Alternative sorptive extraction method for gas chromatography determination of halogenated anisoles in water and wine samples

This is the postprint(accepted manuscript) version of the article published in Analytica Chimica Acta https://doi.org/10.1016/j.aca.2007.07.060 This manuscript version is made available under de CC-BY-NC-ND 4.0 license hppt://creativecommons.org/licenses/by-nc-ck/4.0 Alternative sorptive extraction method for gas chromatography determination of halogenated anisoles in water and wine samples R. Montes, I. Rodríguez, E. Rubí, M. H. Bollaín, R. Cela. Departamento de Química Analítica, Nutrición y Bromatología, Instituto de Investigación y Análisis Alimentario, Universidad de Santiago de Compostela, Santiago de Compostela 15782, Spain. Abstract An alternative sorptive microextraction method for the determination of five halogenated anisoles in water and wine matrices is proposed. Analytes were concentrated in an inexpensive and disposable piece of bulk polydimethylsiloxane (PDMS), desorbed with a small volume of organic solvent, and determined by gas chromatography with electron-capture detection (GC-ECD) or tandem mass spectrometry (GC-MS/MS). The influence of several factors on the efficiency of extraction and desorption steps was investigated in detail and the observed behaviour justified on the basis of thermodynamics and kinetics of the solid-phase microextraction technique. Under optimised conditions, analytes were first extracted in the headspace (HS) mode, at room temperature, for 2.5 hours and then desorbed with 1 mL of n-pentane. This extract was further evaporated to 50 µL. The overall extraction yield of the procedure ranged from 40 to 55% and the limits of quantification remained between 0.5 and 20 ng L-1, depending on the compound considered and the detection technique. Precision and linearity of the method were excellent for all species with both GC-ECD and GC-MS/MS detection. Matrix effects were evaluated with different water and wine samples; moreover, the suitability of the PDMS sorbent for storage of analytes, under different conditions, was demonstrated. Keywords: anisoles, polydimethylsiloxane, sorptive extraction, gas chromatography. 1. Introduction The presence of halogenated anisoles in tap water and wine is a matter of concern for consumers and producers because of their unpleasant taste and flavour. The main sources of these off-flavours in both matrices are the chlorinebased disinfectants added to tap water distribution systems and the bleaching solutions used during production of cork stoppers, washing of oak barrels, etc. In the above situations, breakdown products from humic acids and/or lignin may be first oxidized to chlorinated phenols, as well as brominated ones if a source of bromide is available, and the corresponding anisoles may then be produced through bio-methylation reactions [1,2]. Application of chlorinated pesticides in cork oak forests and the use of 2,4,6-tribromophenol as flame retardant could also contribute to the presence of halogenated anisoles in the environment of wine cellars, with the consequent risk of wine contamination. Sensorial threshold levels of anisoles depend on several factors, such as their degree of halogenation, the type of matrix and the perception skills of each person. It is known that consumers will reject wines containing concentrations of 2,4,6-trichloroanisole (TCA) over 10 ng L-1, whereas, an even lower threshold has been reported for the same compound in tap water [3-5]. Fortunately, the perception limits for other haloanisoles are significantly higher [6,7]. Methods for the determination of halogenated anisoles in water and wine samples rely on the use of gas chromatography with electron-capture detection (GC-ECD) or hyphenated to mass spectrometry (GC-MS) [8-10]. In both cases, a previous extraction step is required to achieve quantification limits in the low ng L-1 region. Therefore, sample preparation remains as the key step that controls the overall performance of most procedures developed for the determination of anisoles. Solid-phase extraction (SPE), closed-loop stripping (CLSA) and pervaporation [11-13] have been successfully applied to the concentration of halogenated anisoles from liquid matrices. Moreover, different microextraction techniques: solid-phase microextraction (SPME) [4,7,10,14], stir-bar sorptive extraction (SBSE) [5,15,16] and more recently single-drop microextraction (SDME) [17] have also proved to be effective enrichment approaches. The main advantages of SPME are the availability of different sorbents and the fact that there is no need for instrumental modifications at the level of the injector in the gas chromatograph. Polydimethylsiloxane (PDMS) and divinylbenzene-Carboxen-polydimethylsiloxane (DVB-CAR-PDMS) coated SPME fibres show the best performance for anisoles extraction. On the other hand, SBSE provides higher extraction efficiencies than SPME, therefore, when used in combination with thermal desorption, it allows obtaining lower quantification limits. In addition to PDMS coated fibres and stir bars (Twisters), the bulk polymer has been used as an effective device for infield monitoring studies aimed at measuring time-weighted average levels of several organic compounds in the aquatic environment [18,19]. Moreover, it has also been applied to direct [20,21] and headspace sorptive extraction (HSSE) of priority pollutants [22], and natural products [23] from discrete samples in the laboratory. The advantages of using bulk PDMS as sorbent are: (1) the possibility of customizing the volume of polymer for each particular application, (2) the availability of different formats (lays, tubes, rods) and, (3) the very low cost of the sorbent in comparison with SPME fibres and Twisters. The latter feature makes it possible to use a new piece of sorbent for each extraction, which avoids problems associated with carry-over and cross contamination, allows simultaneously processing as many samples as desired, and provides the possibility of considering the extraction sorbent for analytes storage purposes. The aim of this study was to assess the suitability of rods of bulk PDMS, produced on an industrial scale and cut in the laboratory, for the extraction of five halogenated anisoles from liquid (water and wine) samples. The effect of different variables on the performance of the process was investigated and a set of working conditions proposed after taking into consideration: (1) their influence on the yield of the extraction and (2) the simplicity of the experimental set-up. Figures of merit of the method were calculated using GC-ECD and gas chromatography combined with tandem mass spectrometry (GC-MS/MS), which are the most commonly used techniques for the determination of halogenated anisoles. 2. Experimental 2.1. Standards and materials Acetone, ethyl acetate, n-pentane, diethyl ether, methanol, ethanol, dichloromethane and isooctane (trace analysis quality), as well as sodium chloride, sodium hydroxide, sodium thiosulphate and tartaric acid were acquired from Merck (Darmstadt, Germany). Standards of 2,4-dichloroanisole (DCA), TCA, 2,4,6-tribromoanisole (TBA), 2,3,4,6-tetrachloroanisole (TeCA) and pentachloroanisole (PCA) were purchased from Aldrich (Milwaukee, WI, USA) and Supelco (Bellefonte, PA, USA). The internal standard, 4-iodoanisole (IA), was supplied by Aldrich. Individual solutions of each compound were prepared in acetone, further dilutions and mixtures of them were made in the same solvent. The stock solution of IA was prepared in acetone and then diluted with isooctane. PDMS rolls (20 m length x 2 mm diameter) were purchased from Goodfellow (Bad Nauheim, Germany). Rods with a length of 10 mm were cut in the laboratory with a sharp blade. After that, they were weighed and those with mass differences higher than 2% were discarded. The nominal volume of each rod was 31 µL, similar to the amount of PDMS incorporated in a Twister (10 mm length, 0.5 mm film thickness). Before being used for analytes extraction, rods were soaked with a mixture of dichloromethane: methanol (1:1) for 15 min and then conditioned overnight at 250 ºC under a flow of nitrogen (ca. 50 mL min-1). 2.2. Sample preparation Spiked and non-spiked water and wine samples were used in this study. In the first case, ultrapure, tap, river and treated wastewater (obtained from an urban sewage plant equipped with primary and secondary treatments) were considered. White and red wine samples, from different geographic areas of Spain, were purchased in local markets. Synthetic wine was prepared by addition of tartaric acid (3.5 g L-1) to 12% ethanol solutions in ultrapure water, followed by adjustment of pH to 3.5 by addition of NaOH 1M [9]. Tap water was spiked with sodium thiosulphate (ca. 40 mg L-1) to remove any residual chlorine. All samples, except ultrapure water, were passed through glass fiber filters. Extractions were carried out in 100 mL capacity glass vessels furnished with PTFE-lined septa and aluminium caps. Vessels and caps were purchased from Sugelabor (Madrid, Spain). A rod of PDMS was fixed to the septum with a stainless steel pin (i.d. 0.7 mm), and dipped in the sample or maintained in the headspace (HS) of the vessel, depending on the selected extraction mode. The effect of temperature on the performance of the process was investigated by placing the extraction vessel, containing the sample and the PDMS sorbent, into an oven with a temperature control precision of ± 2 ºC. After finishing the concentration step, PDMS rods were transferred to 2 mL volume GC vials. Analytes were recovered with 1 mL of an appropriate solvent, using sonication (5 min) to enhance the efficiency of the desorption. Details regarding manipulation of the rods during the desorption step have been provided elsewhere [22]. Under optimised conditions, 80 mL of a spiked sample (water or wine) and 24 g of sodium chloride were placed in each vessel. Analytes were extracted in the HS mode, at room temperature for 2.5 hours. In a further step, they were recovered from the PDMS sorbent with 1 mL of n-pentane. This extract was spiked with the internal standard (a solution of IA in isooctane) and evaporated until a final volume of 50 µL under a gentle stream of nitrogen. A scheme of the sample preparation process is shown in Fig. 1. 2. Desorption 3. Concentration 1. Extraction 80 mL sample (30% NaCl) PDMS rod HS, 2.5 hours Room Tª PDMS rod + 1 mL n-pentane Soaking, 5 min 1 mL extract + 50 µL I.S. in isooctane N 2 evaporation to 50 µL Fig. 1. Scheme of the sample preparation procedure . 2.3. Equipment Anisoles were determined by GC-ECD or GC-MS/MS. In the first case, an Agilent 6890 gas chromatograph (Wilmington, DE, USA) equipped with a micro-electron-capture detector (micro-ECD) was used. The GC-MS/MS system consisted of a Varian CP 3900 gas chromatograph (Walnut Creek, CA, USA) connected to an ion-trap mass spectrometer (Varian Saturn 2100). Both systems were furnished with autosampler injectors. Two different capillary columns, an Agilent HP-5 (30 m x 0.32 mm I.D.; df: 0.25 µm) and an Agilent HP-35 (30 m x 0.25 mm I.D.; df: 0.25 µm), were considered for separation of analytes. Columns were operated at constant helium flow rates of 1.4 and 1.0 mL min-1, respectively. Isooctane standards, containing increased concentrations of target species and a fixed level of IA, and extracts from PDMS rods (2 µL volume) were injected in the splitless mode (1 min), with the oven temperature programmed as follows: 70 ºC (1 min), ramp to 280 ºC at 12 ºC min-1 and hold at 280 ºC for 10 min. Operating under these conditions, both columns provided baseline separations for all species. The most significant difference observed in the corresponding chromatograms was that IA eluted before TCA in the HP-5 column, whereas the reverse order was observed in the HP-35 one. The ion-trap mass spectrometer was operated in the electron impact mode (70 eV). The most intense signals in the spectrum of each anisole, appearing at [M]+ or [M-15]+ m/z units, were chosen as parent ions, isolated with a m/z window of ± 3 units and submitted to collision induced dissociation. The working MS/MS conditions, as well as m/z ratios for parent and product ions, are summarised in Table 1. Compound Molecular ion (m/z) Parent ion (m/z) Excitation storage level (m/z) Excitation amplitude (V) Product ions (m/z)a DCA TCA IA TeCA TBA PCA 178 212 234 246 346 280 163 197 234 246 346 265 70 85 100 105 150 115 1.36 1.62 0.42 1.48 1.50 1.36 133,135 167,169 219 229, 231 329, 331 235, 237 Table 1. Summary of MS/MS working conditions. aUsed as quantification ions 3. Results and discussion 3.1. Desorption conditions Preliminary optimisation experiments were carried out with ultrapure water spiked at the 2 µg L-1 level. On the basis of results reported for SPME fibres [10] and bulk PDMS [22], analytes were extracted overnight at 50 ºC in the HS mode. Rods were then soaked for 5 min with 3 consecutive fractions (1 mL each) of n-pentane or ethyl acetate, and extracts were injected in the GC-ECD system without any further concentration step. For both solvents, relative standard deviations for responses of target species remained below 6% (n= 4 replicates). With ethyl acetate, approximately 85% of the total amount of each compound concentrated in the PDMS rod was recovered with the first mL of solvent, whereas a slightly higher desorption yield (ca. 90%) was achieved with the same volume of n-pentane. Dichloromethane and diethyl ether were also tested as desorption solvents; however, they did not improve the yield provided by n-pentane, which was retained for further experiments. Since rods were employed as disposable sorbents, only 1 mL of solvent was used for analytes desorption. Quantification limits of the method are determined by the efficiency of the extraction step and the volume of solvent required to recover the analytes from PDMS rods. Therefore, the possibility of concentrating the 1 mL n-pentane extracts to a lower volume was investigated. The process was carried out at room temperature, under a gentle stream of nitrogen. Evaporation to dryness, followed by reconstitution of the residue with a small volume of solvent, led to severe losses of DCA and TCA; however, the problem could be overcome by addition of isooctane (50-100 µL, containing IA at the 200 µg L-1 level), as a keeper, to the n-pentane extracts. The internal standard (IA) was used to compensate for differences in the volume of concentrated extracts. Fig. 2 compares the responses obtained for aliquots of the same mixture of anisoles in npentane (TeCA was not included in this study) before and after concentration from 1 to 0.1 or 0.05 mL. As appreciated, the evaporative concentration process did not affect to the ratios between analytes and internal standard peak areas. 0 1 2 3 4 5 6 DCA TCA TBA PCA Peak area/I.S. Peak area Reference Concentrated to 100 uL Concentrated to 50 uL x 10 Fig. 2. Evaluation of potential losses of anisoles after concentration of 1 mL n-pentane extracts to different final volumes, n= 3 replicates . 3.2. Extraction conditions The effect of different factors on the performance of the extraction step was investigated using ultrapure water samples spiked with anisoles at the 0.150 µg L-1 level (1.5 µg L-1 for DCA). Extractions were carried out in 100 mL vessels, and different sample volumes, temperatures, ionic strength values and sampling modes were considered. In order to minimise sorption and cross-contamination problems, and also to simplify the set-up of the extraction step, sample stirring with a magnetic bar was not considered. Analytes were desorbed from PDMS rods with 1 mL of n-pentane, the extract evaporated to a final volume of 50 µL, and 2 µL injected in the GC-ECD system. 3.2.1. Sampling mode Kinetics of sorptive extraction (SE) processes change depending on the position of the sorbent: HS or direct exposure to the sample. Both modes have been proposed for the extraction of anisoles from aqueous samples with SPME fibres [4,7,10] and PDMS coated Twisters [5,15,16]. Considering similar conditions to those proposed for the latter device (2 hours sampling at room temperature), the efficiency of the direct mode represented less than 5% of that achieved in HS. When the sampling step was extended overnight (14 hours), and vessels were maintained at higher temperatures (50 and 80 ºC), at least still twice higher signals were obtained in the HS mode than with direct sampling. 3.2.2. Ionic strength and sample volume The effect of the ionic strength on the efficiency of the HSSE was evaluated with 80 mL samples containing sodium chloride at four different levels, between 0 and 30%. Extractions were performed in duplicate, at room temperature for 2.5 hours. For non-ionic species, their distribution constants between aqueous samples and PDMS sorbents increase with the ionic strength of the sample; however, the kinetics of mass transference decreases at high salt concentrations [24]. Experimental data showed that the amount of each anisole concentrated in the PDMS rod increased slowly, but steadily, with the ionic strength of the sample. On average, twice higher signals were measured for samples containing a 30% of sodium chloride than those found for samples without salt addition. Fig. 3 shows the effect of sample volume on the efficiency of the HSSE. A proportional relationship between anisoles responses and sample volume was observed. This behaviour can be explained on the basis of two different contributions. One is the extraction capability of the rod, which is a function of the high volume of PDMS, as a consequence the extracted amount of each analyte increased with its total mass in the sampling vessel. The second contribution is related to the decrease in the HS of the vessel, which minimised the fraction of analytes remaining in the vapour phase. 0 1 2 3 4 5 6 7 8 9 10 20 30 40 50 60 70 80 90 Sample volume (mL) Peak area/I.S. Peak area DCA TCA TeCA TBA PCA Fig. 3. Effect of sample volume on the yield of the HSSE for water samples containing a 30% of NaCl. Room temperature extraction for 2.5 hours, n= 3 replicates. 3.2.3. Extraction time The effect of sampling time on the yield of the HSSE was evaluated in the range between 0.25 and 8 hours, at room temperature and 50 ºC. Fig. 4 shows the mean results obtained for duplicate extractions. The values on the y axis correspond to the overall efficiency of the extraction procedure, defined as the ratio between the amount of each anisole in the extract from the PDMS sorbent and that added to the sample. The first was determined by external calibration, using standards in isooctane containing increased levels of the analytes and, the same fixed amount of internal standard (IA) as that added to extracts from PDMS rods. At room temperature, equilibrium was achieved after 2.5 hours of sampling, and the yield of the HSSE varied between 40 and 55%. Moreover, there was a direct correlation between the extraction efficiency and the octanol-water partition coefficient of each anisole (log Kow values: 3.4 for DCA, 3.9 for TCA, 4.2 for TBA, 4.5 for TeCA and 5.0 for PCA). At 50 ºC, the kinetics of the extraction was faster and equilibrium was achieved after 1 hour of sampling; however, the yield of the extraction decreased considerably. Taking into account these results and the possibility of simultaneously processing of as many samples as required, 2.5 hours of sampling at room temperature were selected as the final extraction conditions. This sampling period is longer than those reported for SPME fibres when applied to the determination of same analytes [3,7,10,26], however, it remains similar to that proposed for SBSE [5,16]. 0% 10% 20% 30% 40% 50% 60% 70% 0 1 2 3 4 5 6 7 8 Time (hours) Extraction yield (%) DCA TCA TeCA TBA PCA A B 0% 10% 20% 30% 40% 50% 60% 70% 012345678 Time (hours) Extraction yield (%) DCA TCA TeCA TBA PCA Fig. 4. Time course of the HSSE at room temperature (A) and at 50 ºC (B). Average data for duplicate extractions. 3.3. Precision, linearity and quantification limits Figures of merit of the proposed method were evaluated using GC-ECD and GC-MS/MS. Precision studies were accomplished with water samples spiked at different concentration levels between 5 and 50 ng L-1, and extracted the same day (repeatability studies) or on different days (reproducibility evaluation). Relative standard deviations (RSDs) in the responses measured for target species remained below 10%. The only exception was the data obtained for DCA under reproducibility conditions, by GC-MS/MS, Table 2. GC-ECD GC-MS/MS Repeatability Reproducibility Repeatability Reproducibility Added level 10 ng L-1a 50 ng L-1b 20 ng L-1c 5 ng L-1 20 ng L-1 10 ng L-1 DCA TCA TeCA TBA PCA 2.5 7.9 10.1 10.6 8.1 3.1 4.4 4.9 3.7 5.0 8.9 8.0 5.5 7.5 3.7 2.1 1.2 5.5 6.9 8.7 2.5 4.2 4.6 2.7 5.4 15.6 8.5 9.4 9.3 8.9 Table 2. Repeatability (n= 4 replicates in the same day) and reproducibility (n=9 replicates in different days) of the HSSE procedure for water samples spiked at different concentration levels. Relative standard deviations (RSDs) given as percentages. a 100 ng L -1 for DCA, b 500 ng L -1 for DCA, c 200 ng L -1 for DCA. Globally, RSDs were similar to, or even better than those provided by other microextraction techniques (SPME, SBSE, SDME) when applied to samples fortified with similar concentrations of anisoles [5,7,8,17]. Linearity was investigated with samples spiked at six different concentration levels from 2.5 to 250 ng L-1 (25 to 2500 ng L-1 in case of DCA using GC-ECD detection). The peak area of each compound was divided by the signal of the internal standard and plotted against its concentration in the sample. Both GC-ECD and GC-MS/MS, provided calibration graphs with correlation coefficients between 0.9960 and 0.9998, showing the existence of a linear dependence between the amount of each anisole in the extraction vessel and that transferred to the PDMS rod, Table 3. Correlation coefficient (R2) LOQ (ng L-1) Analyte GC-ECD GC-MS/MS GC-ECD GC-MS/MS DCA TCA TeCA TBA PCA 0.9980 0.9994 0.9996 0.9991 0.9998 0.9960 0.9990 0.9996 0.9991 0.9990 20 2.5 1.5 1.3 0.8 0.5 0.8 0.5 1.5 0.5 Table 3. Linearity and limits of quantification (S/N = 10) of the method. Quantification limits (LOQs) of the proposed method ranged from 0.5 to 20 ng L-1. In the case of DCA, GC-MS/MS achieved one order of magnitude lower values than GC-ECD. For the rest of analytes, both techniques provided similar quantification limits, Table 3. Anyhow, on the basis of its higher selectivity, GC-MS/MS is recommended as detection technique. Table 4 summarizes the LOQs reported for TCA (the most commonly determined anisole and also the one exhibiting the lowest threshold level), using different sample preparation approaches followed by GC-MS or GC-MS/MS detection. The value provided by the proposed method was of the same order of magnitude as those achieved with SPME, and slightly lower than those reported for SPE and SDME. Moreover, as only a 4% of the final extract (50 µL volume) was injected in the gas chromatograph, it is expected that combining HSSE with thermal desorption of PDMS rods may compete with SBSE and CLSA in terms of quantification limits.