Abstract
Recent work suggests that copper complexes may serve as a latent source of free hydrogen sulfide (H2S) and other malodorous volatile thiols during wine storage. However, measurements of these complexes require analytical tools that are unavailable to most wineries. To facilitate additional studies of these compounds, we developed an inexpensive and convenient method for detecting copper-complexed H2S in wine with commercially available colorimetric gas detection tubes. Using brine dilution as a pretreatment, we observed that this approach showed acceptable detection limits (0.34 μg/L) and excellent H2S recovery from both model and real wines. Alternative approaches to H2S release from copper sulfide complexes were also investigated, including ethylenediaminetetraacetic acid (EDTA), neocuproine, ascorbic acid, and tris(2-carboxyethyl)phosphine (TCEP), alone and in combination. EDTA resulted in loss of free H2S. Partial recovery (35 to 70%) of H2S from model wines was achieved with the other reagents, with the release induced by ascorbic acid being of particular interest to winemakers. In a survey of seven commercial wines, the fraction of complexed H2S released by the brine was 80 to 95% of the total H2S pool, with lower releases observed for neocuproine and ascorbic acid. However, in several wines, TCEP treatment released higher concentrations of H2S from unknown precursors.
Reductive aromas, also called sulfur-like off-aromas (SLOs), reportedly account for a quarter of faults in wines (Goode and Harrop 2008). These reductive aromas are ascribed to the presence of low-molecular-weight volatile sulfur compounds (VSCs), which include hydrogen sulfide (H2S), methanethiol (MeSH), ethanethiol, dimethyl sulfide (DMS), methyl thioacetate, ethyl thioacetate, dimethyl disulfide, and diethyl disulfide (Ribereau-Gayon et al. 2006, Ugliano 2013). A recent survey has reported that in reductive commercial wines, H2S (“rotten egg” aroma) is the VSC most often present at concentrations above the sensory threshold (1 μg/L) (Siebert et al. 2010). MeSH and DMS were also detected at suprathreshold concentrations in several wines in that study, but other VSCs were rarely (if ever) present above their sensory thresholds (Siebert et al. 2010).
The major pathways responsible for H2S formation during fermentation are reasonably well understood. Sulfide (S2−) is produced by yeast from sulfite or sulfate as an intermediate in the sulfur assimilation pathway during biosynthesis of sulfur-containing amino acids (Ugliano and Henschke 2009). Factors that increase the rate of S2− formation, such as higher yeast sulfite reductase activity (Linderholm et al. 2010), or decrease the rate of S2− assimilation into amino acids, for example, low yeast assimilable nitrogen (Jiranek et al. 1995), are generally reported to increase total H2S formation during the fermentation. Pesticide residues, particularly elemental sulfur (S0) residues, may also be converted to H2S during fermentation, likely through nonenzymatic reaction of S0 with yeast-derived glutathione (Acree et al. 1972). Other pathways, such as cysteine catabolism (Moreira et al. 2002, Ribereau-Gayon et al. 2006), may also contribute H2S during fermentation. Because of its volatility, much of the H2S produced early in winemaking is lost through entrainment in CO2 produced by the fermentation (Park 2008). Residual H2S is generally treated through either sparging with inert gas, aeration to generate coupled oxidation products with quinones (Nikolantonaki and Waterhouse 2012), or addition of Cu(II) (e.g., as cupric sulfate) to form nonvolatile complexes (Zoecklein et al. 1999).
H2S may also accumulate during abiotic storage of some wines, particularly in those stored under very low oxygen conditions (e.g., in sealed ampoules or under screwcaps) (Lopes et al. 2009, Ugliano et al. 2011). H2S is particularly problematic if it is produced postbottling, since there is little opportunity for further remedial action. However, the pathways responsible for H2S formation in bottles and the chemical identity of the latent H2S precursors are still poorly understood (Ugliano 2013). Additions of Cu(II) to wine results in an immediate decrease in free H2S, but can yield higher concentrations of free H2S after bottle storage (Ugliano et al. 2011, Viviers et al. 2013). The authors of these studies proposed that this may be due to metal-catalyzed degradation of cysteine to yield H2S, but this possible process has not been further confirmed.
An alternative hypothesis—but not mutually exclusive with cysteine degradation—is that metal-sulfide complexes, particularly copper sulfide, can serve as a source of latent H2S and other malodorous thiols during anaerobic storage (Franco-Luesma and Ferreira 2014). Following Cu(II) addition, most copper sulfide complexes do not precipitate but instead remain dispersed in the wine, possibly because of inhibition due to tartaric acid (Clark et al. 2015). These complexes can be disrupted through addition of concentrated brine, resulting in a release of VSCs (Franco-Luesma and Ferreira 2014). These brine-releasable bound forms can represent a major portion of total (bound plus free) VSCs in commercial wines—more than 90% of H2S and approximately half of MeSH. Interestingly, the disappearance of these bound forms under anoxic conditions at elevated temperatures inversely correlates with formation of free H2S, suggesting that metal-sulfide complexes (particularly those involving Cu) could serve as a VSC source during storage (Franco-Luesma and Ferreira 2016). The mechanism responsible for the loss of copper sulfide during storage is unclear, but authors of previous studies (Franco-Luesma and Ferreira 2014) have speculated that this loss could result from reactions of Cu(II) with other wine components during storage, for instance, through complexation of copper by polyphenols as demonstrated in model wine systems (Esparza et al. 2007), resulting in the release of free H2S.
Further studies on the ability of copper sulfide or related complexes to serve as latent sources of H2S in both research and wine production environments will be facilitated by simple analytical methods. A previous report on analysis of latent VSCs in wine utilized headspace solid-phase microextraction (HS SPME) followed by gas chromatography (GC) coupled to a pulsed-flamed photometric detector (pFPD) (Franco-Luesma and Ferreira 2014). This approach, along with related sulfur-selective detection approaches such as chemiluminescence (Siebert et al. 2010), is outside of the capabilities of many laboratories. Additionally, the use of a 50-fold brine-dilution step would interfere with the ability of winemakers to perform a quick sniff test for latent SLOs, analogous to bench tests for free H2S and thiols based on copper salt additions (Zoecklein et al. 1999).
Gas detection tubes (GDTs) were originally developed for the mining industry more than 75 years ago and consist of a glass tube packed with a color-sensitive reagent that darkens proportionally to exposure to a target gas analyte. GDTs designed for measuring H2S in air samples have been adapted for quantifying free H2S in wines (Park 2008, Ugliano and Henschke 2010), and more recently for quantifying elemental sulfur in juices or musts after an initial reduction step (Kwasniewski et al. 2011). This work had therefore two aims: 1) to adapt and validate a method using GDTs to quantify H2S released from copper sulfide complexes after brine addition, and 2) to evaluate the ability of alternative copper-binding and reducing agents to release H2S from copper sulfide complexes.
Materials and Methods
Chemicals
Sodium sulfide nonahydrate (98%), sodium chloride, ascorbic acid, sodium hydroxide, tartaric acid, sodium sulfate, sodium thiomethoxide, potassium metabisulfite, copper sulfate pentahydrate, ethylenediaminetetraacetic acid (EDTA), tris(2-carboxyethyl)phosphine (TCEP), and neocuproine were all purchased at ≥99% purity from Sigma–Aldrich. Alka-Seltzer tablets (Bayer Healthcare) were purchased locally. Distilled deionized water was used in all experiments. GDTs for H2S detection (Gastec 4LT and 4LL) were purchased from Nextteq. The detection tubes rely on a colorimetric reaction within the tube between evolved H2S and a metal salt, which is either mercury chloride (Gastec 4LT detection tube) or lead acetate (4LL) adhered to a proprietary, inert matrix.
Preparation of Na2S stock solutions
Thirty mg Na2S•9H2O was dissolved in 100 mL cold distilled water (4°C) to give an S2− concentration of ~1.25 mM. This stock solution was kept refrigerated. A working 0.1-mM Na2S stock solution was prepared daily from the above stock solution in cold water (4°C) in 50-mL amber volumetric flasks and used immediately. This working Na2S stock solution was used for preparing the calibration standards as described below.
Preparation of model and real wine samples
The synthetic wine was made as described by Fang and Qian (2005). Tartaric acid (3.5 g/L) was dissolved into 1 L of a 12% ethanol solution, and the pH was adjusted to 3.5 with 1 M NaOH. The model wine was stored at 4°C when not in use and prepared fresh every two weeks. Commercial wine samples were purchased at a local wine store and cooled to 4°C in their original containers prior to opening. Samples were also handled at 4°C after opening, as previously described (Siebert et al. 2010).
Development of a method to determine free H2S with GDTs
Calibration and figures of merit for free H2S quantification in wine by GDTs
A method for free H2S quantification was adapted from previous work on elemental sulfur quantification using the apparatus shown in Figure 1 (Kwasniewski et al. 2011). Briefly, the apparatus consisted of a 250-mL flask with a 1-hole stopper (Figure 1A), connected via a short piece of plastic tubing fitted through the flask stopper (Figure 1B) to a H2S GDT (Figure 1C). The apparatus was leak-checked regularly.
Schematic of the gas detection tube (GDT) apparatus. Samples are added to a flask (A), and complexed hydrogen sulfide (H2S) is released by addition of brine or other reagents. The H2S evolved is swept through a short piece of tubing (B) by an in situ–generated CO2 stream into an H2S GDT (C). The length of the stain on the GDT is proportional to the original complexed H2S concentration.
H2S standards with concentration ranges of 2.3 to 376.3 μg/L were generated by appropriate additions of the working Na2S stock solution to 30 mL of synthetic wine in a 250-mL Erlenmeyer flask. Two Alka-Seltzer tablets were then added to the flask, the flask was fitted with an appropriate GDT, and the evolved gas was used to sparge H2S from the sample into the detection tube. After gas evolution was complete (~3 min), the ppm concentration reported on the detecting tube was recorded. A standard curve was created by plotting the ppm concentrations reported on the detecting tube versus the corresponding amount of H2S standard solution added. Analyses were performed at ambient temperature; because the H2S is exhaustively purged from the sample, careful control of temperature is not expected to be as important as in methods that require headspace equilibration (Siebert et al. 2010).
Method linearity was determined with six calibration levels, in triplicate, over the concentration ranges of 0.067 to 0.4 μM (2.3 to 13.6 μg/L as H2S) on a 4LT tube and 0.4 to 11.067 μM (13.6 to 376.3 μg/L as H2S) on a 4LL tube. The lower limit of detection (LLOD) and the lower limit of quantification (LLOQ) were determined according to ICH-Q2 (R1) guidelines with the equations LLOD = 3.3 × σ/S and LLOQ = 10 × σ/S, where σ is the standard deviation of the blank concentration calculated from the calibration curve, and S is the slope of the standard curve (Anonymous 2005).
Method precision for the model and real wine samples was determined with triplicate samples. The relative standard deviation (%RSD) for each calibration level was used as a measurement of precision.
Evaluation of interferences for free H2S
Four suspected interferences were evaluated: SO42−, SO2, MeSH, and copper sulfide complexes. All analyses were run in duplicate (and in triplicate for copper sulfide interferences) using the optimized free GDT methodology and 4LT detection tubes. For the first two interference experiments, a model wine and an authentic wine were each spiked with Na2SO4 (final concentration of 560 mg/L as SO42−) and K2S2O7 (final concentration of 250 mg/L as SO2).
For evaluating interferences due to MeSH, a sodium thiomethoxide stock solution was prepared in distilled water (50 mg/L as MeSH) and spiked into 30 mL model wine containing 0.067 μM H2S to yield MeSH concentrations of 0.00, 0.12, 0.24, 0.36, and 0.48 μM. The interference caused by MeSH was then calculated with respect to the H2S standard.
For evaluating the interferences due to copper sulfide complexes (bound H2S), 30 mL model wine and appropriate volumes of the stock H2S solution (all at 4°C) were added to a 500-mL Erlenmeyer flask to yield final H2S concentrations of 6.1, 9.1, and 13.8 μg/L (0.20, 0.27, and 0.41 μM). CuSO4•5H2O (30 mg/L in distilled water) was added to yield final copper sulfide concentrations of 0.0, 0.5, 1.0, and 1.5 mg/L (0, 2, 4, and 6 μM).
Quantification of copper-complexed H2S with GDTs: Method development
General information
For all experiments, Na2S stock solutions were diluted with the model wine to yield appropriate H2S concentrations, immediately followed by addition of CuSO4 stock solution (freshly prepared prior to analysis). After 30 sec, the model wine system was subjected to the treatments (brine, reducing agents, etc.) to disrupt copper sulfide complexes, and the free H2S was measured immediately afterward with the aforementioned GDT method.
Brine addition
Model wine (6 mL) containing H2S at three different levels (0.20, 0.27, and 0.41 μM) and 4 μM CuSO4•5H2O was used to evaluate the effectiveness of brine additions in releasing H2S from copper complexes. Prior to GDT analysis, duplicate wine samples were diluted with NaCl brine (35% w/v) at ratios of 1:3, 1:5, 1:7, or 1:10 (v/v). The released H2S was then immediately quantified with the GDT method. To ensure that all the H2S was sparged from the sample into the detection tubes, three Alka-Seltzer tablets were used if the total volume was larger than 30 mL.
Recovery of copper-complexed H2S from wine using the brine method
To evaluate recovery of copper-complexed H2S, three experimental red wines produced at Cornell University were used. These wines were selected because although they had free H2S, they had no copper additions and no detectable bound H2S after brine treatment. These wines were spiked with known concentrations of Na2S standards (2.3, 4.5, and 9.1 μg/L as H2S), and the total H2S concentrations before and after the addition were measured in triplicate with the brine method. Recovery was then calculated as:
Where C1 was the original total concentration of H2S in the wine before spiking, C2 was the total concentration of H2S after spiking, and C0 was the concentration of spiked H2S.
Addition of copper chelators and reducing agents
The efficacy of different chelators and reducing agents to release H2S from copper complexes was also evaluated. The chelators were EDTA (prepared as 0.01 M in deionized (DI) water and pH adjusted to 8.0 with 1 M NaOH), neocuproine (prepared as 5 mM neocuproine in ethanol), ascorbic acid (1% w/v in DI water), and TCEP (used as is).
The chelators and reducing agents were then tested on a model wine (30 mL) containing 0.267 μM H2S and 0.08 μM CuSO4•5H2O. The reagents were tested at the following concentrations: 1) TCEP (0.01 g was added to yield a final concentration of 1.32 mM), 2) neocuproine (final concentrations ranging from 0.25 to 1.24 mM), 3) EDTA (final concentration of 0.33 and 0.66 mM), and 4) ascorbic acid (final concentration ranging from 1.89 to 15.14 mM). After addition of each reagent, samples were tested with the GDT method within 10 min.
In a second experiment, the combinatorial effect of SO2 with neocuproine and/or ascorbic acid was tested. Model wines (30 mL) containing 0.067 μM H2S (nominal), 250 mg/L SO2 (added as sodium metabisulfite), and 0.02 μM CuSO4•5H2O were prepared. To these mixtures were added neocuproine (0.31 mM final concentration), ascorbic acid (3.78 mM final concentration), or both neocuproine and ascorbic acid (0.31 and 3.78 mM, respectively). Samples were then tested for H2S by the GDT method, as described above.
Quantification of complexed H2S in real wines: Treatment comparison
Seven commercial wine samples were obtained from a local wine store (Ithaca, NY). The free H2S in the original samples was measured as described above, and total H2S was measured by three approaches described below. All samples were analyzed in triplicate. 1) Brine: similar to the procedure described above, except that 50 mL NaCl brine (35% w/v) was added to a 5-mL sample of each wine to increase the sensitivity of the technique; three Alka-Seltzer tablets were also used. 2) Neocuproine and ascorbic acid: the concentrations of ascorbic acid (7.57 mM) and neocuproine (0.99 mM) used were based on those determined in optimization experiments; otherwise, the procedure was the same as described above, except that a 30-mL wine sample was used in place of a standard. 3) TCEP: The procedure was the same as described above, except that a 30-mL wine sample was used instead of a standard.
Statistical analysis
All data are presented as mean ± SD from three independent experiments. In multiple comparisons, the Tukey test was applied as a post-hoc test for the ANOVA one-way analysis of variance on the SPSS software, with p < 0.05 considered statistically significant. For figure of merit data, coefficient of variance (%CV) for each calibration concentration was calculated as (SD / mean value) × 100%. The mean %CV (%RSD) was calculated as the arithmetic mean of %CV values for calibration points that were within the quantification range.
Results and Discussion
Quantification of free H2S with GDTs
The goal of this work was to develop and validate a method for quantifying copper-complexed H2S, based on the premise that complexed H2S could be released by pretreatment of samples with brine, copper chelators, or reducing agents, and that the resulting free H2S could be quantified with GDTs. As an initial step, it was necessary to validate that GDT could be used for quantification of H2S in wine. GDTs previously have been used for H2S quantification in studies of grapes and wine, but under different circumstances. Multiple authors have reported using GDT to quantify total H2S formed during fermentation. However, in these reports, the CO2 gas evolved during fermentation had directed the H2S to the GDT (Park 2008, Ugliano and Henschke 2010). In the present report, we used CO2 generated by addition of antacid tablets to sweep dissolved H2S into the GDT. This approach to measuring free H2S in a finished wine has also been described in a patent (Park 2000), although specific data were not included. A related approach was used by our group to quantify H2S produced via reduction of S0 residues in grape must (Kwasniewski et al. 2011).
Calibration and figures of merit for free H2S method
We evaluated standards of free H2S in model wine with two different GDTs; the linear ranges, LLOQ, LLOD, and other figures of merit for each detection tube are summarized in Table 1. We observed good linearity and precision (r2 > 0.99, average CV < 5%) for both the 4LT (range 2.3 to 13.6 μg/L) and 4LL (range 13.6 to 376.3 μg/L) detection tubes, where the average CV was calculated with calibration points that were only within the quantification range. The LLOD and LLOQ for the more sensitive 4LT tube were 0.34 and 1.03 μg/L, respectively. Both the precision and the detection limits were comparable to those reported in a recent study using GC-pFPD and should be suitable for quantifying H2S at concentrations approaching its sensory limit in wine (1 μg/L) (Siebert et al. 2009). The detection limit is expected to scale with sample size, since detection of H2S is limited by the extent of GDT staining. On a mass basis, the methodological LLOD is 10 ng H2S, taking into account our 30-mL sample size.
Detection limits and quantification ranges for detection tubes.
Evaluation of interferences
The effects of several potential interferences were evaluated. Interferences from SO42− and SO2 were evaluated at the upper levels of typical concentrations for these two compounds in wine (560 and 250 mg/L, respectively). No interferences were observed, in agreement with our previous work on quantification of S0 using the same GDT. Although SO2 can reportedly interfere with the performance of sulfide-detection tubes, it will exist primarily as nonvolatile HSO3− species over the pH ranges encountered during the analysis.
The less sensitive 4LL tubes did not show an interference from MeSH, as expected from the manufacturer’s notes. However, the more sensitive HgCl2-based tubes (4LT) are reported to react with MeSH (Ugliano and Henschke 2010). Using our assay, we observed that MeSH solutions (0 to 0.477 μM) generated a linear response, but with a 3-fold lower molar sensitivity than for H2S, presumably due to the lower volatility of MeSH. Because the concentration of complexed H2S is reported to be 10-fold higher than that of complexed MeSH in the average wine (Franco-Luesma and Ferreira 2016), and because of the greater sensitivity of the assay to H2S, the effects of MeSH on measurements of complexed H2S were expected to be negligible. Other volatile thiols appear to be unlikely to cause major interference, since they typically exist at much lower concentrations, have much lower volatilities, or both. However, concentrations of free H2S and free MeSH in commercial wines are reported to be roughly equal (Franco-Luesma and Ferreira 2016), which would equate to an average interference by MeSH of ~25% on the free H2S signal. A correction for free H2S measurements should be possible by using a thiol-selective GDT, but because the primary goal of this work was to evaluate methods for measuring complexed H2S, this direction was not further explored.
Response of the free-H2S measurement method to copper-complexed H2S in model wine
Copper will form nonvolatile complexes with H2S and thiols, and Cu(II) salts are therefore widely used for treating wines with sulfurous off-aromas (Zoecklein et al. 1999). However, it has recently been observed that copper sulfide complexes do not fully precipitate and may stay dispersed in wine (Clark et al. 2015). Furthermore, some of the common methods used for measuring H2S in wine (e.g., SPME after brine addition and/or anoxic heating) result in disruption of H2S complexes with copper, as well as other transition metals (Franco-Luesma and Ferreira 2016).
To assess whether our simple GDT method measured only free or both free and complexed forms of H2S, copper sulfide complexes were prepared by adding Cu(II) (0.5 to 1.5 mg/L) to H2S-containing model wines at 5 to 30 molar excess. These model wines contained tartaric acid, which was recently shown to decrease copper sulfide particle size in simple model wine systems (Clark et al. 2015). Control samples were prepared without Cu(II) addition. No precipitate was observed in the reaction flask, and we observed detectable H2S only in the samples without Cu(II) addition. Considering the detection limit of the method, free H2S concentrations represented no more than 1.5% of the total H2S in the sample. Comparable results have been observed in real wines, in which on average >90% of H2S exists in complexed form (Franco-Luesma and Ferreira 2014, 2016). The lack of detectable H2S in wines with added Cu(II) indicates that without addition of other reagents, the GDT method measures only free H2S.
Quantification of complexed H2S by brine dilution prior to GDT analysis
Authors of previous reports of brine dilution to release H2S from transition-metal complexes have quantified released H2S with SPME-GC-pFPD (Franco-Luesma and Ferreira 2014, 2016). To simplify the method analysis, we developed a GDT-based method to measure complexed H2S released after brine addition. The test samples were model wines containing 6.1 to 13.8 μg/L H2S and a 10-fold or greater excess of Cu (4 μM or 0.26 mg/L). Dilution factors from 1:3 to 1:10 (wine to 35% w/v NaCl brine) were investigated, and the results are shown in Figure 2. As expected, no signal was observed for samples that were not diluted with brine, but H2S recovery increased with increasing dilution. Acceptable recovery (90 to 100%) was observed at a dilution ratio of 1:10 in the model wines. This dilution was somewhat lower than the 50-fold dilution recommended by a previous report using GC-pFPD detection, although the use of lower dilutions were not included in this previous report (Franco-Luesma and Ferreira 2014). In our work, dilutions of 1:7 or less resulted in unacceptable recoveries (<85%) for one or more of the samples (Figure 2).
Recovery (in %) of hydrogen sulfide (H2S) from copper sulfide complexes in model wine using the brine-dilution method and gas detection tube quantification.
The precision of the brine-dilution method coupled to GDT quantification (average CV <8%) was comparable to that for the measurements of free H2S. The accuracy of the brine method was evaluated by recovery spikes of Na2S (2.3 to 9.1 μg/L as H2S) added to three red wines produced at Cornell University. The concentration of free H2S in the original wines ranged from 3.8 μg/L to 5.8 μg/L prior to addition of the brine, and the complexed H2S (determined through the brine-addition GDT method) ranged from 3.4 μg/L to 4.9 μg/L. The results of the recovery experiments are presented in Table 2. Recoveries ranged from 94.9 to 98.9%, and coefficients of variance were <10% for all samples, again indicating the method’s suitability as a convenient alternative for quantifying complexed H2S.
Recovery of hydrogen sulfide (H2S) spikes from real wines with brine (n = 3).
Evaluating alternative approaches for disrupting copper sulfide complexes
Although the combination of brine dilution and GDT analysis was acceptable for quantifying copper-complexed H2S in model systems, we were interested in investigating additions other than brine that could disrupt these complexes for the following reasons: 1) Potentially, in a winery setting, the release of H2S or other thiols could be detected by a simple sniff test, but this would be complicated by the sample dilution caused by the brine. 2) The use of a high brine-dilution rate required the use of relatively large glassware to achieve similar detection limits as for free H2S.
We therefore evaluated four general strategies for releasing H2S from copper sulfide complexes prior to GDT analysis: copper chelators, reducing agents, and combinations of the two approaches.
The copper chelators EDTA and neocuproine
Using copper sulfide complexes in model wine, we investigated two copper chelators: EDTA, which strongly chelates Cu(II) and to a lesser extent Cu(I), and neocuproine, which predominantly chelates Cu(I). The effect of EDTA on H2S recovery is shown in Figure 3A. Because the binding of EDTA to transition metals is disfavored at low pH (Wiberg 2001), the lack of a significant increase in free H2S was not surprising. Instead, the presence of EDTA resulted in an unexpected decrease in free H2S. We observed a light yellow precipitate at the bottom of the flask after the reaction, suggesting that the EDTA-Cu(II) complex may have facilitated oxidation of H2S to S0, analogous to reactions used for industrial desulfurization of wastewaters or gas streams (Equation 1) (Deshmukh et al. 2013, Steudel 1996).
Eq. 1
(A) Effects of ethylenediaminetetraacetic acid (EDTA) on hydrogen sulfide (H2S) recovery from model wine and real wine. (B) Effects of neocuproine and ascorbic acid on H2S recovery from model wine. (C) Effects of neocuproine and ascorbic acid on H2S recovery from real wines (n = 3). Model wines contained added copper sulfate and H2S, and real wines were spiked with copper sulfate. Recovery was calculated as the percentage of the added H2S quantified by gas detection.
In real wines, we observed no significant change in H2S after the EDTA addition. Although not as problematic as the results in the model wine, we had previously observed that these wines possessed complexed H2S that was releasable by brine treatment. The reason for the non-effect of EDTA (neither increase nor decrease in H2S) in real wines was unclear. Regardless, EDTA appears to be unsuitable for releasing complexed H2S in wine.
Neocuproine, a Cu(I) binder, was also evaluated for its abilities to liberate H2S from copper complexes. Neocuproine was selected instead of the Cu(II) binder, bathocuproine, because Cu(I) is believed to be an important species in reductive wine environments (Danilewicz 2007), although as a caveat, the presence of tartrate may stabilize Cu(II) in wine (Clark et al. 2015). We originally had expected that it would be necessary to add reducing agents to the model wine for neocuproine-induced release of H2S. Interestingly, we observed neocuproine-induced release of H2S from model wines in the absence of reducing agents (Figure 3B), even though copper was added in the form of Cu(II). The maximum recovery of H2S from model wine containing copper sulfide complexes was ~70% using a neocuproine concentration of 1.24 mM. Higher neocuproine concentrations resulted in discoloration of the GDT through an unknown interference. However, even before this point, the recovery of complexed H2S appeared to plateau (Figure 3B). Work in model seawater systems has shown that Cu(II)S complexes react to form stable Cu(I)S nanoclusters, which may stay dispersed in solution (Luther et al. 2002). This reduction of Cu(II) to Cu(I) is coupled to oxidation of sulfur to form S-S bonds. Our observations suggest that neocuproine addition may favor reduction of Cu(II) to Cu(I), resulting in concurrent oxidation (and partial loss) of sulfide.
In spiking experiments with real wines and neocuproine additions, we observed incomplete recovery (58.8%) of complexed H2S (Figure 3C). Although this lower recovery could be due to reasons similar to those affecting recovery from model wines, it may also arise from the selectivity of neocuproine toward copper, as other transition metals have also been implicated in forming complexes with sulfide in wine (Franco-Luesma and Ferreira 2014).
We then evaluated whether addition of reducing agents commonly used in wine production (SO2 and/or ascorbic acid) would increase H2S release (Figure 4). These compounds can reduce Cu(II) to Cu(I) (Danilewicz 2007), which we hypothesized should favor complexation by neocuproine and release of H2S. This hypothesis appeared to be validated for SO2, as significant increases in recovery (75 to 85%) were observed with respect to the neocuproine-only control when SO2 was also added either by itself or in combination with ascorbic acid. However, addition of ascorbic acid with neocuproine did not significantly increase H2S recovery. As described in the next section, ascorbic acid additions may concurrently release complexed H2S and result in partial oxidation of free H2S.
Recovery of hydrogen sulfide (H2S) from model wine and released by combinations of neocuproine and ascorbic acid with or without SO2. Bars with same letters were not significantly different (Tukey test, p < 0.05). Values are mean ± SD, n = 3.
The reducing agents ascorbic acid and TCEP
As an alternative to chelating agents, we also investigated whether reducing agents, including ascorbic acid alone, could be used to release H2S from copper complexes. Ascorbic acid is widely used in the wine industry as an antioxidant (Barril et al. 2016), and is also reported to be capable of reducing Cu(II) to Cu(0) (Xiong et al. 2011). Ascorbic acid–induced release of H2S from copper sulfide complexes in model wines was less efficient on a molar basis than neocuproine (Figure 3B); the highest recovery in model wine was 70% at an ascorbic acid concentration of 15 mM. Similar to neocuproine additions, a plateau was eventually observed with H2S recovery. Recovery of H2S from copper sulfide spikes in real wines (Figure 3C) was somewhat lower; the maximum recovery was 50% and plateaued at ascorbic acid concentrations above 8 mM. The reasons for the incomplete recovery are unclear, but the oxidation product of ascorbic acid (dehydroascorbic acid) is reportedly capable of oxidizing H2S (Hewitt and Dickes 1961).
Interestingly, ascorbic acid is also widely used as part of a bench test for disulfides in wine (Zoecklein et al. 1999). In a typical protocol, ascorbic acid is added, putatively to reduce disulfides to their corresponding thiols, the latter of which have much lower sensory thresholds and will bind to copper. Recent skepticism regarding the presence of disulfides in most wines (Ugliano 2013), coupled with our observation that copper sulfide complexes can be reduced by ascorbic acid, suggests that the target of ascorbic acid in wines may be copper sulfides (or other copper-mercaptide complexes) rather than disulfides. In this case, addition of ascorbic acid followed by copper sulfate would merely regenerate the complexes. This futile cycle of copper sulfide complex release and regeneration, while speculative, has been previously proposed (Franco-Luesma and Ferreira 2016).
Last, the ability of TCEP to reduce copper sulfide complexes was also investigated. TCEP is a relatively safe (and nonodorous) reducing agent. Although TCEP is best known for reducing disulfides to thiols, it also can reduce Cu(II) to Cu(I) (Yost et al. 2001). Since Cu(I) could potentially disproportionate to Cu(0) and Cu(II), we hypothesized that TCEP could be an effective reagent for releasing H2S from copper sulfide complexes while preventing oxidation of S2−. However, in a model wine system, we observed relatively poor recovery of complexed H2S with TCEP (35%, data not shown).
Evaluation of methods to determine complexed H2S release from real wines
Next, the new GDT method was used for quantification of complexed H2S after brine dilution in seven commercial wines (five whites, one red, and one blush), summarized in Table 3. Free H2S ranged from undetectable to 2.5 μg/L and complexed H2S from 2.3 to 14.0 μg/L. The mean concentration of complexed H2S in the blush and white wines (8.2 μg/L) was about one-third that reported in a recent survey of Spanish white and blush wines (25.7 μg/L, n = 8) (Franco-Luesma and Ferreira 2016). The majority of H2S existed in complexed rather than free forms; in the four wines with detectable free H2S, complexed H2S represented between 80 and 95% of the total H2S, similar to a previously reported range (Franco-Luesma and Ferreira 2016). In all wines, the concentration of complexed H2S exceeded the sensory threshold for free H2S. This is of potential concern to wine producers, since this complexed H2S fraction is reported to form free H2S during accelerated anoxic storage. Because our new GDT-based method does not require a coupled chromatography system, it may be well suited for a modestly equipped winery facility to evaluate the potential of their wine to develop off-aromas during storage. As a caveat, our current validation used only seven wines, which were mostly whites, and additional validation on a broader range of wines may be therefore appropriate.
Concentrations of free and releasable hydrogen sulfide (H2S) in commercial wine samples (n = 3 analytical replicates).
The same wines were also evaluated with the optimized neocuproine method. In comparison to brine dilution, lower concentrations of complexed H2S were detected with the neocuproine treatment in all wines. No complexed H2S was detectable in two of the wines, and in the five wines with detectable H2S after neocuproine treatment, the complexed H2S averaged only 33% of the amount released by brine dilution. This value was lower than the recoveries observed with recovery spikes in real wines (Figure 3C). Potentially, this is because real wines contain brine-releasable sources of H2S other than copper sulfide, e.g., complexes of sulfide with other metals such as Zn2+ or Fe2+ (Franco-Luesma and Ferreira 2014). These complexes would be perturbed by the less selective brine-dilution approach, but not by neocuproine.
Last, TCEP was evaluated for its ability to release complexed H2S from wine. Surprisingly, even though TCEP was much less effective than brine at releasing H2S from copper sulfide complexes in the recovery experiments, we observed higher concentrations of H2S released by TCEP in six of the seven wines (up to 25.3 ug/L). These results suggest the presence of additional TCEP-releasable sources of H2S (or thiol interferences) in wine. TCEP is well known to reduce disulfides to their corresponding thiols, and thiols have been reported to be a minor interference for GDT measurements. Although simple disulfide precursors of highly volatile thiols (such as dimethyl disulfide and diethyl disulfide) are reported to be undetectable in most commercial wines (Siebert et al. 2010), mixed disulfides or polysulfides are still a possible latent source of H2S or thiols. For example, we have recently reported that elemental sulfur pesticide residues can be metabolized into polar, wine-soluble intermediates that can be converted into H2S following TCEP addition (Jastrzembski and Sacks 2015). The identity of these intermediates was unclear, but could potentially involve polysulfide adducts formed after nucleophilic addition of glutathione or other thiols to elemental sulfur (Parker and Kharasch 1959). Finally, as mentioned above, conversion of molecular clusters of Cu(II)S to Cu(I)S has been observed in seawater (Luther et al. 2002). This conversion results in oxidation of sulfide through formation of disulfide bonds within the CuS complex. Hypothetically, sulfur in this form may be more efficiently converted to H2S by TCEP than by brine or neocuproine. Future work investigating the release of H2S from CuS over time could help resolve this question.
Conclusion
We have validated a novel, inexpensive approach to measuring copper sulfide complexes in wines based on GDT detection after brine addition. In agreement with previous work, most H2S in wines appears to exist in complexed rather than free forms. Because of the simplicity of this approach, it could be implemented in modestly equipped wineries. Other reagents, such as neocuproine, could also induce release of H2S from complexed forms, but with lower H2S yields than can be achieved with brine. Ascorbic acid could also release H2S from copper sulfide complexes, which puts into question the mechanism behind treatments of wines putatively containing disulfides. Last, although the commonly used biochemical reagent TCEP resulted in poor recovery of complexed H2S in recovery experiments from model wines, it resulted in greater release of complexed H2S from real wines than could be achieved with brine dilution. Further studies will be needed to confirm the identity of these species and whether they can release H2S under normal storage conditions.
Acknowledgments
The authors thank the China Scholarship Council for providing a postdoctoral research scholarship to Y.C. Financial support by the National Natural Science Foundation of China (no. 31471647), Young Scientist Foundation of Jiangxi Province ( 20142BCB23005), and the New York Wine and Grape Foundation is gratefully acknowledged. The authors declare no conflict of interest.
- Received March 2016.
- Revision received June 2016.
- Revision received July 2016.
- Accepted August 2016.
- ©2017 by the American Society for Enology and Viticulture