Method development for the determination of 52 pesticides in tobacco by liquid chromatography–tandem mass spectrometry

Journal of Chromatography A, 1216 (2009) 8953–8959

Contents lists available at ScienceDirect

Journal of Chromatography A journal homepage: www.elsevier.com/locate/chroma

Method development for the determination of 52 pesticides in tobacco by liquid chromatography–tandem mass spectrometry Bernhard Mayer-Helm ∗ R&D, Ökolab Gesellschaft für Umweltanalytik, A Member of the Japan Tobacco International Group of Companies, Hasnerstraße 127, 1160 Vienna, Austria

a r t i c l e

i n f o

Article history: Received 27 April 2009 Received in revised form 1 October 2009 Accepted 23 October 2009 Available online 30 October 2009 Keywords: Pesticides Tobacco Matrix effects Sample cleanup LC–MS/MS

a b s t r a c t A method using reversed phase liquid chromatography–electrospray ionization–tandem mass spectrometry was developed for the determination of 52 pesticides in tobacco. The influence of mobile phase additives was investigated to improve sensitivity and accuracy of the method and to reduce matrix effects. The tobacco extracts were purified via a Chem Elut partition cartridge by consecutive elution with pentane followed by dichloromethane. The two fractions were further purified by Florisil solidphase extraction with acetone or diethyl ether elution. An additional dispersive solid-phase extraction step with primary–secondary amine led to decreased recoveries of several pesticides due to degradation or binding to the sorbent. The method was validated for the tobacco types Burley, Oriental and Virginia. The recovery rates of almost all pesticides ranged between 70 and 120%. The limits of quantification were below or near the 10 ng/g level. Few but significant differences between the tobacco types could be found regarding recovery and sensitivity. © 2009 Elsevier B.V. All rights reserved.

1. Introduction Gas chromatography (GC) and liquid chromatography (LC) coupled to mass spectrometry (MS) enable the analysis of several hundreds of pesticides [1]. Today, the European Union (EU) lists maximum residue levels (MRL) for 497 pesticides in innumerable food and feed products [2]. For tobacco and tobacco products the CORESTA (Cooperation Centre for Scientific Research Relative to Tobacco) published a list of Guidance Residue Levels (GRLs) for 118 pesticides [3]. Over the past few years it turned out that sample preparation can hardly be omitted to ensure reliable and robust methods despite modern instrumentation. Some groups were able to completely omit sample purification by injecting the raw sample extracts. For example, extracts of soya grain and crops were directly injected at a final concentration of 50 and 200 mg matrix/mL [4,5]. Extracts of fruits and vegetables could be injected at even higher sample concentrations, e.g. 500 mg/mL [6] and 2500 mg/mL [7]. Unfortunately, such high matrix concentrations could not be achieved with tobacco extracts, only 12.5 mg/mL could be directly injected [8]. The direct injection of raw extracts fails frequently, as the extracts are too dirty causing unreliable results or as the analyte concentration is too low. Consequently, improved sample purification is required. Several techniques such as solid-

∗ Tel.: +43 1 31342 2253; fax: +43 1 31342 2430. E-mail address: [email protected]. 0021-9673/$ – see front matter © 2009 Elsevier B.V. All rights reserved. doi:10.1016/j.chroma.2009.10.077

phase extraction (SPE), pressurized liquid extraction, liquid–liquid extraction, matrix solid phase dispersion, gel permeation chromatography and dispersive SPE (dSPE) are compared for the determination of pesticides [9,10] and reviewed [11]. One method based on dSPE which attracts a lot of attention is the QuEChERS method [12–15]. The original method with primary–secondary amine (PSA) was used, e.g. for multi-residue methods in grapes [16], baby food [10], fruits and vegetables [17] and barley [9]. Despite this success, some modifications, e.g. the use of different sorbents like graphitized carbon black (GCB) and C18 [12,14,15] was suggested. PSA, GCB and C18 were applied for multi-residue analysis of pesticides in spinach [18], green leafy vegetables [19] and tobacco [20]. However, QuEChERS applied to tobacco samples could reach only low matrix loadings of 15 and 100 mg/mL [21,20]. Improved sensitivity by injecting better purified samples with higher matrix concentration is required to reach the limits for some pesticides. In this study three different sample preparation steps, namely Chem Elut partition, dSPE and SPE were optimized for the purification of tobacco extracts. It was found that a dSPE step with PSA alone was not able to provide suitably clean extracts. Two additional cleanup steps were established to fulfill the requirements for recovery and sensitivity. Additionally, the influence of mobile phase additives on signal intensity of the pesticides was investigated. The method for the determination of 52 pesticides was validated regarding recoveries, matrix effects, sensitivity and stability in three main tobacco types Burley, Oriental and Virginia tobacco.

8954

B. Mayer-Helm / J. Chromatogr. A 1216 (2009) 8953–8959

pentane. The solvent was eluted to waste, whereas the analytes were eluted with two times 8 mL diethyl ether. The DCM fraction was applied to a Florisil column pretreated with 5 mL DCM. The eluate was collected and the pesticides were subsequently eluted with 10 mL acetone. After solvent evaporation, the dry residue of the pentane fraction was redissolved in 0.800 mL MeOH containing 5 mM ammonium acetate and 0.1% acetic acid. The residue of the DCM fraction was filled up to 0.400 mL water and diluted with 0.400 mL 10 mM ammonium acetate and 0.2% acetic acid. Finally, the samples were filtered through a 0.45 ␮m Millex-LCR PTFE filter (Millipore Corp., Bedford, MA, USA). The final sample concentration was 500 mg tobacco/mL. During method development, several experiments with and without PSA were performed. For this, the eluates from the Chem Elut cartridge were subjected to dispersive SPE. After addition of 500 mg magnesium sulfate and 200 mg PSA the eluates were vigorously shaken for 2 min. After centrifugation the supernatant was applied to the Florisil column as described above. Fig. 1. Scheme of sample cleanup.

2. Experimental 2.1. Chemicals and materials All solvents had HPLC grade quality or better. Dichloromethane (DCM), diethyl ether, pentane and acetone were purchased from Sigma–Aldrich/Fluka/Riedel-de Haën/Supelco (St. Louis, MO, USA). Methanol (MeOH) for LC–MS was obtained from Fluka. Water for HPLC was obtained from LGC Promochem (Wesel, Germany). Sodium chloride (≥99.5%) and ammonium acetate (≥98.0%) were obtained from Merck (Darmstadt, Germany). Acetic acid (>99.8%) and ammonium formate (>99.995%) were from Fluka. Magnesium sulfate anhydrous (≥98.0%, Fluka) was baked over night at 500 ◦ C to remove impurities. Primary–secondary amine (PSA, 40 ␮m, Bondesil) and Chem Elut cartridges with 5 mL sample capacity were purchased from Varian Inc. (Palo Alto, CA, USA). Magnesium silicate SPE columns, Supelclean, LC–Florisil, 1 g, were obtained from Supelco. All pesticide reference standards were obtained with the highest purity available from Dr. Ehrenstorfer GmbH (Augsburg, Germany), except dimefox, which was from Riedel-de Haën. 1 mg/mL stock solutions were prepared by dissolving 10 mg of the pure analytical standard in 10 mL acetone. 10 mg carbendazim was dissolved in 50 mL ethanol due to solubility problems. Vamidothion sulfoxide was obtained as a 10 ␮g/mL solution in cyclohexane, where the solvent had to be partly removed before use. Solutions for spiking residue-free tobaccos to generate calibration curves were prepared by mixing the individual stock solutions and by diluting with MeOH. 2.2. Sample extraction and cleanup A scheme of the sample cleanup is depicted in Fig. 1. Test portions of 2 g of ground tobacco were soaked with 6.25 mL of 1% aqueous acetic acid for 5 min. The pesticides were extracted with 12.5 mL MeOH under ultrasound and under shaking for 5 min, respectively. After centrifugation, 4 mL of the supernatant was diluted with 1.33 mL of 20% aqueous sodium chloride solution. After mixing, the whole solution was applied to a Chem Elut partition cartridge. After equilibrium for 5 min, the pesticides were consecutively eluted from the cartridge. Firstly, the apolar pesticides were eluted with three times 8 mL pentane. Secondly, the medium polar and polar pesticides were eluted with three times 8 mL DCM. Subsequently, each fraction was treated separately. The pentane fraction was applied to a Florisil column conditioned with 5 mL

2.3. HPLC–MS/MS analysis The LC system 1200 series consisted of a binary HPLC pump, an online vacuum degasser, a thermostatted automatic sampler ALS and a thermostatted column compartment (Agilent Technologies, Palo Alto, CA, USA). Mobile phase A consisted of 90% water and 10% MeOH, mobile phase B consisted of 10% water and 90% MeOH. Both mobile phases contained 5 mM ammonium acetate. During method development alternative mobile phases were tested, such as a 5 mM ammonium formate and 5 mM ammonium formate acidified with 0.1 or 0.2% formic acid. All mobile phases were filtered with a 0.45 ␮m cellulose nitrate filter and degassed via ultrasound prior to use. The flow rate was 0.4 mL/min. A binary gradient was run from 0 to 21.2 min from 100% mobile phase A to 100% mobile phase B, and held for 7.8 min. Finally the column was reequilibrated to 100% mobile phase A for 6.9 min resulting in a total run time of roughly 36 min. The LC flow was directed into the MS between 1 and 24.1 min using a built-in 6-port valve (Rheodyne LLC, Rohnert Park, CA, USA). The extracted samples were stored at 8 ◦ C in the autosampler tray prior to analysis. 10 ␮L of sample was injected onto the HPLC column. Separation was achieved with a Synergi 4 ␮m Hydro-RP 80 A, 150 mm × 2.0 mm column protected by a 10 ␮m Aqua guard column, 4.0 mm × 2.0 mm (both from Phenomenex, Torrance, CA, USA) thermostatted at 35 ◦ C. The HPLC was coupled to a triple quadrupole mass spectrometer API 4000 via an atmospheric pressure ionisation electrospray probe (Applied Biosystems, Inc., Foster City, CA, USA). The instrument was controlled and data processing was done by software Analyst 1.5 (Applied Biosystems). The analytes were measured in the multiple reaction monitoring (MRM) mode using scheduled time windows. The compound-specific mass spectrometric parameters are summarized in Table 1. Optimization of the declustering potential and collision energy for the individual analytes was done by infusion of the analytes directly into the MS using a syringe pump (Harvard Apparatus, Kent, UK) at a flow rate of 10 ␮L/min in the respective mobile phase composition. The entrance potential and collision cell exit potential were set to 10 and 12 V for all analytes. Two transitions were measured for every analyte. [M+H]+ served as precursor ions in the majority of cases, while [M+NH4 ]+ was used only for methidathion 2 and famoxadone 1. For several analytes fragments served as precursor ions, which were generated during ionization at the Q0 by increased declustering potential, namely thiamethoxam 2, cymoxanil 2, terbufos sulfoxide 2, azoxystrobin 2, terbufos 2 and famoxadone 2. Instrument specific parameters were scheduled MRMs with a detection window of 120 s, a target scan time of 2 s and a pause between mass ranges of 5 ms. The ion source was an electrospray

B. Mayer-Helm / J. Chromatogr. A 1216 (2009) 8953–8959

8955

Table 1 Mass spectrometric parameters for the quantification and confirmation of 52 pesticides in the multiple reaction monitoring (MRM) mode. Pesticide

Methamidophos Acephate Methomyl-oxim Omethoate Vamidothion sulfoxide Demeton-S-methyl sulfoxide Demeton-S-methyl sulfone Thiamethoxam Monocrotophos Pymetrozine Dimefox Imidacloprid Trichlorfon Dimethoate Vamidothion Acetamiprid Cymoxanil Carbendazim Oxadixyl Phosphamidonb Thiophanate-methyl Demeton-S-methyl Fenamiphos sulfoxide Fenamiphos sulfone Fenthion sulfoxide Fenthion sulfone Disulfoton sulfoxide Disulfoton sulfone Thionazin Fensulfothion Methidathion Terbufos sulfone Terbufos sulfoxide Azoxystrobin Fenamidone Acibenzolar-S-methyl Fenitrothion Iprodione Fenamiphos Penconazole Famoxadone Chlorfenvinphosb Fenthion Fonofos Triflumuron Phoxime Pirimiphos-methyl Disulfoton Indoxacarb Fluazifop butyl Profenofos Terbufos a b

Retention time (min)

2.1 2.5 3.6 4.1 6.6 7.2 7.6 7.7 8.3 8.3 9.1 9.8 9.9 10.4 10.6 11.2 11.5 12.3 13.7 13.9 14.8 15.3 15.4 15.6 15.9 16.3 16.4 16.7 17.5 17.6 18.0 18.6 18.7 18.8 19.0 19.5 20.0 20.7 20.8 21.4 21.4 21.7 + 22.1 21.6 21.7 21.9 22.0 22.2 22.4 22.5 23.2 23.3 23.5

1st Transition (quantification)

2nd Transition (confirmation)

Precursor Product ion ion

Declustering potential (V)

Collision energy (V)

Precursor Product ion ion

142 184 106 214 304 247 263 292 224 218 155 256.1 257 230 288 223 199.1 192.1 279.1 300 343 231 320 336 295 311 291 307 249 309 303 321 305 404.1 312.1 211 278 330 304 284 392 359 279 247 359 299 306 275 528.1 384.1 375 289

60 55 41 60 65 60 65 63 50 65 80 45 62 60 45 65 58 35 65 75 67 45 80 46 75 90 60 80 60 80 70 50 52 60 63 70 76 71 70 67 50 75 65 50 70 76 80 40 83 80 85 52

22 13 19 17 25 19 22 18 12 30 33 23 27 15 17 30 14 25 31 18 29 15 33 28 25 32 13 18 17 25 13 12 17 21 39 42 27 23 31 60 14 18 25 17 25 17 30 50 32 28 25 13

142 184 106 214 304 247 263 211 224 218 155 258.1 257 230 288 223 128 192.1 279.1 300 343 231 320 336 295 311 291 307 249 309 320 321 187 372 312.1 211 278 332 304 286 331 361 279 247 359 299 306 275 528.1 384.1 373 233

94 143 58 183 169 169 169 211 193 105 110 209.1 109 199 146 126 128 160.1 133 227 151 89 233 266 280 125 213 153 221 253 145 265 187 372 92 136 125 245 217 70 331 155 169 137 156 129 164 61 218 282 305 103

125 113 42 125 201 109 121 181 127 78 135 211.1 221 171 118 56 111 132 132 127 311 61 171 308 109 109 185 125 97 281 145 171 159 344 236 140 127 247 202 70 238 155 105 109 139 77 108 89 203 328 303 199

Declustering potential (V) 60 55 41 60 65 60 65 96 50 65 80 45 62 60 45 65 75 35 65 75 67 45 80 46 75 90 60 80 60 80 40 50 84 110 63 70 76 71 70 67 85 75 65 50 70 76 80 40 83 80 85 65

Collision energy (V) 20 31 50 30 17 39 22 20 22 60 25 23 17 22 32 31 16 43 45 30 17 45 30 23 43 35 19 25 35 21 18 18 15 29 21 34 27 23 47 60 18 18 35 29 47 45 43 18 55 24 25 11

Relative intensity (%)a

74 (2) 5.1 (2) 2.4 (5) 67 (6) 77 (6) 29 (2) 40 (1) 60 (4) 99 (3) 8.7 (4) 42 (5) 33 (3) 84 (2) 39 (1) 36 (4) 17 (1) 59 (3) 17 (1) 97 (2) 185 (2) 33 (2) 28 (14) 103 (3) 99 (2) 44 (4) 34 (8) 132 (3) 74 (3) 137 (2) 148 (2) 181 (6) 172 (14) 39 (4) 29 (2) 130 (1) 38 (2) 39 (8) 64 (7) 40 (4) 69 (1) 33 (2) 92 (1) 66 (21) 107 (1) 47 (4) 66 (6) 86 (1) 347 (11) 107 (6) 86 (0) 77 (3) 14 (5)

Mean value and relative standard deviation of the tobacco types Burley, Oriental and Virginia. Two isomers. The sum of both peak areas was used for quantification.

(TurboIonSpray) operated in the positive polarity. The interface heater was set to 120 ◦ C and the block source temperature was maintained at 450 ◦ C with a capillary voltage of 5 kV. Nitrogen was generated by a nitrogen generator NM20ZA from Peak Scientific (Inchinnan, UK) was used for the curtain and nebulizer (GS1 and GS2) gases at 20, 40 and 50 psi, respectively. Nitrogen was also used as collision gas at 6 psi.

2.4. Quantification and method validation The method was validated for in the three main tobacco types Burley, Oriental and Virginia. Residue-free tobaccos were spiked with a mix of all pesticides in MeOH. Quantification was performed

with matrix matched external standard calibration. Tobaccos were spiked close to the guidance residue levels (GRLs) for tobacco prior to extraction [3]. The overall recovery was determined by comparing peak area of analytes in the tobacco matrix vs. standard (in 5 mM ammonium acetate plus 0.1% acetic acid in water/MeOH = 50/50, v/v). The linearity was tested with tobaccos spiked at multiples (0.2, 0.5 and 2) of the GRLs. The trueness of the method was investigated by reanalyzing tobacco samples from a proficiency test FT0104 (FAPAS, Sand Hutton, UK). The limits of quantifications (LOQs) were determined from injections of matrix-matched standards at concentration levels corresponding to a signal-to-noise ratio of 10. The EU has proposed the measurement of two transitions for the proof of identity. The relative intensity of these transitions (= Area MRM

8956

B. Mayer-Helm / J. Chromatogr. A 1216 (2009) 8953–8959

Table 2 Recovery rates (%) of 52 pesticides after purification with Chem Elut/pentane or Chem Elut/DCM and Florisil in three main tobacco types. Pesticide

Fraction

Burley

Oriental

Virginia

Mean value (SD)a

Methamidophos Acephate Methomyl-oxim Omethoate Vamidothion sulfoxide Demeton-S-methyl sulfoxide Demeton-S-methyl sulfone Thiamethoxam Monocrotophos Pymetrozine Dimefox Imidacloprid Trichlorfon Dimethoate Vamidothion Acetamiprid Cymoxanil Carbendazim Oxadixyl Phosphamidonb Thiophanate-methyl Demeton-S-methyl Fenamiphos sulfoxide Fenamiphos sulfone Fenthion sulfoxide Fenthion sulfone Disulfoton sulfoxide Disulfoton sulfone Thionazin Fensulfothion Methidathion Terbufos sulfone Terbufos sulfoxide Azoxystrobin Fenamidone Acibenzolar-S-methyl Fenitrothion Iprodione Fenamiphos Penconazole Famoxadone Chlorfenvinphosb Fenthion Fonofos Triflumuron Phoxime Pirimiphos-methyl Disulfoton Indoxacarb Fluazifop butyl Profenofos Terbufos

DCM DCM DCM DCM DCM DCM DCM DCM DCM DCM DCM DCM DCM DCM DCM DCM DCM DCM DCM DCM DCM Pentane DCM DCM DCM Pentane DCM Pentane Pentane DCM Pentane Pentane Pentane DCM Pentane Pentane Pentane Pentane Pentane Pentane Pentane Pentane Pentane Pentane Pentane Pentane Pentane Pentane Pentane Pentane Pentane Pentane

70 76 70 83 85 92 86 92 77 20 26 100 72 87 94 89 79 71 65 74 77 67 150 125 95 52 77 74 113 72 98 109 107 63 112 94 99 108 56 65 117 107 108 105 108 95 113 88 104 152 123 79

56 63 69 78 71 91 97 95 86 9 75 99 87 93 108 99 90 84 91 93 50 78 149 127 98 55 81 88 124 73 108 121 123 63 127 103 123 123 60 69 124 114 122 121 118 90 132 111 103 196 150 91

75 82 71 94 80 101 106 112 103 19 65 107 94 90 97 105 89 79 93 103 81 67 152 139 112 59 95 76 109 85 111 110 108 71 114 102 112 115 58 66 118 117 109 111 119 107 121 95 104 176 144 82

67 (10) 74 (10) 70 (1) 85 (8) 78 (7) 95 (5) 96 (10) 100 (11) 89 (13) 16 (6) 56 (26) 102 (5) 84 (11) 90 (3) 100 (7) 98 (8) 86 (6) 78 (7) 83 (16) 90 (14) 69 (17) 71 (6) 151 (2) 130 (7) 102 (9) 55 (3) 84 (9) 79 (7) 115 (8) 77 (7) 106 (7) 113 (6) 113 (9) 66 (5) 117 (8) 100 (5) 111 (12) 115 (7) 58 (2) 67 (2) 120 (4) 113 (5) 113 (8) 112 (8) 115 (6) 97 (9) 122 (10) 98 (12) 104 (1) 175 (22) 139 (14) 84 (6)

a b

SD, standard deviation. Sum of both isomers.

2/Area MRM 1 × 100) has to be within the maximum permitted tolerances recommended by the EU [22]. 3. Results and discussion 3.1. Mobile phase additives and signal suppression It is well established that the mobile phase composition affects both retention and ionization of the analytes, e.g. [23,24]. In this study the influence of mobile phase additives on signal intensity of pesticides was investigated. All separations were achieved with the same water/MeOH gradient (see Section 2.3). The following additives were added to both mobile phases: 5 mM ammonium acetate, 5 mM ammonium formate and 0.1 or 0.2% formic acid. The peak areas of standard solutions measured with a water/MeOH gradient containing 5 mM ammonium acetate were set to unity and

compared to other mobile phase compositions (see supplementary Table S1). A clear dependance of the peak areas on the mobile phase composition was found. That comparison clearly revealed, that more acidic conditions provided higher peak areas, for example, methamidophos gave 3.5 times higher peak areas in 0.1% formic acid than in 5 mM ammonium acetate. The average increase for all transitions was 2.6. Moreover, the peak area of methamidophos was 12 times higher in 5 mM ammonium formate acidified with 0.1% formic acid than in 5 mM ammonium acetate. The peak areas for almost all pesticides were increased by the addition of ammonium ions, with a mean value of 5.5. That means that both a low pH and ammonium ions support ionization. This is true, although [M+NH4 ]+ ions were used as precursor only in two cases while [M+H]+ ions served as precursor for most transitions. This increase in sensitivity has previously been observed when using ammonium formate [4] or ammonium acetate [25] instead of formic acid. On

B. Mayer-Helm / J. Chromatogr. A 1216 (2009) 8953–8959

8957

the other hand, the addition of 0.2% formic acid to 5 mM ammonium formate was not successful, as no further increase of peak area could be found. Even a serious decrease was found for some pesticides, e.g. acibenzolar-S-methyl and trichlorfon. Obviously, a too high ionic strength leads to partly dramatic signal suppression [7]. Surprisingly, the increase of peak areas from ammonium acetate to ammonium formate plus formic acid depended on retention time: the lower the retention time (the higher the polarity), the higher the peak area increase when using ammonium formate plus formic acid compared to ammonium acetate. Unfortunately, all these attempts to improve the sensitivity by modifying the mobile phase composition did not turn out to be useful. When the overall recovery (recovery of purification steps plus matrix effects [26]) of the pesticides in the tobacco matrices was determined, very low recoveries had to be observed (data not shown). Especially the early eluting pesticides offered in the matrix much lower peak areas than in solvent. That seemed to be caused by signal suppression as the same extracts provided high recoveries in 5 mM ammonium acetate. This means that the increased peak areas in ammonium formate plus formic acid compared to ammonium acetate were observed only in standard solutions and not in the tobacco matrices. Similar signal suppression was reported by Choi et al. who demonstrated that matrix effects of four pesticides depend on the mobile phase composition [24]. Finally 5 mM ammonium acetate was used as an additive, although the response of standards was not optimum compared to more acidic mobile phases. However, most of the analytes offer similar responses in standard solution and tobacco providing low signal suppression and recovery rates of about 100%.

Table 3 Recovery rates of pesticides which are affected by the use of PSA.a .

3.2. Optimization of sample cleanup

Two methods using the QuEChERS method for tobacco purification reached only low matrix loading [20,21]. To improve sensitivity we set out to inject higher concentrated samples. Several experiments with various ratios of sample-to-PSA were performed. A satisfying sample purification and a final concentration of 500 mg tobacco/mL could be achieved only in combination with two additional sample cleanup steps. A method employing Chem Elut with pentane elution and subsequent DCM elution followed by dPSA with PSA and Florisil/diethyl ether or Florisil/acetone purification (see Fig. 1) was used in routine analysis for several months. PSA successfully removed several matrix compounds resulting in almost colorless extracts. However, the recoveries of some analytes were affected by the use of PSA (Table 3). The recoveries were similar for all tobacco types, which is indicated by the low standard deviations. The use of PSA provided largely reduced recoveries for pymetrozine, fenamidone, iprodione, fenamiphos, famoxadone and triflumuron. The low recovery rates must be attributed to degradation or binding to PSA. Reduced recovery rates were reported for cymoxanil and pymetrozin in orange [13]. Furthermore, it is well known that analytes with a carboxylic acid group are strongly retained on PSA sorbent [13]. However, the original dSPE step of the QuEChERS method [12] was not used, especially a higher PSA concentration was used (200 mg PSA per 24 mL solvent with 0.4 g matrix compared to 25 mg PSA per 1 mL solvent with 1 g matrix of the original method). This increased amount of PSA may be responsible for the unintended removal of some pesticides. It must be noted that two pesticides (fluazifop butyl and profenofos) showed signal enhancement without PSA. This gives hints that PSA also removed some matrix compounds which cause signal enhancement. The same sample cleanup steps were also assessed for the determination of some problematic pesticides, which were amenable to PTV–GC–NCI-MS. Only the left branch of Fig. 1 was used for sample purification of apolar pesticides - Chem Elut/pentane, dSPE and Florisil/diethyl ether. Some pesticides, namely captan, folpet and dinocap showed recovery rates below 5%. When PSA was omit-

Three different sample cleanup steps were assessed for the purification of tobacco samples: Chem Elut partition, dispersive SPE and Florisil SPE. The Chem Elut partition step was similar to Ref. [27]. This cleanup step was modified and successfully applied for the purification of apolar pesticides in tobacco in a previous work [28]. Here, the Chem Elut partition was again modified. The elution with DCM and a subsequent Florisil/acetone purification (the right branch of Fig. 1) was not successful, as the baseline of some pesticides showed several seriously disturbing peaks. DCM is a too strong solvent, eluting also many matrix compounds. Only a consecutive elution turned out to be successful. Firstly, the apolar pesticides were eluted with n-pentane. After that the pesticides with higher polarity were eluted with DCM resulting in cleaner extract than after the direct elution with DCM. After the Chem Elut partition, a Florisil purification step was required for further matrix removal. Florisil with acetone was the best solvent for the DCM fraction and Florisil with diethyl ether was the best for the pentane fraction. The combination of both approaches provides very good recovery rates (see Table 2). Almost all recovery rates are within the range of 70 and 120%. Recoveries below 70% are acceptable in most cases because they are consistent and have only low standard deviations [22]. However, one drawback of the consecutive elution from the Chem Elut column is that some pesticides with medium polarity are distributed in both fractions. For straightforwardness, only the fraction with the higher recovery and/or better sensitivity was used for quantification. The unacceptable low recovery rates of pymetrozine between 10 and 20% are caused as it is acid-sensitive [14]. This pesticide provided low recoveries also in orange and lettuce [13,14]. The risk of pymetrozine’s poor accuracy must be taken into account when using acidic conditions during extraction to improve the stability of base-sensitive pesticides. There are obviously some elution zones where matrix effects occur: on the one hand, signal enhancement seems to occur at approximately 15.5 min (fenamiphos sulfoxide, fenamiphos sul-

Pesticide

Pymetrozine Trichlorfon Cymoxanil Fenamidone Iprodione Fenamiphos Penconazole Famoxadone Triflumuron Fluazifop butyl Profenofos a b

Recovery (SD)b Without PSA

With PSA

16 (6) 84 (11) 86 (6) 117 (8) 115 (7) 58 (2) 67 (2) 120 (4) 115 (6) 175 (22) 139 (14)

4.7 (2) 47 (12) 64 (11) 47 (2) 8.4 (1) 20 (3) 46 (3) 7.7 (1) 37 (3) 118 (14) 82 (9)

Sample purification like in Fig. 1, without or with PSA. Mean value and standard deviation from Burley, Oriental and Virginia tobacco.

fone) and 23.2 min (fluazifop butyl, profenofos). On the other hand, losses at 21 min (fenamiphos, penconazole) seem to be caused by signal suppression. Three main tobacco types were used for method validation. Regarding recovery, there were large differences in only two cases. The low recovery of thiophanate-methyl in Oriental tobacco seems to be caused by reduced stability (see Section 3.6). Furthermore, the low recovery of dimefox in Burley seems to be caused by signal suppression. 3.3. Losses during dispersive solid phase extraction

8958

B. Mayer-Helm / J. Chromatogr. A 1216 (2009) 8953–8959

Table 4 Results from the FAPAS proficiency test FT0104, reanalyzed during validation. Pesticide

Fenamiphos sulfoxide Fenamiphos sulfone Fenamiphos Iprodione a

FAPAS Spiked value (␮g/g)

Assigned value (␮g/g)

0.100 0.100 0.100 0.500

0.115 0.104a 0.079 0.595a

Measured (␮g/g)

Calculated z-score

0.104 0.086 0.077 0.545

−0.44 −0.78 −0.12 −0.49

The assigned value was given for information only and not for evaluative purposes as the uncertainty of the robust mean was large.

ted, largely improved recoveries were obtained. Here, degradation of the pesticides is likely, since degradation of captan and folpet occurs during dSPE with PSA [29]. 3.4. Sensitivity The CORESTA recently published a list of GRLs for 118 agrochemicals on tobacco and tobacco products [3]. Almost all pesticides can be quantified below the GRLs and detected below the 10 ng/g level (see supplementary Table S2). Hardly any differences regarding sensitivity between the different tobacco types could be found. Virginia has sometimes lower LOQs than Burley and Oriental, as less baseline distortions are observed. The transitions were chosen so that the first was used for quantification (usually with a higher sensitivity) and the second for confirmation. Nevertheless, the sensitivity of both transitions is in the same range for all pesticides with one exception. The second transition for methomyl-oxim is less sensitive than the first one. However, this is the first report where a second transition for methomyl-oxim is reported. The ionization of methomyl-oxim was reported to be drastically inhibited under ammonium formate conditions [4]. 3.5. Confirmation of identity The unambiguous confirmation of the identity of organic residues was performed by measuring two transitions for each pesticide and by calculating the relative intensities. Positive results from routine samples were confirmed when the relative intensities were within the tolerances recommended by the EU [22]. The mean values of the relative intensities in all three tobacco matrices had low relative standard deviations (see Table 1, last column). This gives evidence that there are hardly any differences between the tobacco types regarding confirmation. 3.6. Stability of pesticides in tobacco extracts The stability of pesticides in water and MeOH may be problematic due to hydrolysis. To investigate the stability of the analytes the extracts were stored light-protected in the autosampler tray at 8 ◦ C for one week. The tobacco extracts were reconstituted after solvent evaporation in water/MeOH (DCM fraction) or MeOH (pentane fraction), both containing 5 mM ammonium acetate. Both solvents were acidified with 0.1% acetic acid, as pesticides are more stable at a lower pH [30]. All pesticides were found to be stable in all three tobacco matrices for six days. The only exceptions were demeton-S-methyl sulfoxide and demeton-S-methyl sulfone which were degraded to 75% within six days in the three tobacco matrices. Furthermore, thiophanate-methyl was significantly unstable in Oriental tobacco with a loss of 50% within six days, while it was stable in Burley, Virginia and standard solution. Thiophanate-methyl is a propesticide that is converted to carbendazim as a biologically active compound [31]. Despite its instability, no conversion to carbendazim could be observed in Oriental tobacco. The rather high stability of the pesticides in tobacco extracts is surprising as some of

them were reported to be unstable: pymetrozine is acid-sensitive [14]. Pesticides with a thioether group (demeton-S-methyl, disulfoton, fenamiphos, terbufos, vamidothion) are prone to be oxidized into the corresponding sulfoxides and sulfones and are unstable in ethyl acetate. Disulfoton degrades even in acetone [30]. Methamidophos was found to be unstable in ethyl acetate, wheat and orange extracts, iprodione was unstable in cabbage extracts [32]. 3.7. Trueness A test sample from a former proficiency test was used to assess the trueness of the method. In April 2008, the CORESTA Sub-Group Agrochemicals Analysis performed the proficiency test FT0104 conducted via FAPAS® (Central Science Laboratory, Sand Hutton, UK). The tobacco sample was spiked with nine pesticides at concentrations near their respective GRLs. This test sample was reanalyzed with the present method. Quantification was performed via matrixmatched calibration curves, which were established with a blank tobacco. Four of the nine pesticides were covered by the present method. The results excellently agree with the assigned values indicated by good z-scores (see Table 4). 4. Conclusions An improved method for the determination of 52 pesticides in tobacco was presented. Two cleanup steps were developed for matrix removal, covering the whole polarity range of pesticides. The combination of consecutive elution from a Chem Elut cartridge with a Florisil SPE enables the injection of higher concentrated samples, i.e. 500 mg/mL. Several important experimental details were reported: (i) mobile phase additives have an impact on ionization. The highest MS response in standard solution does not guarantee the highest sensitivity in the sample matrix due to signal suppression. This complicates the comparison of different MS instruments regarding sensitivity, as both mobile phase and sample matrix must be considered. (ii) Some pesticides are removed by PSA during dSPE sample purification. (iii) The stability of few pesticides can be improved by the addition of acetic acid. (iv) The different tobacco types Burley, Oriental and Virginia are similar regarding recovery, sensitivity, relative intensity and stability. However, some remarkable exceptions must be taken into account. Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j.chroma.2009.10.077. References [1] L. Alder, K. Greulich, G. Kempe, B. Vieth, Mass Spectrom. Rev. 25 (2006) 838. [2] European Union, Pesticide EU-MRLs Database, Regulation (EC) No. 396, 2005, http://ec.europa.eu/sanco pesticides/public/index.cfm. [3] Cooperation Centre for Scientific Research Relative to Tobacco, Guide No. 1: The Concept and Implementation of Agrochemical Guidance Residue Levels, 2008, http://www.coresta.org/Guides/Guide-No1-GRLs(2nd-Issue-0708).pdf. [4] M. Hiemstra, A. de Kok, J. Chromatogr. A 1154 (2007) 3.

B. Mayer-Helm / J. Chromatogr. A 1216 (2009) 8953–8959 [5] I.R. Pizzutti, A. de Kok, R. Zanella, M.B. Adaime, M. Hiemstra, C. Wickert, O.D. Prestes, J. Chromatogr. A 1142 (2007) 123. [6] C.L. Hetherton, M.D. Sykes, R.J. Fussell, D.M. Goodall, Rapid Commun. Mass Spectrom. 18 (2004) 2443. [7] C. Jansson, T. Pihlström, B.-G. Österdahl, K.E. Markides, J. Chromatogr. A 1023 (2004) 93. [8] B. Mayer-Helm, L. Hofbauer, J. Müller, Rapid Commun. Mass Spectrom. 20 (2006) 529. [9] C. Díez, W.A. Traag, P. Zommer, P. Marinero, J. Atienza, J. Chromatogr. A 1131 (2006) 11. [10] A. Hercegová, M. Dömötörová, D. Kruˇzlicová, E. Matisová, J. Sep. Sci. 29 (2006) 1102. [11] C. Soler, Y. Picó, Trends Anal. Chem. 26 (2007) 103. ˇ [12] M. Anastassiades, S.J. Lehotay, D. Stajnbaher, F.J. Schenck, J. AOAC Int. 86 (2003) 412. [13] S.J. Lehotay, A. de Kok, M. Hiemstra, P. van Bodegraven, J. AOAC Int. 88 (2005) 595. [14] S.J. Lehotay, K. Maˇstovská, A.R. Lightfield, J. AOAC Int. 88 (2005) 615. [15] S.J. Lehotay, K. Maˇstovská, S.J. Yun, J. AOAC Int. 88 (2005) 630. [16] K. Banerjee, B.P. Oulkar, S. Dasgupta, S.B. Patil, S.H. Patil, R. Savant, P.G. Adsule, J. Chromatogr. A 1173 (2007) 98. [17] M. Liu, Y. Hashi, Y. Song, J.-M. Lin, J. Chromatogr. A 1097 (2005) 183. [18] L. Li, W. Li, J. Ge, Y. Wu, S. Jiang, F. Liu, J. Sep. Sci. 31 (2008) 3588.

8959

[19] S. Walorczyk, Rapid Commun. Mass Spectrom. 22 (2008) 1. [20] J.-M. Lee, J.-W. Park, G.-C. Jang, K.-J. Hwang, J. Chromatogr. A 1187 (2008) 25. [21] T. Anspach, M. Linkerhägner, QuEChERS Multi Residue Method–An Effective Tool for the Determination of Agrochemical Residues in Tobacco by LC–MS/MS, CORESTA Congress, Abstract PT22, Paris, 2006, http://www.coresta.org/Past Abstracts/Paris2006-SmokeTech-Oct06.pdf. [22] European Union, Method Validation and Quality Control Procedures for Pesticide Residues Analysis in Food and Feed, Document No. SANCO/2007/3131, http://www.ec.europa.eu/food/plant/protection/resources/qualcontrol en.pdf. [23] A.C. Hogenboom, M.P. Hofman, S.J. Kok, W.M.A. Niessen, U.A.T. Brinkman, J. Chromatogr. A 892 (2000) 379. [24] B.K. Choi, D.M. Hercules, A.I. Gusev, Fresenius J. Anal. Chem. 369 (2001) 370. [25] A. Leitner, J. Emmert, K. Boerner, W. Lindner, Chromatographia 65 (2007) 649. [26] A. Kruve, A. Künnapas, K. Herodes, I. Leito, J. Chromatogr. A 1187 (2008) 58. [27] J. Klein, L. Alder, J. AOAC Int. 86 (2003) 1015. [28] B. Mayer-Helm, L. Hofbauer, J. Müller, Talanta 74 (2007) 1184. [29] T. Cajka, J. Hajslova, O. Lacina, K. Mastovska, S.J. Lehotay, J. Chromatogr. A 1186 (2008) 281. [30] K. Maˇstovská, S.J. Lehotay, J. Chromatogr. A 1040 (2004) 259. [31] J.W. Vonk, A. Kaars Sijpesteijn, Pestic. Sci. 2 (1971) 160. [32] V. Kocourek, J. Hajˇslová, K. Holadová, J. Poustka, J. Chromatogr. A 800 (1998) 297.