Anal Bioanal Chem (2004) 379 : 476–483 DOI 10.1007/s00216-004-2587-0
O R I G I N A L PA P E R
Federico Ferrari · Astrid Sanusi · Maurice Millet · Michel Montury
Multiresidue method using SPME for the determination of various pesticides with different volatility in confined atmospheres
Received: 11 September 2003 / Revised: 19 February 2004 / Accepted: 5 March 2004 / Published online: 15 April 2004 © Springer-Verlag 2004
Abstract An analytical method is described for assessing the vapour concentration of 11 pesticides (bioallethrin, chlorpyriphos methyl, folpet, malathion, procymidone, quintozene, chlorothalonil, fonofos, penconazole and trimethacarb) in confined atmospheres (e.g. a greenhouse after pesticide application). This study is a successful extension of a method previously developed by the authors for dichlorvos to much less volatile pesticides. Sampling was performed by using polydimethylsiloxane–solid phase micro-extraction (PDMS–SPME) fibres immersed in a 250-mL sampling flask through which air samples were dynamically pumped from the analysed atmosphere. After a 40-min sampling duration, samples were analysed by GC/MS. Calibration was performed from a vapour-saturated air sample. The linearity of the observed signal versus pesticide concentration in the vapour phase was proved from spiked liquid samples whose headspace concentrations were measured by using the proposed method. This procedure gave calibration curves with regression coefficients (R2) greater than 0.98, and the repeatability of these measurements was found with RSDs of 1.9–7.6%. As a field application test, this analysis procedure was used for the determination of gaseous procymidone concentrations as a function of time in the atmosphere of an experimental 8-m2 and 20-m3 greenhouse. The pesticide was sprayed according to real cultivation conditions, and measurements were made for 80 h after application (8 measurements). The observed concentrations found ranged from 200 to 500 µg m–3, thus indicating the level of contamina-
F. Ferrari · A. Sanusi · M. Montury (✉) Equipe Périgourdine de Chimie Appliquée LPTC, Université Bordeaux 1/CNRS UMR 5472, BP 1043, 24001 Périgueux Cedex, France Tel.: +33 (0) 5-53352429 e-mail:
[email protected] M. Millet Laboratoire de Physico-Chimie de l’Atmosphère CGS, Université de Strasbourg I, CNRS UMR 7517, 1 rue Blessig, 67084 Strasbourg Cedex, France
tion of the air breathed by people in such working conditions. Keywords Pesticide vapours · SPME/GC/MS analysis · Air sampling · Greenhouse atmospheres Abbreviations GC/MS gas chromatography/mass spectrometry · SIM selective ion monitoring · FC43 perfluorotributylamine · RSD relative standard deviation · LOD limit of detection · LOQ limit of quantification
Introduction Recent studies indicate that pesticide contamination is systematically invading all the segments of our biosphere [1]. This can be considered as a consequence of the evolution of human activities in our modern societies. The atmosphere is well known to be a good pathway for the dissemination of pesticides [2, 3] sometimes in zones far away from their emission sites (e.g. in remote Antarctic areas [4] or high-altitude mountain Lakes [5]). Actually, some papers dealing with the characterisation of atmospheric contamination by pesticides are now available. Different methods are used to evaluate the contamination of the atmosphere by pesticides and to describe processes by which these compounds are transferred into the atmosphere (e.g. vaporisation, spraydrift etc.) [4, 6, 7, 8, 9]. Nevertheless, measuring low levels of contaminant concentrations in air samples is still a real challenge for analytical chemists today [9, 10, 11]. Usual sampling consists of collecting high volumes of contaminated air that are filtered through an adapted solid-phase material or washed out with organic solvents. In a further processing step, analytes have to be desorbed and/or concentrated before being chromatographied [12], and this additional work up considerably increases duration and cost of the total analysis. Obtained data therefore give averaged values corresponding to extended sampling times (of at least several hours) and are recommended in the case of studies in
477
which time-weighted average methods (TWA) are used [13, 14]. In cases where the evolution of instant concentrations is required, these types of measurement are not ideal, and a real need exists for rapid methods able to afford more punctual measurements. The situation is different for pollution of confined atmospheres (e.g. greenhouse, stockroom etc.) where worker exposure to pesticides may be intensive during and after application [15, 16] despite recent efforts [17, 18]. In confined atmospheres, there is also a need for relatively rapid sampling methods in order to assess the real worker exposure risk [19]. As an example of the urgent need for such methods, the Common Acceptance Directive 91/414/EEC, which deals with the authorisation of plant protection products (pesticides) and their controlled use, requires that regulators from EU member States evaluate levels of worker exposure to pesticides during their intended use as part of the authorisation process. For sampling purposes, the so called solid-phase microextraction method (SPME), developed by Pawliszyn and his group [20], has proven to be particularly convenient for the determination of traces of organic pollutants in environmental samples. According to the thermodynamically controlled principle of partitioning, organic analytes contained in the analysed matrix accumulate onto the polymeric coating of a silica microfibre probe, which is directly immersed in the aqueous sample. After an optimised exposure time that can be even shorter than the full equilibration time [21], the fibre is withdrawn and desorbed into the injection port of the chromatographic system. This innovative method has been applied for the analysis of many types of water contaminants [22, 23] and was very rapidly and extensively used for the analysis of pesticide traces in all kinds of environmental matrices including waters [24, 25], aqueous solutions and suspensions [26, 27] and fruits and vegetables [28, 29]. Very recently, the authors also demonstrated that dichlorvos, a semi-volatile insecticide, was easily sampled by SPME from the confined atmosphere of a greenhouse and analysed at the level of few µg m–3 by GC/MS [30]. The objective of this paper was to enlarge the application field of the aforementioned analysis method to 11 other pesticides with much lower volatilities at room temperature. Thus, bioallethrin, chlorothalonil, chlorpyriphos methyl, cyanofos, folpet, fonofos, malathion, penconazole, procymidone, quintozene and trimethacarb were selected for this study, since their saturated vapour pressures ranged from 10–1 to 10–5 times that of dichlorvos, and also because they are intensively used to protect products grown in greenhouses. As a consequence, they can also be breathed by workers in the course of their professional activities.
tozene (99.5%) were supplied by Dr. Ehrenstorfer GmbH, Augsburg, Germany; chlorothalonil (98.5%), fonofos (98.0%), penconazole (99.3%) and trimethacarb (98.0%) were supplied by Riedel-de Haen AG, Seelze, Germany; cyanofos was supplied by Chem Service, West Chester, PA, USA. Each pure standard was stored at –18°C. Instrumentation All SPME fibres – polydimethylsiloxane (PDMS) 100 µm; polydimethylsiloxane–divinyl benzene (PDMS–DVB) 65 µm; carboxen– polydimethylsiloxane (CN–PDMS) 75 µm and polyacrylate (PA) 85 µm – were provided by Supelco (USA). A ThermoFinnigan GC/MS coupling system (Polaris ion-trap MS and GC-Q 2000) was used for analysing samples that were directly introduced into the GC injection port with the SPME manifold. The chromatographic column used was a DB-5MS (Supelco), 30-m long with a 0.25-mm ID and a 0.25-µm thickness. A locally commercially available aquarium-type electric pump (5 W, 2 L min–1) was used to generate a homogeneous gas flux through the sampling system. Application of procymidone to field samples was performed in an experimental 8-m2 and 20-m3 greenhouse. Pesticide treatments were applied onto strawberry cultivations according to usual professional procedures. Sample preparation All measurements were made by exposing the selected SPME fibre to air samples. According to the objectives of this study, three different assemblies were used. In the first one, corresponding to laboratory condition experiments (Fig. 1), the 250-mL sampling flask was connected to a 2.5-L glass bottle by using 5-mm-ID stainless steel pipes and an aquarium-type pump. Liquid or solid samples of pure pesticides were introduced into this bottle to obtain a vapoursaturated atmosphere that was regularly circulated throughout the whole assembly and especially the sampling flask. Sampling was performed by immersing SPME fibres into the sampling flask through a septum wrapped with an aluminium foil for a defined duration. In the second assembly (Fig. 2), which was used for assessing the linearity of the method in terms of obtained signals versus concentrations of the analyte in the vapour phase, extractions were carried out in the headspace mode from 5-mL vials containing spiked solutions of the pesticide mixture in water and equipped with a glass-coated stirring bar and a thermostatic bath. The choice of this assembly is due to the difficulty in generating gas-phase calibra-
Experimental Reagents Standards of pure bioalletrin (94%), chlorpyriphos methyl (99.0%), folpet (99%), malathion (99.5%), procymidone (99.7%) and quin-
Fig. 1 Laboratory assembly used for SPME samplings of pesticides vapours
478
Results and discussion Fibre selection
Fig. 2 Assembly used for studying the linearity of HS-SPME samplings
Based on the approach developed in the case of dichlorvos [30], the four most usual coatings, namely 100-µm PDMS, 65-µm PDMS–DVB, 75-µm CN–PDMS and 85-µm PA, were successively tested for extracting procymidone, which was considered as representative of the selected compounds with a saturated vapour pressure in the middle of the selected range (Ps=1.8×10–2 Pa at 25°C). Corresponding extraction profiles are presented in Fig. 5; these indicate that the 100-µm PDMS coating was the most sensitive fibre for extraction of procymidone. As a consequence and because this result was in agreement with those obtained for dichlorvos, 100-µm PDMS was selected for the following study. Extraction profile
Fig. 3 Proposed assembly for SPME air sampling tion samples of defined concentration, especially for pesticides that have very low vapour pressures. This approach is in accordance with the process model for the three-phase system involved in headspace solid-phase microextraction (HS–SPME) as described by Zhang and Pawliszyn [31]. Fibres are used to extract only insignificant portions of the target analytes in a given phase (HS or liquid) without affecting their distribution in the whole system. Based on external calibration, they can give the analyte concentrations in the phase of interest. So, measurements performed in the HS from perfectly agitated spiked aqueous solutions are representative of the gradient of their initial concentrations. In a third assembly, which was used for greenhouse measurements, the sampling flask was only connected through the output knob to the pump with a stainless pipe, and the input knob was just open to the greenhouse atmosphere (Fig. 3). According to most of the procedures described in the literature in this field, all sample vapours extracted by the fibre were considered as perfect gases (low concentration) and thus the influence of humidity has not been considered. After sampling, fibres were then withdrawn from the flask and directly introduced into the injector port of the GC system. In all experiments, stable conditions were reached before performing SPME (by performing several measurements at a stable level according to time). GC sample analysis Fibres were desorbed into the split/splitless insert of the GC at the temperature of 270°C. The He flow rate was fixed at 1.5 mL min–1, and the injector was used in the splitless mode for 3 min. The GC oven temperature was programmed as follows: 50°C (3 min), 25°C min–1 to 160°C, 8°C min–1 to 240°C, 150°C min–1 to 300°C (2 min). The transfer line was held at 250°C and the detector at 220°C. The MS was tuned to FC43 (perfluorotributylamine), and mass spectra were collected over the mass range (m/z) 50–650. Pesticide identifications were based on the comparison of their retention times with those of standard samples and their mass spectra (Fig. 4 and Table 1) obtained in the electron impact mode (EI). Quantification was made in the selective ion monitoring mode (SIM) by measuring peak areas of the fragment ions selected for each pesticide and by comparing their values with the corresponding external calibration curves described in Table 1.
To visualise the partition equilibria between the fibre and the atmospheric air samples for each of the 11 selected molecules (Table 1), extraction profiles were successively performed by analysing 12 samples (i.e. including a blank one) according to the procedure described above and represented schematically in Fig. 1 and with exposure times ranging from 3 to 193 min. The first and main observation made was that relative chromatographic peaks intensities were very high for all compounds (Table 1), even after few minutes of extraction. Nevertheless, several types of curves have been obtained as indicated in Fig. 6, in which pesticides have been gathered according to the order of magnitude of observed signals. Independently of this feature, chlorothalonil, cyanofos, malathion and folpet appeared nearly at equilibrium after 40–60 min of exposure to the PDMS fibre. In contrast, bioalletrin, fonofos, penconazole and quintozene, did not reach equilibrium within the 193 min of the experiment. The behaviours of chlorpyriphos methyl, procymidone and trimethacarb were different: quasi-linear extraction profiles were found over the duration of the experiment. This means that the 100-µm PDMS fibre can accumulate these compounds for a long time (i.e. at least approximately 200 min). Taking into account the above features, an extraction time of 40 min was chosen to obtain significant signals within reasonably short analysis durations for air samples. Repeatability A series of six measurements were performed from vapour saturated samples in the sampling flask according to the conditions described above (Fig. 1). Each extraction was performed for 40 min, and relative results are presented in Table 1. These showed RSDs ranging from 1.9% to 7.6%; four of the measurements were made on one day and the two others on the following day under the same experi-
479 Fig. 4a–e Chromatograms obtained in single ion monitoring mode (123+266 amu, 136+ 137+237 amu, 127+286 amu, 127+283 amu, 161+260 amu, for a–e (top to bottom), respectively); bold numbers refer to pesticides listed in Table 1
Table 1 Identification and repeatability Compound
Number Class
Bioallethrin 1 Trimethacarb 2 Quintozene 3 Cyanofos 4 Fonofos 5 Chlorothalonil 6 Chlorpyriphos methyl 7 Malathion 8 Penconazole 9 Folpet 10 Procymidone 11 am/z
Pyrethroid Carbamate Aromatic HC derivative Organophorphorus Organophorphorus Halocyanobenzene Organophorphorus Organophorphorus Azole N-Trihalomethylthio Dicarboximide
Molecular Vapour weight pressure (amu) Pa (25°C) 302 193 295 243 246 266 286 330 248 297 283
values used for identification of compounds are peaks corresponding to fragments used to quantify peak areas using the SIM mode
bValues
4.4×10–2 6.8×10–3 1.3×10–2 1.1×10–1 2.8×10–2 7.6×10–5 5.6×10–3 5.3×10–3 2.1×10–4 1.3×10–3 1.8×10–2
Identification fragment ionsa (amu)
Retention Observed time signalc (min) (au)
RSDc (%)
136, 123b, 91, 79 136b, 121, 91, 77 265, 249, 237b, 214 243, 127b, 125, 109 246, 137b, 109, 81 266b, 229, 168, 124 286b, 271, 210, 197 173, 127b, 143, 99 248, 161b, 159 297, 260b, 130, 104, 283b, 255, 96, 67
9.01 13.64 14.82 15.05 15.14 15.34 16.26 17.23 18.27 18.50 18.57
4.18 2.93 4.27 3.54 3.21 5.42 4.79 1.92 2.83 7.59 4.32
1,028,142 19,717,664 19,970,922 8,553,016 48,014,491 241,656 8,540,794 608,508 475,646 13,187 53,560
cAverage of 6 measurements from previously vapour-saturated HS samples extracted for 40 min as described in Fig. 1
480
Linearity and calibration
Fig. 5 Selection of the fibre: extraction profiles from saturated vapour air samples for procymidone by using 100-µm PDMS (◊), 85-µm Polyacrylate (V), 75-µm Carboxen-PDMS (–) and 65-µm PDMS-DVB (O) fibres
mental conditions. This particular experiment also proved that in all cases the saturated vapour pressure is renewed between two experiments, within less than 40 min, which was the time needed for chromatographic analysis of the preceding extracted sample.
Fig. 6 Extraction profiles of the 11 selected pesticides from saturated vapour air samples classified according to observed signal intensities: fonofos (O), trimethacarb (◊), quintozene (V), chlorpyriphos methyl (–), cyanofos (T), bioallethrin (x), malathion (+), penconazole (E), chlorothalonil ( ] ), procymidone (N), folpet (K)
Routine standard air samples containing pesticide vapours are not commercially available. In a first approach and to verify the linearity of observed signals versus pesticide concentrations in air samples under atmospheric pressure, a series of headspace samplings were carried out at 25°C according to the described protocol (Fig. 2). As was justified in the “Experimental”, the analyte concentration in the headspace is assumed to be proportional to its concentration in the aqueous solution regardless of the extraction temperature. Solutions of pesticides were then prepared according to the range indicated in Table 2 by diluting the most concentrated one with water, in the ratio 1, 0.75, 0.50 and 0.25, respectively. Observed signals were plotted against corresponding HS concentrations expressed in the same ratio. As a result of the very low vapour pressure of some of the compounds at very diluted concentrations, no signals were observed for them at this temperature. Another series was successfully analysed at 65°C for all the selected compounds. Figure 7 illustrates these results in the cases of bioallethrin, chlorothalonil, cyanofos and procymidone. Under these conditions, regression coefficients greater than 0.99 were found for all of the compounds except penconazole and cyanophos, for which values greater than 0.98 were obtained (curves were obliged to go through the origin as long as blank samples gave no signal). Once the linearity of the extracting method was verified, calibration curves relative to the sampling procedure indicated in Fig. 1 were realised by measuring a unique point corresponding to a definite and known concentration for each of the selected pesticides. This point was given by recycled saturated vapours obtained from a mix-
481 Table 2 Linearity and quantification Compound
Bioallethrin Trimethacarb Quintozene Cyanofos Fonofos Chlorothalonil Chlorpyriphos methyl Malathion Penconazole Folpet Procymidone aValues
Program Number
1 2 3 4 5 6 7 8 9 10 11
Linearity R2 (65°C)
R2 (25°C)
Conc. Calibration in saturated curve slopea Range of solution vapour phasea (au µg–1m3) concentration (µg L) (µg m–3)
0.997 0.990 0.997 0.98 0.998 0.989 0.998 0.999 0.988 0.998 0.992
0.999 – 0.994 0.98 0.989 0.98 0.98 – – – –
4.0–200 2.0–100 0.5–25 2.0–100 3.7–185 3.6–180 8.8–440 5.0–250 3.2–160 2.4–120 1.6–80
192 37200 13200 830 17200 29600 13200 861 22600 84 26
5,351 530 1,512 10,298 2.780 8.16 646 706 21.0 156 2,056
LODb (µg m–3)
LOQb (µg m–3)
10.4 0.11 0.11 2.41 0.17 0.03 0.11 1.74 0.09 5.91 76.7
18.2 0.23 0.22 6.26 0.32 0.07 0.22 3.83 0.19 15.3 172.7
obtained by Marriott law at room temperature condition (25°C) starting from vapour pressure listed in Table 1 limit of detection and quantification
bEstimated
ture of these compounds introduced into the large bottle used in the first assembly. Under these conditions, the effective concentration of each pesticide was calculated according to the Marriott’s law equation: 3V = &[ 57
(1)
where Cx is the concentration of the pesticide x in a saturated gas sample, T the absolute temperature (in K), R the perfect gas constant, and Ps is the corresponding equilibrium vapour pressure of this pesticide at this temperature. Slopes of the calibration curves thus established are indicated in Table 2 as are the limits of detection (and quantification) that have been estimated on the basis of a ratio higher than 3 (and 10) between the peak height and the background noise for each analyte, respectively. Under such conditions, these limits ranged from 0.03 for chlorothalonil to 77 µg m–3 for procymidone. Surprisingly, the lowest LOD (30 ng m–3) was found for chlorothalonil although its saturated vapour pressure was very low (7.6×10–5 Pa at 25°C). For comparison, the LOD for Cyanofos was found to be much higher (80 times), whereas its Ps was 1,500 times more important than that of chlorothalonil. From this observation, it appears that vapour pressure is not the predominant factor for explaining either the slopes of calibration curves or the observed detection limits. In fact, some very low volatile compounds (then with higher distribution level in the vapour phase) exhibit higher extraction efficiencies than others with higher vapour pressures. In other words, the adsorption of compounds onto the probe is not merely dependent on their concentration in the vapour phase. The corresponding partitioning coefficients between vapours and the polymeric solid phase must also be considered. Actually, pesticides are partitioned between the polymeric probe and the vapour phase according to their partitioning coefficients, respectively, and independently of their vapour pressures. Under these conditions, even a compound characterised by a low vapour pressure can accumulate onto the fibre mainly because the
corresponding partitioning coefficient largely favours the condensed form. As a consequence, the accumulation of the pesticide onto the fibre coating can be the result of an important transfer of compound from the sample to the probe through the vapour phase, independently of its concentration in this phase. Application to confined air samples in an experimental greenhouse In the course of this study, a validation experiment was realised with procymidone under real-life conditions in a small 8-m2 and 20-m3 experimental greenhouse in which strawberry cultivation was performed. Air concentrations of this pesticide was assessed according to the procedure described with the third assembly (Fig. 3) and plotted versus time after treatment in Fig. 8. Ambient sampling temperatures were also noted and indicated on the same diagram. During the 80 h of this experiment, 8 measurements were taken at the rate of twice a day, and procymidone concentration levels were observed between 200 and 500 µg m–3 in strong correlation with corresponding temperatures which varied between 10 and 20°C in the same period. These results are in agreement with those obtained by Sieber and Mattusch [32] for the assessment of parathion and pirimicarb vapours under similar conditions in greenhouse air samples. The main difference with the case of dichlorvos previously reported [30] is that no decrease of the procymidone concentration in air was observed according to the time spent after treatment. In fact, the concentration that is measured in this case, corresponds to vapours emitted by procymidone sprayed onto the vegetal matter but long after the sedimentation of the spraying aerosol has occurred. Under these conditions and even for a low-volatile pesticide, it is not surprising to observe concentration values mainly correlated to ambient temperatures. Nevertheless, a concentration of 324 µg m–3 was
482
Fig. 8 Observed air concentrations of procymidone (K) and corresponding air temperatures (T) in the greenhouse as a function of duration after treatment
Conclusion
Fig. 7a–d Linearity curves obtained by headspace SPME technique at 65°C (S) and 30°C (K) for bioallethrin (a), cyanofos (b), chlorothalonil (c) and procymidone (d). Relative concentrations in the HS are expressed in the ratio of the corresponding solutions
found 2 h after treatment, and this particular result exemplifies first the level of potential contamination that can be found in greenhouse, and second, the use that can be made of this method for assessing the amount of pesticide breathed by people working in the greenhouse.
The previously described method for assessing dichlorvos contamination of confined atmospheres has been successfully extended to 11 other pesticides selected from different chemical families with a large range of saturated vapour pressures. The observed efficiency of the method was mainly related to the ability of the PDMS fibre to accumulate the pesticide via an important transfer from the “infinite” sample of air onto the coating. No correlation appeared with the saturated pressure of the compounds. Because of this, the method should also be useful for the assessment of a large scope of other organic air pollutants. As was previously found for dichlorvos, which is considered as quite a volatile molecule, the observed limits of detection with this method are in the µg m–3 level and far below the concentration that can be found in real professional working conditions. In addition, sampling time is reduced to 40 min (i.e. about 10 times shorter than for other usual methods). This should be of real interest for assessing pesticide concentrations in air samples in many different situations, as long as the studied atmosphere is stable and considered as a homogenous medium. In all cases, the method is direct and solvent-free. It can therefore only reduce the global cost of analyses, which at present corresponds to a universal need. Acknowledgements The authors acknowledge the financial support from the Conseil Général de la Dordogne, the Centre National de la Recherche Scientifique and the European Community Marie Curie Training Site Program.
References 1. Oehme M, Haugen JE, Schlabach M (1996) Environ Sci Technol 30:2294–2304 2. Eisenrich SJ, Looney BB, Thorston JD (1981) Environ Sci Technol 15:30–38 3. Bidleman TF, Leonard R (1982) Atmos Environ 16:1099–1107 4. Sanusi A, Millet M, Mirabel Ph, Wortham H (2000) Sci Total Environ 263:263–277
483 5. Carrera G, Fernandez P, Gimault JO, Ventura M, Camarero L, Catalan J, Nickus U, Thies H, Psenner R (2002) Environ Sci Technol 36:2581–2588 6. Alegria HA, Bidleman TF, Shaw TJ (2000) Environ Sci Technol 34:1953–1958 7. Clément M, Arzel S, Le Bot B, Seux R, Millet M (2000) Chemosphere 40:49–56 8. Briand O, Millet M, Bertrand F, Clément M, Seux R (2002) Anal Bioanal Chem 374:848–857 9. Sanusi A, Millet M, Wortham H, Mirabel Ph (1997) Analusis 25:302–308 10. Lambropoulou DA, Albanis TA (2001) J Chromatogr A 922: 243–255 11. Murayama H, Mukai H, Mitobe H, Moriyama N (2000) Anal Sci 16:257–263 12. Garcia-Alonzo S, Perez-Pastor RM, Quejido-Cabezas AJ (2002) Talanta 57:773–783 13. Martos PA, Pawliszyn J (1999) Anal Chem 71:1513–1520 14. Khaled A, Pawliszyn J (2000) J Chromatogr A 892:455–457 15. Capri E, Alberici R, Glass CR, Minuto G, Trevisan M (1999) J Agric Food Chem 47:4443–4449 16. Cruz Marquez M, Arrebola FJ, Egea Gonzalez FG, Castro Cano ML, Martinez Vidal JL (2001) J Chromatogr A 939:79– 89 17. Elflein L, Berger Preiss E, Levsen K, Wünsch G (2003) J Chromatogr A 985:147–157
18. Yoshida T, Matsunaga I, Oda H (2004) J Chromatogr A 1023: 255–269 19. Koziel JA, Martos PA, Pawliszyn J (2004) J Chromatogr A 1025:3–9 20. Pawliszyn J (1997) Solid phase micro-extraction: theory and practice. Wiley-VCH, New York 21. Ai J (1997) Anal Chem 69:1230–1236 22. Yi H, Yan W, Kee LH (2000) J Chromatogr A 874:149–154 23. Doong RA, Chang SM, Sun YC (2000) J Chromatogr A 879: 177–188 24. Eisert R, Levsen KJ (1995) Fresenius J Anal Chem 351:555– 562 25. Correia M, Matos CD, Alves A (2000) J Chromatogr A 889: 59–67 26. Urruty L, Montury M (1996) J Agric Food Chem 44:3871– 3877 27. Hernandez F, Beltran J, Lopez FJ, Gaspar JV (2000) Anal Chem 72:2313–2322 28. Falqui-Cao C, Wang Z, Urruty L, Pommier JJ, Montury M (2001) J Agric Food Chem 49:5092–5097 29. Kataoka H, Lord HL, Pawliszyn J (2000) J Chromatogr A 880: 35–62 30. Sanusi A, Ferrari F, Millet M, Montury M (2003) J Environ Monit 5:574–577 31. Zhang ZY, Pawliszyn J (1995) Anal Chem 67:34–43 32. Siebers J, Mattusch P (1996) Chemosphere 33:1597–1607
