Matrix solid-phase dispersion and solid-phase microextraction applied to study the distribution of fenbutatin oxide in grapes and white wine

This is the postprint(accepted manuscript) version of the article published in Analytical and Bioanalytical Chemistry https://doi.org/10.1007/s00216-009-3175-0 This manuscript version is made available under de CC-BY-NC-ND 4.0 license hppt://creativecommons.org/licenses/by-nc-cd/4.0 Matrix solid-phase dispersion and solid-phase microextraction applied to study the distribution of fenbutatin oxide in grapes and white wine R. Montes • P. Canosa • J. Pablo Lamas • A. Piñeiro • I. Orriols • R. Cela • I. Rodríguez Abstract The fate of the acaricide fenbutatin oxide (FBTO) during the elaboration of white wine is evaluated. Matrix solid-phase dispersion (MSPD) and solid-phase microextraction (SPME) were used as sample preparation techniques applied to the semi-solid and the liquid matrices involved in this research, respectively. Selective determination of FBTO was achieved by gas chromatography with atomic emission detection (GC–AED). GC coupled to mass spectrometry was also used to establish the identity of FBTO by-products detected in must and wine samples. MSPD extractions were accomplished using C18 as dispersant and co-sorbent. Sugars and other polar interferences were first removed with water and water/acetone mixtures, then FBTO was recovered with 8 mL of acetone. When used in combination with GC–AED, the MSPD method provided limits of quantification (LOQs) in the low nanogram per gram range, recoveries around 90% and relative standard deviations below 13% for extractions performed in different days. Performance of SPME for must and wine was mainly controlled by the extraction temperature, time and fibre coating. Under final conditions, FBTO was extracted in the headspace mode for 45 min at 100 °C, using a 100 μm poly(dimethylsiloxane)-coated fibre. The achieved LOQs remained around or below 0.1 ng mL−1, depending on the type of sample, and the inter-day precision ranged from 10% to 13%. FBTO residues in grapes stayed mostly on the skin of the fruit. Although FBTO was not removed during must and white wine elaboration, it remained associated with suspended particles existing in must and lees, settled after must fermentation, with a negligible risk of being transferred to commercialised wine. On the other hand, two by-products of FBTO (bis and mono (2-methyl-2-phenylpropyl) tin) were identified, for first time, in must and final white wines obtained from FBTO treated grapes. Found values for the first species ranged from 0.03 to 0.9 ng mL−1. Keywords Fenbutatin oxide . Grapes . Wine . By-products . Sample preparation . Organotin species Introduction Organotin compounds, particularly tri-substituted species, are well-known biocides. One of their most important applications was as antifouling additives in paints used to prevent the attachment of marine organisms to the hull of vessels. The widespread use of this practice has led to the ubiquitous presence of tributyltin (TBT) and its degradation products, in commercial harbours, estuaries and other aquatic environments with intense ship traffic [1–3]. This fact, added to the persistence of butyltin species in sediments [4] and their adverse effects on some aquatic organisms [5, 6], has caused the international ban of organotin-containing paints [7], although in a smaller scale some organotin compounds are also used as biocides in agriculture. Among them, fenbutatin oxide (bis [tris (2- methyl-2-phenylpropyl) tin] oxide (FBTO)) is one of the most relevant and less studied ones. This species is mainly used as acaricide, being particularly effective against red mites (Tetranychus urticae). On the other hand, it shows a much lower toxicity towards predator mites (e.g. Phytoseiulus persimilis), which are considered as potential biological control agents [8, 9]. Because of its extremely high octanol– water partition coefficient (log Kow 12.8) and chemical stability, FBTO might remain for long time in the treated food stuffs. In the European Union, relatively high FBTO residues are allowed in vegetables (up to 5 μg g−1 for some commodities) [10]; therefore, the risk of violating this threshold is limited. However, this implies that consumers could be exposed to concentrations of FBTO six orders of magnitude higher than the level at which other organotin compounds, e.g. TBT, exert endocrine disruption effects on invertebrates [11]. Surprisingly, the determination of FBTO in foodstuff samples has received very little attention. Gas chromatography (GC)-based methodologies require a previous derivatisation step in order to break down the ether bond in the structure of FBTO rendering two lower molecular weight, tetra-substituted organotin derivatives [12, 13]. Devos and co-workers have demonstrated that this reaction can be accomplished in aqueous media, using sodium tetraethyl borate (NaBEt4) [14], as previously reported for other organotin species [15]. Liquid chromatography (LC)-based methodologies avoid the aforementioned derivatisation step; however, the extreme non-polar character of FBTO makes difficult its separation using reversed-phase columns [16, 17]. As regards sample preparation, Soxhlet and manual shaking with organic solvents have been the most resorted approaches for the extraction of FBTO from environmental matrices, e.g. sediments [12, 13], soils [18] and vegetables [16, 17]. High time and solvent consumption, in addition to limited selectivity, are some of the most significant drawbacks of these approaches. As recently published for soil matrices [19], modern sample preparation techniques allow to overcome some of these limitations. Aims of this research are (1) to evaluate the distribution of FBTO in white grapes from vineyards fumigated with this acaricide and (2) to assess its persistence during the winemaking process. For this, two sensitive methods, based on the use of matrix solid-phase dispersion (MSPD) and solid-phase microextraction (SPME), for the extraction and concentration of FBTO from grapes and semi-solid residues generated during wine elaboration (lees), must and wine samples have been optimised. After extraction and ethylation, FBTO and its by-products were selectively determined by GC with atomic emission detection (AED). Although MSPD is a popular technique for the extraction of agrochemicals in fruits and vegetables [20, 21], as far as we could trace, it has never been applied neither to FBTO nor to any other organotin species. On the other hand, the suitability of SPME for the extraction of organotin compounds from complex liquid matrices (e.g. wine) has been reported in several works [22–24]; however, in the case of FBTO, SPME has been just applied to water samples [14], and some important parameters, such as the effect of the fibre coating on the efficiency of the extraction and on mass transfer kinetics, have not been evaluated yet. Experimental Solvents, reagents and materials Acetone, methanol and isooctane, trace analysis grade solvents, were obtained from Merck (Darmstadt, Germany). FBTO, triphenyltin chloride (Ph3SnCl), NaBEt4, glacial acetic acid and sodium acetate were purchased from Aldrich (Milwaukee, WI, USA). A standard solution of triphenyl pentyltin (Ph3SnPe) in n-hexane (10 μg mL−1) was provided by IRMM (Geel, Belgium) with an occasion of a former intercomparison exercise. Stock solutions of FBTO (200 μg mL−1) and Ph3SnCl (1000 μg mL−1), used as internal surrogate in the extraction of wine and must samples, were prepared in acetone and methanol, respectively. Further dilutions of both species were made in acetone. The stock solution of Ph3SnPe was diluted with isooctane and used as internal standard (IS) during the determination of FBTO in grapes and lees, sedimented after wine fermentation. Concentrations of organotin species are referred as tin throughout the manuscript. Sodium acetate 0.2 M solutions were prepared in ultrapure water, obtained from a Milli-Q system (Millipore, Billerica, MA, USA) and adjusted at pH 5.5 with glacial acetic acid. The derivatisation reagent, NaBEt4, was also diluted in ultrapure water (0.1 M in NaOH) at a 1% (w/v) concentration. C18 (70–230 mesh), used in the MSPD process, was acquired from Aldrich. Polypropylene solid-phase extraction syringes (15 mL capacity) and 20 µm polyethylene frits were purchased from International Sorbent Technology (Mid Glamorgan, UK). A manual SPME holder and fibres coated with different polymers: poly(dimethylsiloxane) (PDMS, 100 μm film thickness), polyacrylate (PA, 85 μm film thickness) and PDMS–divinylbenzene (PDMS–DVB, 65 μm film thickness) were obtained from Supelco (Bellefonte, PA, USA). Before being used for the first time, fibres were thermally conditioned following instructions provided by the supplier. Samples and sample preparation White grapes (Dona Branca variety), proceeding from two vineyards fumigated with miticides containing FBTO as active ingredient (a.i.), were supplied by EVEGA (Galicia, NW Spain). Must, wine and lees, obtained from FBTO treated grapes, were also provided by EVEGA. White grapes, not fumigated with FBTO, as well as white (Palomino) and red (Grenache) wines, free from FBTO residues, were also used throughout this study. Grapes, must and lees were stored at −18 °C until being analysed. Wine was kept at room temperature in amber bottles. Grapes were removed from the bunch, freeze-dried and thoroughly homogenised using first a domestic blender (Moulinex, Spain) and then a mortar. Lees were also lyophilised before extraction. Under final working conditions, 0.2 g fractions of lyophilised samples (whole grapes, skin, pulp or lees) were mixed with 1 g of C18 in a mortar with a pestle. After that, this blend was poured into a polypropylene syringe containing a polyethylene frit and 1 g of the same sorbent. Another frit was placed on top before slight compression. The packed sample was rinsed with 15 mL of ultrapure water followed by 5 mL of water/acetone (8:2), by gravity. After that, FBTO was recovered with 8 mL of acetone. This extract was concentrated to 1 mL, using a gentle stream of nitrogen at room temperature, and derivatised under conditions described in a previous work [19]. Briefly, 0.5 mL of acetone extracts and 10 mL of buffer solution at pH 5.5 were poured in a glass tube together with 1 mL of NaBEt4 (1% w/v, aqueous solution). The tube was capped and shaken vigorously for 5 min. The ethylated form of FBTO was then extracted with 2 mL of isooctane containing a known concentration of Ph3SnPe (IS). The upper phase was injected directly in the chromatographic system or concentrated to a lower volume (ca. 0.2 mL) if required. In the latter case, variations in the final volume of the extract were compensated with the response measured for Ph3SnPe. Optimization of MSPD conditions was carried out using white grapes treated with FBTO before harvest. Recoveries of the proposed sample preparation method were assessed with samples fortified in the laboratory. The spiking procedure consisted of adding a standard of FBTO, made in acetone, to a known mass of freeze-dried grapes. After thorough homogenisation, the slurry was left in a hood, at room temperature, for 1 week before extraction. Wine and must, corresponding to treated and non-treated FBTO grapes, were concentrated using the SPME technique. Unless otherwise is stated, both matrices were processed as received, without further filtration in the laboratory. Aliquots of 1 mL were diluted with 10 mL of the sodium acetate– acetic acid buffer (0.2 M, pH 5.5) in 20 mL volume glass vessels, furnished with PTFE layered silicone septa and aluminium crimp caps. Ph3SnCl (50 μL of a 10 ng mL−1 standard solution) and 1 mL of NaBEt4 were added. Vessels were closed and shaken vigorously for 5 min in order to convert FBTO and the internal surrogate into the corresponding ethyl derivatives. After that, vessels were equilibrated at a given temperature, for 10 min, and a SPME fibre was exposed to the sample. Optimisation of SPME conditions was carried out with white wine samples fortified with FBTO and Ph3SnCl at 5 and 0.5 ng mL−1, respectively. Under final working conditions, extraction vessels were introduced in a water boiling bath, and compounds were concentrated in a PDMS-coated fibre exposed to the headspace (HS) of the vial for 45 min. Determination Levels of FBTO in the processed samples were determined by GC–AED using an Agilent (Wilmington, DE, USA) 6890 GC system, equipped with a split/splitless injector and connected to a G2350A atomic emission detector. Separations were carried out with an Agilent HP-5 type capillary column (30 m×0.32 mm i.d., df 0.25 μm) operated at a constant helium flow of 1.4 mL min−1. The GC oven was programmed as follows: 80 °C (held for 2 min), 15 °C min−1 to 300 °C (held for 5 min). Injections (1–2 μL) were made in the pulsed splitless mode (40 psi, 2 min), using an autosampler. The injection port was maintained at 270 °C, whereas the transfer line between the GC and the helium microwave-induced plasma and the plasma cavity were set at 300 °C. Helium was used as make-up gas in the plasma at 270 mL min−1. O2 and H2 were also added to the plasma as auxiliary gases at pressures of 30 and 20 psi, respectively [19]. Chromatograms were recorded at 303 nm. GC combined with mass spectrometry (MS) was also used to establish the identity of some organotin species (FBTO by-products) detected in must and wine samples. The employed system was comprised of a Varian (Walnut Creek, CA, USA) 450 GC instrument connected to an iontrap Varian 240 MS. Separations were carried out in a Varian Factor Four VF-5 ms type capillary column (30 m× 0.25 mm i.d., df 0.25 µm) using a constant helium flow of 1.2 mL min−1. The temperature of the inlet, the injection mode and the oven temperature program were the same as in the GC–AED system. Transfer line, electron impact ionisation source and trap were set at 285, 200 and 150 °C, respectively. MS spectra were recorded in the scan (80–600 m/z units) and also in the sim mode, selecting a reduced number of ions for each FBTO by-product. Levels of FBTO in whole grapes, pulp, skin and lees were quantified by comparison to standards derivatised under same conditions as sample extracts. Calibration curves, in the range from 5 to 1,000 ng mL−1, were obtained by plotting the ratio between the peak areas of FBTO and Ph3SnPe (IS) versus the concentration of the first species. Within the above range, a correlation coefficient (R2) of 0.9994 was obtained, whereas the instrumental limit of quantification (LOQ) for FBTO was 2 ng mL−1, as tin. Efficiency of SPME for wine and must was matrix dependant; therefore, standard addition was used as quantification technique. Calibration graphs corresponded to the representation of the ratio between FBTO and Ph3SnCl (IS) peak areas versus the concentration of FBTO added to the liquid sample contained in the SPME vessel. The same quantification methodology was used to calculate the concentrations of bis (2-methyl-2-phenylpropyl) tin. Results and discussion MSPD of semi-solid samples Optimisation of extraction conditions Preliminary MSPD extractions were performed using frozen grapes homogenised with a blender. Fractions (0.5 g) of this matrix were blend with 3 g of C18, in a mortar with a pestle, until a dry powder was obtained, which was then poured over 1 g of the same sorbent in a polypropylene syringe. Acetone, which had been employed for the extraction of FBTO from soil samples [19], was selected as elution solvent. Operating under these conditions, a viscous extract, with a high content in sugars, was obtained. Sugars led to the formation of an emulsion, making impossible the separation between the aqueous phase and the upper isooctane layer, added after the ethylation of FBTO. This limitation has overcome the rinsing of the dispersed samples first with 15 mL of ultrapure water and then with 5 mL of a water/acetone (8:2) mixture. Sugars were eluted with two of these fractions, whereas the highly lipophilic FBTO remained in the sample. In a further step, it was recovered using 15 mL of acetone. However, the repeatability of the above procedure was extremely poor, with relative standard deviations (RSDs) around 40% for extractions carried out in the same day. On the other hand, contamination problems were not detected, extractions appeared to be quantitative, and the repeatability of the derivatisation step, for aliquots of the same extract, stayed below 4%; thus, lack of repeatability seemed to be a consequence of dealing with heterogeneous samples. In order to overcome this drawback, freeze drying was considered as sample pre-treatment. Approximately 100 g of frozen grapes (whole grape, pulp or skin) was first lyophilised and further thoroughly grounded using first a blender and then a mortar. MSPD extractions were repeated with 0.2 g of the resulting powder, equivalent to ca. 2 g of fresh sample. In this case, only 1 g of C18 was used as dispersant. Sugars were removed using same conditions as those reported for fresh (not freeze-dried) grapes, whereas methanol and acetone were considered as extractants. The latter solvent provided higher responses than the former one; thus it was maintained as extractant. By collecting consecutive fractions from the MSPD syringe (2 mL each), it was verified that 8 mL of acetone sufficed to recover 99% of the FBTO contained in the sample (Fig. 1). This represents a considerable reduction in the consumption of organic solvents when compared to other methods based on manual shaking of vegetable commodities with dichloromethane or ethyl acetate to extract FBTO [16, 17]. MSPD performance Precision of the MSPD method was estimated with freezedried samples corresponding to whole grapes obtained from a vineyard fumigated with FBTO. RSDs from 10% to 12% were attained for extractions (n =4 replicates) performed in the same day, whereas a RSD of 13% was achieved for extractions carried out in three consecutive days (n=12 replicates; Table 1). Accuracy was evaluated using grapes (Palomino variety) spiked a two different concentrations (70 and 200 ng g−1 referred to the fresh sample), below the maximum allowable residue of FBTO in this commodity (450 ng g−1 as tin) [10]. Obtained recoveries, estimated against calibration standards prepared as described in the “Experimental” section, were 96±7% and 89±11% for samples fortified at 70 and 200 ng g−1, respectively. The LOQ of the proposed method, defined as the concentration of FBTO producing a signal ten times higher than the standard deviation of the blank, was 2 ng g−1, referred to freeze-dried samples, which is equivalent to 0.2 ng g−1 in fresh grapes. This value is ten times lower than those previously reported for the determination of FBTO in different vegetables and fruits using LC–MS [16]. SPME of liquid matrices Optimisation study Departure conditions for SPME of FBTO from wine and must samples were those previously reported by Devos and 100 90 80 70 60 50 40 30 20 10 0 1 2 3 4 5 Fraction number (2 mL each) Fig. 1 Normalised responses in the consecutive fractions (2 mL of acetone each) collected from the MSPD syringe, n=4 replicates 5 4 3 2 1 0 3 4 5 Normalized signal (%) Normalized signal (%) Table 1 Performance of MSPD for the extraction of FBTO from freeze-dried grapes Precision (RSD%) Recovery (%) ± SD LOQ (ngg−1 ) Day 1a Day 2a Day 3a Inter-dayb 70ngg −1c 200ngg− 1c 10 12 11 13 96±7 89±11 2 a n =4 replicates b n =12 replicate extractions performed in three consecutive days c Added concentration referred to wet weight co-workers for water [14]. Taking into account the limited volatility of the ethyl derivative of FBTO, the extraction temperature was the first factor considered in the optimisation study. After the derivatisation step, SPME vessels were first equilibrated at different temperatures, using a water bath, and then fibres were exposed to their HS for 30 min. The efficiency of the extraction was negligible below 60 °C, and then it increased dramatically until 100 °C. As example, 10-fold higher responses were noticed when the bath temperature rose from 80 to 100 °C. Comparison between FBTO responses attained in direct and HS modes, at 100 °C, proved the higher efficiency of the latter one; data not shown. Figure 2 shows the responses obtained using three different fibre coatings. Carboxen fibres were not tested since they have been reported as unsuitable for the extraction of large molecular weight compounds [25]. As observed, the highest efficiency was achieved with the PDMS fibre (Fig. 2). Probably, the non-polar character of the ethyl derivative of FBTO, added to its higher diffusion coefficient in the liquid PDMS versus PA and PDMS–DVB sorbents, are responsible for the depicted trend. Sample stirring, using PTFE-coated stir bar, led to a slight reduction in the efficiency of the extraction. Adsorption of FBTO on the surface of the stir bar might explain this behaviour. Increasing the ionic strength of the sample in the SPME vessel with sodium chloride, after performing the ethylation reaction, also reduced the yield of the microextraction. In this case, it can be assumed that the rate of mass transfer from the liquid sample to the HS of the vial decreases with the increase in the viscosity of the matrix. Figure 3 shows the time course of the HS SPME process at 100 °C. Around 2 h was required to achieve equilibrium conditions. In further extractions, the sampling step was limited to 45 min as a compromise between sensitivity and sample throughput. Memory effects were evaluated by desorbing PDMS fibres in consecutive steps of 2 min at 270 °C. The response of FBTO in the second desorption represented less than 3% of that measured in the first one. In order to prevent crosscontamination problems, PDMS fibres were left in the injector of the gas chromatograph for 5 min. After 2 min, the solenoid valve was passed from the splitless mode to the split one to sweep any trace of FBTO remaining in the coating. SPME figures of merit Table 2 shows some data related to linearity, precision and LOQs of the SPME method for white and red wine, as well as must obtained from white grapes. The correlation coefficient (R2) values were higher than 0.999 for the three matrices. RSDs obtained for samples spiked at different levels, and performed in the same or in consecutive days, varied from 4% to 15%. Achieved LOQs were matrix dependant, ranging from 0.05 to 0.1 ng mL−1. These LOQs 500 450 400 350 300 250 200 150 100 50 0 PDMS PDMS-DVB PA 400 350 300 250 200 150 100 50 0 0 50 100 150 200 Time (min) Fig. 2 Comparison of FBTO responses using different SPME-coated fibres, n=4 replicates. Data for HS SPME at 100 °C for 30 min Fig. 3 Kinetics of the HS SPME at 100 °C for white wine using a PDMS-coated fibre. Average values for duplicate extractions FBTO peak area FBTO peak area Table 2 Performance of the HS SPME method for the determi- White wine Red wine Musta nation of FBTO in wine and must samples Linearity (R2,0.1–20 ng mL−1)b 0.9994 0.9991 0.9990 Calibration levels 6 6 6 Replicates per level 2 2 2 Slope 0.057 0.042 0.040 Standard deviation of slope 0.001 0.001 0.002 Intercept 0.011 -0.002 0.017 Standard deviation of intercept 0.004 0.004 0.006 Intra-day precision (RSD%,n=4 replicates) a Must from white grapes b Ratios between FBTO and Ph3SnCl (0.5 ng mL−1 ) peak areas were plotted versus the concentration of FBTO c Added concentration 0.5 ng mL−1c 7 4 9 Inter-day precision (RSD%,n=10 replicates) 2 ng mL−1c 12 13 10 LOQ (ng mL− 1 ) 0.05 0.1 0.1 are similar to those attained for water samples using the same concentration technique [14], and 100 times lower than those previously reported for the determination of FBTO in wine [26]. Matrix effects were assessed by comparing the normalised responses (ratios between peak areas for the ethyl derivatives of FBTO and Ph3SnCl) measured for different samples fortified at the same concentration level. Similar responses were noticed for water and white wine, whereas a significant reduction was observed for red wine and must (Fig. 4). This behaviour is in agreement with differences among sensitivities (slopes of standard addition curves) of the SPME method for the three latter matrices (Table 2). On the basis of these results, quantification has to be performed with addition of standards over must and wine samples. Field studies White grapes, from two different vineyards fumigated with FBTO containing acaricides (55 g of a.i. per 100 L of sprayed water) as well as the corresponding must and wine samples, were processed. Table 3 summarises the results 1.2 1 0.8 0.6 0.4 0.2 0 Water White wine Red wine Must Fig. 4 Relative extraction efficiencies of HS SPME for different liquid matrices. Normalised values to ultrapure water, n=4 replicates obtained for the first matrix. In addition to whole grapes, skin and pulp were also processed. In the three cases, around 100 g of frozen samples was accurately weighed and lyophilised before extraction. Given values are expressed as wet weight, after correction with the percentage of water removed during freeze drying. In agreement with the high lipophility of FBTO (calculated log kow 12.6), its concentration in the skins was one order of magnitude higher than in the whole fruit, whereas the residue found in the pulp was very low, in the case of grapes from vineyard 1, or it even remained below the LOQ or the method (ca. 0.2 ng g−1) in grapes from vineyard 2. Thus, FBTO behaves as other contact fungicides, e.g. folpet, which stay mostly on the skin of grapes, showing a low capability to migrate to the pulp [27]. The large difference between concentrations found in both grape samples is not well understood. It might be related with the surface of grapes when the treatment was applied and the time passing between application and harvest. Figure 5 shows the GC–AED chromatogram obtained for must elaborated with grapes containing a relatively high Table 3 Distribution of FBTO in white grapes fumigated with this acaricide Vineyard code Wet weight concentration (ngg−1±standard deviation) Whole grapes Pulp Skin 1 a 115±19 25±2 1,570±120 2b 1.8±0.1 n.q. 35±4 Average values, referred to wet weight, for n =4 replicates n.q. below quantification limits a Treated with FBTO 55 days before harvest b Treated with FBTO 77 days before harvest Normalized response 2.0 ng mL −1c 15 8 10 5.0 ng mL− 1c 4 10 8 Fig. 5 GC–AED chromatogram for must elaborated from grapes containing 115 ng g−1 of FBTO Counts 175 150 125 100 75 50 25 0 10 12 14 16 18 min concentration of FBTO (115±19 ng g−1). The residue of FBTO found in this sample was 0.33±0.06 ng mL−1. When must was additionally filtered in the laboratory, using a 0.45 μm cellulose acetate membrane, the concentration of FBTO decreased below the LOQ of the method (less than 0.1 ng mL−1). This reduction indicates that the miticide stays mainly associated with suspended particles (basically crushed skin pieces) remaining in must after coarse filtration in the cave. FBTO was not be detected either in must from grapes with the lower concentration of the acaricide (1.8±0.1 ng g−1), or in wine elaborated with grapes from vineyards codes 1 and 2 (Table 3). The concentration of FBTO in lees, settled after the fermentation of must elaborated with grapes belonging to vineyard code 2, was 31±6 ng g−1, a value very close to the concentration determined in the skin of grapes (35 ± 4 ng g−1). Unfortunately, lees were not available for grapes harvest in vineyard code 1. Anyhow, the above finding, combined with the distribution of FBTO between skin and pulp of grapes, suggests that although FBTO is not removed during must fermentation; it remains adsorbed on solid wastes that settled at the bottom of the stainless steel containers. Thus, physical operations normally performed in caves, e.g. filtration and sedimentation, prevent the presence of FBTO in bottled white wines. In addition to FBTO, the chromatogram depicted in Fig. 5 presents two extra peaks with similar intensities to that of the internal surrogate (added to the sample at 0.5 ng mL−1). Both peaks were also detected considering additional emission lines of Sn, at 301 and 271 nm, figure not shown; thus, very likely, they correspond to organotin species. Their MS spectra, obtained by operating the iontrap-type GC–MS system in the scan mode, and also selecting a limited number of ions (sim mode), are shown in Fig. 6. Both species were identified as FBTO by-products, resulting from cleavage of the ether bond between tin atoms, followed by losses of one or two methylphenylpropyl moieties, leading to bis and mono (2- methyl-2-phenylpropyl) tin cationic species (Fig. 7). In the presence of NaBEt4, they are converted into the corresponding ethyl derivatives and transferred to the PDMS fibre, when exposed to the HS of the extraction vessel. The base peaks in their MS spectra, appearing at m/z values of [M-29]+, reflected the loss of one of the ethyl groups introduced in the reaction with NaBEt4. Other fragments corresponded to further losses of ethyl and 2- methyl-2-phenylpropyl moieties or their replacement by hydrogen (Fig. 6). This pattern has been previously reported for the ethyl derivative of FBTO [14], and other organotin species, under electron impact ionisation. Finally, some cluster of ions, e.g. those at m/z ratios of 359, 227 and 197, suggest a rearrangement of the 2-methyl-2-phenyl- propyl moiety, with a loss of the alkyl chain and migration of the phenyl group to the atom of tin. In addition to the above information, the isotopic pattern for the cluster of ions corresponding to base peaks in the spectra of bis and mono (2-methyl-2-phenylpropyl) tin, acquired in the sim mode, matched with the natural abundance of most intense tin isotopes (120Sn, 100%; 118Sn, 73%; and 116Sn, 43%; Fig. 6). The identity of by-product number 2 was finally confirmed with a standard of bis (2-methyl-2-phenylpropyl) tin dichloride (CAS number 14208-42-5) acquired from ABCR (Karlsruhe, Germany) and derivatised under the same conditions as FBTO. The GC–AED system showed a similar linear response range (5–1,000 ng mL−1) for the ethyl derivative of this compound than for FBTO, with a R2 value of 0.9990. However, a twice higher sensitivity (defined as the slope of the calibration curve) and a lower instrumental LOQ (1 ng mL−1) was achieved for bis (2-methyl-2-phenyl- propyl) tin. Probably, these differences are related to a higher efficiency of the splitless injection process for the by-product than for the parent acaricide, due to the higher volatility of the former species. Linearity details of the SPME method for bis (2-methyl-2-phenylpropyl) tin, operating under conditions optimised for FBTO, are summarised in Table 4. Slopes of standard addition curves are around 20 times higher than those reported in Table 3 for FBTO, indicating a higher yield for the SPME of bis (2- methyl-2-phenylpropyl) tin. Consequently, lower LOQs, By-product 1 Ph3SnEt (IS) By-product 2 FBTO Fig. 6 Scan and sim MS spectra of FBTO by-products. R 2- methyl-2-phenylpropyl, Ph phenyl 100% 75% 50% m/z 25% 0% 100 200 300 400 500 600 m/z 100% 75% 50% m/z 25% 0% 100 150 200 250 300 350 m/z around 0.01 ng mL−1, can be achieved. On the other hand, linearity cannot be insured for concentrations over 5 ng mL−1, due to saturation of the AED detector. As compared to FBTO, its by-products could not be completely removed from must with additional filtration of this matrix in the laboratory, using 0.45 μm cellulose acetate membranes. After this operation, their responses underwent a reduction of 60±9% and 16± 2% for bis and mono (2-methyl-2-phenylpropyl) tin, respectively. Both organotin species were also detected in wine. Concentrations of bis (2-methyl-2-phenylpropyl) tin, calculated with the standard addition method, in the four processed liquid samples ranged from 0.033 to 0.9 ng mL−1, expressed as tin (Table 5). Levels of 2-methyl-2-phenylpropyltin could not be quantified due to the lack of commercial available standards. Ratios between the SPME response for this compound and the internal standard, added to samples at 0.5 ng mL−1, ranged from 0.2 to 1.15 (Table 5). Thus, its concentration is expected to stay in the low or subnanogram per gram level. By-product 2 SIM spectrum 415 [R2SnEt] + Sn-Et2 [RSnEt2] + [RSnH 2 ]+ [PhSnEtH] + 311 [RPhSnEt]+ 359 [PhSn]+ 197 227 255 Scan spectrum SIM spectrum By-product 1 311 Sn-Et3 [RSnEt2] + 227 [PhSnEtH]+ [RSnH 2 ]+ 255 Scan spectrum Fig. 7 Chemical structures and proposed formation pathway for FBTO by-products Sn-O-Sn FBTO H2O 2 Sn-OH Bis (2-methyl-2-phenylpropyl) tin Sn2+ Sn3+ Mono (2-methyl-2-phenylpropyl) tin Conclusions The suitability of MSPD and SPME for the extraction of FBTO from grapes, must and wine samples has been demonstrated for first time. MSPD provides LOQs far below the maximum allowable residue of FBTO in grapes; moreover, it consumes a low volume of organic solvents in comparison to other sample preparation methodologies applied to the extraction of this miticide from fruits. SPME, followed by GC–AED, allows the sensitive detection of FBTO in must and wine at concentrations below the nanogram per millilitre level; however, the yield of the extraction is sample dependant; thus standard addition is recommended as quantification technique. Results obtained for grapes fumigated with FBTO before harvest, and the corresponding must and wine samples, showed that FBTO stays mostly on the skin of grapes, and suspended particles existing in must and solid wastes settled after must fermentation. Therefore, although FBTO does not disappear during wine elaboration, the risk of this miticide to reach Table 4 Data for standard addition curves corresponding to the SPME of bis (2-methyl-2-phenylpropyl) tin from different matrices Table 5 Concentrations of bis (2-methyl-2-phenylpropyl) tin and ratios between responses of mono (2-methyl-2-phenylpropyl) tin and the internal surrogate (Ph3SnCl, spiked concentration 0.5 ng mL−1) in White wine Red wine Musta must and wine samples elaborated from grapes treated with FBTO Linearity (R2,0.05– 5 ng mL−1)b 0.9980 0.9950 0.9995 Sample type Vineyard code Bis (2-methyl-2- phenylpropyl) tin Mono (2-methyl-2- phenylpropyl) tin Calibration levels 6 6 6 Conc. (ngg−1) ±SD Analyte/IS peak area ± SD Replicates per level 2 2 2 a Must from white grapes b Ratios between bis (2-methyl-2-phenylpropyl) tin and Ph3SnCl (0.5 ng mL−1 ) peak areas were plotted versus the concentration of bis (2-methyl-2-phenylpropyl) tin HS SPME extraction using conditions optimised for FBTO, n=4 replicates. Must was additionally filtered in the laboratory, using 0.45 cellulose acetate membranes, before extraction. Wine was processed as received 3 3 3 Slope 1.08 0.825 0.81 Must 1 0.9±0.1 0.45±0.02 Standard deviation of slope 0.03 0.004 0.02 Wine 1 0.07±0.01 0.021±0.001 Intercept 0.07 0.05 0.016 Must 2 0.07±0.01 1.15±0.02 Standard deviation of intercept 0.03 0.02 0.006 Wine 2 0.033±0.004 0.42±0.02