Identification of Hydrogen Disulfanes and ... - ACS Publications

22 downloads 0 Views 986KB Size Report
Nov 6, 2016 - Firmenich SA, Corporate R&D Division, P.O. Box 239, CH-1211 Geneva 8, Switzerland. §. Changins − Viticulture and Oenology, University of ...
Article pubs.acs.org/JAFC

Identification of Hydrogen Disulfanes and Hydrogen Trisulfanes in H2S Bottle, in Flint, and in Dry Mineral White Wine Christian Starkenmann,*,† Charles Jean-Francois Chappuis,† Yvan Niclass,† and Pascale Deneulin§ †

Firmenich SA, Corporate R&D Division, P.O. Box 239, CH-1211 Geneva 8, Switzerland Changins − Viticulture and Oenology, University of Applied Sciences and Arts Western Switzerland, CH-1260 Nyon 1, Switzerland

§

S Supporting Information *

ABSTRACT: Through the accidental contamination of a gas cylinder of H2S, the importance of polysulfanes for flint, gun powder, and match odors was discovered. The hydrogen disulfane was prepared from disulfanediylbis[methyl(diphenyl)silane], and its odor descriptor was evaluated in the gas phase from a gas chromatograph coupled to an olfaction port. The occurrence of this compound in flint and pebbles was confirmed by analyses after derivatization with pentafluorobromobenzene. The occurrence of this sulfane was also confirmed in two dry white Swiss Chasselas wines, sorted by a large-scale sensory analysis from 80 bottles and evaluated by 62 wine professionals. The occurrence of disulfane was confirmed for the two wines described as the most mineral. Polysulfane comprises a class of compounds contributing to the flint odor and that may contribute to the wine mineral odor descriptor. Due to the high volatility and instability pure HSSH was not isolated but kept in solution and its odor profile was described by gas chromatography coupled to an olfaction port as flint, matches, and fireworks with a higher odor intensity compared to H2S. KEYWORDS: disulfane and trisulfane, wine minerality, flint odor, bis(methyldiphenylsilane) disulfane



aroma is benzene methanthiol.23 Tominaga discussed the minerality linked to a smoky, empyreumatic aroma and the contribution of benzene methanethiol and 4-methyl-4-sulfanylpentan-2-one, which constitute the boxwood smell in their definition of gun flint aroma.23 No other reference exists other than scraping rocks together for the flinty odor. South African and New Zealand enologists agree with this definition of the flinty/mineral odor, which is mainly used for Corban and Sauvignon blanc wines.17−20 Green et al. studied the influence of mercaptans on minerality,21 whereas Esti et al. associated mineral odor with flint, sulfur, and tuff, which is a type of volcanic rock.22 With regard to the taste aspect, the perceived acidity driven by the ratio between malic and lactic acid is one type of minerality.17 Flint nodules are formed in an anaerobic environment. The sediments are rich in H2S produced by the bacterial reduction of sulfates. The presence of iron from clay minerals then immobilizes the H2S as iron sulfides.24 We can speculate that other sulfur forms, such as H2Sn and a range of polysulfides, are encapsulated and released when these rocks are scraped together. This research aimed to analyze the odor contaminants in a pressurized H2S bottle. We also searched for sulfanes in flint and conglomerate pebbles, as well as in Swiss dry white Chasselas wine.

INTRODUCTION Volatile organic sulfur compounds are important odorant contributors found in the environment, in food, and in the chemical signals of animals.1−4 Odorant molecules used to flavor food matrices are now well understood, but some pieces are still missing, mainly in terms of understanding unstable compounds. In the course of our research on toilet malodors, a gaseous reconstitution of malodors was delivered in olfactometers.5,6 Diluted H2S in N2 was blown into olfactometers with toilet malodor constituents.5 Over time, we noted that the smell of the hydrogen sulfide (H2S) changed. When H2S was the only gaseous compound in the olfactometer, the odor was no longer eggy or sewage, but changed to a flint-like odor, similar to the cold smell of fireworks or the smell associated with a dentist drilling teeth. Our hypothesis was that the odor originated from oxidized H2S because the source of organic compounds should be limited in a certified H2S/N2 gas cylinder. Di- and trisulfanes have been reported in the literature since the early 1900s but not their odors. Publications associated with these compounds are linked to the rubber, latex science, sensor development, and atmospheric and geological chemistry industries, but to the best of our knowledge not to the food products industry. The wellknown odors of flint have thus not been explained in detail.7−12 The misleading definition of minerality has been stressed by several authors,13,14 and a sensory consensus seems difficult to find.15−17 This descriptor is used mainly for dry white wines from cool climates such as Loire Valley and Burgundy, as well as for Riesling wine from Alsace and Germany. A mineral wine may be associated with odors such as stony, wet rock, chalky, or like rocks being scraped together.18−22 Flint odor is not well understood despite being widely used by enologists.17 The root of the problem is that the only odor reference for gun flint © 2016 American Chemical Society

Received: Revised: Accepted: Published: 9033

September 2, 2016 November 2, 2016 November 6, 2016 November 6, 2016 DOI: 10.1021/acs.jafc.6b03938 J. Agric. Food Chem. 2016, 64, 9033−9040

Article

Journal of Agricultural and Food Chemistry



°C/min to 300 °C (15 min); split 50:1, injection, 1 μL; MS emission tune fixed in μA; Cl, 5.1/100 (source temperature, 300 °C). Gas Chromatography Flame Photometric Detector Plus (FPD Plus). The GC was an Agilent 6890 equipped with flame photometric detector plus (FPD Plus). The column was a HP-5 capillary column (30 m × 0.25 mm), with a film thickness of 0.32 μm (Agilent). The temperature program was as follows: 45 °C, 5 min, then raised at 10 °C/min to 250 °C. Carrier gas flow (He) was 1.1 mL/min; split ratio, 1:10, injection volume, 1 μL of neat sample. High-Pressure Liquid Chromatography HPLC-UV. The acids were quantified using an HPLC Agilent 110 equipped with detector UV (DAD G1315B). The column was a Phenomenex Aqua 5 μm C18 125 Å, LC column, 250 × 2.0 mm. The elution type was isocratic with 50 mM KH2PO4, pH 2.9, with a flow rate of 0.25 mL/min. The flow and column diameter were changed according to application note Phenomenex App ID 14171. The dosage was performed exactly as described in the application note. Gas Analysis via NEM Derivatization. The gas was bubbled (350 L, 200 L/h) in the same conditions as described previously5 in a 500 mL solution of NEM (5 mg/L) with a KH2PO4/K2HPO2 buffer at pH 8 (0.01 M). The organic compounds were trapped on a solid phase extraction cartridge (OASIS) and desorbed with 10 mL of Et2O containing 1 μg of methyl octanoate as the IS. Preparation of HSSH via Disulfanediylbis[methyl(diphenyl)silane].25 The methyldiphenylsilane (25 g, 126 mmol) was heated at 185 °C for 24 h in the presence of elemental sulfur (4 g, 126 mmol). The crude mixture was cooled and distilled (125−135 °C, 0.1 mm) to give 10.4 g (yield = 35.8%). This compound was diluted in toluene (40 mL) and added dropwise (30 min) to NaH (2.2 g, 55% in oil, 50 mmol) in toluene (90 mL) in an ice bath (5−10 °C). The reaction mixture was filtered. Iodine (5.1 g, 20 mmol) in toluene (200 mL) was added dropwise (2 h), keeping the temperature at 20 °C with a cool water bath. As soon as the yellow color disappeared, the reaction was stopped. The solution was filtered and the solvent removed at 40 °C under vacuum on a rotary evaporator. The oil (4 g) (pH 6−7) was kept at −18 °C. To release HSSH, the crude oil (1 g, 2.2 mmol) was diluted in toluene (5 mL), and TFA (0.22 g, 2.0 mmol) was added and stirred for 12 h. A Vigreux column (1 cm) was then placed above the reactor, and about 1.0 mL of toluene enriched in HSSH was distilled under a 10 mmHg vacuum. The distillate was collected in a flask cooled to −78 °C. GC-MS-high resolution (HR): HSSH M + H = 65.9653 Δ = 1.97 ppm. Preparation of H2S Mother Solution. Pure H2S from a cylinder was bubbled to saturation in EtOH (100 mL). The outlet flask was connected to a bleach scrubber. When H2S absorption stopped, the flask was closed and weighed. The weight gain at 22 °C was precise and reproducible at 1.4 g/100 mL. This solution was used for quantitation. The solution was typically prepared Monday, used on Tuesday and Wednesday, and then discarded in diluted bleach solution. The saturated solution of H2S in toluene was prepared the same way. The solubility was 0.85 g/100 mL. Preparation of H−Sn−H and R−Sn−H in EtOH. The H2S/EtOH saturated solution (2 mL) was placed in a 20 mL vial and about 200 mg of Fe° was added. The solution can be used after 15 h at 22 °C as such or extracted with pentane (20 mL) and HCl (0.1 M), 2 × 10 mL. The ratio between H2S, HSSH, and HSSSH was not stable. The H2S/EtOH saturated solution (10 mL) was also mixed with CH3SH (1% EtOH, 4.8 mL, 1 mmol), C2H5SH (62 mg, 1 mmol), Fe° powder (100 mg, 18 mmol), and H2O2 (2 mmol, 30% in H2O, 230 μL). After 15 h at 22 °C, the mixture was extracted with pentane and washed with HCl (0.1 M), dried on anhydrous Na2SO4, filtered, and concentrated gently under argon. GC-MS-HR: CH3SSH M + H = 80.9831 Δ = 4.7 ppm, CH3CH2SSH 94.9988 Δ = 4.6 ppm. Sulfite Solution Treated with Iron. Na2SO3 (0.2 g, 1.6 mmol) and iron powder (10 g) were added to a white wine model system containing kL-malic acid (1.5 g), DL-tartaric acid (1.5 g), water (900 mL), and EtOH (100 mL) adjusted to pH 3.3 with NaOH (1 M). Every hour, 100 mL of the solution was extracted as described below for analysis.

MATERIALS AND METHODS

Chemicals. Commercially available reagents and solvents of adequate quality were used without further purification. The H2S for reactions was in a 227 g cylinder. C 2 H 5 SH (97%), 1,8diazabicyclo[5.4.0]undec-7-ene (DBU) 99%, trifluoroacetic acid (TFA) 98%, NaH 55% in oil, elemental sulfur (S8), the internal standard (IS) methyl octanoate, ethanethiol, propanethiol, iron powder (Fe°) (catalog 44900), salts for buffers, triethyl amine, 1Hpyrrole-2,5-dione, and 1-ethyl-(N-ethylmaleimide) (NEM) were from Sigma-Aldrich (Buchs, Switzerland). CH3SH in ethanol (1%) was made from in-house ingredients (Firmenich S.A., Geneva, Switzerland). Pentafluorobenzyl bromide (PFBBr) and methyldiphenylsilane were from Alfa Aesar (Heysham, UK). The contaminated H2S pressurized cylinder was a dilution containing 52.5 μL/m3 H2S and 15.4 × 103 μL/m3 N2 (Carbagas, Carouge, Switzerland). Solid phase extraction OASIS HLB cartridges (1 g) were purchased from Waters (Montreux-Chailly, Switzerland). Wines. A generic Chasselas wine was used for method development: Chasselas de Romandie 12% (Coop Basel, Switzerland). Four other dry white Chasselas wines were selected on the basis of their minerality exemplarity: Brez La Colombe 2012, Raymond Paccot (VDm54); Sélection Comby Héritage, 2012, Yann Comby (VSm72); Lu, 2012 Guy et Mathurin Ramu (GEa14); and Le Cellier du Mas, Grand Cru, Mont-sur-Rolle 2012, David et Françoise Blanchard (VDa60). Gas Chromatography (GC)−Electron Impact Mass Spectrometry (EI-MS). The analytical GC-MS was an Agilent GC-6890 system connected to an Agilent MSD-5973 quadrupole mass spectrometer (Palo Alto, CA, USA) operated at ca. 70 eV. Helium was the carrier gas set at a constant flow rate of 0.7 mL/min. Separations were performed on fused-silica capillary columns, coated with SPB-1 (Supelco, Buchs, Switzerland, 30 m × 0.25 mm i.d., 0.25 μm). The standard oven program was as follows: 40 °C for 8 min, increased to 250 °C at 15 °C/min; for derivatives, 100 °C for 5 min, increased to 250 °C at 10 °C/min, and then held at 250 °C. Gas Chromatography−Mass Spectrometry Olfaction (EI-MSO). The GC-MS-O analyses were carried out using an Agilent GC7890A system equipped with an Agilent 5975C mass spectrometer and a Gerstel Olfactometry Port (ODP3). The column was a DB-1 capillary column (60 m × 0.25 mm), with film thickness 0.25 μm (Agilent). The temperature program was as follows: 45 °C, raised to 120 °C by 5 °C/min, then by 10 °C/min to 250 °C, 5 °C isothermal, then raised by 15 °C/min to 300 °C, isothermal 20 min. Carrier gas flow (He): 1.1 mL/min; split ratio, 1:10, injection volume, 1 μL of neat sample. The crosspiece was installed in the GC oven to provide a 1:1 ratio between the sniffing port and the MS. Mass spectra were generated at 70 eV at a scan range from m/z 40 to 450. The compound identifications were made by using Firmenich MS and linear retention index (LRI) libraries and by comparison with reference samples. LRIs were determined after injection of a series of n-alkanes (C5−C31) under identical conditions on the first dimension (nonpolar). The LRIs were calculated by linear interpolation from the retention times of the analytes and the two closest alkanes. Mass spectra are available in the Supporting Information. Gas Chromatography Quadrupole Time of Flight (QTOF). The high mass resolutions were measured with an Agilent 7200 GC− quadrupole time-of-flight (QTOF) mass spectrometer operated in positive chemical ionization mode and controlled by MassHunter Acquisition B.07. The GC system was equipped with a DB-1 MS Ultra Inert capillary column (60 m, 250 μm i.d., 0.25 μm thick). The GC oven temperature was programmed to begin at 50 °C (held for 5 min) to increase to 120 °C by 5 °C/min, and then to increase to 300 °C by 10 °C/min (held for 15 min). The MMI inlet was set at 250 min °C in split mode with a 50:1 split ratio and an injection volume of 1 μL. Helium was used as a carrier gas at a flow rate of 1.5 mL/min. The QTOF MS was operated at 2 spectra/s in an m/z range of 30−500, with a resolution of approximately 15,000 at m/z 150−500. The column was a DB-1 UI (60 m, 250 μm i.d., 0.25 μm thick), operated at 50 °C (5 min), increased by 5 °C/min to 120 °C, and increased by 10 9034

DOI: 10.1021/acs.jafc.6b03938 J. Agric. Food Chem. 2016, 64, 9033−9040

Article

Journal of Agricultural and Food Chemistry

Figure 1. Structural formulas of NEM derivatives present in a contaminated H2S gas pressurized bottle.

Figure 2. Synthetic route used to prepare HSSH from its silanyl precursor. Wine Analysis of H2S and HSSH after Derivatization. Wine (100 mL) was extracted with a mixture of pentane (50 mL), diethyl ether (20 mL), and methyl octanoate (0.1 mL) as the IS (solution: 1 g/100 mL EtOAc). The free thiols were derivatized by adding PFBBr (10 μL) and DBU (10 μL). The solvent was removed by distillation to 5 mL and washed with H2O (2 × 2 mL), the organic phase was dried by using anhydrous Na2SO4 and filtered, and the solvent volume was reduced to 0.3−0.5 mL. Analysis of Flint and Pebble. Two were struck and immediately rinsed in a pentane solution containing PFBBr and DBU. The pentane was washed with water to remove DBU and dried and the solvent was injected on GC-MS. General Method for Quantitation Estimations. The Chasselas de Romandie wine (100 mL) was spiked with 0.1 mL of IS (solution: 1 g/100 mL EtOAc) and H2S, from the saturated solution in EtOH, 1400, 140, and 14 μg and no addition. The wine was extracted as described above. The extract was injected, and the peak area in SIM of 394 was recorded as well as the peak area of 74 (IS). The calibration curve was based on the ratio of 394/74 to give y = 0.0003x + 0.0004, RSQ = 0.997. The same wine was spiked with HSSH in toluene, 10 and 1 μL and no addition. The wine was extracted as described above. The extract was injected, and the peak area in SIM of 426 was recorded as well as the peak area of 74 (IS). The calibration curve was based on the ratio of 426/74 to give y = (3.37 × 10−5)x + 1.06 × 10−5, RSQ = 0.992. HSSH Sensory Evaluation. Subjects for GC-MS-O and odor profiles of H2S compared to HSSH were conducted by Firmenich S.A. (Geneva, Switzerland) employees from perfumery, flavor, and R&D. They were asked to participate in the GC-MS-O (12 subjects) study and to give a profile for the struck stones. Selection of Two Groups of Chasselas Contrasted on Their Minerality Level by Sensory Analysis. The wines came from a specific sensory analysis about minerality in wine performed by researchers at ChanginsUniversity of Applied Sciences and Arts Western Switzerland. Eighty Chasselas wines (vintage 2012) were selected in equal numbers from the four French-speaking parts of Switzerland (Vaud, Valais, Genève, and Neuchâtel) and spread over two sessions (40 wines per session). To perform this selection, 62 wine professionals participated at the two tasting sessions. For each wine presented, they had to answer the following question: “Do you think that this wine is a good example or a poor example of what a mineral wine is?”26−28 The minerality level evaluation was scored on a linear scale from 0 = “poor example” to 10 = “good example”.

Evaluation was global without any distinction between orthonasal and mouthfeel perception. A total of 30 mL of each wine was poured at 15 ± 2 °C into an official Institut National d’Appellation d’Origine (INAO) glass and coded with three random digits. The presentation was monadic and balanced, based on a Williams Latin square design. A 2 min break was imposed after every 10 wines and a 10 min break in the middle of the session (after 20 wines). Two significant distinct groups were identified and selected, seven with a poor level of minerality exemplarity and seven with a high level of minerality.16 Within each subgroup, two wines were selected for this study (Table 1 in the Supporting Information).



RESULTS AND DISCUSSION Analysis of the Contaminated H2S/N2 Cylinder. The GC-MS analysis of the contaminated gas cylinder displayed two major signals with a mass of 284, corresponding to the molecular formula C12H16N2O4S with m/z 127 (C6H9NO2), typical for the NEM moiety. These two signals were assigned to the NEM-S-NEM diastereoisomers.5,29,30 Additional compounds with a fragment corresponding to m/z 127 were detected with masses corresponding to 173, 187, 205, 219, and 316 (MS in the Supporting Information). The possible chemical structures should be limited to compounds that are closely related to H2S. From the masses and the fragmentation patterns of NEM derivatives, M+ 173 was attributed to methylmercaptan, M+ 187 to ethylmercaptan, and M+ 205 and 219 to their disulfides; M+ 316 could well be attributed to NEM-S-S-NEM. A GC-MS signal with a mass of M+ 159 and a fragmentation of m/z 126 and 60 was also obtained. This MS spectrum was never observed during our previous work when we focused on quantifying H2S and methylmercaptan in latrine headspace in developing countries. The mass of M+ 159, corresponding to C6H9NO2S, was tentatively assigned to NEMSH (Figure 1; Figures S1−S7). The flint smell could not come from methylmercaptan and ethylmercaptan because their odor profiles do not correspond to the smell liberated by the cylinder in the toilet model; therefore, the only possible origin of the smell was from the disulfane. To verify this hypothesis, we prepared the disulfane. 9035

DOI: 10.1021/acs.jafc.6b03938 J. Agric. Food Chem. 2016, 64, 9033−9040

Article

Journal of Agricultural and Food Chemistry

Figure 3. Mass spectra (EI-MS) of HSSH and HSSSH obtained via the silanyl precursors.

Figure 4. (A) GC-MS trace of pentafluorobenzene HSSH and HSSSH derivatives obtained via the silanyl precursor deprotected toluene solution (R = pentafluorobenzene). (B) Mass spectra (EI-MS) of RSR, RSSR, and RSSSR.

−18 °C without further purification. Sulfane HSSH (H2S2) (LRI 561) was generated when needed by using TFA and codistilled with toluene as described.25 It was not possible to avoid the formation of HSSSH (H2S3) (LRI 817). The proportion of trisulfane was around 10% of the disulfane in terms of peak area. The di- and trisulfanes are stable in crude distilled toluene solution. They can be injected on a GC apolar column but decomposed on a polar column (Figure 3). To avoid having the disulfane in toluene and TFA, we investigated a Fenton-type oxidation method. The H2S/EtOH saturated solution was added to a suspension of iron (Fe°) or a

Preparation of Hydrogen Sulfanes (H2Sn). The disulfane was prepared in 19309 by dissolving elemental sulfur in boiling water with Na2S, and then this solution was added to concentrated HCl. A mixture of sulfane was obtained by distillation. Reproducing this procedure failed in our laboratories, maybe because we worked on a too-small scale. A cleaner method was described by Becker and Wojnowski31 and Hahn and Altenbach (Figure 2).25 The disulfanediylbis[methyl(diphenyl)silane] was not stable on SiO 2 and decomposed during flash chromatography to give silane oxide derivatives. For this reason, the filtered solution was kept at 9036

DOI: 10.1021/acs.jafc.6b03938 J. Agric. Food Chem. 2016, 64, 9033−9040

Article

Journal of Agricultural and Food Chemistry

Figure 5. Structural formulas of PFBBr sulfur derivatives of the other main compounds found in stones and wine.

obvious even if H2S was the major compound in the headspace. The only way to evaluate the odor profile of HSSH pure was by GC-O. Its retention time on the GC apolar column is the same as that of EtOH, and it decomposes on polar column. When pentane is used as an extraction solvent, the hydrogen disulfane (LRI of 561) is in the peak tailing of the solvent. The pentane solution obtained by the extraction of the H2S oxidation in EtOH is not stable, probably because of the coextraction of radical intermediate species from the reaction in ethanol, but the crude HSSH solution obtained from the silanyl precursor in toluene is stable for several weeks at 4 °C. Therefore, we reasoned that the best estimate of odor intensity could be obtained by comparing the odor impact of HSSH with that of H2S by GC-O. We also supposed that smelling highly diluted H2S over milliseconds by GC-O did not affect our sense of smell, as we assumed that the time is too short for habituation. A saturated solution of H2S in toluene was prepared (8 ± 0.3 mg/mL) from the H2S cylinder and the concentration determined by weight. We added up to 500 μL of this H2S solution, portion-wise, to 400 μL of HSSH toluene solution and 100 μL of toluene to get the same peak area for H2S and HSSH by GC-MS in TIC. The same solution injected on a GC equipped with a pulsFPD gave a peak ratio of H2S 65% and HSSH 35%. This was our best guess for obtaining approximatively quantitative data, and we assumed that the concentrations of H2S and HSSH was 2/1 (4 mg/mL H2S and 2 mg/mL in toluene). The GC-O was performed with eight subjects. The sniffing times were predetermined for H2S (6.08 min) and HSSH (7.09 min), and the sniffing was stopped just before the elution of toluene. Only one sniffing per day was performed by each subject. The sniffing was repeated for four dilution steps corresponding to 1.3, 0.44, 0.15, and 0.05 mg/ mL. The focus was to describe the odor intensity on a linear scale from 0 to 5 and to describe the odor profile of hydrogen disulfane. The descriptors were flint, matches, firework, and cold ashes for HSSH. The intensity of HSSH, in considering all of the difficulties, was roughly estimated as about 5−10 times higher than H2S by GC-O (Figure 6). This value indicates that HSSH has a stronger impact than H2S, but in our conditions it was not possible to determine an odor threshold. Disulfane from Struck Conglomerate Pebbles and Flint. Flint is a hard sedimentary cryptocrystalline form of the mineral quartz. We obtained two pieces from Dordogne, France, close to the “Grotte de Lascaux.” When two rocks are struck against themselves, sparks can be produced along with a typical smell. Twenty subjects were asked to describe the odor immediately after striking. The descriptors are listed in order of frequency of use by the panelists: flint, struck matches, cold fireworks, and dentist drill odor. Dark pebbles from a conglomerate, a common and widely distributed type of rock formed by metamorphic processes, were obtained from Mont Pelerin (Canton de Vaud, Switzerland). The odor was close to that obtained from flint. No blind sensory panels were organized to assess whether the differences observed by direct comparison are significant. The odor

solution of FeSO4 in water at pH 3.5 containing 1 molar equiv of H2O2. The addition of this solution to H2S instantaneously produces a white solid, likely corresponding to the formation of elemental sulfur. Using a catalytic amount of H2O2 (1% mol) promoted the formation of sulfanes, but over time they disappeared. The other problem was that ethanol was slightly oxidized in acetaldehyde with the Fenton solution. Acetaldehyde reacts with H2S and gives a range of sulfur compounds having a garlic character. Therefore, the preparation of H2Sn in ethanol as an authentic sample was abandoned. The GC-MS obtained after the addition of iron powder (Fe°) to ethanol saturated with H2S, followed by extraction and derivatization, is disclosed in Figures S8−S10. The addition of methylmercaptan and ethylmercaptan to H2S and iron gave a very complex mixture having a garlic odor profile. From this complex mixture, we obtained references for CH3SnH and C2H5SnH (n = 1, 2, 3). These compounds were characterized by MS, HRMS, GC-O, and LRI measurements. CH3SSH (LRI 634) was described as matches, burnt eggs, and sewage. CH3CH2SSH (LRI 726) eluted just after dimethyl disulfide (LRI 715) and was described as garlic, rubber, matches, and green (Figures S11−S18). Derivatization with Pentafluorobenzyl Bromide (PFBBr). Derivatization with NEM is not appropriate for H2S or for H2Sn.5,29 For this reason, we changed the derivatization agent and used PFBBr.32 The HSSH freshly distilled in toluene was derivatized with PFBBr and DBU (Figure 4). This method is preferred over that with NEM, mainly for chromatographic reasons (Figure 4A) as the peaks are better defined. A typical mass spectrum fragment of PFBBr in MS is m/z 181, which should allow easy tracking in single ion monitoring mode (Figure 4B), but the problem is that PFBBr is a hard electrophile and can react with electrophiles. In the case of wine, it reacts with organic acids, and the ester between octanoic acid and PFBBr has a close LRI (1580 ± 5) on an apolar column compared to PFB-S-PFB (LRI 1617 ± 5). For this reason the molecular ion was recorded for the quantifications and not m/z 181 (Figure 4B). Different options have been considered for preparing H2S and H2Sn derivatives in wine. The first option was to raise the wine pH to 8 with NaOH and then add PFBBr/DBU, but changing the pH could harm H2S and HSSH by oxidation or loss due to other reactions. Therefore, the wine was extracted with a mixture of pentane and diethyl ether, and the derivatization was performed in the organic extract. On the polar GC column the ratios of PFB-S-PFB (LRI 2082 ± 5) and PFB-S-S-PFB (LRI 2345 ± 5) were different and PFB-S-S-S-PFB decomposed. An authentic sample was also prepared by mixing H2S, CH3SH, and C2H5SH in ethanol in the presence of iron (Fe°) and H2O2 to confirm the occurrence of these compounds in flint and pebble (Figure 5). Odor Impact Estimation of HSSH over H2S by GC-O. HSSH could not be obtained pure; it is unstable and always contains traces of H2S and H2S3 When the contaminated gas was blown in the toilet model system, the flint smell was 9037

DOI: 10.1021/acs.jafc.6b03938 J. Agric. Food Chem. 2016, 64, 9033−9040

Article

Journal of Agricultural and Food Chemistry

Extraction and derivatization were performed in the organic phase, and then the ratio of m/z 394/74 in SIM mode was recorded to establish the calibration curve. The wine is a difficult matrix; thiol can be oxidized, complexed, or reacted with electrophiles.33,34 Ferreira noted that in some wine 97% of H2S is complexed.33 The calibration curve was directly determined in white wine, and the semiquantitative values obtained are not considered if it is free or bound H2S. The same calibration curve was established for HSSH; the ratio m/z 426/74 was plotted, and we obtained a perfect linearity. We experienced many difficulties; for instance, we used propanethiol as standard 29 but, unfortunately, HSSH reacted immediately to give H2S and C3H7S2H. The goal of this paper is not to develop a precise quantitative method but to demonstrate that the olfactives differences observed between mineral and nonmineral wines can be seen at the level of H2S and HSSH. Due to the importance reported of malic acid for the mineral character, the concentrations of malic acid−lactic acid were measured. The most mineral wine had indeed a higher proportion of malic acid compared to the nonmineral wines. The nonmineral wines such as VDa60 had 0.06 g/L malic acid and 3.3 g/L lactic acid (pH 3.9), and GEa14 had 0.25−2 g/L (pH 3.6) for the least mineral wines. The values found were 2 g/L malic acid and 0.2 lactic acid g/L (pH 3.7) for VDm54, the most mineral wine, and 1.38−0.75 g/L (pH 3.3) for VSm72. This analysis confirmed that HSSH is more abundant in wines defined as mineral and that H2S was present in all wines (Figure 8). Is it possible to promote the formation of HSSH by adding iron (Fe°)? When the white wine Chasselas de Romandie was treated with iron, both H2S and HSSH concentrations increased by a factor of 10. This was difficult to explain if it was due to oxido-reduction reaction or any other interaction with sulfur complexes. To gain some understanding, Na2SO3 was added to a wine model system containing L-malic acid, DLtartaric acid, water, and EtOH, adjusted to pH 3.3. When iron was added, H2S was first generated, but after 6−8 h, HSSH was

Figure 6. GC-O sensory evaluation (mean ±95% CI) of pure HSSH and comparative odor intensity rated on a linear scale from 0 to 5 between H2S and HSSH.

descriptors are listed in order of frequency of use by the panelists as follows: gun powder, skunks, burnt hairs, mushroom shiitake, struck matches, cold fireworks. The struck pebble odor was more culinary and more complex compared with flint, for which fewer descriptors were used. The smell disappeared in