Abstract
Studies in wine and model systems have established that iron is an essential catalyst that mediates the reaction of polyphenols with oxygen. This investigation examined how this metal exerts its action. When wine is protected from air, iron exists in its reduced ferrous state, Fe(II), which is rapidly oxidized to the ferric state, Fe(III), on exposure to oxygen. This rapid transformation is observed when Fe(II) is added to model wine saturated with aerial oxygen, but the reaction slows to a very slow rate before completion due to the inhibitory action of Fe(III). 4-Methylcatechol was found not to be oxidized by Fe(III) or as Fe(II) was reacting with oxygen. It was apparent, therefore, that the catechol did not react with intermediate oxygen radicals. Consequently, it is proposed that hydroperoxyl radicals are not produced in wine conditions and a revised mechanism for the reaction of Fe(II) with oxygen to produce hydrogen peroxide is proposed. However, sulfite addition, which is known to promote catechol oxidation, resulted in rapid Fe(III) reduction and attainment of an Fe(III)/Fe(II) redox equilibrium. Benzenesulfinic acid, which does not react with oxygen, produced the same effect and it is proposed that nucleophiles, which react rapidly with quinones, allow the oxidation of catechols to proceed. Examination of the Fenton reaction showed that the reaction of Fe(II) with hydrogen peroxide was rapid and resulted in the uptake of oxygen. A comparison of the rates of Fe(II) oxidation and Fe(III) reduction in the presence of different ligands showed a dependence on reduction potentials.
Wine can take up a considerable amount of oxygen, red more than white since polyphenols with vicinal phenolic functions, which are most readily oxidized, are primary reactants. Limited controlled oxidation is beneficial for red wine, but the quality of white wine is generally not improved by this process (Boulton et al. 1996). An important exception is exposure to oxygen to remove reductive odors. However, oxygen cannot react directly with polyphenols to oxidize them because of its electronic configuration. It can only accept electrons singly, which it is able to do by interacting with transition metals or free radicals (Danilewicz 2003). Consequently, although present at low concentration, iron and copper are essential catalysts in wine oxidation (Danilewicz 2007). In wine which has been long protected from air, iron is in the ferrous state (Fe(II)). Introduction of oxygen initiates the oxidative process by first rapidly oxidizing Fe(II) to the ferric state (Fe(III)), a process which is accelerated by copper (Danilewicz 2011, Danilewicz and Wallbridge 2010). It is envisaged that O2 is reduced in one-electron steps to produce hydroperoxyl radicals (Scheme 1, reaction A) and a further Fe(II) reduces this radical to hydrogen peroxide (reaction B). It is then proposed that the Fe(III) which is produced coordinates with catechols and oxidizes them first to semiquinones and then to quinones (reactions C and D). Therefore, in this first phase of the redox process, two Fe redox-cycle to transfer two electrons from a catechol to O2. More recent work has shown that the oxidation of Fe(II) occurs most rapidly and reaches an apparent equilibrium but the oxidation of catechols, such as (+)-catechin, by Fe(III) does not occur spontaneously at a significant rate (Danilewicz 2011). It requires the presence of substances that share the common property of being capable of reacting rapidly with quinones. These substances, which include sulfite, are proposed to draw forward an otherwise unfavorable equilibrium and may be termed “oxidation promoting nucleophiles” (Danilewicz 2011).
It has also been proposed that catechols are the primary reacting species with hydroperoxyl radicals (Scheme 1, reaction E) and that the resulting semiquinone may disproportionate (reaction F) (Waterhouse and Laurie 2006). However, initial findings showed that the reduction of oxygen by Fe(II) occurs independently and is not altered by the presence of (+)-catechin, which indicated that catechols do not participate directly in O2 reduction (Danilewicz 2011).
Fe(II) rapidly reduces hydrogen peroxide to produce what is thought to be the hydroxyl radical, which oxidizes ethanol to acetaldehyde by way of the Fenton reaction (Scheme 2). The hydroxyl radical is a powerful oxidant, which attacks the first reductant it encounters. Ethanol is, therefore, preferentially oxidized because of its higher concentration, despite the presence of Fe(II) or polyphenols, which are more powerful reductants. It is proposed that the hydroxyl radical first abstracts a hydrogen atom from ethanol to produce a hydroxyethyl radical, which reacts with oxygen, so providing a second mechanism for the reduction of oxygen (Bielski and Arudi 1983, Danilewicz 2003, Waterhouse and Laurie 2006). One Fe(II) was found to produce one acetaldehyde (Elias and Waterhouse 2010), and a possible explanation is that the intermediate hydroxyethylperoxyl radical eliminates a hydroperoxyl radical, which disproportionates to regenerate half a mole of oxygen and half a mole of hydrogen peroxide. Consequently, for every mole of hydrogen peroxide that is reduced, one mole of oxygen is taken up and two moles of Fe(II) and two moles of ethanol are oxidized. The net result is that overall (Schemes 1 and 2), when oxygen is reduced to water by accepting four electrons, two Fe(II) would be oxidized for each oxygen that reacts, ethanol providing two further electrons to complete the reduction. Concurrently, one catechol would be oxidized as two Fe(III) are reduced back to Fe(II), so completing the recycling process.
One function of sulfite is to promote the oxidation of polyphenols and so accelerate the removal of oxygen from wine. This action results from the rapid reduction of the quinones back to the original catechols (Scheme 1, reaction G), preventing the formation of further reaction products, some of which are colored (Danilewicz and Wallbridge 2010). Sulfite also removes the hydrogen peroxide, preventing the oxidation of ethanol to acetaldehyde (Wildenradt and Singleton 1974, Waterhouse and Laurie 2006). In effect, sulfite counteracts its own pro-oxidant activity and overall prevents the damaging action of oxygen.
Iron plays a key role in the whole oxidative process and the aim of this paper is to examine its mechanism of action more closely by examining (1) the reduction of oxygen by Fe(II) and oxidation of catechols by Fe(III), (2) the role of tartaric acid in determining the reactivity of Fe(III)/Fe(II), (3) the uptake of O2 in the Fenton reaction, and (4) the role of sulfite in catechol oxidation.
Materials and Methods
Materials.
Water (Emsure; Merck, Darmstadt, Germany), Fe ≤1 μg/L, Cu ≤0.4 μg/L), Cu(II) sulfate pentahydrate, Fe(III) chloride hexahydrate, sodium hydroxide, (+)-tartaric acid (BDH AnalaR grade), and ethanol (96% GRP grade) were obtained from VWR International (Lutherworth, UK). Benzenesulfinic acid (BSA) sodium salt, 4-methylcatechol (4-MeC), 2,2′-bipyridyl (Bipy), diethylenetriaminepentacetic acid (DTPA), ferrozine, and Fe(II) sulfate heptahydrate (99+% ACS reagent) were obtained from Sigma-Aldrich (Poole, Dorset, UK). Potassium metabisulfite (Kadifit) was from Erbslöh Geisenheim AG (Geisenheim, Germany). UV-vis spectra were taken with a Jenway 7315 spectrometer (Bibby Scientific, Stone, UK).
Preparation of model wine solutions.
(+)-Tartaric acid (16.0 g) was dissolved in water (~1.6 L) in a 2 L volumetric flask. Ethanol was added to give a 12% (v/v) final concentration. The pH was increased to 3.60 with 2.5 N sodium hydroxide, adding water progressively to the mark as the required pH was approached. The concentration of tartaric acid was selected so that it would simulate the action of all the wine acids and produce a final titratable acidity of ~4 g/L.
Reaction of Fe(II) with oxygen in model wine.
Procedure 1.
A two-necked round-bottomed flask containing a magnetic stir bar was almost filled with model wine (217 mL), which had been saturated with aerial oxygen. Reactants such as 4-MeC, Bipy, and DTPA were added as solids. Fe(III) was added as a solution of FeCl3 in 250 μL H2O. The side neck was closed with a septum and a Clark electrode was inserted in the other neck and sealed with a rubber O-ring. The volume of liquid introduced was just sufficient to reach the O-ring seal and expel any air as the electrode was inserted, with the tip positioned 3 to 5 mm from the stir bar at the bottom. The flask was placed in an insulated water bath, which was stirred in the dark with the same magnetic stirrer. Stirrer speed was increased to ensure that adequate solution passed over the electrode membrane and maximum O2 readings were obtained. When the system stabilized, Fe(II) and Cu(II) were introduced as solutions of their sulfates in a small volume of water (20 to 250 μL) via the septum with a syringe. When hydrogen peroxide was added, it was dissolved in water (50 or 130 μL), the solution having been standardized by adding a known volume to a standard sulfite solution in water and determining the amount of SO2 that had reacted by the modified Ripper procedure. The O-ring rubber seal prevented the entry of any air but allowed small volumes of the solution to escape as reactants were introduced. Oxygen concentration was followed over time and Fe(III) and Fe(II) concentrations were measured at the end of the experiment. Experiments were performed in duplicate. However, because of the difficulty in achieving exactly the same conditions each time, the results were not meaned but those of only one experiment are shown. The duplicate gave essentially the same results.
Procedure 2.
Model wine (1 L) was shaken in air until saturated with O2 and O2 concentration was measured with a Clark electrode. Solid Fe(II) sulfate heptahydrate (49.8 mg) was added to give 10.0 mg/L Fe(II) and when dissolved this was followed by CuSO4 solution (100 μL) to give 0.6 mg/L Cu(II). The solution was poured into 14 ~58 mL brown bottles, which were sealed immediately with the exclusion of any air bubbles and stored in the dark. Metal addition and filling of the bottles took ~11 min. O2, Fe(II), and Fe(III) concentrations were followed taking three bottles at each time point.
Procedure 3.
A 200 mL volumetric flask was filled with model wine. Fe(II)SO4.7H2O (250 μL of a solution containing 398.2 mg dissolved in 10 mL H2O) was added to give 10 mg/L (1.79 × 10−4 mol/L) Fe(II), followed by Cu(II)SO4.5H2O (100 μL of a solution containing 47.2 mg dissolved in 10 mL H2O) to give 0.6 mg/L Cu(II). The solution (66 mL) was poured into each of three 250 mL flasks, which were closed with plastic film, stored in the dark, and shaken periodically to maintain O2 saturation. Fe(II) and Fe(III) concentration was followed in triplicate over time. When the reaction was repeated in the presence of 4-MeC, 17.8 mg were added first before the metals to give 88.9 mg/L (7.16 × 10−4 mol/L) and only Fe(III) concentration followed over time, as ferrozine could not be used to measure Fe(II) concentration in the presence of the catechol.
Reaction of Fe(III) with 4-MeC in model wine.
Procedure 4.
When 4-MeC (88.9 mg/L) was used it was dissolved initially in model wine in a 200 mL volumetric flask. FeCl3 dissolved in 200 μL H2O and Cu(II)SO4.5H2O in 92 μL H2O were added to give Fe(III) (10 mg/L) and Cu(II) (0.6 mg/L). The solution was poured into three 250 mL flasks as described in procedure 3. SO2 was added as K2S2O5 in 200 μL H2O at the times indicated and free SO2 and Fe(III) concentrations were followed over time in triplicate.
Procedure 5.
Three 20 mL volumetric flasks were almost filled with model wine. BSA sodium salt (117.5 mg/L, 7.16 × 10−4 mol/L) was added followed by 4-MeC (88.9 mg/L, 7.16 × 10−4 mol/L) when required, both dissolved in 100 μL H2O. This was followed by Fe(III) (10 mg/L) as a solution of FeCl3.6H2O in water (25 μL) and Cu (0.6 mg/L) as a solution of Cu(II)SO4.5H2O in water (9.5 μL). Solutions were made up to the mark. Fe(III) concentration was measured immediately and the solutions poured into 58 mL brown bottles closed with a screwcap, so leaving an air space above the model wine. The bottles were stored in the dark and shaken periodically to maintain oxygen saturation as Fe(III) concentration was determined over time in triplicate.
Fe(II) and Fe(III) measurement.
Fe(III) concentration was measured spectroscopically at 330–334 nm and Fe(II) by the ferrozine method (Stookey 1970). Ferrozine solution (200 μL; 79.5 mg in 5 mL H2O) was added to 10 mL of the reacting solution and absorbance measured at 562 nm.
O2 and SO2 measurements.
Solutions were sealed in ~58 mL brown glass bottles with screwcaps fitted with a plastic cone liner, ensuring the exclusion of any air bubbles and then stored in the dark. The small air space between the liner and the cap was filled with a polymeric filler to ensure that no O2 passed through the plastic cone into solutions during the course of experiments. Three bottles were taken at each time point so as to measure O2 concentrations in triplicate.
An HI-9146 dissolved O2 meter fitted with a HI-76408 Clarke-type electrode was used (Hanna Instruments, Leighton Buzzard, UK). The manufacturer specifies a resolution and limit of detection (LOD) of the meter as 0.01 mg/L O2. For measurements the bottle caps were quickly removed and a small stir bar inserted followed by the electrode, which displaced ~13 mL of liquid. The electrode tip was lowered to ~5 mm of the briskly stirred magnetic bar. Readings stabilized within 30 sec and then remained stable for more than 5 min, showing that, although the system was not sealed during measurement, no measurable amount of external O2 reached the measurement area during that time.
Free SO2 concentration was measured with the modified Ripper procedure, using potassium iodate and starch-KI (Ough and Amerine 1988).
Data analysis.
When measurements were taken in triplicate, mean values (±SD) were calculated and figures drawn using Excel software (Microsoft, Redmond, WA). Where error bars denoting ±SD are not shown, they were smaller than the data point symbol dimensions. Experiments were conducted at ambient temperature (18.5 to 20.5°C).
Results and Discussion
Reactions of Fe(II) and Fe(III).
Studies were initiated by measuring O2 consumption when metals were added to air-saturated model wine, sealed to prevent the ingress of any further air. The introduction of Fe(II) (10 mg/L) alone produced a relatively slow oxygen uptake, which was much accelerated by the addition of Cu(II) (0.6 mg/L) (Figure 1, curve a). The ability of small additions of Cu to synergize with Fe to accelerate the oxidation of model wine and wine appears, therefore, to be due to facilitation of Fe redox cycling by acceleration of Fe(II) oxidation (Danilewicz and Wallbridge 1010). Studies were consequently conducted at a fixed 19:1 Fe:Cu molar concentration ratio.
Addition of the two metals together resulted in a rapid initial oxygen consumption, which slowed and appeared almost to stop when 1.66 mg/L had been taken up (Figure 1, curve b). Fe(III) and Fe(II) concentrations were then 6.64 and 3.19 mg/L, respectively, giving a Fe(III):Fe(II) concentration ratio of 2.08:1. It might be anticipated that in the presence of 4-MeC, some of the Fe(III) produced might be reduced and a lower Fe(III):Fe(II) concentration ratio might be observed. However, that was not the case. Oxygen uptake was 1.67 mg/L and Fe(III) and Fe(II) concentrations were then 7.28 and 2.72 mg/L, respectively, giving a higher Fe(III):Fe(II) concentration ratio of 2.6:1 (Figure 1, curve c). The apparent lack of interaction between Fe(III) and the catechol was unexpected, as it has been shown that 4-MeC is oxidized extremely rapidly by Fe(III) in model wine (Elias and Waterhouse 2010). However, these authors included ferrozine to monitor Fe(II) production and, as will be discussed below, ferrozine raises the reduction potential of the Fe(III)/Fe(II) couple substantially, from well below to well above that of the quinone/catechol couple, making Fe(III) a much stronger oxidant. It was confirmed that the addition of ferrozine to an otherwise stable mixture of Fe(III) and 4-MeC resulted in the rapid production of Fe(II) and showed that ferrozine cannot be used in the presence of a catechol. Fe(III) concentration was measured directly using the broad absorption band at 330 to 334 nm (ɛmax = 1,920 L/mol.cm). It was well separated from that for 4-MeC at 282 nm (ɛmax = 2,483 L/mol.cm) (Figure 2). The absorption at ~230 nm was at the cut-off point due to absorption by the model wine solution. When Fe(III) and 4-MeC were combined, the resulting UV spectrum did not change significantly over several days, confirming that they do not interact or at least do so extremely slowly in model wine. The absorbance of Fe(II) is not significant in the above spectral range.
In previous work with model wine, it was shown that there was a rapid initial oxygen uptake due to the oxidation of Fe(II), which then stopped or at least became extremely slow even in the presence of (+)-catechin (Danilewicz 2011). Assuming a 2:1 Fe(II):O2 reaction ratio, it was consequently deduced from the oxygen uptake that only 60 to 70% of the Fe(II) was oxidized, which had suggested the attainment of an Fe(III)/Fe(II) equilibrium that was not displaced by a catechol. When monitoring Fe(III):Fe(II) concentration as described above, Fe(II) oxidation almost stopped when a ~2:1 ratio was indeed reached (Figure 1, curve b), which might suggest the approach of an equilibrium such as previously assumed (Scheme 3). It has been generally accepted that the oxidation of catechols involves the initial stepwise reduction of oxygen by Fe(II) to produce hydroperoxyl radicals followed by hydrogen peroxide (Scheme 1) (Danilewicz 2003, Waterhouse and Laurie 2006, du Toit et al. 2006). However, the involvement of such a reversible process requires reconsideration. As will be discussed below, hydrogen peroxide is reduced extremely rapidly by Fe(II) and is anticipated to be too unstable in model wine to coexist with Fe(II) at equilibrium concentrations (Scheme 3). The reduction of O2 by Fe(II) would appear to be irreversible and an alternative explanation is required to explain the slowing of the reaction before completion.
However, as described above, when 10 mg/L Fe(II) (1.79 × 10−4 mol/L) was oxidized in model wine, for the 1.66 mg/L (5.19 × 10−5 mol/L) of oxygen that was taken up, 6.64 mg/L (1.19 × 10−4 mol/L) of Fe(III) was produced, giving a 2.3:1 Fe(II):O2 reaction ratio (Figure 1, curve b). In the presence of 4-MeC, the ratio increased further to 2.5:1 (Figure 1, curve c). There appears, therefore, to be an additional pathway for Fe(II) oxidation. When model wine containing the same amount of Fe(II) and Cu(II) as above was sealed saturated with air and oxygen uptake followed over a longer time, it essentially ceased after 10 hr, but oxidation of a small amount of Fe(II) slowly continued (Figure 3). At 23.5 hr the uptake of O2 was 2.0 ± 0.1 mg/L and was 2.16 mg/L at 60.75 hr. At 13 hr, the Fe(II):O2 reaction ratio was 2.37:1, increasing to 2.42:1 at the end of the experiment. It appeared that Fe(II) was reacting with an oxidizing intermediate, perhaps an iron-oxygen or iron-hydrogen peroxide complex.
The initial fast oxidation of Fe(II) followed by a slow gradual oxidation was also evident when model wine was maintained at aerial oxygen saturation (Figure 4). In the absence of 4-MeC, Fe(II) concentration was measured using the ferrozine method (curve a) and Fe(III) spectroscopically at 330 nm (curve b). This latter method was then used to follow Fe(III) concentration in the presence of 4-MeC (curve c). The catechol had no effect on the rate of formation of Fe(III), and from the UV absorbance at 282 nm its concentration did not appear to change, allowing for the increase in background absorbance due to Fe(III). From seven readings taken over the four days of the experiment, the 4-MeC concentration remained in the range 89.1 ± 0.4 mg/L.
Although Fe(III) oxidizes 4-MeC in the presence of sulfite and other nucleophiles in model wine, the above results show that no or at least very little reaction occurs in the absence of these oxidation promoters, which are proposed to advance the oxidative process by reacting with quinones (Danilewicz et al. 2008, Danilewicz and Wallbridge 2010). Furthermore, catechols are not oxidized by intermediates generated during the reduction of oxygen, such as by hydroperoxyl radicals (Danilewicz 2003, Waterhouse and Laurie 2006) or hydroxyl radicals, which are proposed to react principally with ethanol. Either catechols do not react with hydroperoxyl radicals, or, more likely, they are not produced when oxygen is reduced, first to hydrogen peroxide (Scheme 1) and subsequently in the Fenton reaction (Scheme 2).
Attempts have been made to show that hydroperoxyl radicals are produced during wine oxidation by electron paramagnetic resonance (EPR) spectroscopy using 5-tert-butoxycarbonyl-5-methyl-1-pyrroline N-oxide (BMPO), which is known to form stable spin adducts with hydroperoxyl radicals (Elias et al. 2009). However, when a wine was oxidized in the presence of added Fe(II) and Cu(II), only the spectrum of the BMPO-CH3C•HOH spin adduct was observed, showing that oxygen had been reduced to hydrogen peroxide, which had reacted further to oxidize ethanol. This finding led the authors to suggest that the intermediate formation of the hydroperoxyl radical might not occur in wine conditions.
Superoxide, the deprotonated form of the hydroperoxyl radical (pKa 4.8), is known to react with catechols at pH 7 and their ability to do so depends on their reduction potentials (Kitagawa et al. 1992, Jovanovic et al. 1994). The reduction potential of the oxidant and reductant increase in parallel as pH is reduced (Danilewicz 2012). Consequently, ΔE for the superoxide-hydroperoxyl radical/catechol system should remain constant at ~220 mV in the pH range 0 to 9 and the reaction should, therefore, be thermodynamically just as favorable at pH 3.6 as at pH 7.
When 5.0 mg/L Fe(II) and 0.3 mg/L Cu(II) were added to model wine, again a rapid initial oxygen consumption was observed, which slowed and appeared almost to stop when 0.67 mg/L had been taken up (Figure 5, curve a). Fe(III) and Fe(II) concentrations were 3.03 and 1.99 mg/L, respectively. It appears, therefore, that the rate of Fe(II) oxidation becomes extremely slow at a Fe(III):Fe(II) concentration ratio of ~3:2. Since an equilibrium is unlikely due to the instability of hydrogen peroxide, an alternative possibility is that Fe(III) may be competing with Fe(II) for an intermediate, increasingly preventing Fe(II) oxidation as Fe(III) concentration rises. This action was confirmed when Fe(II) (5.0 mg/L) and Cu (0.3 mg/L) were added to model wine containing Fe(III) (11.65 mg/L), giving a starting Fe(III):Fe(II) molar ratio of 70:30. The rapid initial oxygen uptake was no longer evident, only a much slower steady decline in oxygen concentration (Figure 5, curve b).
A proposed mechanism of Fe(II) autoxidation should, therefore, allow for the slow irreversible net production of hydrogen peroxide, which would be determined by the Fe(III):Fe(II) concentration ratio, without the intermediacy of hydroperoxyl radicals. Such a mechanism has been proposed for the autoxidation of the Fe(II)-ethylenediaminetetraacetic acid (EDTA) complex (Scheme 4) (Siebig and van Eldik 1997, Yurkova et al. 1999).
It is proposed that electron transfer within the initially formed Fe(II)-O2 adduct produces an Fe(III)-superoxo complex (Scheme 4, reaction 1), omitting ligands and metal-bound water for simplicity. Reduction by Fe(II) then yields the diiron(III),(III)-dioxygen complex (reaction 2), which hydrolyzes to produce hydrogen peroxide. Hydrogen peroxide is rapidly and irreversibly reduced by Fe(II) to give hydroxyl radicals in the Fenton reaction (reaction 3). It has also been proposed that this powerful intermediate oxidant is the ferryl ion, (Fe(IV)=O)2+ (Rush and Koppenol 1986, Yamazaki and Piette 1991). However, under acid conditions and low Fe(II) concentration, it is argued that the hydroxyl radical (HO•) is the dominant product (Koppenol and Leibman 1984). Although some uncertainty may still remain, the oxidizing radicals produced in the Fenton reaction react with ethanol at similar rates to “pure” hydroxyl radicals generated by photolysis, where iron is not present and so the exact nature of this oxidant may not be important as far as wine oxidation is concerned. If it is then proposed that the Fe(III)-superoxo complex can also be oxidized by Fe(III) to regenerate oxygen, then that would explain the inhibitory action of Fe(III) in reducing the net reduction of oxygen (reaction 4). The formation of hydroperoxyl radicals, which was proposed to be energetically disfavored (reaction 5), would be even less likely in model wine, since the reduction potential of the Fe(III)/Fe(II) couple is much higher in the presence of tartaric acid than EDTA. The reverse reaction, the oxidation of hydroperoxyl radicals by Fe(III) (reaction 5) to produce Fe(II) and oxygen, is thermodynamically highly favored in wine conditions (ΔE = 440 mV) (Danilewicz 2012).
The role of tartaric acid.
Since tartaric acid binds more strongly to Fe(III) than to Fe(II), it will displace Fe(III)/Fe(II) equilibria to favor Fe(III), making Fe(II) a stronger reductant (Timberlake 1964). Consequently, the reduction potential of the Fe(III)/Fe(II) redox couple has been found to be reduced to 385 mV by cyclic voltammetry in model wine at pH 3.3 (P. Kilmartin, private communication, 2012), which is the same as the value previously obtained in the presence of 3 mol equivalents of tartaric acid (Green and Parkins 1961). The latter authors found that the reduction potential decreased by ~130 mV per pH unit increase in the wine pH range, giving a value of 345 mV at pH 3.6. EDTA reduces this potential substantially to 117 mV at pH 7, which is little changed down to pH 3.5 (Ilan and Czapski 1977). Diethylenetriaminepentacetic acid (DTPA) reduces the potential even further to 30 mV at pH 7 (Galey 1997). These strong Fe(III)-selective ligands, therefore, make Fe(II) a much stronger reductant and addition of Fe(II) (10 mg/L) to model wine containing 3 equivalents of DTPA resulted in a rapid oxygen consumption. The ferrozine assay showed that no Fe(II) remained after 2 hr (Figure 5, curve c) and the Fe(II):O2 molar reaction ratio was 2.3:1. The curve initially obtained without DTPA, when Fe(III) and Fe(II) are complexed to tartaric acid, is shown for comparison (curve d).
Conversely, Fe(II)-selective ligands raise the reduction potential of the Fe couple, making Fe(II) a weaker reductant but Fe(III) a stronger oxidant. The reduction potential of the 1,10-phenanthroline complexes is raised to +1.15 V at pH 7 (Buettner 1993), which will be similar at pH 3.5, as the extent of protonation should not change (E0 = 1.12 V). The ferrozine complexes should have a similar reduction potential, explaining why Fe(III) becomes a much stronger oxidant. Addition of 5 equivalents of 2,2′-bipyridyl (Bipy), which is of that ligand type (E0 = 1.03 V) (Housecroft and Sharpe 2005), weakens the reducing ability of Fe(II) to the point that it no longer reduces oxygen (Figure 5, curve e).
The reduction potential of the O2/H2O2 couple ( E3.6) is 565 mV in model wine conditions (Danilewicz 2012), which places it ~220 mV above that of the Fe(III)/Fe(II) couple. Since the difference in reduction potential of the couples ΔE = −ΔG/nF, the reduction of oxygen by Fe(II) results in a reduction in free energy (ΔG = −42.4 KJ/mol) and is thermodynamically favorable. The reaction is found to proceed rapidly until an opposing reaction by Fe(III) intervenes. DTPA reduces the reduction potential of the Fe(III)/Fe(II) couple, increasing the difference in the above potentials still further (ΔE = 448 mV, ΔG = −86.5 KJ/mol) and the rate of reduction of oxygen by Fe(II) is found to accelerate markedly. Bipy, which raises the reduction potential of the Fe(III)/Fe(II) couple above that of the oxygen system, prevents Fe(II) oxidation, as the change in free energy would become positive (ΔE = −465 mV, ΔG = +89.7 KJ/mol). A further observation is that the reduction potential of the Fe(III)/Fe(II) couple in a NaClO4/H2SO4 system, where Fe(III) and Fe(II) exist as the hexaaquo-complexes, was found to be 687 mV at pH 3.3 by cyclic voltammetry (P. Kilmartin, private communication, 2012), and now being above that of the O2/H2O2 couple (E3.3= 582 mV), Fe(II) no longer reduced oxygen under these conditions when tartaric acid is absent (data not shown).
Though tartaric acid increases the reducing power of Fe(II), it reduces the oxidizing power of Fe(III). The reduction potential of the Fe(III)/Fe(II) couple is reduced below that of the 4-MeC couple in model wine (E3.6 = 577 mV) (Danilewicz 2012), resulting in ΔE = −227 mV and ΔG = +43.8 KJ/mol. As observed, it is no longer oxidized by Fe(III), unless an oxidation promoting nucleophile is present. However, when the reduction potential of the Fe(III)/Fe(II) couple is raised above that of the 4-MeC couple, such as by addition of ferrozine, oxidation proceeds rapidly. The reaction forms the basis of the FRAP (ferric reducing antioxidant power) assay, where Fe(III) oxidizes polyphenols in the presence of the Fe(II) selective ligand, 2,4,6-tripyridyl-s-triazine, to produce a blue Fe(II) complex (Katalinic et al. 2004). It is therefore evident that the reactivity of the Fe, O2 and catechol redox couples is dependent on their respective reduction potentials. Clearly, tartaric acid and presumably malic acid, aside from their effect on wine acidity, have a crucial function to adjust the reduction potential of the Fe(III)/Fe(II) couple so as to determine the rates of oxidation that are observed in wine. Therefore, in working with model wines, attention should be given not only to pH but also to these acids and to Fe and Cu concentrations.
Uptake of oxygen in the Fenton reaction.
As described in the introduction, it is proposed that oxygen is also taken up during the Fenton phase of ethanol oxidation (Scheme 2). To examine this possibility, Fe(II) (5 mg/L, 8.95 × 10−5 mol/L) and Cu(II) were added to model wine containing Fe(III) (11.65 mg/L, 2.09 × 10−4 mol/L) to achieve a slow, steady oxygen uptake (Figure 6, curve a). Addition of hydrogen peroxide (1.79 × 10−4 mol/L) at 266 min resulted in an immediate oxygen uptake of 1.42 mg/L (4.44 × 10−5 mol/L), 90% of which was taken up within 3 min. Ferrozine showed that no Fe(II) then remained and so the reaction was limited by Fe(II) availability, stopping when all the Fe(II) was oxidized, as previously observed (Elias and Waterhouse 2010). Fe(III) concentration was 16.6 mg/L by UV-spectroscopy, confirming that this analytical procedure accounted for all the Fe that was present. The Fe(II):O2 molar reaction ratio was therefore 2:1 as required by the proposed mechanism, as half the oxygen that initially reacts is proposed to be regenerated (Scheme 2).
However, it appears unlikely that hydroperoxyl radicals are produced, since 4-MeC does not react with intermediates generated as oxygen is reduced. Hydroxyethyl radicals are undoubtedly produced when wine is oxidized (Elias and Waterhouse 2010), and it seems likely that it is these carbon-centered radicals that react with oxygen. The precise mechanism by which the resulting hydroxyethylperoxyl radicals decompose to produce acetaldehyde remains to be determined. A possibility is that the hydroxyethyl radical, which is a strong reductant, may reduce hydroxyethylperoxyl radical, which may then eliminate H2O2 directly with formation of acetaldehyde.
When the reaction was repeated in the presence of 4-MeC, the same initial steady slow rate of oxygen reaction was observed, with an immediate uptake when hydrogen peroxide was added. However, the uptake then continued at a slower rate until all the hydrogen peroxide had reacted, since the addition of a further small amount of hydrogen peroxide (6.88 × 10−5 mol/L) at 358 min resulted in a further uptake of oxygen (Figure 6, curve b). Evidently, the catechol reduced Fe(III) to regenerate Fe(II) to allow the reaction to continue (Elias and Waterhouse 2010). However, hydrogen peroxide is required to be present, since 4-MeC alone does not reduce Fe(III) in model wine. Oxygen uptake was 2.47 mg/L (7.72 × 10−5 mol/L) for the addition of 1.79 × 10−4 mol/L of hydrogen peroxide, giving an H2O2: O2 molar reaction ratio of 2.3:1. If the sole function of the catechol was to reduce Fe(III) and H2O2 was not involved in the process, then this ratio would be 1:1 (Scheme 2). It appears that H2O2 is consumed during the regeneration of Fe(II), possibly by the reaction of the catechol with an Fe(III)-H2O2 complex.
When H2O2 (1.82 × 10−4 mol/L) was added to model wine containing Fe(III) and 4-MeC in the absence of Fe(II) (Figure 6, curve c), oxygen (1.52 mg/L, 4.75 × 10−5 mol/L) was taken up at a rate comparable to that observed in the second slower phase previously noted, when all the Fe(II) had reacted (curve b, between 253 and 358 min). Evidently, Fe(II) was being produced and the reaction proceeded till all the hydrogen peroxide reacted, which was confirmed by the addition of a further small amount (6.88 × 10−5 mol/L) at 208 min, which resulted in an uptake of oxygen. The H2O2:O2 molar reaction ratio was 3.8:1.
These studies with hydrogen peroxide confirm that its reaction with Fe(II) is very fast and certainly faster than its rate of formation (Elias and Waterhouse 2010). Therefore, as discussed above, hydrogen peroxide would not attain equilibrium concentrations, reacting as soon as it is formed, since Fe(II) would always be present. The total conversion of Fe to the ferric state would require prolonged oxygen saturation. Also, for the first time, it is demonstrated that oxygen is consumed in the process. The reduction of Fe(III) by 4-MeC in the presence of hydrogen peroxide is an interesting observation but is unlikely to occur in wine since hydrogen peroxide would react more rapidly with Fe(II). A revised reaction sequence is tentatively proposed for the reduction of oxygen, which does not involve the production of hydroperoxyl radicals, as previously assumed (Scheme 5).
Role of sulfite in catechol oxidation.
The ability of sulfite to promote the oxidation of catechols was then investigated. SO2 was added to the air-saturated model wine containing Fe(III) (10 mg/L) and Cu(II) (0.6 mg/L). Solutions were placed in flasks with a large air space, closed, and shaken periodically to maintain air saturation. In the absence of the catechol, there was a slight reduction in Fe(III) concentration (~0.5 mg/L) and a slow progressive reduction in free SO2 concentration, starting at 33.3 ± 0.3 mg/L, which reduced to 29.5 mg/L after 29 hr (Figure 7, curve a), showing that alone in the absence of a catechol SO2 was indeed very slowly oxidized. As already discussed, 4-MeC alone does not reduce Fe(III), but in the presence of SO2, Fe(III) concentration was rapidly halved while free SO2 concentration fell more rapidly, starting at 32.2 ± 1 mg/L, and reduced to 14.5 mg/L after 29.3 hr (Figure 7, curve b). It is concluded that Fe(III) concentration steadied initially at a ~1:1 Fe(III):Fe(II) concentration ratio, when the rate of reduction of Fe(III) by the catechol equaled the rate of Fe(II) oxidation. The function of sulfite is to react with the quinone and the hydrogen peroxide that are produced as the metal redox cycles and the catechol is oxidized. It has been shown that sulfite reacts very rapidly with quinones and in the case of 4-MeC, about half the quinone is reduced back to the catechol and most of the remainder reacts to form the 5-sulfonic acid adduct (Danilewicz et al. 2008). In the case of (+)-catechin, all the quinone is reduced back to the catechol (Danilewicz and Wallbridge 2010). The reaction of SO2 with hydrogen peroxide is also rapid, as is evident from the use of this reaction in the aeration-oxidation method for the determination of SO2 concentration in wine.
It has been proposed that this ability of sulfite to accelerate polyphenol oxidation is due to its high nucleophilic reactivity with products of catechol oxidation, thus displacing forward an otherwise very slow reaction (Danilewicz 2011). An alternative explanation is that these are coupled reactions, where the initial unfavorable step (catechol oxidation, ΔG > 0) is coupled to a subsequent more favorable transformation (quinone reaction, ΔG << 0), which results in the net reduction in free energy, allowing the catechol to be oxidized with the regeneration of Fe(II). The chelating action of tartaric acid then allows Fe(II) to react rapidly with oxygen so allowing the metal to redox-cycle. However, it might be argued that as a reductant the activity of sulfite might be due to some other reaction, possibly involving oxygen directly. This possibility seems unlikely as two other nucleophiles, benzenesulfinic acid (BSA) and azide, which are known to react with quinones but not with oxygen, also accelerate the oxidation of catechols (Danilewicz 2011).
In previous work it was found that when 4-MeC was oxidized in aerial oxygen-saturated model wine containing an equivalent amount of BSA in the presence of Fe(II) (5.0 mg/L) and Cu(II) (0.15 mg/L) and the reaction followed by HPLC, there was a steady fall of 4-MeC concentration. The sulfone-quinone adduct was formed in high yield, until after a few days it began to crystalize out. The sulfone was also prepared in minutes by simply adding Fe(III) to a mixture of 4-MeC and BSA in model wine (Danilewicz et al. 2008). Evidently, 4-MeC is rapidly oxidized in the presence of BSA and the reaction of BSA with the resulting quinone is also very fast. To check that BSA does not react with Fe(III) directly, BSA was added to model wine containing Fe(III) (10.0 mg/L) and Cu(II) (0.6 mg/L) and the solution poured into brown bottles, which were sealed leaving a large air space, so that when shaken periodically air saturation was maintained (procedure 5). No change in Fe(III) concentration was observed. However, when 4-MeC was present, Fe(III) concentration fell rapidly and steadied at a ~6:4 Fe(III)/Fe(II) molar ratio in a similar manner to the effect of SO2 shown in Figure 7. As for SO2, Fe redox-cycled as Fe(III) was reduced by the catechol and Fe(II) reacted with oxygen. The reaction progressed as the quinone reacted by forming the BSA adduct, as previously shown by HPLC (Danilewicz et al. 2008). Odorant thiols could also be lost in this manner in wine (Danilewicz et al. 2008, Nikolantonaki and Waterhouse 2012).
Conclusions
A close examination of Fe(II) oxidation in model wine has now confirmed that oxygen consumption slows markedly and almost ceases before all the Fe(II) is oxidized. This apparent Fe(III)/Fe(II) equilibrium is shown to be due to the inhibitory action of Fe(III), which increasingly prevents Fe(II) oxidation as Fe(III) concentration increases. However, following the reaction over a longer time, although oxygen consumption ceased, a small amount of Fe(II) continued to be oxidized, suggesting the presence of an as yet unidentified oxidizing intermediate, which, however, does not react with catechols.
Comparison of the reduction potential of the Fe(III)/Fe(II) couple, which was altered using different ligands, with those of the quinone/4-MeC and O2/H2O2 couples showed their relative magnitudes were predictive and the reactivity of Fe(II) with oxygen and Fe(III) with the catechol. Tartaric acid coordinates strongly with Fe(III) and determines the reduction potential of the Fe(III)/Fe(II) couple, which is further affected by pH, as this determines the extent of ligand ionization. In effect, tartaric acid and presumably malic acid control the rates of redox processes in wine.
The model catechol 4-MeC did not affect the rate of Fe(II) oxidation and did not react while Fe(II) was oxidized. Consequently, it appears that the catechol neither reacts with Fe(III) nor reacts with any oxygen-derived intermediates produced as oxygen is reduced by Fe(II). It is therefore concluded that the hydroperoxyl radical is not formed and hydroxyl radicals react principally with ethanol. It was also shown that oxygen is taken up during the very rapid reduction of hydrogen peroxide by Fe(II) in the Fenton reaction, confirming the existence of this second entry point for oxygen in the overall oxidative process.
Although 4-MeC does not appear to react with Fe(III) in model wine, addition of sulfite or BSA, termed “oxidation promoting nucleophiles,” results in its very rapid reduction and establishment of an Fe(III)/Fe(II) redox equilibrium as oxidation of the catechol proceeds and oxygen is reduced by Fe(II).
- Received November 2012.
- Revision received April 2013.
- Accepted May 2013.
- Published online August 2013
- ©2013 by the American Society for Enology and Viticulture