Abstract
Wine oxidation is catalyzed by iron (Fe). Fe(II) is oxidized by O2 to produce Fe(III), which oxidizes reductants, such as polyphenols, thereby returning to the ferrous state. Therefore, Fe-redox cycles and the [Fe(III)]:[Fe(II)] ratio depend on the relative rates of Fe(II) oxidation and Fe(III) reduction. Under reducing conditions, Fe(II) dominates, but with increasing O2 exposure, the proportion of Fe as Fe(III) increases. Reduction potentials have been used for many years to determine the redox state of wines. However, it has now been realized that these potentials are mainly due to oxidation of ethanol on platinum electrodes and not due to the oxidative processes that normally occur in wine. It is proposed that [Fe(III)]:[Fe(II)] ratios may provide an alternative way of estimating the redox state. These ratios were obtained with ferrozine to determine Fe(II) and total Fe concentrations spectroscopically in white wines. However, ferrozine cannot be used in red wine because of color interference. A similar method using 2-(5-Bromo-2-pyridylazo)-5-[N-propyl-N-(3-sulfopropyl)amino] phenol disodium salt dihydrate (Br-PAPS), which is water soluble and the Fe(II) complex of which absorbs outside the red wine absorption range, was therefore developed with minimal wine disturbance so as not to alter ratios. High Fe(II) content (~97%) was observed in bottled wine with screwcap, technical, and plastic closures. Lower proportions of Fe(II) were found with natural corks and wine boxes. The [Fe(III)]:[Fe(II)] ratio increased on O2 exposure. Calculated reduction potentials of the Fe couple did not correspond to those that would be measured for wines, providing further evidence that these wine potentials are not true reduction potentials.
Modern wines generally contain between 1 and 6 mg/L total iron (Fe) (Riganakos and Veltsistas 2003, Ferreira et al. 2007, López-López et al. 2015). The use of stainless steel equipment and avoiding contact of musts and wine with mild steel during processing has no doubt tended to reduce Fe concentrations, but concentrations as high as 9 mg/L are still found (Riganakos and Veltsistas 2003). The most important action of this metal is to catalyze the reaction of O2 with reducing constituents of wine, principally polyphenols (du Toit et al. 2006, Waterhouse and Laurie 2006). Addition of Fe to wine accelerates wine oxidation, whereas Fe removal slows it (Danilewicz and Wallbridge 2010). The electronic configuration of O2 prevents its direct reaction with substances containing paired electrons, such as polyphenols and sulfite (Danilewicz 2003). O2 acquires electrons singly and is initially reduced by Fe(II) to produce hydrogen peroxide (Scheme 1), a reaction that is markedly accelerated by small amounts of copper (Cu) (Ribéreau-Gayon 1931, Danilewicz 2013). However, Fe(III) inhibits Fe(II) oxidation, which slows markedly as Fe(III) concentration increases (Scheme 2) (Danilewicz 2013). Hydrogen peroxide is further reduced by Fe(II) to produce hydroxyl radicals. These powerfully oxidizing radicals will rapidly oxidize ethanol unless the hydrogen peroxide is intercepted by sulfite (Elias and Waterhouse 2010). The Fe(III) produced on Fe(II) oxidation then in turn oxidizes polyphenols with the assistance of nucleophiles such as sulfite, thereby regenerating Fe(II). Sulfite assists this reaction by removing quinones by either forming adducts or reducing them back to the original polyphenol (Danilewicz 2011, Nikolantonaki and Waterhouse 2012). Wine oxidation is therefore mediated by the redox cycling of the Fe(III)/Fe(II) couple, a process first proposed some 86 years ago (Ribéreau-Gayon 1931).
So-called reduction potentials have been measured to determine the redox state of wines; however, now it has been realized, as further discussed below, that these potentials, although dependent on O2 concentration, are mainly generated by ethanol oxidation on platinum electrodes (Kilmartin and Zou 2001, Danilewicz 2012), and not by redox processes involved in wine oxidation (Scheme 1). Consequently, an alternative approach could be useful to determine the redox state. The relative concentration of Fe(II) to Fe(III) may provide such an approach. The rate of Fe(II) oxidation depends principally on O2, Fe, and Cu concentrations, and the rate of Fe(III) reduction depends on the concentration and reactivity of polyphenols as well as on sulfite concentration (Danilewicz 2007, Danilewicz et al. 2008). When allowed to equilibrate, the [Fe(III)]:[Fe(II)] ratio will depend on the extent to which a wine is exposed to O2, and on the relative rate of its removal by polyphenols (Danilewicz 2016). A low [Fe(III)]:[Fe(II)] ratio would indicate that polyphenols in a wine are capable of removing the O2 that is entering the system so as to maintain a reductive state. An increasingly raised ratio would indicate that a wine is oxidizing at faster rates.
The [Fe(III)]:[Fe(II)] ratio in red wine has been investigated spectrophotometrically using Fe(II)-selective ligands (Ferreira et al. 2007, López-López et al. 2015). However, in these studies, wines were exposed to O2 and their compositions changed substantially, which as discussed below, will change the relative concentrations of Fe(III) and Fe(II). Importantly, addition of Fe(II)-selective ligands facilitates Fe(III) reduction, which can result in a rapid increase in Fe(II) concentration. This effect has been overcome in white wine with use of ferrozine by measuring Fe(II) concentration over 10 minutes after ligand addition and extrapolating back to zero time, that is, to the moment of ferrozine addition (Danilewicz 2016). Unfortunately, ferrozine cannot be used for red wine due to wine color interference, as the absorption maximum of the Fe(II)-complex is at 562 nm. A modified simple procedure is therefore described here, which allows the determination of the [Fe(III)]:[Fe(II)] ratios in red wines with minimal sample disturbance using 2-(5-Bromo-2-pyridylazo)-5-[N-propyl-N-(3-sulfopropyl)amino] phenol disodium salt dihydrate (Br-PAPS) (Figure 1). This ligand is both water soluble and forms an Fe(II) complex with an absorption maximum (λmax 740 nm) that is outside the absorption range of red wines (Homsher and Zak 1985).
Materials and Methods
Materials
Water (Emsure, Fe ≤ 1μg/L, Cu ≤ 0.4μg/L, E. Merck), sodium hydroxide, l-(+)-tartaric acid (BDH AnalaR grade), and ethanol (96% GPR grade) were obtained from VWR International. Ascorbic acid, Fe(II) sulfate heptahydrate (99+%, ACS reagent), and Br-PAPS were from Sigma-Aldrich (Poole). UV-vis spectra were taken with a Jenway 7315 spectrometer.
Wines
Merlot (12.5% alcohol, pH 3.31, Groupo de Bodegas, Vinartis, Spain) and Shiraz (14.5% alcohol, pH 3.17, WO Western Cape, South Africa) were obtained from Sainsbury’s Supermarket Ltd., London EC1N 2HT. Shiraz (13.5% alcohol, pH 3.15, Vacaranda Hill, SE Australia) and Tempranillo (2011, 13.5% alcohol, pH 3.34, Felix Solis, Albali, Valdepeñas, Spain) were obtained from Co-operative Group Ltd., Manchester M60 0AG. The above wines were from bag in boxes and nonvintages, except for the Tempranillo. Bottled wines were the following: Cabernet Sauvignon (2016, 12.5% alcohol, pH 3.2, Vale Central, Chile) imported by Co-operative Group Ltd., Manchester M60 0AG; Côte du Rhône, Village (2015, 14% alcohol, Les Chais Reunis, Saint Priest, France); Château Vieux Manoir (2015, 14% alcohol, Targon, Gironde, Bordeaux, France); and Chianti (2015, 12.5% alcohol, Vin Sen srl., Lamporecchio, Italy).
Measurement of the Fe(III):Fe(II) concentration ratio in red wine
A calibration curve was obtained by adding FeSO4.7H2O to model wine (12% ethanol, 8.0 g/L tartaric acid, pH 3.3 with NaOH), containing an excess of Br-PAPS (5.2 × 10−4 mol/L) and 2 mole equivalents of ascorbic acid to ensure no Fe(III) was present. The ligand was added first to ensure that the added Fe(II) did not oxidize. Addition of Fe(II) first would result in the rapid oxidation of the tartrate complex. As with ferrozine, the Fe(II)–Br-PAPS complex would not react with O2. The relationship: absorbance = 0.5377 × C + 0.117 (R2 = 0.9996), where C = Fe(II) concentration in mg/L, was obtained for Fe(II) concentrations ranging from 1.0 to 4.0 mg/L (1.79 to 7.16 × 10−5 mol/L). Absorbance was measured at 740 nm, using 10 mm quartz cells.
Wine bottle closures were removed, and a gentle nitrogen steam was quickly introduced into the headspace. To take wine samples, a cork was fitted with a thin plastic tube reaching to the bottom of the bottle and extending some 40 cm from the top, together with a short tube that extended only into the headspace. The tubes were flushed with nitrogen, and the assembly was inserted into the bottle; applying a gentle nitrogen pressure into the headspace by way of the short tube allowed wine (50 mL) to be expelled into a flask also filled with nitrogen, with an initial ~20 mL being discarded. When the wine was taken from wine boxes, a small amount was first flushed through the tap and the sample dispensed directly into the nitrogen-filled flasks. The wine was maintained in a nitrogen atmosphere as 2.5 mL samples were transferred to nitrogen-filled reaction tubes. The samples were stirred briskly under nitrogen, and Br-PAPS (5.2 × 10−4 mol/L; 25 μL of a solution containing 14 mg/500 μL H2O) was rapidly added. Absorbance at 740 nm was measured after 30 sec and at 1 min intervals for 6 min against wine blanks. The results were then plotted for each determination to obtain trend lines, which were extrapolated to zero time to obtain the absorbance at the moment of mixing (t0). Determinations were conducted in triplicate, and mean t0 (±SD) values were calculated. To measure total Fe concentration, absorbance could be measured once stabilized, which was after 7 to 24 hrs. More conveniently, excess ascorbic acid (9.4 mg in 100 μL 0.01 N H2SO4) was added to 50 mL of wine under nitrogen, and Br-PAPS (25 μL of the above solution) was added to 2.5 mL of wine under nitrogen, making a correction for the small increase in volume. Absorbance, corresponding to total Fe concentration, stabilized after 2 to 4 hrs, and was measured in triplicate. To validate the method, results for four wines were compared with values obtained by flame atomic absorption spectrometry. Some wines were saturated with air, and the above determinations were repeated to monitor the effect of O2 on the [Fe(III)]:[Fe(II)] ratio over 24 hrs.
O2 measurements
O2 concentration was measured with an HI-9146 dissolved-O2 meter fitted with an HI-76408 Clarke type electrode (Hanna Instruments Ltd.). Wine was transferred under nitrogen to small bottles, and measurements were made under nitrogen as previously described (Danilewicz 2016).
Fe and Cu measurements
Fe and Cu analyses were done by Campden BRI, using a Perkin Elmer PinAAcle 500 flame atomic absorption spectrometer (FAAS) (Perkin Elmer Corp). Values given are from identical duplicate runs.
Results and Discussion
The [Fe(III)]:[Fe(II)] ratios in red wines have been investigated spectrophotometrically with the Fe(II)-selective ligands 2-(5-bromo-2-pyridylazo)-5-diethylaminophenol (Br-PADAP) (Ferreira et al. 2007) and 2,2′-dipyridylketone picolinoylhydrazone (DPKPH) (Figure 1) (López-López et al. 2015). However, in these studies, wines were substantially diluted, their pH increased, and they were exposed to air for undefined periods during sample preparation and measurement of Fe(II) concentration, so that ratios undoubtedly would have changed from their original values. The reduction potential of the Fe couple is altered by changing the protonation state of the ligand, and, therefore, changes in pH should be avoided. Fe(II) also reacts rapidly with O2 under wine conditions (Danilewicz 2013), so the O2 concentration should not be altered during measurement of Fe(II) concentration.
Importantly, addition of an Fe(II)-selective ligand substantially increases the reduction potential of the Fe(III)/Fe(II) couple, making Fe(III) a much stronger oxidant (Danilewicz 2013). For example, Fe(III) is very rapidly reduced by a catechol in model wine on addition of ferrozine (Scheme 1) (Elias and Waterhouse 2010). The oxidizing power of Fe(III) is also much increased by addition of the Fe(II)-selective ligand 2,4,6-tripyridyl-s-triazine in the ferric ion reducing anti-oxidant power assay (Huang et al. 2005). Any Fe(III) present in a wine would be quickly reduced, especially in red wine, which contains high polyphenol concentrations, so that Fe(II) concentrations would increase rapidly on addition of the ligand. This effect was overcome with the use of ferrozine in white wine by measuring Fe(II) concentration over a short period after ligand addition and extrapolating back to zero time, that is, to the moment of ferrozine addition.
Ferrozine is particularly useful, as it is readily soluble in wine and forms a magenta Fe(II) complex with a strong absorption band at 562 nm (Danilewicz 2016). The ligand can be added directly to white wine under nitrogen with minimal changes in composition. Total Fe concentration is measured by first adding ascorbic acid under nitrogen to ensure that all the Fe was present as Fe(II). The values obtained for total Fe agree well with those obtained by FAAS. Addition of ferrozine to white wine results in a slow increase in absorbance as the Fe(III) present was reduced by polyphenols. By following absorbance over 10 min, it is possible to extrapolate back to the moment of addition to obtain the original Fe(II) concentration.
Unfortunately, ferrozine cannot be used for red wine because of color interference. Initial experiments were conducted with Br-PADAP (Figure 1), the Fe(II) complex of which has an absorption band at 748 nm, which is well outside the color range of red wine (Ferreira et al. 2007). However, the ligand proved insufficiently soluble in undiluted wines, but this problem was readily overcome by using the water-soluble analog Br-PAPS (Figure 1), the Fe(II) complex of which has an absorption maximum at 740 nm (Figure 2). Total Fe concentration was measured in four wines by first adding ascorbic acid, and the values were compared with those obtained by FAAS. A similar value was obtained at 2.16 mg/L, but values diverged to a small extent as Fe concentration increased (Table 1, Figure 3). To compare like with like, total values obtained with Br-PAPS were used to calculate % Fe(II) concentrations.
A bottled wine of Chilean Cabernet Sauvignon fitted with a screwcap was first examined. In this sample, 97.3 ± 0.3% of the Fe was present as Fe(II) (Table 1), and when saturated with air, the Fe(II) concentration fell progressively over a day to 61.6% (Figure 4). As mentioned above, Fe(II)-selective ligands substantially increase the reduction potential of the Fe(III)/Fe(II) couple such that Fe(III) becomes a much stronger oxidant. With ferrozine in white wine, this reaction is relatively slow. However, in red wine, which has much higher polyphenol concentrations, the reduction of Fe(III) is much faster on adding the ligand to the wine. Absorbance was measured 30 sec after adding the ligand to the wine, stirring briskly to ensure rapid mixing. Absorbance was then followed for 6 min (Figure 5). Initially, when little Fe(III) was present, the increase in absorbance was only slight (T = 0 hrs), but as the Fe(III) concentration increased as the wine oxidized, such as after 23.25 hrs, absorbance increased rapidly on ligand addition, and the accuracy of the extrapolated value was likely to be less certain. As anticipated, three wines (Shiraz [from SE Australia], Merlot, and Tempranillo) from wine boxes had lower initial Fe(II) % concentrations, ranging from 74.6 ± 0.1 to 78.6 ± 2.0 and 85.7 ± 0.7% (Table 1). The Fe(II) concentration fell similarly in these three wines when they were saturated with air (Figure 4). However, Fe(II) concentration appeared to stabilize for a time after ~3 hrs, and then continued to fall. This unexpected behavior needs to be examined further.
The effect of closures on % Fe(II) content was then examined. A Chianti wine closed with a technical cork (Diam 2) gave a value of 97.2 ± 1.1%, which was similar to that for Cabernet Sauvignon, the bottle of which was closed with a screwcap (Table 1). This result was consistent with the known low O2 transmission rate of this type of bottle closure (Godden et al. 2001). However, a Côte du Rhône in a bottle closed with a plastic closure (Nomacorc, 38 mm), which should allow greater ingress of O2, also very unexpectedly gave a high value (96.0 ± 1.0%) (Godden et al. 2001). It appeared that the polyphenols in red wine can maintain reductive conditions despite a certain amount of O2 ingress.
Three bottles of the same Bordeaux wine closed with natural corks were compared. Some variation was observed, as expected with this type of closure, but on average, % Fe(II) content was lower than in bottles fitted with a screwcap, Diam 2, or Nomacorc closures, suggesting that the wines were oxidizing more quickly. Three boxes of the same South African Shiraz were also compared. % Fe(II) concentrations were only slightly lower, varying from 84.0 ± 2.6 to 90.9 ± 2.0%. Interestingly, no measurable amounts of O2 (<0.01 mg/L) were found in these three wines. Consequently, if the so-called “reduction potentials” of these wines were measured, very low values would be obtained, which would misleadingly indicate a highly reductive state. In fact, the [Fe(III)]:[Fe(II)] ratios indicated that these wines were oxidizing at a rate limited by the rate of O2 ingress, so that there was no build-up of O2. The aforementioned three wines taken from wine boxes, Shiraz (SE Australia), Merlot, and Tempranillo, contained lower Fe(II) % concentrations and also contained no measurable amounts of O2. In contrast, the wine boxes containing white wine previously studied (Danilewicz 2016) contained 0.2 and 0.29 mg/L O2, with the lower polyphenol levels being unable to remove O2 as it was entering the wine.
Reduction potentials have been measured to determine the redox state of wines (Vivas et al. 1992, 1993). These authors prefaced their work by stating, “The reduction potential (EH) is a measure of the state of oxidation or reduction of a system, in the same way that pH is a measure of acid/base balance. EH expresses the level of oxidation or reduction.” However, there is a fundamental problem with wine. Reduction potentials can be generated only when both components of redox couples are present and when they are in thermodynamic equilibrium, as generated potentials are determined by differences in Gibbs free energy between two redox states. In wine, both O2 and polyphenols react irreversibly, and so redox couples associated with the two main redox systems are not present at equilibrium. O2 is reduced irreversibly to water, and polyphenol oxidation products are very unstable in wine conditions (Nikolantonaki and Waterhouse 2012). However, it is found that the observed so-called reduction potential is largely determined by O2 concentration, as shown for a red wine (Vivas et al. 1993). At very low O2 concentrations, the observed “reduction potential” dropped to ~200 mV, increasing to ~535 mV on aerial O2 saturation (~7 mg/L). Model wine, which only contains ethanol and tartaric acid at pH 3.6, gives very similar values, which are not changed by gallic acid addition (Kilmartin and Zou 2001), indicating that polyphenols are not involved in generating these potentials. Red and white wines behave similarly with respect to these potentials, despite greatly differing in polyphenol concentrations. Most telling, however, is that sulfite addition has little effect on the measured potential (Vivas et al. 1993), strongly supporting this conclusion, as sulfite would ensure the removal of oxidation products so that polyphenol redox couples could not be present.
Despite the substantial literature on the subject, it was finally realized that the generated potential at platinum electrodes largely relies on the oxidation of ethanol coupled to reduction processes such as O2 reduction, which have little to do with the initial redox processes that occur during wine oxidation centered on metal-catalyzed oxidation of polyphenols (Scheme 1). O2 reduction occurs very readily at platinum electrodes, and as the O2 levels increase, the cathodic current generated will be matched by an equal and opposite anodic current involving oxidation of the oxidizable species, including ethanol, which is oxidized also directly at platinum electrodes (Kilmartin and Zou 2001). Although measured potentials may give a general indication of how much O2 is present, O2 concentration can be much more reliably and easily measured with an O2 meter (Tomlinson and Kilmartin 1997, Kilmartin and Zou 2001, Danilewicz 2012).
The formal reduction potential of the Fe(III)/Fe(II) couple is measured as 770 mV (versus the standard hydrogen electrode, SHE), but, by definition, this is obtained at pH 0. Tartaric acid coordinates preferentially with Fe(III) and thereby shifts the Fe-redox equilibrium toward Fe(III), making Fe(II) a stronger reductant. Also, the tartrate ligand undergoes greater deprotonation at wine pH, particularly with Fe(III), and the negative charges surrounding the Fe(III) ion further preferentially stabilize this complex. Consequently, the formal reduction potential of the Fe couple is lowered to 345 mV (versus SHE) in the presence of tartaric acid at pH 3.6 (Danilewicz 2012). As a result, the reduction potential of the Fe couple may be estimated under wine conditions using the Nernst equation (1), where T[Fe(III)] and T[Fe(II)] are total ion concentrations.
Eq. 1If redox couples in wine were at equilibrium, all the couples would have adopted the same potential, so the Fe couple would be expected to have equilibrated to the overall measured wine potential. The Cabernet Sauvignon wine with a screwcap initially contained 97.3% Fe(II), which stabilized at 61.2% on air saturation after 23 hrs. From Equation 1, a 39:61 [Fe(III)]:[Fe(II)] ratio corresponds to a reduction potential of 334 mV, considerably lower than the ~520 mV that would be expected for such an air-saturated wine. The three South African Shiraz wines from boxes, which contained no measurable amounts of O2, showed between 84.0 and 90.9% Fe(II) at the onset. These values corresponded to reduction potentials between 302 and 287 mV, considerably higher than the ~200 mV expected for a wine containing no appreciable amounts of O2. Clearly, the moderate change in the reduction potential of the Fe(III)/Fe(II) couple is greatly exceeded by the change in reduction potentials that would be measured in wines after oxygenation.
In a previous study with white wines (Danilewicz 2016), an Argentinian Pinot Grigio wine with a screwcap and containing no measurable O2 (< 0.01 mg/L) contained 95% Fe(II), which stabilized down to 57% on air saturation after four days (Figure 6). The resulting [Fe(III)]:[Fe(II)] ratio corresponded to an initial reduction potential of 270 mV, increasing to 343 mV on oxidation. These values again differed considerably from the expected values of ~200 and ~520 mV, respectively. A French Chardonnay wine closed with a natural cork equilibrated down to 36.2% Fe(II) on air saturation after three days, while a South African Sauvignon blanc from a wine box equilibrated at 27.3% Fe(II) on air saturation after two days. These values corresponded to reduction potentials of 360 and 370 mV, again considerably less than the ~520 mV expected for an air-saturated wine. A further consideration is that all wines would be expected to adopt reduction potential close to 500 mV over time on air saturation. [Fe(III)]:[Fe(II)] ratios would then be ~426:1, which is 99.8% Fe(III). Clearly, this is not the case, as in the three white wines discussed above, the Fe couples adopted different and much lower reduction potentials ranging from 343 to 370 mV (Figure 6).
Conclusion
The [Fe(III)]:[Fe(II)] ratio may be measured spectroscopically, but the procedure needs to be performed quickly, extrapolating back to the moment of ligand addition, as any Fe(III) present is very rapidly reduced in red wine in the presence of an Fe(II)-selective ligand. Wine composition must be disturbed as little as possible and without exposure to any additional O2, as Fe(II) is rapidly oxidized before ligand addition. The [Fe(III)]:[Fe(II)] ratio increases on O2 exposure and varies for different wine closures. This simple procedure may therefore be useful for determining the redox status of wines; for instance, it may provide an additional way to examine the efficiency of wine bottle closures and of different storage systems as previously proposed for white wine. [Fe(III)]:[Fe(II)] ratios were used to calculate the reduction potentials of the couple, which is the central mediator of wine oxidation. However, the values did not correspond to those that would be observed in wines. It is proposed that this is a consequence of wine oxidation not having achieved equilibrium conditions and that wine “reduction potentials” are not true reduction potentials, but are mixed potentials largely arising from ethanol oxidation in addition to other electroactive species, and, therefore, cannot be used to determine a wine’s redox state. In fact, reduction potentials could misleadingly indicate that red wines, which contained very low O2 concentrations (<0.01 mg/L), were in a highly reduced state, but [Fe(III)]:[Fe(II)] ratios showed that such wines were actually oxidizing.
- Received September 2017.
- Revision received November 2017.
- Accepted November 2017.
- Published online March 2018
- ©2018 by the American Society for Enology and Viticulture