when extracting with n-heptane. In the case of PCB 180 this effect was less pronounced, since a recovery rate of 101% was obtained with n-heptane/acetone, which was only slightly decreased compared with the trials using n-heptane, in which 104% were obtained. The experiments did not reveal a significant effect of the temperature on the recovery rate of the PCBs. This is more clearly seen in Fig. 2. Looking at the overall variability of the obtained results revealed (Table 3) that all experiments with the exception of trial 6 gained high values for the pooled recovery of all PCBs, since the values were at least 90% or higher. This also hinted at the
The results for the vegetable cattle feed are seen in Table 4. The statistical evaluation revealed that only for PCB 52 the effect of the solvent was significant. The recovery rate obtained with n-heptane was 77% compared to 70% when using Hept/Ac. Surprisingly, for the other PCBs, the interaction of the solvent and the temperature and not the solvent was significant. The interaction was observed, since at 50°C n-heptane/acetone gave better results whereas the opposite applied when extracting at higher temperatures. The obtained recovery rate was somewhat decreased for PCB 28 and PCB 52, but acceptable values for the recovery rate of the other PCBs were achieved. The weak influences of the extraction temperature and the composition of the solvent are also shown in Fig. 3.
Discussion The results from fishmeal are similar to those observed by Chen et al. for chlorinated pesticides [20]. Relatively high temperatures are needed to extract the analytes from the matrix. However, the positive effect of a polar solvent composition on the extraction efficiency was less strong. The fact that the experiments conducted at 150°C resulted in concen-
Table 3. Results from the experimental design for feed for poultry expressed in terms of mean values of the duplicate experiments performed on each parameter combination Exp. No.
Solvent
1 Hept/Ac 2 Hept/Ac 3 Hept/Ac 4 Hept/Ac 5 Hept/Ac 6 Hept/Ac 7 Hept 8 Hept 9 Hept 10 Hept 11 Hept 12 Hept Relative standard deviation
394
Temp. (°C)
50 50 100 100 150 150 50 50 100 100 150 150 due to analytical
Volume (%)
50 150 50 150 50 150 50 150 50 150 50 150 error (%)
Recovery rate (%) of each PCB congener (n = 2)
Average
28
52
101
118
153
138
180
(7 PCBs)
94 82 94 86 82 82 122 93 100 97 99 100 11
83 85 84 84 86 86 101 99 96 95 96 96 2.4
92 91 91 91 92 89 100 98 97 97 100 100 1.9
91 91 91 90 92 90 98 94 94 94 95 97 1.5
92 92 92 92 92 89 97 95 93 94 96 96 1.4
91 90 91 90 91 89 97 95 94 95 96 97 1.3
103 101 102 101 101 99 104 105 103 103 104 106 1.4
92 90 92 91 90 89 103 97 97 96 98 99
Chromatographia 2005, 61, April (No. 7/8)
Original
Fig. 2. Effects of solvent and temperature on PCB recoveries for feed for poultry (fortified sample). The results in each bar is the average recovery from four experiments at a constant temperature and solvent but with different flush volumes, using the average recovery for the different PCBs as an estimate of the recovery in a single experiment (error bars indicate the standard error, n ¼ 4.)
trations even higher than the target concentrations could indicate that the PLE procedure applying this temperature is somewhat harsher than any of the conditions applied by the different laboratories in the intercomparison study for this fishmeal [22] and that the values in that study might be underestimated. The result for the feed for poultry is different from the results obtained with fishmeal. The fact that n-heptane gave quantitative recoveries is logical since the PCBs in the fish oil spiked to the feed for poultry should be rather easily accessible. More striking is the fact that the presence of the polar solvent had a negative impact on the results. The difference in extraction behaviour between the fishmeal and the feed for poultry reflects the way the PCBs were introduced into the matrix (natural or spiked). Frequently the contamination of feed with PCBs occurs due to a sort of spiking procedure and for these situations, pure n-heptane might be
sufficient to extract out the analytes. In the specific case of feed for poultry the high polarity of the solvent even seem to hinder the extraction. In contrast, for naturally contaminated material addition of acetone gives slightly better values since the polar solvent probably aids in extracting more hardly bound analytes by penetrating polar region of the matrix. However, the results also indicated that a high extraction temperature in combination with pure n-heptane could gain a similar extraction efficiency. In the case of vegetable cattle feed the influence of the different factors in general was very low, and reasonable recoveries could be obtained with nearly all conditions. In all experiments the flush volume seems of low importance and the commonly applied flush volume of 60% is most likely sufficient. When analysing feed samples of unknown provenance the way in which
PCBs are incorporated in the matrix cannot be established prior to analysis. Suitable extraction conditions should therefore result in acceptable recovery rates of the PCBs irrespective of the matrix type. Based on the results of the current study the preferred solvent would be n-heptane in combination with an extraction temperature of 150°C as applied in trials 11 and 12. Comparing the results obtained in trial 11 and 12 of all matrices with the results from the other trials showed that this parameter combination gained very high extraction efficiencies of the target analytes.
Conclusion The results indicated that PLE allows for high extraction efficiency of PCBs from various feed matrices irrespective of the extraction conditions applied. However, the extraction should be conducted at elevated temperatures above 100°C, especially when dealing with samples in which the PCBs have been incorporated in the matrix under natural conditions. This optimisation also demonstrated that including different types of matrices – preferably naturally contaminated – is important when investigating the impact of extraction parameters on the extraction efficiency.
Acknowledgement This work was part of the DIFFERENCE project (no. G6RD-CT-200100623) funded by the European Commission.
Table 4. Results from the experimental design for vegetable cattle feed expressed in terms of mean values of the duplicate experiments performed on each parameter combination Exp. No
Solvent
1 Hept/Ac 2 Hept/Ac 3 Hept/Ac 4 Hept/Ac 5 Hept/Ac 6 Hept/Ac 7 Hept 8 Hept 9 Hept 10 Hept 11 Hept 12 Hept Relative standard deviation
Original
Temp. (°C)
50 50 100 100 150 150 50 50 100 100 150 150 due to analytical
Volume (%)
50 150 50 150 50 150 50 150 50 150 50 150 error (%)
Recovery rate (%) of each PCB congener (n=2)
Average
28
52
101
118
153
138
180
(7 PCBs)
88 95 84 69 75 55 74 75 81 69 90 84 13
66 72 68 79 70 67 79 73 77 79 79 78 3.2
84 88 86 86 87 82 88 83 89 86 88 89 2.7
88 90 86 87 89 85 87 83 89 87 89 90 2.4
91 92 92 90 92 88 91 87 94 90 92 93 2.4
89 90 89 88 90 86 88 85 91 88 89 91 2.0
110 113 110 108 112 107 107 105 112 108 110 112 2.7
88 91 88 86 88 81 88 85 90 87 91 91
Chromatographia 2005, 61, April (No. 7/8)
395
Fig. 3. Effects of solvent and temperature on PCB recoveries for vegetable cattle feed (fortified sample). The results in each bar is the average recovery from four experiments at a constant temperature and solvent but with different flush volumes, using the average recovery for the different PCBs as an estimate of the recovery in a single experiment (error bars indicate standard error, n ¼ 4.)
References 1. Bernard A, Hermans C, Broeckaert F, De Poorter G, De Cock A, Houins G (1999) Nature 401:62–70 2. Bernard A, Broeckaert F, De Poorter G, De Cock A, Hermans C, Saegerman C, Houins G (2002) Environ Research Sect A 88:1–18 3. De Vos S, Maervoet J, Schepens P, De Schrijver R (2003) Chemosphere 51:7–11 4. Horst K, Lehmann I, Oetjen K (1998) Chemosphere 36:2819–2832 5. Rappe C, Bergek S, Fiedler H, Cooper KR (1998) Chemosphere 36:2705–2720
396
6. Eljarrat E, Caixach J, Rivera J (2002) Chemosphere 46:1403–1407 7. Jacobs MN, Covaci A, Schepens P (2002) Environ Sci Technol 36:2797–2805 8. Bell JG, McEvoy J, Webster JL, McGhee F, Millar RM, Sargent JR (1998) J Agric Food Chem 46:119–127 9. Lyndon AR (1999) Mar Biotechnol 1: 376– 379 10. Martin SAM, Vilhelmsson O, Me´dale F, Watt P, Kaushik S, Houlihan DF (2003) Biochim Biophys Acta 1651:17–29 11. Turchini GM, Mentasti T, Caprino F, Panseri S, Moretti VM, Valfre´ F (2004) J Appl Ichtyol 20:71–75
Chromatographia 2005, 61, April (No. 7/8)
12. Jacobs MN, Santillo D, Johnston PA, Wyatt CL, French MC (1998) Chemosphere 37:1709–1721 13. Fiedler H, Cooper KR, Bergek S, Hjelt M, Rappe C (1997) Chemosphere 34:1411– 1419 14. Jacobs M, Ferrario J, Byrne C (2002) Chemosphere 47:183–191 15. Sparr Eskilsson C, Bjo¨rklund E (2000) J Chromatogr A 902:227–250 16. Turner C, Sparr Eskilsson C, Bjo¨rklund E (2002) J Chromatogr A 947:1–22 17. Bjo¨rklund E, Nilsson T, Bøwadt S (2000) Trends Anal Chem 19:434–445 18. Ahmed FE (2003) Trends Anal Chem 22:170–185 19. Saito K, Sjo¨din A, Sandau C, Davis M, Nakazawa H, Matsuki Y, Patterson D (2004) Chemosphere 57 (5):373–381 20. Chen S, Gfrerer M, Lankmayr E, Quan X, Yang F (2003) Chromatographia 58:631– 636 21. Gfrerer M, Chen S, Lankmayr E, Quan X, Yang F (2004) Anal Bioanal Chem 378:1861–1867 22. Takeuchi Y, Ono Y, Hisanaga N, Kitoh J, Sugiura Y (1980) Br J Ind Med 37:241–247 23. von Holst C, Mu¨ller A (2001) Fresenius J Anal Chem 371:994–1000 24. FAPASÒ Series 6, round 12 (September 2002), Environmental Contaminants, Report 0612, Central Science Laboratory, Sand Hutton, York YO41 1LZ (UK) 25. von Holst C, Mu¨eller A, Bjo¨rklund E, Anklam E (2001) Eur Food Res Technol 213:154–160
Original