Journal of Chromatography A, 1140 (2007) Eva Campo, Juan Cacho, Vicente Ferreira

Transcription

1 Journal of Chromatography A, 1140 (2007) Solid phase extraction, multidimensional gas chromatography mass spectrometry determination of four novel aroma powerful ethyl esters Assessment of their occurrence and importance in wine and other alcoholic beverages Eva Campo, Juan Cacho, Vicente Ferreira Laboratory for Flavor Analysis and Enology, Department of Analytical Chemistry, Faculty of Sciences, University of Zaragoza, Zaragoza, Spain Received 12 June 2006; received in revised form 7 November 2006; accepted 14 November 2006 Available online 29 November 2006 Abstract A method for the quantitative determination of four powerful aromatic ethyl esters recently identified in some wines has been developed, validated and applied to the determination of these compounds in different samples of wine, whisky and brandy. Ethyl 2-, 3-, and 4-methylpentanoate and ethyl cyclohexanoate are extracted from 100 ml of sample by solid phase extraction (SPE) on a 200 mg LiChrolut EN bed. Major compounds are eliminated by rinsing with a water methanol (50:50) solution containing 1% sodium bicarbonate, and analytes are eluted with 1.5 ml of dichloromethane. Fifty microlitres of this extract are then injected in a multidimensional gas chromatography mass spectromety (GC GC MS) system. Recoveries in the SPE are quantitative. Method repeatability is satisfactory (5 12% for a 5 10 ng l 1 level, and less than 7% for ng l 1 level), the method linearity holds along the whole range of occurrence of analytes ( ng l 1 ), and the signal is independent on the matrix. Method detection limits are below 1 ng l 1 in all cases. Results suggest that these compounds are formed by the slow esterification with ethanol of the corresponding acids formed by different microorganisms. The levels of these compounds are above the corresponding thresholds in most samples of aged wines or distillates, but are particularly high in some sweet wines, whiskeys and brandies where they may constitute the most important contributors to the sweet-fruity notes reaching concentrations up to times higher than the corresponding odor thresholds Elsevier B.V. All rights reserved. Keywords: Wine; Distilled beverages; Whisky; Brandy; Cognac; Aroma; Flavour; Ethyl 2-, 3- and 4-methylpentanoate; Ethyl cyclohexanoate; SPE; GC GC; GC MS 1. Introduction Esters are abundant volatile constituents of different foods and beverages such as fruits and fruit juices, olive oil, fermented dairy derivates, beer, wine or distilled alcoholic beverages [1 6]. During alcoholic fermentation, a great number of esters can be generated as a result of yeast metabolism, but the most abundant are essentially ethyl esters of organic acids, acetates of fusel alcohols and ethyl esters of fatty acids [7]. The ethyl ester content of alcoholic beverages increases during aging, as a consequence of the slow esterification of different organic acids with ethanol. Some of the ethyl esters can be found at concentrations above their odor threshold and their aroma role is well documented in Corresponding author. Tel.: ; fax: address: vferre@unizar.es (V. Ferreira). the scientific literature. This is particularly true for the major ethyl esters of fatty acids [5,7], although some other minor ethyl esters, such as the ethyl esters of 2-methyl and 3-methylbutyric acids may also play some role [8 11]. Recent research by gas chromatography-olfactometry has given first evidence that some aged fortified wines contain other novel and naturally rare ethyl esters which may have some impact on their aroma [12,13]. Such compounds are ethyl 2-, 3- and 4-methylpentanoate, and ethyl cyclohexanoate, which exhibit pleasant strawberry-liquorices-like odors. In the early nineties, Takeoka et al. published different papers studying the odor properties of several unsaturated, branched and cyclic esters, including the four above mentioned [14,15]. The authors concluded that these ethyl esters possess remarkably low odor thresholds, ranging from to 0.01 gl 1 (in water), well below the odor threshold of their major linear cousin, ethyl hexanoate (1 5 gl 1 ) [16]. The presence of three of these com /$ see front matter 2006 Elsevier B.V. All rights reserved. doi: /j.chroma

2 E. Campo et al. / J. Chromatogr. A 1140 (2007) pounds in some other natural foodstuffs has been documented. Ethyl 3-methylpentanoate was first reported by Dumont and Adda in Beaufort cheese [17] and several years later was identified by Cormier et al. [18] in milk affected by Pseudomonas fragi bacteria, where ethyl cyclohexanoate was also present. The latter was first identified in rum by Heide et al. [19] and then in refined (virgin quality) olive oil by Guth and Grosh [20]. As for ethyl 4-methylpentanoate, Van der Wal et al. [21] found this compound in roasted cocoa more than thirty years ago whereas Ong and Acree [22] reported its presence in both fresh and canned lychee fruit (Litchi chinesis) in the late nineties. More recently, Steinhaus and Schieberle [23], and Fritsch and Schieberle [24] have identified ethyl 4-methylpentanoate in dried hop cones (Humulus lupulus) used by the brewing industry and in Bavarian Pilsner-type beers, respectively. However, and in spite of the aroma potential of this group of compounds, their exact sensory role remains for the most unknown because of the lack of analytical data and analytical methods for their determination. To our knowledge, only Guth and Grosch [20], and Fritsch and Schieberle [24] have quantified ethyl 4-methylpentanoate in olive oil and beer samples, respectively, by means of a stable isotope dilution essay. Solid-phase extraction (SPE) is widely used in analytical laboratories for either sample extraction or sample clean up procedures. Many benefits of SPE methods have been commonly cited including its robustness, potential for automation, capacity for providing clean extracts, selective isolations and even a fractionation of the different sample components. For these reasons, SPE is a powerful pre-concentration technique which can be easily adapted for routine analysis and, in fact, many studies based on SPE procedures for monitoring different compounds in wine samples have been published in the last years [25 28]. However, as the SPE systems have a low number of chromatographic plates [27] the selectivity (measured as the ratio between the chromatographic retention factors of analytes and interferences) must be high in order to get good separations. Such selectivity can be easily achieved if the target molecule is strongly non-polar, such as TCA [28], or if it is rather polar, such as furaneol [29], but in the case that both analytes and interferences are ethyl esters, the SPE system cannot provide a sufficient separation. The outcome is that even if polar interferences, such as fusel alcohols or fatty acids, have been removed, the chromatographic profile is still too complicated to get an adequate mass spectrometrical (MS) signal of compounds present at very low level (ng l 1 )if they elute in the section of the chromatogram where most major esters and non-polar interferences are also found. Additional selectivity can then be provided by multidimensional gas chromatography (MDGC) or by tandem mass spectrometry (MS MS). While the later is particularly useful if the target compounds have abundant high mass ions, the former should be preferred if analytes are small molecules with relatively poor and unspecific MS fragmentation patterns. An additional advantage of heart-cutting multidimensional gas chromatography (GC GC) is that it is possible to work in conditions near to mass overload (including large volume injection of a relatively concentrated extract) in the first chromatographic column and still obtain a perfect chromatographic separation in the second analytical column, so that method detection limits can be improved [30]. MDGC techniques have already been used in wine research with different purposes such as quality and authenticity control [31 33], identification of novel aroma compounds [12,34,35] or quantification of trace odorants [36]. The main purpose of the present paper is, therefore, the development and validation of an analytical method based on selective SPE and further multidimensional gas-chromatographic analysis for the determination of four novel ester compounds in wine and other alcoholic beverages. A second objective of this work is to assess the potential importance of these compounds in the aroma and flavor of different wines and alcoholic beverages. 2. Materials and methods 2.1. Chemicals and reagents Dichloromethane, methanol, and ethanol, LiChrosolv quality, were from Merck (Darmstadt, Germany). Pure water was obtained from a Milli-Q purification system (Millipore, Bedford, MA, USA). Polypropylene cartridges (3 ml) prepacked with LiChrolut EN resins were also obtained from Merck whereas NaHCO 3 was supplied by Panreac (Barcelona, Spain). The chemical standards used for quantitative analysis were from Lancaster (Frankfurt, Germany) or Sigma (St. Louis, MO, USA). The internal standard solution contained 4-hydroxy-4-methyl- 2-pentanone at 1500 gml 1 in dichloromethane Samples A total of 31 samples were analyzed. This set included wines submitted to biological and/or oxidative aging from Andalusia (Southern Spain), fortified Porto wines (Portugal), wines made with grapes affected by Botrytis Cinerea (noble rot) from Sauternes (France) and Tokaji (Hungary), aged red wines, young red and white wines from northern Spain and a Cava (Spanish sparkling wine). Four distilled alcoholic spirits, two Sherry brandies and two Scotch whiskeys, were also included in this study. All the samples were purchased from a wine-retailer in Zaragoza. Some details on the sample origins, aged or alcoholic content are given in Table 1. Wine Pazo Ribeiro (sample 3 in Table 1), virtually free of the analytes (only 14 ng l 1 of 4-methylpentanoate were found), was used for validation purposes SPE modelling and optimization Solid liquid distribution coefficients of the four analytes between LiChrolut EN resins and wine or different rinsing systems were measured as explained in [27]. For these measurements, a Saturn 2200 GC-Ion Trap MS from Varian (Walnut Creek, CA, USA) was employed using standard conditions. Chromatographic retention factors were calculated from the corresponding solid liquid distribution coefficients with the expression k = K sl φ; being φ the phase ratio in the SPE bed. This was defined as m o /V m, being m o the mass of sorbent in the bed, and V m the dead volume. For a 200 mg bed filled with

3 182 E. Campo et al. / J. Chromatogr. A 1140 (2007) Table 1 Sample type, brand, origin, grape variety and ethanolic content of the samples analyzed No. Sample type Brand AOC Grape varieties Ethanol, % (v/v) 1 White young Viña Albada Calatayud Macabeo 14 2 White young Marqués de Riscal Rueda Sauvignon Blanc 13 3 White young Pazo Ribeiro Treixadura Red young Montesierra Somontano Tempranillo, Cabernet Red young Viñas del Vero Somontano Tempranillo Red young Borsao Campo de Borja Garnacha, Tempranillo, Cabernet Red Crianza Lan Rioja Tempranillo 13 8 Red Reserva Viña Pomal Rioja Tempranillo, Graciano Red Gran Reseva Faustino Rioja Tempranillo Porto White Pousada Porto Malvazia, Gouveio Porto Tawny Pousada Porto Cao, Barroca Porto Ruby Taylor s Select Porto Cao, Barroca Sauternes Château Laribotte Sauternes Semillon, Muscadelle, Sauvignon Blanc Sauternes Baron Philippe Sauternes Semillon, Muscadelle, Sauvignon Blanc Sauternes Aureus Sauternes Semillon, Muscadelle, Sauvignon Blanc Tokaji Late Harvest Oremus Tokaji Aszù (40%) Cava Brut Gramona Cava Xarello, Macabeo, Chardonnay Pale Cream Cartojal Málaga Moscatel de Alejandría Cream Ibérica Sherry Palomino Fino, Pedro Ximénez Cream Canasta Sherry Palomino Fino Fino Tío Pepe Sherry Palomino Fino Fino Quinta Sherry Palomino Fino Fino Cobos Montilla-Moriles Pedro Ximénez Pedro Ximénez Don PX 1971 Montilla-Moriles Pedro Ximénez Pedro Ximénez Don PX 1975 Montilla-Moriles Pedro Ximénez Pedro Ximénez Leyenda Sherry Pedro Ximénez Pedro Ximénez Duquesa Sherry Pedro Ximénez Brandy Marqués de Misa Sherry Airén, Palomino Fino Brandy Gran Duque de Alba Sherry Airén, Palomino Fino Whisky Cardhu Malt Whisky Knockando Malt 43 LiChrolut EN resins, φ was determined to be 0.57 mg sorbent per millilitre of mobile phase. Breakthrough volumes were finally estimated with the help of the expression V b =1/z (1 + k) V m, being z the correction factor proposed by Lovkist-Jonsson [37] for chromatographic beds with a low number of plates. In our conditions, the number of plates is around 7, z for a 1% breakthrough level is 0.76 and V m was determined to be 0.15 ml [27] Proposed SPE method SPE cartridges (in 3 ml reservoirs) filled with 200 mg LiChrolut EN resins were put in the extraction unit (VAC ELUT 20 Station from Varian) and conditioned by passing 10 ml of the following solvent systems: dichloromethane, methanol and a 15% hydro alcoholic solution (v/v). After this, 100 ml of sample were loaded. In the case of distillates, the samples were previously diluted with pure water to adjust their alcoholic content to 15% (v/v). Major compounds were washed out by rinsing with 25 ml of an aqueous solution containing 50% methanol (v/v) and 1% NaHCO 3. After this, the cartridge was dried by applying vacuum (negative pressure 0.6 min) for 10 min, analytes were eluted with 1.5 ml of dichloromethane, and 20 l of the internal standard solution were added to this volume. The vial was then capped and stored at 20 C until analysis PTV GC GC MS instrumentation GC GC mass spectrometry conditions. All analysis were carried out with a multidimensional gas chromatograph from Varian constituted by two independent gas chromatographs interconnected by means of a thermo regulated transfer line kept at 200 C. Chromatograph 1 (GC-1): The chromatograph was a CP 3800 model equipped with a 1079 PTV injector and a flame ionization detection (FID) system. This GC was retrofitted with a Deans pressure-driven switch (Valco Instruments, Houston, TX, USA), which enables to selectively transfer heart cuts eluting from the first column directly into the analytical column placed in the second chromatograph. The carrier gas (Helium) was delivered at a constant pressure of 30 psi. During the two first minutes of each run an auxiliary He flow (Deans switch) was maintained at 15 psi, then it was raised to 20 psi. The column was a DB-WAX (polyethylene glycol) from J&W (Folsom, CA, USA), 30 m 0.32 mm I.D. with 0.50 m film thickness, and was connected to both the injector and the Deans switch. An uncoated, deactivated fused-silica column (30 m 0.32 mm I.D.) from Supelco (Bellefonte, PA, USA) was used as interface between the Deans switch and the FID detector. The oven temperature program was 40 C during 7 min, then raised at 4 C min 1 up to 100 C, at 6 C min 1 up to 140 C and at 20 C min 1 up to 200 C. The FID was kept at 300 C.

4 E. Campo et al. / J. Chromatogr. A 1140 (2007) Chromatograph 2 (GC-2): The chromatograph was a CP 3800 model coupled to an ion trap mass spectrometricdetector (Saturn 2200). The column was a FactorFour-VF 5MS (polymethylsiloxane 5% diphenyl) from Varian (30 m 0.32 mm 1 m film thickness). The column was directly connected to the Deans switch placed in the first chromatograph via the thermostated transfer line. The oven temperature program was the following: 40 C during 16 min, then raised to 130 C at 5 C min 1 and finally to 250 at 20 C min. The MS-parameters were: transfer line at 170 C; ion trap at 150 C and trap emission current 10 A. The global run time (34 min) was recorded in full scan mode ( m/z mass range), 1 scan being performed each second. The m/z fragments chosen for quantitative purposes were 102, 88, and 156 for, ethyl 2-, 3-, and 4- methylpentanoate and ethyl cyclohexanoate, respectively. FID and MS data were registered and processed with Workstation 6.30 software equipped with NIST 98 (US National Institute of Standards and Technology) MS library (NIST, Gaithersburg, MD, USA). All Deans switching operations were software controlled Programmable injector conditions. The insert (internal diameter 3.4 mm) was filled with 50 mg of silane treated glass wool (Supelco) previously washed with 2 ml of dichloromethane and conditioned by heating under a stream of nitrogen at 200 C. Large volume injections (50 l) were optimized for extracts in dichloromethane as solvent. The injection was carried out in the solvent split mode. During the injection the PTV was kept at low temperature (40 C) and the split valve was opened (split ratio = 30) to promote solvent evaporation. The chromatographic process took place at a constant head pressure of 30 psi, to facilitate the smooth operation of the Deans switch. At such pressure, the carrier gas flow rate during the injection was high (5 ml min 1 ). After solvent elimination (0.25 min), the split valve was closed. The injector was then heated to 250 C at a rate of 200 C min 1. After three minutes the split valve was opened again (split ratio = 20) and the injector was held at 250 C. The different injection parameters were carefully optimized and the performance of the system was measured to ensure a complete transfer of analytes and good retention time reproducibility Selected heart-cutting windows. These parameters were optimized in different preliminary assays in order to ensure a complete transfer of analytes between columns. Two different heart-cutting windows were performed in a single chromatographic run. In the first cut (from to min), ethyl 2-, 3- and 4-methylpentanoates were transferred to the second column. In the second cut (from to min) ethyl cyclohexanoate was transferred. The internal standard was monitored by FID (retention time = min) Method validation Method repeatability was essayed by the repeated analysis of wine Pazo Ribeiro spiked at two different concentration levels (n = 3 in both cases). The repeatability of the GC GC MS process was evaluated by the replicate injection (n =4)of an extract from wine Pazo Ribeiro spiked with 50 ng l 1 of ethyl 2-, 3- and 4-methylpentanoate and 25 ng l 1 ethyl cyclohexanoate. The amount of analyte not recovered during the sample preparation was evaluated comparing the areas obtained in the analysis of this same wine spiked with 1 gl 1 of analytes with those obtained in the analysis of an extract from the same wine spiked with an equivalent mass of analytes. Method linearity was evaluated by the analysis of wine Pazo Ribeiro spiked with known amounts of the standards and by the direct GC GC MS analysis of dicloromethane solutions containing known amounts of the analytes covering the range of occurrence of these compounds. The existence of matrix effects was checked by comparing the slopes obtained in the previous experiment and by determining the recovery of analytes by the analysis of different spiked samples. Method detection limits were calculated by the analysis of this wine spiked with small amounts of analytes (5 10 ng l 1 ). 3. Results and discussion 3.1. Selective Solid Phase Extraction (SPE) A main goal of the present paper is to develop an analytical method to quantify some minor esters in wine and alcoholic beverages. The first step of the method setup was the design of the isolation strategy. A selective extraction on a SPE bed Table 2 Solid Liquid distribution coefficients and breakthrough volumes (ml) of the target analytes between LiChrolut EN resins and wine or different rinsing solutions Compound Wine/30% a 40% 50% 60% 70% Ksl Vb Ksl Vb Ksl Vb Ksl Vb Ksl Vb Et. 2mp >8000 > ,1 Et. 3mp >8000 > Et. 4mp >8000 > Et. Cyclo >8000 > Breakthrough volumes are given in ml and refer to a SPE bed formed by 200 mg of resins packed in a 3 ml standard reservoir. The rinsing solutions were in all cases water methanol solutions with 1% NaHCO 3. Abbrevations: Et. 2mp (ethyl 2-methylpentanoate); Et. 3mpe (ethyl 3-methylpentanoate); Et. 4mp (ethyl 4-methylpentanoate); Et. Cyclo (Ethyl cyclohexanoate). a Retention in wine and in the 30% methanol rinsing solution was too high to measure the Solid Liquid distribution coefficients.

5 184 E. Campo et al. / J. Chromatogr. A 1140 (2007) Table 3 Method quality parameters 1 SPE-Recovery GC GC MS repeatability Method repeatability Compound (%) RSD (%) a (n =4) Low b (n = 3) High a (n =3) Ethyl 2-methylpentanoate Ethyl 3-methylpentanoate Ethyl 4-methylpentanoate Ethyl cyclohexanoate Recovery in the solid phase extraction (in %), repeatability of the GC GC MS operation and overall method repeatability (both in RSD (%)) at two concentration levels. a Determined in an extract from a wine spiked with 50 ng l 1 ethyl 2-, 3-, and 4-methylpentanoates and with 25 ng l 1 ethyl cyclohexanoate. b Determined in an extract from a wine spiked with 10 ng l 1 ethyl 2-, 3-, and 4-methylpentanoates and with 5 ng l 1 ethyl cyclohexanoate. filled with LiChrolut EN was selected, since previous experiences confirm that highly satisfactory recoveries and selectivity can be obtained with this system [25,27]. The strategy followed to optimize the SPE was presented in previous papers [27,38] and it is based on the determination of the solid liquid distribution coefficients of the analytes between the sorbent and wine or different rinsing solvents. Table 2 shows these values for wine and different washing up solvents, together with the corresponding estimated breakthrough volumes for a 1cm-SPE bed filled with 200 mg of resins. It can be observed that these compounds are strongly retained in the sorbent, and that large breakthrough volumes are obtained in wine or in solvent systems containing less than 30% methanol. According to data in the table, a 200 mg bed should be able to quantitatively retain nearly all the analytes present in more than 1000 ml of wine. However, and in order to ensure a consistent and matrix independent extraction, the sample volume was fixed as 100 ml. This figure should also make it possible to introduce an exhaustive washing step to remove major compounds. Results in the table indicate that a washing solution containing 50% of methanol and 1% of NaHCO 3 is a good choice since this is a high elution power phase for which breakthrough volumes for analytes are still high enough to ensure that they are not going to be removed during the washing step. A washing volume of 25 ml was selected. This volume makes it possible to get rid of major fusel alcohols and fatty acids and thus to obtain better chromatograms. A standard recovery experiment was carried out to determine the amount of analyte recovered during the isolation step. This experiment confirmed a quantitative recovery for the four analytes considered as it can be seen in Table 3. higher the solvent split time, the lower the retention time. This is consequence of the reduction in the amount of solvent introduced in the column. High amounts of solvent in the column exert a weak but noticeable displacement effect on the early eluting peaks, and the effect is more pronounced at low split times. This would not represent a major problem if such effect was not so irreproducible and dependent on the particular composition of the extract. The relative standard deviations (in %) of analyte retention times in the replicate injection of an extract ranges from less than 0.5% (for solvent split times of 0.30 and 0.25 min) to more than 8% (for solvent split time of 0.15 min). This last figure becomes still worse if different extracts are considered. On the other hand, the loss of analytes by co-evaporation is also noticeable but only at a solvent split time of 0.30 min. The selected solvent split time of 0.25 min represents therefore the optimum. Under these conditions, the mass transfer was also highly repetitive (area RSD (%) <2%). The heart cutting-windows were selected to ensure a complete transference of analytes to the second column. The three methylpentanoates (ethyl 2-, 3- and 4-methylpentanoate) were recovered in a single cut (160 s) and ethyl cyclohexanoate was isolated in a second cut. The ion chromatograms obtained in the analysis of one of the studied wines can be seen in Fig. 2, which shows that a neat peak separation is achieved in all cases. The possibility of a direct GC MS analysis of the aforementioned extracts was also considered. The problem found was that most of the target analytes coelute with relatively 3.2. Large volume injection in the GC GC system One of the most critical parameters in the large volume injection using the solvent split mode is the split time or vaporization time. This parameter has a great influence not only on the total amount of solvent vaporized and on the amount of analyte lost by co-evaporation (or by injector overload), but also on the retention times of the compounds eluting at the beginning of the chromatogram. This is particularly critical in GC GC, since retention times must be known with accuracy beforehand. The effect of the solvent split time on the retention time for three of the target analytes is shown in Fig. 1. As can be observed, the Fig. 1. FID chromatograms of ethyl 2-, 3-, and 4-methylpentanoate and ethyl cyclohexanoate at four different split times.

6 E. Campo et al. / J. Chromatogr. A 1140 (2007) Table 4 Method quality parameters 2 Compound DL QL Linear range a r 2 Ethyl 2-methylpentanoate Ethyl 3-methylpentanoate Ethyl 4-methylpentanoate Ethyl cyclohexanoate Method detection and quantification limits (in ng l 1 ), linear range and coefficient of determination (n = 10). a These concentrations are concentrations of analytes in wine in ng l 1. In the case of alcoholic distillates, these figures should be multiplied by the dilution factor (ca. 2.5) used to adjust the alcoholic degree to 15% (v/v). than 100 ng l 1 for ethyl cyclohexanoate, which cannot be considered sufficient for the levels found in wine Method validation Fig. 2. Example of the ionic chromatograms obtained for ethyl 2-, 3- and 4- methylpentanoate and ethyl cyclohexanoate in Cava wine. similar compounds present at high concentration, even in a carbowax column. For instance, ethyl 2-methylpentanote coelutes with ethyl pentanoate in the carbowax column and with ethyl hexanoate in a non-polar phase. Although ethyl pentanoate is not a wine major compound, it still can be found at concentrations times higher than ethyl 2-methylpentanoate, which causes serious spectral interferences and impedes the correct quantification of this compound at low levels. The use of MS MS technique improves but not solves these problems. So far, the detection limits that can be achieved by direct GC MS analysis range from 8 ng l 1 for ethyl 3-methylpentanote to more A wine free from the target analytes was spiked at two different concentration levels (see Table 3) and was analyzed in triplicate following the complete proposed procedure. Results of the experiment are given in that table which shows that for the 5 10 ng l 1 level, the method repeatability ranges from 6.4 to 11.7%, while at the ng l 1 level, such figure is in all cases below 6.4%. At this concentration level, the repeatability of the chromatographic process was evaluated separately by the replicate injection of different extracts. The results of this experiment, found in the second data column of Table 3, indicate that the GC GC MS operation is a major contributor to the overall imprecision, except in the case of ethyl 3-methylpentanoate. Other quality parameters for the proposed procedure are given in Tables 4 and 5. The method detection limits (Table 4) were determined as the concentration of analyte in wine which gives a signal to noise ratio equal to 3 and in all the cases are below the ng l 1. More important, in all the cases these figures are below the odor thresholds of these compounds, which should make it possible to determine them at the concentrations in which they are potentially odor-active. The method linearity was checked by the injection of both solutions of standards in dichloromethane and by the analysis of wines spiked with known amounts of the analytes. Method linearity (Table 4) was studied attending to the natural range of occurrence of these compounds in wine and distilled beverages. Linearity holds along the whole Table 5 Method quality parameters 3 Compound Slope 1 a Slope 2 b Recovery 1 c Recovery 2 d Ethyl 2-methylpentanoate ± 9 96 ± 7 Ethyl 3-methylpentanoate ± ± 11 Ethyl 4-methylpentanoate ± ± 7 Ethyl cyclohexanoate ± ± 5 Evaluation of matrix effects. Comparison between the slopes found in the analysis of dichloromethane standard solutions and those found in a standard addition experiment. Average recoveries determined in the analysis of different spiked samples. a Slope of the straight regression line found in the analysis of synthetic standards in dichloromethane. b id. in wine 3 spiked with known amounts of analytes. c Average recovery found in the analysis of 4 different wines spiked with 150 ng l 1 of the three branched esters and with 25 ng l 1 of ethyl cyclohexanoate. d Average recovery found in the analysis of a whisky and a brandy spiked with 500 ng l 1 of the three branched esters and with 50 ng l 1 of ethyl cyclohexanoate.

7 186 E. Campo et al. / J. Chromatogr. A 1140 (2007) Table 6 Levels of ethyl 2-, 3- and 4-methylpentanoate and ethyl cyclohexanoate (expressed as ng l 1 ) found in the studied samples and odor activity values (OAV a ) Sample type Year Brand Et. 2mp Et. 3mp Et. 4mp Et. ciclo conc. OAV conc. OAV conc. OAV conc. OAV White young 2004 Pazo <q.l. <q.l <q.l Viña Albada <q.l. <q.l <q.l Marqués de Riscal <q.l. <q.l <q.l. Red young 2005 Montesierra <q.l. <q.l <q.l Borsao <q.l. <q.l <q.l Viñas del Vero <q.l. <q.l <q.l. Red barrel aged 1999 Lan Viña Pomal Faustino Porto Ruby Tawny White Noble rot 2002 Saut. Laribotte <q.l Saut. Baron <q.l Saut. Aureus Tokaji Oremus Cava 2001 Gramona Fino 3 b Cobos b Quinta <q.l. 5 b Tío Pepe Cream 5 b Cream Canasta b Cream Ibérica Pale Cream 2002 Cartojal <q.l. Pedro Ximénez 8 b Duquesa b Leyenda <q.l. <q.l <q.l Don PX Don PX Brandy 8 b Marqués Misa b Duque de Alba Whisky 12 b Knockando b Cardhu a Odor thresholds were taken from references [14] and [15]. b Sample with no attributable vintage date on the bottle. Instead, the aging period (years) is indicated <q.l.: below the quantification limit. studied range and can be considered satisfactory. Finally, as data in the Table 5 shows, it can be concluded that the method is free from matrix effects since the slopes obtained in the analysis of synthetic samples do not differ significantly from those obtained in a standard addition experiment. The overall recovery obtained in the analysis of different spiked samples is also shown in Table 5, and in all cases the means obtained did not significantly differ from 100%. This was also expected from the high recovery obtained in the SPE procedure. It should be noted that, in spite of the sophisticated instrumental approach involved in the method, the chromatographic analysis takes just 34 min (45 from run to run) and that the sample preparation is simple and straightforward Occurrence of these compounds in different wines and distilled alcoholic beverages Table 1 summarizes the samples analyzed in the study and provides some information about the type, brand, origin, age, and ethanol content. The 31 samples belong to a quite broad spectrum of samples in order to have a first idea about the types in which the target compounds may be important odorants. Quantitative results are given in Table 6, arranged by wine type. The first observation which must be made is that these odorants are more characteristic of samples which have undergone an aging period. Only one of the four esters, ethyl 4-methylpentanoate, can be found in young wines, and at really low levels. On the contrary, the four compounds are present in the rest of samples studied, and samples with longer aging tend to have higher content on these compounds. This suggests that these compounds are formed by slow esterification of the corresponding acids, probably formed during the fermentation. The second and third observations are that the content of these compounds seems to be related: (1) to the ethanol content of the sample; and (2) to the complexity of the fermentation processes used to produce such sample. The effect of the ethanol content can be seen in the highest levels of most of these compounds found in the brandies and whiskeys analyzed. The effect of the complexity of the fermentation can be seen by comparing samples with the same age and different fermentation processes, for instance young reds versus young whites or cava versus aged red wines. The fermentation of reds is more complex than that of whites, since the presence of the solids in the fermenting media adds more nutrients and brings about the concomitant presence of many different strains of yeast and bacteria in the first stages of the alcoholic fermentation. The fermentation of cava is still more complex, since in

8 E. Campo et al. / J. Chromatogr. A 1140 (2007) its elaboration there are two different alcoholic fermentations. Something similar can be observed when comparing Finos with aged reds of with noble rot wines. Finos undergo a long aging under a particular kind of yeast (flor yeast). The importance of the fermentation can also explain the high variability found in the levels of these compounds between the studied andalusian sweet wines (Cream or Pedro Ximénez). These wines are made of sun-dehydrated grapes (raisins) which have such a high sugar content (up to 500 g Kg 1 ) that in some samples the alcoholic fermentation hardly takes place (in most cases only 4 8% of ethanol is produced during the fermentation, the rest is supplied). The microbiological origin of these compounds is also coherent with previous scientific reports [17,18]. These compounds can be found in a wide range of concentrations. The levels of ethyl 2-methyl pentanoate range from less than 2 ng l 1 (the method quantification limit) to more than 1 gl 1. The maximum amounts of this compound are found in the most aromatic Pedro Ximénez sample (Don PX 1975), in whiskeys and in cava. A similar range is found in the case of ethyl 3-methylpentanoate for which maximum values are found again in whisky, the Pedro Ximénez aromatic sample, and in one of the Fino wines (Tío Pepe). Ethyl 4-methylpentanoate could be quantified in all the samples in a range from 14 to 2724 ng l 1. Maximum values are found in the same samples but in this case a Cream wine also showed a very high level. The levels of this compound in some of the samples are well above the concentration determined by Fritsch and Schieberle in Bavarian Pilsner-type beers (280 ng l 1 ). These authors have suggested that such compound may possibly be transferred from hops into the beer during the boiling of wort. Finally, ethyl cyclohexanoate is the rarest compound (and the one with the lowest odor threshold) and the maximum amount found is only 85 ng l 1 in one of brandies. In this case, brandies, the aromatic Pedro Ximénez sample, and one of the Porto wines showed maxima values Odor activity values (OAVs) The ratio odorant concentration/odor threshold is known as odor activity value (OAV). This parameter makes it possible to make a first assessment of the potential importance of a specific compound to the aroma of a product. Those coefficients for the four ethyl esters in the different samples studied are also found in Table 6 and suggest that these compounds can be very important odorants of wine and alcoholic beverages. The maximum OAV is 355, reached by ethyl 2-methylbutyrate in the Don PX 1975 sample, but also ethyl 3-methylbutyrate and ethyl 4-methylbutyrate can reach more than 100 OAV units. Ethyl cyclohexanoate reaches 85 OAV. In addition, it can be thought that these compounds are going to produce an additive or even synergic effect, since their odors are quite similar. This means that this group of compounds may constitute the biggest pool of sweet-fruity odorants of many aged wines and distillates, which makes that these compounds are probably important quality factors of these samples. This finding contrasts with the fact that their presence has been reported only recently. 4. Conclusions The method proposed combining a selective SPE, a large volume injection and GC GC MS determination, enables to accurately and conveniently quantify ethyl 2-, 3- and 4- methylpentanoate and ethyl cyclohexanoate in wines and in distilled beverages. The method has made it possible to provide for the first time an assessment of the levels and potential sensory importance of these compounds in different beverages. Results suggest that these compounds are formed by the slow esterification with ethanol of the corresponding acids formed by different microorganisms. The levels of these compounds are above the corresponding threshold in most samples of aged wines or distillates, but are particularly high in some sweet wines, whiskeys and brandies where they may constitute the most important contributors to the sweet-fruity notes. Acknowledgements This work has been funded by the Spanish CICYT (project AGL /ALI). E.C. has received a grant from the FPI program of the Spanish Government. References [1] E. Guichard, M. Souty, Z. Lebens, -Unters. Forsch. 186 (1988) 301. [2] A.K. Kiritsakis, J. Am. Oil Chem. Soc. 75 (1998) 673. [3] S.Q. Liu, R. Holland, V.L. Crow, Int. Dairy J. 14 (2004) 923. [4] H. Maarse, Volatile compounds in food and beverages, Marcel Dekker, New York, [5] L. Nykänen, H. Suomalainen (Eds.), Aroma of Beer, Wine and Distilled Alcoholic Beverages, Reidel, Dordrech, [6] M. Riu-Aumatell, E. Lopez-Tamames, S. Buxaderas, J. Agric. Food Chem. 53 (2005) [7] P.X. Etievant, in: H. Maarse (Ed.), Wine in Volatile compounds in foods and beverages, Marcel Dekker, New York, 1991, p [8] E. Campo, V. Ferreira, A. Escudero, J. Cacho, J. Agric. Food Chem. 53 (2005) [9] H. Guth, J. Agric. Food Chem. 45 (1997) [10] J. Petka, V. Ferreira, M.A. Gonzalez-Vinas, J. Cacho, J. Agric. Food Chem. 54 (2006) 909. [11] P. Schlich, L. Moio, Sci. Aliments 14 (1994) 609. [12] E. Campo, V. Ferreira, R. López, A. Escudero, J. Cacho, J. Chromatogr. A 1122 (2006) 202. [13] E. Campo, V. Ferreira, A. Escudero, J.C. Marques, J. Cacho, Anal. Chim. Acta 563 (2006) 180. [14] G.R. Takeoka, R.G. Buttery, J.G. Turnbaugh, M. Benson, Lebensm. -Wiss. Technol. 28 (1995) 153. [15] G.R. Takeoka, R.G. Buttery, J.G. Turnbaugh, R. Teranishi, Lebensm. -Wiss. Technol. 24 (1991) 569. [16] L.J. van Gemert, A.H. Nettenbreijer, Compilation of odour threshold values in air and water, TNO, Zeist, [17] J.P. Dumont, J. Adda, J. Agric. Food Chem. 26 (1978) 364. [18] F. Cormier, Y. Raymond, C.P. Champagne, A. Morin, J. Agric. Food Chem. 39 (1991) 159. [19] T. Heide, H. Schaap, H. J. Wobben, P. J. De Valois, R. Timmer, G. Charalambous, G. Inglett (Ed.), The Quality of Foods and Beverages, New York, p [20] H. Guth, W. Grosch, J. Am. Oil Chem. Soc. 70 (1993) 513. [21] B. van der Wal, D.K. Dettenes, J. Stopffelsma, G. Sipma, A.T.J. Semper, J. Agric. Food Chem. 19 (1971) 276. [22] P.K.C. Ong, T.E. Acree, J. Agric. Food Chem. 47 (1999) 665. [23] M. Steinhaus, P. Schieberle, J. Agric. Food Chem. 48 (2000) 1776.

9 188 E. Campo et al. / J. Chromatogr. A 1140 (2007) [24] H.T. Fritsch, P. Schieberle, J. Agric. Food Chem. 53 (2005) [25] R. López, M. Aznar, J. Cacho, V. Ferreira, J. Chromatogr. A 966 (2002) 167. [26] V. Ferreira, L. Culleré, R. López, J. Cacho, J. Chromatogr. A 1028 (2004) 339. [27] V. Ferreira, I. Jarauta, L. Ortega, J. Cacho, J. Chromatogr. A 1025 (2004) 147. [28] S. Insa, E. Anticó, V. Ferreira, J. Chromatogr. A 1089 (2005) 235. [29] V. Ferreira, I. Jarauta, R. López, J. Cacho, J. Chromatogr. A 1010 (2003) 95. [30] L. Mondello, A.C. Lewis, K.D. Bartle (Eds.), Multidimensional Chromatography, Wiley, Chichester, [31] M. Herraiz, G. Reglero, T. Herraiz, E. Loyola, J. Agric. Food Chem. 38 (1990) [32] A. Hollnagel, E.M. Menzel, A. Mosandl, Z. Lebensm, -Unters. Forsch. 193 (1991) 234. [33] A. Mosandl, Proceedings of the 1st Symposium in Vino Analytica Scientia, Bordeaux, June p. 31. [34] E. Campo, V. Ferreira, J. Cacho, J. Chromatogr. A 1137 (2006) 223. [35] P. Darriet, M. Pons, R. Henry, O. Dumont, V. Findeling, P. Cartolaro, A. Calonnec, D. Dubourdieu, J. Agric. Food Chem. 50 (2002) [36] B. Martin, P. Etievant, J. High Resolut. Chromatogr. 14 (1991) 133. [37] P. Lovkvist, J.A. Jonsson, Anal. Chem. 59 (1987) 818. [38] L. Ortega, R. Lopez, J. Cacho, V. Ferreira, J. Chromatogr. A 931 (2001) 31.