Article pubs.acs.org/JAFC
Study and Modeling of the Evolution of Gas−Liquid Partitioning of Hydrogen Sulfide in Model Solutions Simulating Winemaking Fermentations Jean-Roch Mouret,†,‡,§ Jean-Marie Sablayrolles,†,‡,§ and Vincent Farines*,†,‡,§ †
INRA, UMR 1083, 2 Place Viala, 34060 Montpellier Cedex 1, France Montpellier SupAgro, UMR 1083, 2 Place Viala, 34060 Montpellier Cedex 1, France § Université Montpellier, UMR 1083, 2 Place Viala, 34060 Montpellier Cedex 1, France ‡
ABSTRACT: The knowledge of gas−liquid partitioning of aroma compounds during winemaking fermentation could allow optimization of fermentation management, maximizing concentrations of positive markers of aroma and minimizing formation of molecules, such as hydrogen sulfide (H2S), responsible for defects. In this study, the effect of the main fermentation parameters on the gas−liquid partition coefficients (Ki) of H2S was assessed. The Ki for this highly volatile sulfur compound was measured in water by an original semistatic method developed in this work for the determination of gas−liquid partitioning. This novel method was validated and then used to determine the Ki of H2S in synthetic media simulating must, fermenting musts at various steps of the fermentation process, and wine. Ki values were found to be mainly dependent on the temperature but also varied with the composition of the medium, especially with the glucose concentration. Finally, a model was developed to quantify the gas−liquid partitioning of H2S in synthetic media simulating must to wine. This model allowed a very accurate prediction of the partition coefficient of H2S: the difference between observed and predicted values never exceeded 4%. KEYWORDS: hydrogen sulfide, wine fermentation, volatility, partition coefficient, gas−liquid transfer, modeling
■
INTRODUCTION Reduced sulfur compounds (molecules formed in the absence of oxygen) are a recurrent problem in winemaking because most smell foul (rubber, swamp, cabbage, garlic); they are also detrimental to mouthfeel and enhance bitterness.1 In particular, hydrogen sulfide (H 2 S) is produced naturally during fermentation but can leave a wine with the undesirable characteristic odor of rotten eggs. The threshold of perception of this molecule in wine is very low, in the range of 8−10 μg/ L.2 Several factors may be responsible for the presence of H2S in wine. It is well established that the amounts of H2S produced during fermentation depend on the strain of Saccharomyces cerevisiae used.1,3−6 H2S production also varies with the growth conditions of the yeast. Environmental and nutritional factors, including the availability of elemental sulfur,7 the presence of sulfur dioxide8,9 and organic compounds containing sulfur,3,10,11 nitrogen composition and limitation,5,10,12−15 timing of nitrogen addition,16 and vitamin deficiency,15,17−19 have all been found to be associated with the production level of hydrogen sulfide. The relationship between fermentation rate and rate of H2S production has been investigated:20 the concentration of H2S in wine at the end of fermentation depends primarily on the rate of its biosynthesis by the yeasts but may also be significantly affected by losses in the exhaust CO2.20 Estimating the transfer of H2S from the liquid to the gas phase is essential for calculating mass balances. The ability to calculate the total production (sum of the quantities accumulated in the liquid and lost in the gas) of this volatile compound and the amount remaining in the liquid is a major issue for optimizing fermentation control. From a technological point of view, the concentration of H2S remaining in the wine is © 2015 American Chemical Society
very important. Indeed, mastering fermentation, in particular through control of temperature, can help to minimize the H2S concentration in wines at the end of fermentation. The distribution of a volatile compound (i) between the liquid and gas phases depends on the vapor−liquid equilibrium (VLE), which is defined by the gas−liquid partition coefficient. The gas−liquid partition coefficient, also called the dimensionless Henry’s law coefficient, is defined as Ki =
cg(i) c l(i)
(1)
Ki is expressed as the ratio between the concentration of the compound in the gas phase [cg(i) in (mol or g) m−3] and that in the liquid phase [cl(i) in (mol or g) m−3] at equilibrium. The gas−liquid partition coefficient and Henry’s law constant Hi are related to each other by Hi = K i × R × T
(2)
where R (8.314 m3·Pa mol−1 K−1) is the gas constant and T (K) is the ambient temperature. Many of the data in the literature are presented as partition coefficient, Ki, or Henry’s law coefficient, Hi. However, strictly speaking, Hi is the partition coefficient at infinite dilution of species i. The gas−liquid partition coefficient can also be calculated using molar fractions or the activity coefficient (γi), Received: Revised: Accepted: Published: 3271
December March 12, March 12, March 12,
2, 2014 2015 2015 2015 DOI: 10.1021/jf505733a J. Agric. Food Chem. 2015, 63, 3271−3278
Article
Journal of Agricultural and Food Chemistry
132.5 g/L glucose and 5.3% (v/v) ethanol, or (c) 87.5 g/L glucose and 8% (v/v) ethanol. The resulting five media corresponded to 0, 20, 40, 60, and 100% progression of fermentation. Determination of Gas−Liquid Partition Coefficients. The gas−liquid partition coefficient of H2S was determined by a “semistatic” method: a gas phase for which the H2S concentration was perfectly known and constant was continuously bubbled through the stirred liquid phase until the system was driven to equilibrium using the device shown in Figure 1. The concentration of H2S in the
representing the deviation from ideality. These two coefficients are linked by the equation Ki =
yi xi
=
γi × Pi0(T ) PT
(3)
where xi and yi are molar fractions in the liquid and gas phases, 0 respectively, Pi(T) is the vapor pressure of pure component i at a given temperature T (Pa), and PT is the total pressure (Pa). Generally, Ki is measured at fixed temperature, pressure, and volume after the system reaches equilibrium. The most widely used methods for determining the partition coefficient can be divided into three groups: differential methods,21−23 dynamic headspace methods,24,25 and static methods.26−33 No available methods are valid for complex matrices or for all aroma compounds regardless of their volatility. Moreover, some determination methods are technically difficult to implement or require external calibration.34 H2S as a pure compound is very volatile, with a vapor pressure of 1698 kPa at 290 K.35 Therefore, existing methods are not appropriate for the determination of its partition coefficient. Indeed, its very high volatility makes precise calibration and quantification by chromatographic methods very difficult, so the first step of this study was the development of a novel method that could be described as a “semistatic method” specifically to measure the partition coefficient of H2S. The values of KH2S were first determined in water to validate the method and for comparison with published data. The method was then applied to synthetic media that simulated must, fermenting musts at different stages of fermentation, and wine. The effect of temperature on the partition coefficient of H2S (KH2S) was also assessed. Finally, a model was established to calculate KH2S in enological conditions throughout alcoholic fermentation.
■
Figure 1. Schematic representation of the stirred reactor and associated sensors used for determination of the gas−liquid partition coefficient of H2S.
liquid phase was continuously measured with the microsensor. A specially designed glass reactor with a total volume of 0.75 L (shown in Figure 1) was first filled with the liquid phase (500 mL) and regulated to the desired temperature (10, 15, 20, or 25 °C), within ±0.1 °C, by circulating water through the heating/cooling jacket fed by a cryostat Polystat cc1 (Huber, Offenburg, Germany). When the desired temperature was reached in the liquid phase, the gas supply to the reactor was opened with the lowest concentration of H2S in the gas (H2Sg; 112 ppmV at ambient temperature). All pipes connecting gas outside and inside the reactor were Sulfinert. Gas was sparged at 2000 NmL min−1 in the reactor through a stainless steel 316 L cylindrical porous distributor with a mean pore diameter of 0.5 μm. The liquid phase was stirred magnetically at 250 rpm. The concentration of H2S in the liquid phase (H2Sl) was recorded every minute for 20 min. Equilibrium was reached in the liquid phase after 15 min. The value of H2S at equilibrium (H2Sl*) was calculated as the mean of five instantaneous values (one value per minute for 5 min after the first 15 min required to reach equilibrium). The procedure for measuring the equilibrium concentration of H2S in the liquid phase was then repeated for four other values of H2Sg: 135.4, 158.7, 205.4, and 263.7 ppmV of H2S at ambient temperature. The H2S concentrations in the gas studied were thus varied from 112 to 263.7 ppmV, which corresponds to a partial pressure from 1.13 × 10−2 to 2.67 × 10−2 kPa (respecting Henry’s law,
