Effect of Wine Matrix Composition on the Quantification of Volatile Sulfur Compounds by Headspace Solid-Phase Microextraction-Gas Chromatography-Pulsed Flame Photometric Detection.

The analysis of volatile sulfur compounds using headspace solid-phase microextraction (HS-SPME) is heavily influenced by matrix effects. The effects of a wine matrix, both non-volatile and volatile components (other than ethanol) were studied on the analysis of several common sulfur volatiles found in wine, including hydrogen sulfide (H2S), methanethiol (MeSH), dimethyl sulfide (DMS), dimethyl disulfide (DMDS), dimethyl trisulfide (DMTS), diethyl disulfide (DEDS), methyl thioacetate (MeSOAc), and ethyl thioacetate (EtSOAc). Varying levels of devolatilized wine and common wine volatiles (acids, esters, alcohols) were added to synthetic wine samples to act as matrices. Sulfur standards were added and analyzed using gas chromatography with pulsed-flame photometric detection (GC-PFPD). Five internal standards were used to find best representatives of each compound despite matrix effects. Sensitivity remained stable with the addition of devolatilized wine, while addition of volatile components decreased sensitivity. DMS was found to be best measured against EMS; DMDS and the thioacetates were best measured against DES; H2S, MeSH, DEDS, and DMTS were best measured against DIDS. The method was used to quantitate the volatile sulfur compounds in 21 wines with various ethanol contents and volatile profiles.


Introduction
Volatile sulfur compounds (VSC), including H 2 S, methanethiol, ethanethiol, thiol esters dimethyl sulfide, dimethyl disulfide, as well as dimethyl trisulfide are frequently present in wine. These VSCs pose problems for winemakers as they exhibit off-odors of onion, garlic, cabbage, cheese, and rotten egg even at very low concentrations in wine, due to their very low sensory thresholds [1]. Wine makers need to know their concentrations at various stages of wine making process so proper mitigation actions can be taken. However, the analysis of VSCs is challenging because of their high volatility and low concentrations [2,3].
VSCs analysis typically needs to isolate these compounds from the sample, then separate them by gas chromatography before detection and quantification. Many conventional extraction techniques such as solvent extraction, static headspace sampling, or purge-and-trap are not quite suitable for the analysis of VSCs in wine. Solvent extraction causes loss of analytes during the concentration stage, particularly compounds with high volatility; headspace sampling often does not provide insufficient sensitivity for trace components; and purge-trap has great potential of thermal artifact formation. In addition, alcohols in wine further complicate the extraction and concentration. the target analytes. Six devolatilized wines (DVWs) served as non-volatile matrix standards, and mixtures of most-prominent non-sulfur-containing volatiles in wine were used as volatile matrix standards, including acids, alcohols, and esters, based on reported ranges in wines [19]. In addition, the method was used to analyze volatile sulfur compounds in 21 different wines with various volatile and nonvolatile composition.

Non-Volatile Matrix Effects
Results of DVW effects on sulfur extraction from chardonnay wine are shown in Figure 1. Though a very gradual decrease can be seen in all compounds, there is very little effect seen in chardonnay wine. Similar curves are seen for all other wine matrices ( Figure 2). The slight decrease as DVW content rises is likely due to a decrease in salt in the system, as 40% DVW reduces salt water content to less than 6 mL. 3 volatile compounds, limiting their ability to enter the headspace. Furthermore, not all sulfur volatiles are affected equally by matrix parameters [18].
This study aims to understand the influences of wine non-volatile and volatile components other than ethanol on the analysis of sulfur compounds using HS-SPME-GC-PFPD, and to develop a method to compensate for the matrix effect by selecting internal standards that behave most similarly to the target analytes. Six devolatilized wines (DVWs) served as non-volatile matrix standards, and mixtures of most-prominent non-sulfur-containing volatiles in wine were used as volatile matrix standards, including acids, alcohols, and esters, based on reported ranges in wines [19]. In addition, the method was used to analyze volatile sulfur compounds in 21 different wines with various volatile and nonvolatile composition.

Non-Volatile Matrix Effects
Results of DVW effects on sulfur extraction from chardonnay wine are shown in Figure 1. Though a very gradual decrease can be seen in all compounds, there is very little effect seen in chardonnay wine. Similar curves are seen for all other wine matrices ( Figure 2). The slight decrease as DVW content rises is likely due to a decrease in salt in the system, as 40% DVW reduces salt water content to less than 6 mL.

Volatile-Matrix Effects
The analyses of volatile sulfur compounds with varying levels of other (non-sulfur) volatiles are shown in Figure 4. Data is arranged by volatile-matrix level, ranging from 0 (no additional volatiles added) to 6. These correlate with the aforementioned concentrations of each compound in each set (acids, esters, alcohols). Analysis of the total mixture was performed foremost, in order to gauge effects; the total mixture most closely reflects that of a wine, which would not be completely deficient in one category. Thus, within a wine, the volatiles would have a cumulative effect as measured. Results from this total mixture best exemplify the effects of other volatile constituents on SPME adsorption of sulfur compounds.

Volatile-Matrix Effects
The analyses of volatile sulfur compounds with varying levels of other (non-sulfur) volatiles are shown in Figure 4. Data is arranged by volatile-matrix level, ranging from 0 (no additional volatiles added) to 6. These correlate with the aforementioned concentrations of each compound in each set (acids, esters, alcohols). Analysis of the total mixture was performed foremost, in order to gauge effects; the total mixture most closely reflects that of a wine, which would not be completely deficient in one category. Thus, within a wine, the volatiles would have a cumulative effect as measured. Results from this total mixture best exemplify the effects of other volatile constituents on SPME adsorption of sulfur compounds.

Volatile-Matrix Effects
The analyses of volatile sulfur compounds with varying levels of other (non-sulfur) volatiles are shown in Figure 4. Data is arranged by volatile-matrix level, ranging from 0 (no additional volatiles added) to 6. These correlate with the aforementioned concentrations of each compound in each set (acids, esters, alcohols). Analysis of the total mixture was performed foremost, in order to gauge effects; the total mixture most closely reflects that of a wine, which would not be completely deficient in one category. Thus, within a wine, the volatiles would have a cumulative effect as measured. Results from this total mixture best exemplify the effects of other volatile constituents on SPME adsorption of sulfur compounds.  As seen with ethanol, a strong decrease in the adsorption of volatile sulfur compounds is seen with increasing volatile-profile concentration. This suggests a competitive mechanism, as the volatile matrix components will fill the headspace and adhere to the fiber. The concentrations of each volatile added are insufficient to act as co-solvents as ethanol might, though may affect the equilibrium of volatiles in the headspace as more compounds become present [4,17].
The analyte-to-internal-standard ratios ( Figure 5) showed high variation. DMS still closely follows EMS. In ethanol studies, MeSOAc and EtSOAc both resemble EIS and DES, suggesting they might be accurate internal standards. However, the volatile-matrix data suggests that EIS loses its similarity at higher concentrations of volatiles. While EMS seems to match closely with both, ethanol studies showed it did not function well with varied ethanol content. Thus, DES is the internal standard of choice for the thioacetates. DEDS and DMTS, similar to the thioacetates, show good correlation with EMS. However, ethanol studies also suggested that DIDS was the only viable internal standard. H2S and MeSH are not well-represented by any of the internal standards, though they seem to correlate with EMS and DIDS. EMS was not found to correlate well with shifting alcohol contents, however, so DIDS remains the most viable internal standard for both.  As seen with ethanol, a strong decrease in the adsorption of volatile sulfur compounds is seen with increasing volatile-profile concentration. This suggests a competitive mechanism, as the volatile matrix components will fill the headspace and adhere to the fiber. The concentrations of each volatile added are insufficient to act as co-solvents as ethanol might, though may affect the equilibrium of volatiles in the headspace as more compounds become present [4,17].
The analyte-to-internal-standard ratios ( Figure 5) showed high variation. DMS still closely follows EMS. In ethanol studies, MeSOAc and EtSOAc both resemble EIS and DES, suggesting they might be accurate internal standards. However, the volatile-matrix data suggests that EIS loses its similarity at higher concentrations of volatiles. While EMS seems to match closely with both, ethanol studies showed it did not function well with varied ethanol content. Thus, DES is the internal standard of choice for the thioacetates. DEDS and DMTS, similar to the thioacetates, show good correlation with EMS. However, ethanol studies also suggested that DIDS was the only viable internal standard. H 2 S and MeSH are not well-represented by any of the internal standards, though they seem to correlate with EMS and DIDS. EMS was not found to correlate well with shifting alcohol contents, however, so DIDS remains the most viable internal standard for both. As seen with ethanol, a strong decrease in the adsorption of volatile sulfur compounds is seen with increasing volatile-profile concentration. This suggests a competitive mechanism, as the volatile matrix components will fill the headspace and adhere to the fiber. The concentrations of each volatile added are insufficient to act as co-solvents as ethanol might, though may affect the equilibrium of volatiles in the headspace as more compounds become present [4,17].
The analyte-to-internal-standard ratios ( Figure 5) showed high variation. DMS still closely follows EMS. In ethanol studies, MeSOAc and EtSOAc both resemble EIS and DES, suggesting they might be accurate internal standards. However, the volatile-matrix data suggests that EIS loses its similarity at higher concentrations of volatiles. While EMS seems to match closely with both, ethanol studies showed it did not function well with varied ethanol content. Thus, DES is the internal standard of choice for the thioacetates. DEDS and DMTS, similar to the thioacetates, show good correlation with EMS. However, ethanol studies also suggested that DIDS was the only viable internal standard. H2S and MeSH are not well-represented by any of the internal standards, though they seem to correlate with EMS and DIDS. EMS was not found to correlate well with shifting alcohol contents, however, so DIDS remains the most viable internal standard for both. The analysis of sulfur compounds using HS-SPME is heavily influenced by the presence of other volatiles. Little effect is seen from non-volatile matrix components. Based on the results of both alcohol effects and volatile effect, ideal internal standards to compensate for variation of these parameters in multiple wines are EMS (for DMS), DES (for DMDS, MeSOAc, and EtSOAc), and DIDS (for H2S, MeSH, DEDS, and DMTS). The analysis of sulfur compounds using HS-SPME is heavily influenced by the presence of other volatiles. Little effect is seen from non-volatile matrix components. Based on the results of both alcohol effects and volatile effect, ideal internal standards to compensate for variation of these parameters in multiple wines are EMS (for DMS), DES (for DMDS, MeSOAc, and EtSOAc), and DIDS (for H 2 S, MeSH, DEDS, and DMTS).
A typical chromatogram for wine analysis is shown in Figure 6. This chromatogram represents a Cabernet Sauvignon wine. Standard curves were constructed to represent a range of concentrations near the odor threshold, as well as potential levels in wines (Table 1). Good linearity was seen for all curves, with R 2 values greater than 0.99 for DMS, DMDS, MeSOAc, EtSOAc, DEDS, and DMTS. Highly volatile compounds H 2 S and MeSH achieved R 2 values greater than 0.97 (Figure 7).

8
A typical chromatogram for wine analysis is shown in Figure 6. This chromatogram represents a Cabernet Sauvignon wine. Standard curves were constructed to represent a range of concentrations near the odor threshold, as well as potential levels in wines (Table 1). Good linearity was seen for all curves, with R 2 values greater than 0.99 for DMS, DMDS, MeSOAc, EtSOAc, DEDS, and DMTS. Highly volatile compounds H2S and MeSH achieved R 2 values greater than 0.97 ( Figure  7).   The results of the analysis of 21 California wines are seen in Table 1. Traces of most compounds were found in all samples. Many wines had quantifiable levels of each sulfur compound. H2S was found frequently in trace amounts, though it may still be present in perceivable concentrations. Due to the broad range reported for its odor threshold value [20,21], it may be perceived at levels beneath its limit of detectability. White varietals like Chardonnay exhibit greater levels of H2S and MeSH than reds. DMS was found in slightly higher concentrations in red varietals, particularly Cabernet Sauvignon and Merlot. DMS and DMTS were the only compounds found consistently in all wines. Levels for DMS suggest a possible impact on the flavor of the wines, as concentrations slightly above the odor threshold are said to impart a beneficial fruity aroma [22]. The results of the analysis of 21 California wines are seen in Table 1. Traces of most compounds were found in all samples. Many wines had quantifiable levels of each sulfur compound. H 2 S was found frequently in trace amounts, though it may still be present in perceivable concentrations. Due to the broad range reported for its odor threshold value [20,21], it may be perceived at levels beneath its limit of detectability. White varietals like Chardonnay exhibit greater levels of H 2 S and MeSH than reds. DMS was found in slightly higher concentrations in red varietals, particularly Cabernet Sauvignon and Merlot. DMS and DMTS were the only compounds found consistently in all wines. Levels for DMS suggest a possible impact on the flavor of the wines, as concentrations slightly above the odor threshold are said to impart a beneficial fruity aroma [22].

Calibration of Sulfur Compounds
Hydrogen sulfide standards were prepared using equivalents of sodium sulfide (Na 2 S) dissolved in distilled water, and further diluted with cold (−15 • C) methanol. MeSH standards were prepared by bubbling the pure gas over cold methanol and recording gained mass. All other standards were prepared by dilution with cold methanol. A standard mixture (mix 1) was prepared containing DMS (3000 µg/L), MeSOAc (1285 µg/L), DMDS (218 µg/L), EtSOAc (564 µg/L), DEDS (55 µg/L), and DMTS (47 µg/L). Because MeSH readily oxidizes to DMDS, and the higher affinity for DMDS on the SPME fiber causes much greater peak responses, the two compounds were not calibrated simultaneously. A separate mixture (mix 2) was thus prepared containing MeSH (37 µg/L) and H 2 S (31 µg/L). A mixture containing EMS (5 mg/L), DES (1 mg/L), MIS (1.5 mg/L), and DIDS (25.9 µg/L) was used for internal standards. Calibration samples consisted of 2 mL synthetic wine (3.6 g/L tartaric acid) diluted to 10 mL with saturated salt water and ethanol, for a final ethanol content of 3%. Vials were flushed with argon and internal standards mixture (10 µL) and analyte calibration levels (20 µL) were added through the septum.

Volatile-Matrix Effect
Four separate sets of volatile-matrix standards were prepared; these consisted of acids (acetic, hexanoic, octanoic, decanoic), alcohols (2-methyl-1-propanol, 3-methyl-1-butanol, phenethyl alcohol), esters (ethyl acetate, 3-methyl-1-butyl acetate, ethyl hexanoate, ethyl octanoate, ethyl decanoate) and a total mixture of all three. Each set was prepared by diluting the respective compounds in cold (4 • C) ethanol. Final concentrations of each compound in the acid and alcohol mixtures (after added to synthetic wine to reflect base wine concentration) were 1, 2, 3, 4, 5, and 6 mg/L. Final concentrations of each compound in the ester mixture were 0.1, 0.25, 0.5, 1, 2, and 3 mg/L. Final concentrations of each compound in the total mixture of acids, alcohols, and esters, were the same as in their respective mixtures. The cumulative concentration of compounds in the highest level (level 6) of the total mixture consisted of four acids each at 6 mg/L, three alcohols each at 6 mg/L, and five esters each at 3 mg/L, thus 57 mg/L total. Wines were devolatilized as follows: 300 mL of wine was boiled using a rotary evaporator (Büchi, Switzerland) under vacuum at 40 • C and 85 rpm. Each wine was boiled until 40% remained (120 mL), then distilled water was added back to original concentration. This ensured all volatile compounds had been evaporated, including ethanol.