Pressurised hot water extraction–microporous membrane liquid–liquid extraction coupled on-line with gas chromatography–mass spectrometry in the analysis of pesticides in grapes

Abstract

Pressurised hot water extraction–microporous membrane liquid–liquid extraction was coupled on-line with gas chromatography–mass spectrometry (PHWE–MMLLE–GC–MS) for the analysis of pesticides in grapes. MMLLE serves as a trapping step after PHWE. Water from PHWE is directed to the donor side of the membrane unit and the analytes are extracted to the acceptor solution on the other side. The role of MMLLE is to clean and concentrate the extract before on-line transfer to the GC via a sample loop and an on-column interface using partially concurrent solvent evaporation. The extraction conditions were investigated, and then the quantitative features such as linearity, limit of quantification (LOQ), extraction yield and enrichment factors. LOQs in the range 0.3–1.8 µg kg⁻¹ were achieved. Procymidone and tetradifon were found in the skins of the grapes. The results were in good agreement with those obtained by liquid–solid and ultrasonic extractions.

---

1. Introduction

The extraction techniques utilised in many standardised analytical procedures for determining pesticides in food and environmental matrices are tedious, time and solvent consuming, and waste producing. Many of the solvents are toxic, flammable and expensive. New approaches for sample preparation are thus continuously being sought. Promising extraction techniques of current interest are automated solvent extraction, sonication extraction, microwave-assisted solvent extraction (MASE), supercritical fluid extraction (SFE), pressurised hot water extraction (PHWE),^1–5 and pressurised fluid extraction with organic solvents.^6,7

Sample preparation techniques can also be connected on-line to chromatographic techniques. Coupling of extraction and analysis offers several advantages, and many of the problems associated with more traditional approaches can be avoided. The analysis is typically faster, less solvent is needed, and the cost of analysis decreases. The reliability and repeatability of the analysis are improved as well, since the analysis and sample clean-up take place in a closed, usually automated system, and the risks of sample loss and contamination are decreased.

PHWE has proven to be well suited for the extraction of a variety of samples, including soil and sediment^8–10 and plant materials.^11–13 Pressurised liquid extraction with acidified water has been applied to the extraction of anthocyanins and total phenolics from red grape skin¹⁴ and for the determination of pesticides in soil and sediments.^15–18 Water is an appealing choice for the extraction medium because it is cheap and non-toxic, and its physico-chemical properties can easily be adjusted by temperature. The properties of water are significantly altered at high temperatures, which leads to dramatic increase in the solubility of less polar compounds.¹⁹ After the extraction the analytes are usually trapped in a solid-phase trap or solvent. The trapping to solvent can also be done in an automated way with a membrane extraction technique, such as solid-phase microextraction (SPME) or the microporous membrane liquid–liquid extraction (MMLLE) technique employed here.

Microporous membrane liquid–liquid extraction (MMLLE) has the potential for extracting, cleaning and concentrating aqueous samples;^20,21 it can be used in PHWE in place of a solid-phase trap and as an interface connecting PHWE and gas chromatography.^22,23 MMLLE allows continuous liquid–liquid extraction to be made in a closed system. The aqueous sample is pumped to the donor side of the microporous membrane, while the organic acceptor on the other side remains stagnant. The pores of the hydrophobic membrane are also filled with the organic solvent. MMLLE eliminates many of the problems of classical continuous liquid–liquid extraction, such as the need for phase separators and segmentors. The donor and acceptor phases are not mixed and the mass transfer between the phases takes place on the membrane surface. In addition to the liquid–liquid extraction, also size exclusion takes place, which increases the selectivity of the extraction. Appropriate choice of the material of the membrane and its pore size can thus be used to advantage. In previous studies, we applied PHWE–MMLLE–GC in the determination of PAHs in soil and sediment.^22,23 MMLLE–GC systems have also been applied in the determination of pesticides in liquid samples, such as water and wine.^24,25

In this work, we developed a PHWE–MMLLE–GC–MS method for the determination of pesticides in grapes. The effects of temperature and extraction time were investigated, and quantitative features such as the linearity, repeatability and the limits of detection and quantification compared with results obtained by liquid–solid and ultrasonic extractions. As well, recovery and enrichment in PHWE–MMLLE factors were studied. The method was then applied to the analysis of pesticides in green grapes. Procymidone was found in the skins in the concentration of 0.040 ± 0.008 mg kg⁻¹ and traces of tetradifon were also detected.

2. Experimental

2.1. Reagents and samples

All solvents were of high-performance LC quality. Toluene was from Lab Scan Analytical Sciences (Dublin, Ireland). Water was distilled and deionised. Pesticide standards included lindane, vinclozolin, quinalphos, procymidone, endosulfan sulfate and tetradifon. These pesticides were selected because they had been detected in wine in earlier studies.²⁵ They were purchased from Accustandard (New Haven, USA). Two internal standards, diphenylamine (extraction standard) from Merck (Darmstadt, Germany) and 1,1′-binaphthyl (GC-MS standard) from Acros Organics (New Jersey, USA), were employed. Stock solutions (1 mg ml⁻¹) of pesticides were in isooctane or toluene and were diluted via isopropanol to water. A 10 µg l⁻¹ solution of pesticides was prepared in deionised water.

Green grapes purchased from a local supermarket (Helsinki) were used as samples. Samples were analysed unwashed. The sample amount was 50 mg for both skin and pulp, except for ultrasonic and liquid–solid extraction where the amount was 500 mg (skin). The grape skins and pulp were crushed in a mortar. The samples were always weighed as wet, before possible drying.

2.4. Apparatus

The PHWE–MMLLE–GC apparatus has been described earlier.²² The PHWE system consisted of a GC oven (HP 5790A, Palo Alto, CA, USA) and a Jasco PU-980 pump (Tokyo, Japan). Special laboratory-made high-temperature vessels of stainless steel (volume 2.8 ml, id 10 mm) were used in the extractions.²⁶ Connections between pumps, valves and extraction vessel were made of 1/16″ stainless steel tubing of 0.5 mm id.

The microporous membrane liquid–liquid extraction unit consisted of two blocks of Teflon and PEEK with grooves of 11 µL volume (61 × 1.78 × 1 mm, l × w × d) in both blocks. A porous polypropylene membrane (Celgard 2400, Hoechst Celanese, Charlotte; NC, USA) of 25.4 µm thickness was used. The pore dimensions were 0.05 × 0.125 µm and the porosity was 0.4. The membrane was wetted with acceptor solvent by pumping the solvent through the acceptor channel. The membrane was changed after every 20 extractions to avoid possible problems due to adsorption. The inlet and outlet tubings of the membrane unit were made of 1/16″ teflon tubing (0.3–0.5 mm id).

In the preliminary experiments the GC was a Fisons Instruments Dualchrom 3000 Series on-line HPLC–HRGC (CE Instruments, Milan, Italy) operating with a Phoenix 30CU pump. In the GC, a 10 m × 0.53 mm id DPTMDS (1,2-diphenyl-1,1,3,3-tetramethyldisilazane) deactivated retention gap (BGB Analytik AG, Zürich, Switzerland) was connected to a 20 m × 0.25 mm id analytical column (HP-5) of 0.25 µm phase thickness (Agilent Technologies, USA) and to a solvent vapour exit (SVE) via a glass pressfit Y-piece. The detection was made by a flame ionisation detector (FID) at 300 °C, detector gas pressures being 150 kPa for air and 50 kPa for hydrogen. The carrier gas was helium at 150 kPa. The oven was programmed from 85 °C (8 min) to 150 °C (2 min) at 40 °C min⁻¹ and then to 300 °C (10 min) at 5 °C min⁻¹.

An HRGC 5300 from CE Instrumentation (Milano, Italy) with on-column interface was used in the PHWE–MMLLE–GC–MS studies. The GC columns were identical with those used in the MMLLE–GC–FID studies. A quadrupole MS (Automass Solo, Thermoquest, Argenteuil Cedex, France) was used as the detector. Electron ionisation at 70 eV was applied and fragmented ions were monitored with total ion current (TIC) from 50 to 500 amu. In the quantification the ions of interest were filtered from the TIC. Carrier gas and oven conditions were as described above.

Before the connection of PHWE–MMLLE to GC–MS via large-volume injection, preliminary experiments were carried out with off-line injections to another GC–MS (Hewlett Packard 6890N gas chromatograph, 5973N quadrupole mass spectrometer (USA)), also equipped with an on-column injector. The MS analysis was carried out in scan mode (scan range 50–550 amu) with electron impact ionisation (EI, 70 eV). The temperature of the GC–MS interface was 300 °C, that of the ion source 230 °C and that of the analyser 150 °C. The analytical column of the gas chromatograph was a 20.0 m HP–5MS (Agilent Technologies, USA) with 0.25 mm id and 0.25 µm phase thickness. A 3.0 m retention gap (BGB Analytik AG, Zürich, Switzerland) with 0.53 mm id and deactivation was connected to the analytical column with a press-fit connector (BGB Analytik AG, Zürich, Switzerland). The oven was programmed from 85 °C with the same program as used with GC–FID.

2.5. Procedures

Ultrasonic²⁷ and liquid–solid extraction were selected as reference methods for the extraction of pesticides in grapes. These reference methods were chosen because they have traditionally been used in the determination of pesticides in various matrices. The methods are also straightforward, the equipment is easily available and the repeatability is good. Grape skin samples (500 mg) were dried in a dessiccator for 24 h and then placed into a test tube. After addition of 3 ml toluene, samples were extracted in an ultrasonic bath for 30 min. The grape skins were removed and the toluene was dried with sodium sulfate. The sample was filtered and analysed with GC–MS equipped with a large-volume injector. Sample preparation was otherwise the same as in ultrasonic extraction but the extraction was carried out in a shaker for 3 h.

Samples (50 mg) were put into an extraction vessel together with sea-sand. Water flow-rate of 1 ml min⁻¹ was applied in the PHW extraction. The sample was extracted at 120 °C for 40 min. The extract from PHWE was led to the donor side of the MMLLE unit. This means that the PHW extract was also the donor solvent for MMLLE, and MMLL extraction took place at the same time with PHWE. During the extraction the acceptor solvent remained stagnant and the analytes were enriched to the acceptor solvent by diffusion via the pores of the membrane. After 40 min, the PHW extraction and thus MMLLE donor flow was stopped and the extract was eluted by pumping acceptor solvent out of the channel. Elution was carried out with a flow-rate of 0.2 ml min⁻¹ for 45 s, leading to 0.150 ml extract volume.

The samples were injected off-line or transferred on-line to GC at 85 °C under conditions allowing partially concurrent solvent evaporation, and the oven was programmed as described above. The introduction was made with the help of an LC pump with flow-rate of 0.2 ml min⁻¹ and injection time of 1 min. The introduction volume exceeded the volume of the extract to ensure the transfer of the whole extract to GC and to flush the sample loop with fresh solvent. The solvent vapour exit was kept open during the introduction and it was closed 30 s after the transfer was complete.

3. Results and discussion

Pesticides are sprayed on the skins of grapes, which means that the pesticide concentration is highest on the surface. We studied grape skins and fruit pulp separately, with our main interest the grape skins. For the pesticides included in this study, the maximum residue limits (MRL) established by the EU for grapes are in the range 0.05–5 mg kg⁻¹.²⁸ The LODs required for analytical methods of the investigated pesticides range from 0.02 to 0.05 mg kg⁻¹.²⁸ Grape samples for pesticide analysis have often been pre-treated by liquid–liquid or solid-phase extraction after liquid–solid extraction with organic solvents.²⁹ The aim of the present study was to develop a closed on-line coupled PHWE–MMLLE–GC–MS system for selective extraction and analysis of pesticides in grapes.

3.1. Investigation of extraction conditions

The main parameters affecting PHWE–MMLLE extraction are the extraction temperature (in PHWE), the flow rate and the extraction time. Investigation of extraction temperature and time was made with spiked samples comprising 0.2 ml of pesticide solution with a pesticide concentration of 1 µg ml⁻¹. 0.150 ml of the extract were injected off-line to the GC for analysis. Extractions were carried out with a flow-rate of 1 ml min⁻¹, which has been found to be a good compromise for efficient extraction in both PHWE and MMLLE.^30–33 In PHWE the solubility of the analytes, not desorption,³⁴ tends to be the limiting factor in the extraction and high flow-rates are therefore preferred. The MMLLE conditions (extraction solvent, flow-rate and time) were adopted on the basis of previous studies where conditions were optimised for the extraction of pesticides in aqueous samples.^24,25

3.1.1. Extraction temperature in PHWE. Temperatures ranging from 50 to 300 °C have been employed in the extraction of pesticides from a variety of matrices.^14,16,35 A temperature of 120° has been used for the extraction of selected pesticides in fruit samples.¹⁶ For less polar and thermally stable pesticides, temperatures up to 300 °C have been applied for the extraction of soils and sediments, without degradation of analytes.³⁵ Very high temperatures should be used with caution, however, because excessively high temperatures may cause degradation of analytes prone to hydrolytic attack. Dramatic degradation of monolinuron and linuron, for example, occurred when extraction was performed at 130 °C.¹⁷ Moreover, the selectivity of the extraction decreases at high temperatures because of the larger amounts of unwanted low-polar matrix species that are coextracted. The solubility of some pesticides in water, it should be noted, is relatively poor at low temperatures.

We investigated PHW extraction temperatures of 100 °C, 120 °C and 150 °C with the extraction time of 30 min to determine the best temperature at which efficient extraction of all analytes could be obtained. As can be seen in Fig. 1, a temperature of 120 °C yielded the best recovery for most compounds, and 120 °C was used in further studies. The poorer recoveries of all analytes at 150 °C indicated some degradation.

Fig. 1 Investigation of PHW extraction temperature. Flow-rate 1 ml min⁻¹, extraction time 30 min. Samples spiked in sea-sand at concentration 200 ng. X-axis: compounds, Y-axis: total peak area.

3.1.2. Extraction time in PHWE–MMLLE. Typical times required for extraction in PHWE range from 15 to 60 min.^1,2 We studied extraction times in the range of 20–50 min with extraction temperature of 120 °C and extraction flow-rate of 1 ml min⁻¹. The effect of extraction time on peak areas of the analytes can be seen in Fig. 2. Increasing the extraction time enhanced the efficiency of extraction of the standard compounds up to 40 min, after which levelling off occurred. The recoveries of procymidone and vinclozolin even decreased with longer extraction times, probably because of the MMLL extraction, as the same trend has been noticed in the MMLL extraction of pesticides in wine.²⁵ An extraction time of 40 min was selected for the further experiments.

Fig. 2 Investigation of PHW extraction time. Elution profile obtained with the flow-rate of 1 ml min⁻¹ and the extraction temperature of 120 °C. Samples spiked with 0.2 ml of a pesticide solution of 1 µg ml⁻¹. X-axis: extraction time, Y-axis: total peak area representing elution profile.

3.2. Quantitative analysis

The extraction yield was studied by comparing the peak areas obtained from extractions with those obtained from standard on-column injections of known concentrations. This was done with spiked samples as described in sect. 3.1. Under the selected conditions (1 ml min⁻¹ donor flow-rate, 120 °C extraction temperature, 40 min extraction) the total recoveries of the system were in the range 9–26% (Table 1). We assume that the relatively low recoveries are due to the MMLLE part of the system, because similar to solid-phase microextraction (SPME), 100% extraction recoveries are not usually obtained in this membrane-based technique. Moreover, the extraction efficiency for pesticides in MMLLE alone has been in a comparable range (9–35%).²⁵ The recovery of the PHW extraction was tested using grape skin in the extraction and collection into organic solvent (toluene). Unfortunately, owing to strong emulsion formation, accurate determination of the recovery did not prove possible.

Table 1 Quantitative features of PHWE–MMLLE–GC–MS analysis

Compound	Recovery (%)	E e	R ^2	LOQ/µg kg^−1
Lindane	10 ± 2	27	0.998	1.8
Vinclozolin	28 ± 6	75	0.987	1.3
Quinalphos	23 ± 5	61	0.974	0.3
Procymidone	26 ± 5	69	0.978	0.3
Endosulfan sulfate	9 ± 1	24	0.973	0.6
Tetradifon	10 ± 2	27	0.980	0.6

---

The performance of the MMLLE system can best be evaluated by means of the enrichment factor (E_e), which is the calculated ratio of the analyte concentration in the acceptor compared to that in the sample and describes the concentration effect achieved in the extraction (E_e = V_donor/V_acceptor × recovery). The enrichment factors ranged from 24 to 69, showing efficient concentration of the target compounds in the extract.

The linearity of the method was tested in the range 0.015–3 mg kg⁻¹ and was observed to be good. The correlation coefficients were in the range 0.973–0.998 for all the compounds studied. Repeatability of peak areas for spiked samples was on average 19% (as RSD%: range 10–23%). The limits of quantification (LOQs, for authentic samples) ranged from 0.3 to 1.8 µg kg⁻¹ calculated as the analyte concentration, and gave a signal equal to ten times that of the blank signal when calculated with using the regression equations. The LOQs were at the same level as or even lower than values found for organochlorine pesticides and chlorobenzenes in fruit and vegetables in a study done with accelerated solvent extraction–SPME.¹⁶ LOQs were also lower than those obtained for acidic pesticides in fruits with off-line SPME–CE–MS.³⁶ The LODs required by the European Union for the pesticides of interest in this study are 0.02–0.05 mg kg⁻¹.

The amount of sample to be used in PHWE–MMLLE–GC–MS should be considered, to avoid overloading of the GC system due to the very efficient concentration provided by the on-line system. A suitable amount of sample was about 50 mg. Samples were weighed as such; no drying of the skin or pulp was required. Only liquid–solid and ultrasonic extractions were made with dried samples (for 24 h), to avoid emulsion formation during the extractions. The moisture content was determined by drying several 500 mg samples in a desiccator for 10 days. The moisture content of skin samples was found to be 76%.

After establishment of the qualitative aspects of the method, grape samples were analysed for pesticides with the PHWE–MMLLE–GC–MS system. The analyte concentrations were calculated based on calibration curves. In the determinations the peak areas of the analytes relative to the peak areas of internal standards diphenylamine and 1,1′-binaphthyl were fitted to the curves. The procymidone concentration in the skins was 0.040 ± 0.008 mg kg⁻¹ and traces of tetradifon were detected in two of four samples (Fig. 3). For samples treated by ultrasonic extraction, the concentration determined for procymidone was 0.052 ± 0.012 mg kg⁻¹ and for those treated by liquid–solid extraction the concentration was 0.040 ± 0.001 mg kg⁻¹. Tetradifon was not detected in these other extracts. The concentration of procymidone determined by PHWE–MMLLE–GC–MS in the pulp was 0.0048 ± 0.001 mg kg⁻¹, which means that the concentration was lower in the grape pulp than in the skin. This was as expected. The amounts of procymidone determined were below the MRL set by the European union (5 mg kg⁻¹).²⁸ No MRL was found for tetradifon. The results obtained with the different extraction techniques were similar, confirming the reliability of the PHWE–MMLLE–GC–MS method. As can be seen in Fig. 3, the chromatographic behaviour was significantly better with the PHWE–MMLLE–GC–MS system than with the liquid–solid and ultrasonic extractions with off-line analysis, demonstrating the selectivity of MMLLE trapping of the extract.

Fig. 3 Comparison of extraction methods: (A) PHWE–MMLLE–GC–MS chromatogram of 50 mg grape sample. Extraction conditions as reported in the Experimental. (B) GC–MS chromatogram recorded after liquid–solid extraction of 500 mg sample. (C) GC–MS chromatogram recorded after ultrasonic extraction of 500 mg sample. In (A) the ions for procymidone (284) and tetradifon (356) were filtered. In (B) and (C) only procymidone was detected.

4. Conclusions

The PHWE–MMLLE–GC–MS method performed well in the determination of pesticides in grapes. It was linear over a wide range and sensitive, and the results agreed well with those obtained with other (off-line) extraction methods in food analysis. Procymidone was found in the extracts and its concentration was lower in the grape pulp than the skin, as expected. The LOQs were well below the requirements set by EU regulations. The method could suitably be applied to the determination of pesticides in other plant material.

Acknowledgements

Financial support was received from the Academy of Finland (project 48867).

References

1. L. Ramos, E. M. Kristenson and U. A. Th. Brinkman, J. Chromatogr. A, 2002, 975, 3–29.
2. R. Smith, J. Chromatogr. A, 2002, 975, 31–46.
3. V. Janda, K. D. Bartle and A. A. Clifford, J. Chromatogr., 1993, 642, 283–299.
4. S. B. Hawthorne and D. J. Miller, Anal. Chem., 1994, 66, 4005–4012.
5. S. Bowadt and S. B. Hawthorne, J. Chromatogr. A, 1995, 703, 549–571.
6. S. B. Hawthorne, C. B. Grabanski, E. Martin and D. J. Miller, J. Chromatogr. A, 2000, 892, 421–433.
7. J. Pörschmann and J. Plugge, Fresenius’ J. Anal. Chem., 1999, 364, 643–644.
8. Y. Yang, S. Bowadt, S. B. Hawthorne and D. J. Miller, Anal. Chem., 1995, 67, 4571–4576.
9. K. Hartonen, G. Meissner, T. Kesälä and M.-L. Riekkola, J. Microcol. Sep., 2000, 12, 412–418.
10. L. N. Konda, G. Füleky and G. Morovján, J. Agric. Food Chem., 2002, 50, 2338–2343.
11. E. Ibañez, A. Kubátová, F. J. Señorans, S. Cavero, G. Reglero and S. B. Hawthorne, J. Agric. Food Chem., 2003, 51, 375–382.
12. M. Ollanketo, A. Peltoketo, K. Hartonen, R. Hiltunen and M.-L. Riekkola, Eur. Food Res. Technol., 2002, 215, 158–163.
13. M. M. Jiménez-Carmona, J. L. Ubera and M. D. Luque de Castro, J. Chromatogr. A, 1999, 855, 625–632.
14. Z. Y. Ju and L. R. Howard, J. Agric. Food Chem., 2003, 51, 5207–5213.
15. C. Crescenzi, G. D'Ascenzo, A. Di Gorcia, M. Nazzarri, S. Marchese and R. Samperi, Anal. Chem., 1999, 71, 2157–2163.
16. L. Wennrich, P. Popp and J. Breuste, Chromatographia, 2001, 53 Suppl., S-380–S386.
17. C. Crescenzi, A. Di Gorcia, M. Nazzarri and R. Samperi, Anal. Chem., 2000, 72, 3050–3055.
18. P. Richter, B. Sepúlvedaa, R. Olivaa, K. Calderóna and R. Segue, J. Chromatogr. A, 2003, 994, 169–177.
19. D. J. Miller and S. B. Hawthorne, Anal. Chem., 1998, 70, 1618–1621.
20. J. Å. Jönsson and L. Mathiasson, J. Chromatogr. A, 2000, 902, 205–225.
21. J. Å. Jönsson and L. Mathiasson, J. Sep. Sci., 2001, 24, 495–507.
22. K. Kuosmanen, T. Hyötyläinen, K. Hartonen, J. Å. Jönsson and M.-L. Riekkola, Anal. Bioanal. Chem., 2003, 375, 389–399.
23. K. Kuosmanen, T. Hyötyläinen, K. Hartonen and M.-L. Riekkola, Analyst, 2003, 128, 434–439.
24. K. Lüthje, T. Hyötyläinen and M.-L. Riekkola, Anal. Bioanal. Chem., 2004, 378, 1991–1998.
25. T. Hyötyläinen, T. Tuutijärvi, K. Kuosmanen and M.-L. Riekkola, Anal. Bioanal. Chem., 2002, 372, 732–736.
26. T. Andersson, K. Hartonen, T. Hyötyläinen and M. -L. Riekkola, Anal. Chim. Acta, 2002, 466, 93–100.
27. M. Palma and C. G. Barroso, Anal. Chim. Acta, 2002, 458, 119–130.
28. Maximum residue limits in the annexes to European Council Directives 76/895/EEC, 86/362/EEC, 86/363/EEC & 90/642/EEC, 03/2001.
29. R. Rial Otero, B. Cancho Grande and J. Simal Gándara, J. Chromatogr. A, 2003, 992, 121–131.
30. M. Sandahl, L. Mathiasson and J. Å. Jönsson, J. Chromatogr. A, 2002, 975, 211–217.
31. M. Sandahl, L. Mathiasson and J. Å. Jönsson, J. Chromatogr. A, 2000, 893, 123–131.
32. S. B. Hawthorne, Y. Yang and D. J. Miller, Anal. Chem., 1994, 66, 2912–2920.
33. Y. Yang, S. B. Hawthorne and D. J. Miller, Environ. Sci. Technol., 1997, 31, 430–437.
34. A. Kubatova, B. Jansen, J.-F. Vaudoisot and S. B. Hawthorne, J. Chromatogr. A, 2002, 975, 175–188.
35. P. Richter, B. Sepúlvedaa, R. Olivaa, K. Calderóna and R. Segue, J. Chromatogr. A, 2003, 994, 169–177.
36. R. Rodriguez, J. Mañes and Y. Pico, Anal. Chem., 2003, 75, 452–459.

---