Static Headspace Analysis Using Low-Pressure Gas Chromatography

Article Cite This: Anal. Chem. XXXX, XXX, XXX−XXX

pubs.acs.org/ac

Static Headspace Analysis Using Low-Pressure Gas Chromatography and Mass Spectrometry, Application to Determining Multiple Partition Coefficients: A Practical Tool for Understanding Red Wine Fruity Volatile Perception and the Sensory Impact of Higher Alcohols Margaux Cameleyre, Georgia Lytra, and Jean-Christophe Barbe*

Anal. Chem. Downloaded from pubs.acs.org by DURHAM UNIV on 08/29/18. For personal use only.

University of Bordeaux, Unité de Recherche Oenologie, EA 4577, USC 1366 INRA, ISVV, 33882 Villenave d’Ornon Cedex, France S Supporting Information *

ABSTRACT: To evaluate the partition coefficients of volatiles between the liquid and gas phases, an analytical method was developed and optimized using static headspace analysis and lowpressure injection gas chromatography coupled to mass spectrometry (SHS-LP-GC/MS). Two different types of analytical columns were coupled for low-pressure chromatography injection: a narrow restriction microbore column on the inlet side and a mega-bore column on the mass spectrometer side. Coupling these two columns and static headspace analysis to gas chromatography and mass spectrometry resulted in a simple, fast, sensitive, and accurate approach. Several points have been optimized: time to reach the thermodynamic equilibrium in the gas phase, syringe filling rate, gas injection rate, and volume ratio between the gas and liquid phases. This new method was used to determine partition coefficients between the liquid and gas phases and study multicomponent mixtures for which particular perceptive interactions had previously been highlighted. The partition coefficients of 9 esters and 5 higher alcohols were determined in dilute alcohol solution (12% v/v) and dearomatized red wine. These partition coefficients revealed modifications in ester headspace release in the presence of higher alcohols for the first time in this type of matrix. The correlation of these results with sensory data highlighted the role of physicochemical, presensory effects on sensory modifications for the first time, suggesting that this type of interaction may partly modulate qualitative and quantitative fruity perception.

W

The perception of volatiles during sensory analysis is highly related to the distribution of aromas between the matrix and the gas phase and may be related to perception during food consumption.20,21 Headspace release depends on the affinity of these aroma compounds for the matrix, which considerably influences the rate of transfer.22 The distribution of one volatile is characterized by its partition coefficient, ki, describing the aroma concentration ratio between the gas and liquid phases. Further partition coefficient studies have elucidated the role of wine matrix composition on aroma volatility. For instance, increasing levels of ethanol from 10 to 20% v/v have been shown to cause a decrease in volatility of some aroma compounds by increasing their solubility.22−25 Moreover, Aznar et al. demonstrated that this decrease in volatility with increasing ethanol levels was only applicable to polar compounds, with no effect on the most apolar ones (Log P > 3).26 An increase in proteins and polysaccharides at levels

ine is a very complex aromatic matrix, composed of over a thousand volatile compounds.1 This diversity of chemical families gives wine its tremendous aromatic complexity.2,3 In red wines, Pineau et al. showed that at least part of the fruity aroma of red wines resulted from perceptive interactions between several aromatic compounds, particularly ethyl esters and acetates, even if they were present at concentrations below their olfactory thresholds.4 Thus, these authors identified a 13ester fruity pool that characterized Bordeaux red wine aromas. Many other examples of perceptive interactions have been highlighted, with synergistic5−12 or masking effects13−17 on the fruity aromatic expression of wines. For instance, Cameleyre et al. showed the remarkable masking effect of higher alcohols on fruity note perception.18 Berglund et al. suggested that perceptive interactions occurred on four different levels.19 The first was described as presensory interactions, such as chemical or physicochemical interactions in the gas phase or in the nasal pathway and mucosa. The three others referred to interactions in the peripheral and central nervous system at the receptor surface or in the olfactory bulb. © XXXX American Chemical Society

Received: April 27, 2018 Accepted: July 16, 2018

A

DOI: 10.1021/acs.analchem.8b01896 Anal. Chem. XXXX, XXX, XXX−XXX

Article

Analytical Chemistry between 5 and 20 g/L also led to a decrease in the volatility of certain compounds.27 Sugars have several effects on the release of some aroma compounds, causing either a reduction or an increase in their volatility.28,29 Other studies focused on the effect of glycerol,30,31 phenolic compounds,32,33 and the impact of temperature variations.34,35 Various experimental methods have already been developed to determine gas−liquid partition coefficients, such as the differential method, as well as dynamic or static headspace.25,36 The latter is calculated at the thermodynamic equilibrium between gas and liquid phases. Some methods, such as Vapor Phase Calibration37 (VPC) or Liquid Calibration-Static Headspace29,38 (LC-SH) use an external calibration standard, unlike the Equilibrium Partitioning in Closed System39 (EPICS) or Phase Ratio Variation (PRV).34,40 The latter measures the partition coefficient using the fact that headspace concentration changes as a function of the ratio between gas and liquid phases. The PRV method presents various advantages: it requires no calibration, and sample preparation is easy; it is capable of measuring several partition coefficients in multicomponent systems and also offers good reproducibility.25 Headspace analysis by gas chromatography coupled to mass spectrometry requires a highly sensitive detector, due to the very low concentrations of volatiles in the gas phase. In the 1960s, several authors introduced a new concept, low-pressuregas chromatography (LP-GC), which consisted of generating a vacuum in the analytical column with the outlet at atmospheric pressure.41,42 This technique reduced gas viscosity and, consequently, accelerated the analysis. In 2000, Zeeuw et al. proposed a new design for LP-GC/MS analysis, with a short, narrow restriction capillary column at the inlet (acting as a guard-column) connected to an analytical mega-bore column, finishing at the MS ion source.43 This assembly maintains the analytical column at the same vacuum as the mass spectrometer, whereas the guard-column keeps the inlet at normal GC pressure.44 In comparison to the microbore column, this approach reduces analysis time by a factor of 2, increases resolution, handles larger injection volumes, and reduces the Limits of Detection and Quantification.45 Although the impact of matrix composition is well-known, only a few studies have investigated the origins of perceptive interactions.46 Moreover, even if more and more perceptive interactions among volatiles impacting wine aromatic expression have recently been described thanks to sensory analysis, no previous study had determined whether these perceptive interactions were due to presensory, peripheral, or central effects. In this context, the aim of this work was to develop and optimize a method for calculating multiple partition coefficients in complex volatile mixtures, in order to investigate how presensory effects impact aroma perception.

butyl acetate, ethyl 3-methylbutanoate, hexyl acetate, 2methylbutan-1-ol, 3-methylbutan-1-ol, 2-methylpropan-1-ol, propan-1-ol, and butan-1-ol from Sigma−Aldrich, SaintQuentin-Fallavier, France; and 3-methylbutyl acetate from VWR-Prolabo, Fontenay-sous-Bois, France. Ethyl (2S)-2methylbutanoate, ethyl (2R)- and (2S)-2-hydroxy-4-methylpentanoate (95/5, m/m) were synthesized by Hangzhou Imaginechem Co., Ltd. (Hangzhou, China). All compounds used in this work were olfactorily pure, as confirmed by the three judges who performed gas chromatography analysis with olfactometric detection of reference compounds. Moreover, flame ionization detector analysis confirmed the products’ very high purity. Sample. A red Vin de Pays d’Oc (Cabernet Sauvignon) wine was used for optimization and validation of the method as well as preparing the dearomatized red wine. Aromatic Reconstitution Preparation. Dilute alcohol solution was prepared using ethanol and microfiltered water to obtain an ethanol level of 12% vol (v/v) and 5 g/L tartaric acid (pH adjusted to 3.5 with sodium hydroxide). Dearomatized red wine (DRW) was prepared according to the method described by Lytra et al., by evaporating red wine to two-thirds of its original volume using a Rotavapor (Laborota 4010 digital Rotary Evaporator, Heidolph, Germany) with a 20 °C bath temperature.16 The liquid was then mixed with ethanol and microfiltered water to reproduce the alcohol concentration and volume of the original wine. The DRW was then supplemented with 5 g/L LiChrolut EN resin (40−120 μm), stirred for 12 h to eliminate all traces of esters and higher alcohols, and filtered twice. The resulting DRW did not contain any trace of the volatile compounds included in this study. The fruity aromatic reconstitution (FAR) was prepared with 13 ethyl esters and acetates in dilute alcohol solution or DRW at the average concentrations found in red wine (Supporting Information, Table S1).10 Complete reconstitution was achieved by mixing these 13 ethyl esters and acetates (at the average concentrations found in red wine) in dilute alcohol solution or DRW, with the five higher alcohols at levels representative of the low, medium, and high concentrations found in wine, as described in Table 1.

EXPERIMENTAL SECTION Reagents. Absolute ethanol (analytical grade, 99.97%) and sodium sulfate (99%) were provided by Scharlau Chemie S.A, Barcelona, Spain. Microfiltered water was obtained using a Milli-Q Plus water system (resistivity, 18.2 MΩ cm, Millipore, Saint-Quentin-en-Yvelines, France). Tartaric acid and sodium hydroxide were purchased from VWR-Prolabo, Fontenay-sousBois, France. Standard-grade purity compounds were obtained from commercial sources as follows: ethyl propanoate, ethyl 2methylpropanoate, ethyl butanoate, ethyl hexanoate, ethyl octanoate, ethyl 3-hydroxybutanoate, 2-methylpropyl acetate,

Chromatographic Conditions. The chromatographic conditions were based on Mastovskà et al. and Merabtine et al.45,47 The partition coefficients were evaluated using static headspace coupled to low-pressure GC-MS (SHS-LP-GC/ MS) under equilibrium conditions at 20 °C. Samples were poured into glass vials (22.8 mL, Chromoptic, France) and then incubated at 20 °C until thermodynamic equilibrium was reached. A 750 μL sample of the headspace was withdrawn using a 2.5 mL thermostatic gastight syringe, preheated to 35 °C on a Gerstel autosampling device (Mülheim an der Ruhr,

Table 1. Low, Medium, and High Concentrations of Higher Alcohols Used for Partition Coefficient Calculations in mg/ L (Cameleyre et al.18) low concentration medium concentration high concentration

2MBa

3MBa

2MPa

Pa

Ba

56 85 105

208 286 355

52 81 115

16 22 36

1.3 2 2.7

a

2MB, 2-methylbutan-1-ol; 3MB, 3-methylbutan-1-ol; 2MP, 2methylpropan-1-ol; P, propan-1-ol; B, butan-1-ol.

■

B


Article

Analytical Chemistry

concentration of the compound in the vial, and β is the ratio between headspace (Vg) and liquid (Vl) volumes. By plotting 1/A against β, eq 2 gives a linear relationship between 1/A and β, as follows:

Germany), and each vial was analyzed once. Gas chromatography analyses were carried out on an HP 5890 GC system coupled to an HP 5972 quadrupole mass spectrometer (Agilent). Injections were in splitless mode, using a 4 mm i.d. deactivated gooseneck splitless liner transfer (injector temperature, 220 °C; injector pressure, 3 psi; septum purged flow deactivated). Oven temperature was programmed from 35 °C (for 3 min) to 40 °C at 1 °C/min and from 40 to 220 °C at 10 °C/min. The carrier gas was Helium N5.5 (Air Liquide, France) with a constant flow of 1.3 or 12 mL/min for microand mega-bore columns, respectively. The mass spectrometer was operated in electron ionization mode at 70 eV in selectedion-monitoring (SIM) mode. Monitored ions are listed in Table S1 (Supporting Information). Optimizing Headspace Operating and GC/MS Parameters. The impact of various operating conditions on headspace analysis was evaluated. Five parameters were considered: (i) time to reach the thermodynamic equilibrium in the gas phase (from 0 to 2880 min); (ii) syringe filling rate (250, 500, and 750 μL/s); (iii) gas injection rate (250, 500, and 750 μL/s); (iv) volume ratio between the gas and liquid phases, from 455 to 1.28, corresponding to liquid volumes from 50 μL to 5 mL; and (v) the type of analytical column used (microbore BP21 capillary column (50 m × 0.32 mm i.d., film thickness, 0.25 μm, SGE) or mega-bore BP21 capillary column with low-pressure chromatography (30 m × 0.53 mm, film thickness, 0.5 μm, SGE, connected by means of a Siltite μunion (SGE) to a 7 m × 0.25 mm i.d. deactivated column (SGE) at the inlet end). Each of the above parameters was optimized separately. For esters and higher alcohols, the equilibrium time was evaluated in dilute alcohol solutions containing the mix of esters and higher alcohols, all at the average concentrations found in red wine (Supporting Information, Table S1).7,18 All the solutions were prepared at room temperature (20 °C); the vials were filled with 1 mL of each solution and loaded onto a tray cooler at 20 °C. The headspace was analyzed at 15 different times, from 0 to 2880 min, and the chromatographic peak area of each compound of interest was evaluated each time. For the syringe filling rate, gas injection rate, and the type of analytical column, vials were filled with 1 mL of a solution containing the higher alcohols and esters prepared at room temperature (20 °C). Calculating Partition Coefficients. The partition coefficient (kg/m) represents the ratio of concentrations between the gas phase (Cgas) and the liquid matrix (Cliq) of a volatile compound at the thermodynamic equilibrium: kg/m =

1 = a + bβ A

where a =

(1)

Partition coefficients were determined using the Phase Ratio Variation (PRV) method developed by Ettre et al.,34 who established the following equation, where the concentration of volatiles in the headspace is proportional to the sample volume in the vial: 1 1 1 = + ×β liq A fi × Ci × kg/m fi × Ciliq

1 fi × Ciliq × kg/m

and b =

1 fi × Ciliq

.

The value of the partition coefficient kg/m is thus equal to the ratio between a and b, where kg/m = b/a, expressed as a concentration ratio. Partition coefficients were determined by plotting the inverse of the chromatographic peak areas against the phase ratio, β, to obtain values for a and b. Glass vials (22.8 mL, Chromoptic, France) were filled with six different volumes of solutions containing volatiles in dilute alcohol or in dearomatized red wine (0.05, 0.1, 0.5, 1, 1.5, and 2 mL), with phase ratios from 227 to 10.4 (according to the liquid sample volumes). Method Validation. The linearity of the SHS-LP-GC/MS method was determined using the approach described by Green.48 Concentration linearity was achieved by spiking dilute alcohol solution with esters and higher alcohols at eight different concentrations. The peak area in the headspace of each target compound was plotted against concentration. Syringe filling linearity was evaluated by using eight syringes to collect volumes ranging from 250 to 2000 μL from the headspace of a vial containing esters and higher alcohols at the average concentrations found in red wines. The peak area ratios were plotted against sampling volume. Linearity was evaluated by correlation coefficients and by determining the slope of the response factor plot. Concentration and syringe filling linearity was evaluated for a vial content of 1 mL liquid. Precision injection repeatability was assessed using dilute alcohol solution containing esters and higher alcohols at the average concentrations found in red wine as well as a red wine. Fifteen headspace injections were carried out on the same day and precision was evaluated by calculating the relative standard deviation (% RSD, corresponding to the ratio between standard deviation and the mean multiplied by 100). Reproducibility was assessed using dilute alcohol solution and red wine. All esters and higher alcohols were analyzed in quadruplicate by the same person, using the same equipment, on three different days over a 4-week period. Dilute alcohol solution and the red wine were aliquoted and stored at 4 °C prior to the analyses described above. Statistical Analysis. The Kruskall−Wallis test was used to evaluate the difference in chromatographic peak area variation to determine the equilibrium time and variations in the partition coefficients of the compounds of interest. The statistically significant level was 5% (XLSTAT software, version 2014.5.03, Addinsoft; p < 0.05).

Cgas C liq

(3)

■

RESULTS AND DISCUSSION Development and Optimization of the SHS-LP-GC/ MS Method for Determining the Partition Coefficient of Esters and Higher Alcohols. It was not possible to detect all the molecules targeted by this study using the optimized conditions described above, as hexyl acetate and ethyl 3hydroxybutanoate were not detected. In view of its Log P value (0.31), the latter compound may have a high affinity for the

(2)

kg/m is the partition coefficient between the gas and the matrix, A is the chromatographic peak area at the thermodynamic equilibrium, f i is the detector response factor, Cliq i is the initial C


Article

Analytical Chemistry matrix (dilute alcohol solution) and thus be retained.49 Moreover, it was hypothesized that this compound was added at concentration below its detection limits. The latter hypothesis is also applicable to hexyl acetate, in view of its concentration in the matrix (2 μg/L), as confirmed by evaluating concentration linearity. Chromatographic conditions and, more precisely, the column phase used (BP21 capillary column, SGE, Nitroterephthalic acid modified polyethylene glycol) were not capable of separating 2- and 3methylbutan-1-ol. Other types of columns, such as the CP-Wax 57 (50 m × 0.32 mm i.d.; film thickness, 0.25 μm; Agilent), were capable of separating these two molecules but only detected 5 of the esters used in the aromatic reconstitutions. Consequently, 2- and 3-methylbutan-1-ol were studied as a single peak area during optimization and validation of the method. In conclusion, the method was optimized and validated for 12 ethyl esters and acetates, with the exception of ethyl 3-hydroxybutanoate and hexyl acetate, as well as 5 higher alcohols. Injection on micro- (0.32 mm i.d.) and mega-bore capillary columns (0.53 mm i.d.) was compared, with low-pressure chromatography on the latter. Otherwise, conditions were identical on both columns: 750 μL of gas phase injected, 3 psi pressure in the injector, 5 mL liquid phase containing esters and higher alcohols in the vial. An effect on chromatographic peak area was observed for 11 of the 15 molecules analyzed (p < 0.05) (Supporting Information, Figure S1). More precisely, low-pressure chromatography with the mega-bore column produced significantly higher chromatographic peak areas for 3 higher alcohols and 7 ethyl esters and acetates than on the microbore column (p < 0.05). The mega-bore column also detected butyl acetate and ethyl 2-hydroxy-4-methylpentanoate. This correlated with previous findings, which highlighted a decrease in detection limits with this type of capillary column.45 Moreover, for all the compounds studied, the megabore column significantly reduced analysis time by a factor of 2.2. This result was in accordance with other results obtained for pesticide and essential oil analyses, where the mega-bore column reduced analysis time and produced taller chromatographic peaks.45,50 Moreover, injections using the microbore also resulted in poorer resolution, with splits in the chromatographic peaks of 8 of the 15 molecules (results not shown). In view of these results, as well as easier sample loadability and ruggedness, the low-pressure, mega-bore capillary column was used for all the chromatographic analyses. To our knowledge, this was the first time that static headspace analysis was coupled to a low-pressure mega-bore column. In view of the large volumes injected (about 1 mL), the impact of pressure in the injector was evaluated using three different values: 3, 5, and 7 psi. As shown in Figure S2a (Supporting Information), among the esters, injector pressure only impacted ethyl octanoate, with a higher peak area at 7 psi, while best results for higher alcohols were obtained at 3 psi. Previous work had demonstrated that higher inlet pressures did not produce higher relative responses.51 This is in accordance with our results using LP-GC/MS injections, except for ethyl octanoate. Finally, we decided to use 3 psi inlet pressure for the analysis. Sampling and injection rates are two important parameters in headspace injection. The practical effect of the filling rate (sampling the headspace from the chromatographic vial) was evaluated by using three different rates: 250, 500, and 750 μL/ s. The rate tested did not influence the chromatographic peak

areas (results not shown). This observation was in accordance with previous research on ethyl hexanoate.25 Injection at three different rates (injection from the syringe to the chromatographic injector) was also tested: 250, 500, and 750 μL/s. As shown in Figure S2b (Supporting Information), injection at 250 and 500 μL/s was best suited to three esters and all the higher alcohols, producing significantly higher peak areas compared to injection at 750 μL/s. In contrast, injection at 500 and 750 μL/s produced higher peak areas for butyl acetate. In view of this observation, injection at 500 μL/s was preferred to facilitate reproducibility of the chromatographic peaks, with lower standard deviations. Athès et al. reported similar results for ethyl hexanoate after testing rates from 100 μL/s to 600 μL/s.25 They reported that increasing rates of injection led to a decrease in chromatographic peak areas, with a maximum detection of the volatile compound at 100 μL/s. We finally decided to use 500 μL/s sampling and injection rates for this study. Partition coefficients were determined using variable volumes. In order to evaluate whether a saturation in the gas phase was observed for the compounds in this study, 8 volumes, from 0.05 to 5 mL dilute alcohol solution containing esters and higher alcohols at the average concentrations found in red wines, were put into 22.8 mL glass vials. For ethyl esters and acetates, saturation of the gas phase was observed in vials containing over 2 mL. As shown in Figure S3a (Supporting Information) for ethyl propanoate, the peak areas increased gradually and significantly between 0.05 and 2 mL liquid phase (p = 0.05), but no significant difference was observed between 3 and 5 mL. This result was not in agreement with previous research, demonstrating a significant increase in headspace concentrations of ethyl hexanoate at volumes between 0.05 and 5 mL, while the headspace peak area of this ester remained unchanged at liquid phase volumes above 5 mL.25 Nevertheless, this difference may be due to the concentrations used in the earlier study, which were 10-fold higher than in this work. For higher alcohols, no significant variation was observed above 2 mL solution, which may indicate a saturation of the headspace at this volume (p = 0.05) (Supporting Information, Figure S3b). In conclusion, volumes used to determine the partition coefficient ranged from 0.05 to 2 mL liquid phase for ethyl esters and acetates as well as higher alcohols. For all molecules studied, the partition coefficients determined using static methods were analyzed at the thermodynamic equilibrium. Therefore, the time to reach this equilibrium inside the vial was also studied for the 13 ethyl esters and acetates as well as the five higher alcohols. When the esters and higher alcohols were analyzed, separately or in a mixture, in water or dilute alcohol solution, no statistically significant difference in peak area variation was observed after over 7 h of static equilibrium (420 min, results not shown) (p < 0.05). Moreover, the peak areas for all the esters and higher alcohols were significantly lower in dilute alcohol solution than in water, with lower standard deviations (p < 0.001) (results not shown). This observation was consistent with the solubility of these esters in these two matrices, as reported in previous work, demonstrating that partition coefficients for these molecules decreased significantly with increasing ethanol levels in the matrix.22,23,25 Consequently, the partition coefficients for the esters and higher alcohols were analyzed after at least 7 h of equilibration at 20 °C. D


Article

Analytical Chemistry Method Validation. The linearity of ester and higher alcohol concentrations was evaluated by adding 1 mL dilute alcohol solution containing a range of concentrations observed in red wines to different vials. Calibration curves were produced by plotting peak areas against compound concentrations. The functions were linear for higher alcohols, with correlation coefficients above 0.997 (Supporting Information, Table S2). Correlation coefficients were above 0.996 for ethyl esters and acetates, except for butyl and hexyl acetates and ethyl 2-hydroxy-4-methylpentanoate, which had coefficients between 0.82 and 0.96. For these three compounds, the functions were linear at concentrations rarely found in red wines, i.e., above 32 and 80 μg/L for hexyl and butyl acetate, respectively, and 3.2 mg/L for ethyl 2-hydroxy-4-methylpentanoate (results not shown). For hexyl acetate, this result confirmed that this compound had been added at levels below its detection limit. The linearity of headspace volume collected was also studied using a range from 0.25 to 2 mL. The calibration curves were established by plotting the peak area against the collected volumes. The functions were linear with correlation coefficients above 0.99 for almost all esters and higher alcohols (Supporting Information, Table S2). As observed for concentration linearity, the functions were not linear for ethyl 2-hydroxy-4-methylpentanoate and butyl acetate (0.8 and 0.79, respectively), confirming that concentrations of these compounds were below their detection limits. To evaluate repeatability, 15 identical samples of dilute alcohol solution containing esters and higher alcohols at the average concentrations found in a commercial red wine were analyzed as well as a red wine. The relative standard deviations were below 5% for all compounds in dilute alcohol solution and red wine, except butyl acetate (18% and 20% in dilute alcohol solution and red wine, respectively) and ethyl 2hydroxy-4-methylpentanoate (20% and 40% in dilute alcohol solution and red wine, respectively) (Supporting Information, Table S2). To evaluate reproducibility, a dilute alcohol solution and a red wine were analyzed over a 4-week period. In dilute alcohol solution, the relative standard deviations were below 5% for the higher alcohols and 4% for the esters, except butyl acetate and ethyl 2-hydroxy-4-methylpentanoate (20 and 18%, respectively). The same trends were observed in red wine (Supporting Information, Table S2). In view of the nonlinear results obtained for butyl acetate, hexyl acetate, and ethyl 2hydroxy-4-methylpentanoate, their partition coefficients were not determined. Application of the New SHS-LP-GC/MS Method to Characterizing Sensory Effects. The new SHS-LP-GC/MS method developed and optimized in this work was used to calculate the partition coefficients of various ethyl esters and acetates, as well as higher alcohols, particularly to evaluate whether perceptive interactions observed by sensory analysis were explained by presensory effects. Partition coefficients were calculated for nine esters in dilute alcohol solution alone or supplemented with low, medium, or high concentrations of the five higher alcohols, for which a sensory effect had previously been demonstrated (Table 1).18 As shown in Figure 1a, in dilute alcohol solution, the addition of higher alcohols led to a significant decrease in ester partition coefficients (p = 0.05), except in the case of ethyl propanoate (p > 0.05). Higher alcohol partition coefficients were also calculated at low, medium, and high concentrations found in

Figure 1. Impact of various concentrations of higher alcohols found in red wines on the partition coefficients of ethyl esters and acetates in (a) dilute alcohol solution (12% v/v) and (b) dearomatized red wine. Letters refer to statistically different groups, identified by the KruskallWallis test; error bars indicate standard deviation. kg/m, partition coefficient between gas and liquid phase; C3C2, ethyl propanoate; C4C2, ethyl butanoate; C6C2, ethyl hexanoate; C8C2, ethyl octanoate; 2MeC3C2, ethyl 2-methylpropanoate; 2MeC4C2, ethyl 2-methylbutanoate; C2iC4, 2-methylpropyl acetate; C2iC5, 3methylbutyl acetate; 3MeC4C2, ethyl 3-methylbutanoate.

red wines, in dilute alcohol solution alone, or supplemented with a pool of 13 esters at average levels (Supporting Information, Table S1). In contrast to esters, the partition coefficients of higher alcohols were not impacted by this addition (p > 0.05) (results not shown). As the partition coefficient represents the distribution of molecules between the gas and liquid phases (eq 1), a decrease in this parameter indicated a decrease in volatilization into the gas phase.52 These results, therefore, indicated that the addition of higher alcohols led to a decrease in ester concentrations in the gas phase, proportional to the increase in alcohol concentrations, except isoamyl acetate. These observations may be explained by the addition of these five higher alcohols to the dilute alcohol solution at concentrations outside the infinite dilution region, which may change the thermodynamic properties of the mixture. In this study, the 5 higher alcohols were added at total molar fraction ranging from 6.7 × 10−2 to 1.2 × 10−1 for higher alcohols at low and high levels, respectively. Alessi et al. introduced the concept of “infinite dilution”, corresponding to the conditions where “the addition of an infinitesimal amount of the component 1 does not modified the thermodynamic behaviour of the mixture, that is like the component 2 does not notice the E


Article

Analytical Chemistry addition of the component 1”.53 It was also specified that the range of infinite dilution in mixtures started at a molar fraction less than 10−4.53−55 These data may explain why, in our context, the addition of higher alcohol, at concentrations above infinite dilution modified volatility of the esters. Previous sensory analyses demonstrated that adding higher alcohols raised the olfactory threshold of the pool of 13 esters, attenuating the perception of fruity notes and increasing the perception of butyric and solvent notes.18 The reduction in volatility of esters (contributors to the fruity character in red wines) in the presence of higher alcohols is a physicochemical fact, which may explain the attenuation of olfactory fruity perception after higher alcohols addition. The concomitant increase in the perception of solvent and butyric notes, common descriptors of higher alcohols, may be explained by the fact that their concentrations were not impacted by the presence of esters, for which concentrations in the gaseous phase decreased, facilitating the direct perception of the higher alcohols. In the past, various authors have explained the role of nonvolatile compounds in modulating the perception of aromatic nuances in various matrices;21,56−59 this was, to our knowledge, the first time that volatile-volatile interactions had been analyzed, thanks to partition coefficient determination, and correlated with sensory analysis data to elucidate sensory perception phenomena. The same experiment was conducted using a dearomatized red wine. As shown in Figure 1b, the ester partition coefficients were significantly lower in this matrix than in dilute alcohol solution (p < 0.01). In this matrix, the addition of higher alcohols also led to a decrease in ester partition coefficients, except for ethyl octanoate and isobutyl acetate (p > 0.05), with varying effects depending on the concentrations of higher alcohols added (p = 0.05). As observed in dilute alcohol solution, the addition of esters did not affect the release of higher alcohols in the gas phase (results not shown). This difference may be due to the presence of various nonvolatile molecules in the matrix, including phenolic compounds, sugars, and polysaccharides, which impact the release of volatile compounds from the liquid phase.27,29−33,60 Moreover, the reduction in ester levels and the increase in higher alcohols in the headspace may be responsible for the sensory results reported by previous authors and, more precisely, the intensification of solvent and butyric notes.18

time, the role of presensory interactions on the perceptive interactions reported in previous work. This new methodology may be validated for other volatiles involved in previously described perceptive interactions and used to assess whether physicochemical effects may, at least partly, modulate the release of volatiles in the gas phase, thus changing the perception of foods.

■

ASSOCIATED CONTENT

S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.analchem.8b01896. Comparison of injection on a micro-bore or lowpressure mega-bore column on the chromatographic peak area of higher alcohols, as well as ethyl esters and acetates; practical effect of injector pressure and injection rate on chromatographic peak areas of ethyl esters and acetates, as well as higher alcohols, in dilute alcohol solution; impact of liquid volume in the vial on the chromatographic peak area of ethyl propanoate and propan-1-ol; data summary of average concentrations and physicochemical properties of ethyl esters and acetates, as well as higher alcohols, used for partition coefficient calculations; and data summary for validation of the SHS-LP-GC/MS method for higher alcohols and esters (PDF)

■

AUTHOR INFORMATION

Corresponding Author

*Phone: +33557575863. Fax: +33557575813. E-mail: [email protected]. ORCID

Jean-Christophe Barbe: 0000-0001-6013-4770 Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS This work received financial support from Bordeaux Sciences Agro, the Regional Council of Nouvelle Aquitaine, France AgriMer, and the Bordeaux Wine Council (CIVB).

■

■

CONCLUSIONS A new approach was developed and validated with the aim of calculating multiple partition coefficients in complex mixtures, particularly for esters and higher alcohols. For the first time, we combined different methods usually used for headspace analysis and characterization on the one hand (static headspace) and pesticides analysis on the other hand (lowpressure gas chromatography). This methodology consisted of analyzing vial headspace at the thermodynamic equilibrium using a short guard capillary column connected to a mega-bore analytical column coupled to the MS detector. Associating these two techniques increased injection volume, making it possible to detect more compounds than a microbore column as well as decrease the run time. This method made it possible to calculate partition coefficients for multicomponent mixtures and revealed the decrease in ester release into the headspace from different matrices on addition of higher alcohols. By correlating these results to sensory analysis, this work highlighted, for the first

REFERENCES

(1) Genovese, A.; Dimaggio, R.; Lisanti, M. T.; Piombino, P.; Moio, L. Ann. Chim. 2005, 95 (6), 383−394. (2) Etievant, P. X. Wine. In Volatile Compounds of Food and Beverages; Maarse, H., Ed.; Marcel Dekker: New York, 1991; pp 483− 546. (3) Rapp, A.; Mandery, H. Experientia 1986, 42 (8), 873−884. (4) Pineau, B.; Barbe, J. C.; Van Leeuwen, C.; Dubourdieu, D. J. Agric. Food Chem. 2009, 57 (9), 3702−3708. (5) Kotseridis, Y. Etude de l’Arôme Des Vins de Merlot et de Cabernet Sauvignon de la Région Bordelaise. Thèse de Doctorat en Sciences Biologiques et Médicales, Bordeaux 2, Bordeaux, France, 1999. (6) Pineau, B.; Barbe, J.-C.; Van Leeuwen, C.; Dubourdieu, D. J. Agric. Food Chem. 2007, 55 (10), 4103−4108. (7) Lytra, G.; Tempere, S.; Le Floch, A.; de Revel, G.; Barbe, J.-C. J. Agric. Food Chem. 2013, 61 (36), 8504−8513. (8) Lytra, G.; Tempere, S.; de Revel, G.; Barbe, J.-C. J. Agric. Food Chem. 2014, 62 (22), 5005−5010. (9) Lytra, G.; Tempere, S.; Zhang, S.; Marchand, S.; De Revel, G.; Barbe, J.-C. J. Int. Sci. Vigne Vin 2014, 48, 75−85. F


Article

Analytical Chemistry (10) Lytra, G.; Cameleyre, M.; Tempere, S.; Barbe, J.-C. J. Agric. Food Chem. 2015, 63 (48), 10484−10491. (11) Cameleyre, M.; Lytra, G.; Tempere, S.; Barbe, J.-C. Food Chem. 2017, 237 (Supplement C), 364−371. (12) Renault, P.; Coulon, J.; de Revel, G.; Barbe, J.-C.; Bely, M. Int. J. Food Microbiol. 2015, 207, 40−48. (13) Atanasova, B.; Thomas-Danguin, T.; Langlois, D.; Nicklaus, S.; Etievant, P. Flavour Fragrance J. 2004, 19 (6), 476−482. (14) Atanasova, B.; Thomas-Danguin, T.; Langlois, D.; Nicklaus, S.; Chabanet, C.; Etiévant, P. Food Qual. Prefer. 2005, 16 (6), 504−510. (15) Atanasova, B.; Thomas-Danguin, T.; Chabanet, C.; Langlois, D.; Nicklaus, S.; Etiévant, P. Chem. Senses 2005, 30 (3), 209−217. (16) Lytra, G.; Tempere, S.; Revel, G. de; Barbe, J.-C. J. Agric. Food Chem. 2012, 60 (50), 12260−12269. (17) Chaput, M. A.; El Mountassir, F.; Atanasova, B.; ThomasDanguin, T.; Le Bon, A. M.; Perrut, A.; Ferry, B.; Duchamp-Viret, P. Eur. J. Neurosci. 2012, 35 (4), 584−597. (18) Cameleyre, M.; Lytra, G.; Tempere, S.; Barbe, J.-C. J. Agric. Food Chem. 2015, 63 (44), 9777−9788. (19) Berglund, B.; Berglund, U.; Lindvall, T. Psychol. Rev. 1976, 83 (6), 432−441. (20) Guyot, C.; Bonnafont, C.; Lesschaeve, I.; Issanchou, S.; Voilley, A.; Spinnler, H. E. J. Agric. Food Chem. 1996, 44 (8), 2341−2348. (21) Guth, H.; Rusu, M. Food Chem. 2008, 108 (4), 1208−1216. (22) Voilley, A.; Beghin, V.; Charpentier, C.; Peyron, D. Lebensm. Wiss. Technol. 1991, 24 (5), 469−472. (23) Whiton, R. S.; Zoecklein, B. W. Am. J. Enol. Vitic. 2000, 51 (4), 379−382. (24) Conner, J. M.; Paterson, A.; Piggott, J. R. J. Agric. Food Chem. 1994, 42 (10), 2231−2234. (25) Athès, V.; Peña y Lillo, M.; Bernard, C.; Pérez-Correa, R.; Souchon, I. J. Agric. Food Chem. 2004, 52 (7), 2021−2027. (26) Aznar, M.; Tsachaki, M.; Linforth, R. S. T.; Ferreira, V.; Taylor, A. J. Int. J. Mass Spectrom. 2004, 239 (1), 17−25. (27) Dufour, C.; Bayonove, C. L. J. Agric. Food Chem. 1999, 47 (2), 671−677. (28) Friel, E. N.; Linforth, R. S. T.; Taylor, A. J. Food Chem. 2000, 71 (3), 309−317. (29) Nahon, D. F.; Harrison, M.; Roozen, J. P. J. Agric. Food Chem. 2000, 48 (4), 1278−1284. (30) Swiegers, J. h.; Bartowsky, E. j.; Henschke, P. a.; Pretorius, I. s. Aust. J. Grape Wine Res. 2005, 11 (2), 139−173. (31) Lubbers, S.; Verret, C.; Voilley, A. LWT - Food Sci. Technol. 2001, 34 (4), 262−265. (32) Pozo-Bayón, M. Á .; Reineccius, G. Interactions Between Wine Matrix Macro-Components and Aroma Compounds. In Wine Chemistry and Biochemistry; Moreno-Arribas, M. V., Polo, M. C., Eds.; Springer: New York, 2009; pp 417−435. (33) Dufour, C.; Bayonove, C. L. J. Agric. Food Chem. 1999, 47 (2), 678−684. (34) Ettre, L. S.; Welter, C.; Kolb, B. Chromatographia 1993, 35 (1− 2), 73−84. (35) Jouquand, C.; Ducruet, V.; Giampaoli, P. Food Chem. 2004, 85 (3), 467−474. (36) Wen, Y.; Lopez, R.; Ferreira, V. J. Chromatogr. A 2018, 1534, 130−138. (37) Kolb, B.; Welter, C.; Bichler, C. Chromatographia 1992, 34 (5), 235−240. (38) Buttery, R. G.; Ling, L.; Guadagni, D. G. J. Agric. Food Chem. 1969, 17 (2), 385−389. (39) Gossett, J. M. Environ. Sci. Technol. 1987, 21 (2), 202. (40) Ettre, L. S.; Kolb, B. Chromatographia 1991, 32 (1−2), 5−12. (41) Giddings, J. C. Anal. Chem. 1962, 34 (3), 314−319. (42) Palamand, S. R.; Thurow, D. O. J. Chromatogr. A 1969, 40, 152−155. (43) de Zeeuw, J.; Peene, J.; Jansen, H.-G.; Lou, X. J. High Resolut. Chromatogr. 2000, 23 (12), 677−680. (44) Sapozhnikova, Y.; Lehotay, S. J. Anal. Chim. Acta 2015, 899, 13−22.

(45) Maštovská, K.; Lehotay, S. J.; Hajšlová, J. J. Chromatogr. A 2001, 926 (2), 291−308. (46) Tempere, S.; Schaaper, M. H.; Cuzange, E.; de Lescar, R.; de Revel, G.; Sicard, G. Food Qual. Prefer. 2016, 50, 135−144. (47) Merabtine, Y.; Lubbers, S.; Andriot, I.; Tromelin, A.; Guichard, E. J. Sci. Food Agric. 2010, 90 (9), 1403−1409. (48) Green, J. M. Anal. Chem. 1996, 68 (9), 305A−309A. (49) Mirhosseini, H.; Tan, C. P. Food Chem. 2009, 115 (1), 324− 333. (50) Poynter, S. D. H.; Shellie, R. A. J. Chromatogr. A 2008, 1200 (1), 28−33. (51) Cunha, S. C.; Fernandes, J. O.; Alves, A.; Oliveira, M. B. P. P. J. Chromatogr. A 2009, 1216 (1), 119−126. (52) Heilig, A.; Heimpel, K.; Sonne, A.; Schieberle, P.; Hinrichs, J. Int. Dairy J. 2016, 56, 101−107. (53) Alessi, P.; Fermeglia, M.; Kikic, I. Fluid Phase Equilib. 1991, 70 (2), 239−250. (54) Gruber, D.; Krummen, M.; Gmehling, J. Chem. Eng. Technol. 1999, 22 (10), 827−831. (55) Scheuren, H.; Baldus, M.; Methner, F.-J.; Dillenburger, M. J. Inst. Brew. 2016, 122 (1), 181−190. (56) Brauss, M. S.; Linforth, R. S. T.; Cayeux, I.; Harvey, B.; Taylor, A. J. J. Agric. Food Chem. 1999, 47 (5), 2055−2059. (57) Bücking, M.; Steinhart, H. J. Agric. Food Chem. 2002, 50 (6), 1529−1534. (58) de Roos, K. B. Int. Dairy J. 2003, 13 (8), 593−605. (59) Robinson, A. L.; Ebeler, S. E.; Heymann, H.; Boss, P. K.; Solomon, P. S.; Trengove, R. D. J. Agric. Food Chem. 2009, 57 (21), 10313−10322. (60) Linforth, R. S. T.; Friel, E. N.; Taylor, A. J. Modeling Aroma Release from Foods Using Physicochemical Parameters. In Flavor Release; ACS Symposium Series, Vol. 763; American Chemical Society, 2000; pp 166−178.

G


Static Headspace Analysis Using Low-Pressure Gas Chromatography

Recommend Documents