Food Chemistry 229 (2017) 319328

Contents lists available at ScienceDirect

Food Chemistry
journal homepage: www.elsevier.com/locate/foodchem

The impact of aging wine in high and low oxygen conditions on the
fractionation of Cu and Fe in Chardonnay wine
Nikolaos Kontoudakis a,b, Anque Guo a,c, Geoffrey R. Scollary a,d, Andrew C. Clark a,b,
a
National Wine and Grape Industry Centre, Mambarra Drive, Wagga Wagga, NSW 2678, Australia
b
School of Agricultural and Wine Sciences, Charles Sturt University, Locked Bag 588, Wagga Wagga, NSW 2678, Australia
c
College of Enology, Northwest A&F University, Yangling, Shaanxi 712100, China
d
School of Chemistry, The University of Melbourne, Vic. 3010, Australia

a r t i c l e i n f o a b s t r a c t

Article history: Solid-phase extraction has previously been used to fractionate copper and iron into hydrophobic, cationic
Received 23 November 2016 and residual forms. This study showed the change in fractionated copper and iron in Chardonnay wines
Received in revised form 3 February 2017 with 1-year of bottle aging under variable oxygen and protein concentrations. Wines containing protein
Accepted 14 February 2017
in low oxygen conditions induced a decrease (2050%) in total copper and increased the proportion of the
Available online 15 February 2017
hydrophobic copper fraction, associated with copper(I) sulfide. In contrast, protein stabilised wines
showed a lower proportion of the hydrophobic copper fraction after 1-year of aging. In oxidative storage
Keywords:
conditions, the total iron decreased by 60% when at high concentration, and the concentration of the
Wine
Copper
residual fraction of both copper and iron increased. The results show that oxidative storage increases
Iron the most oxidative catalytic form of the metal, whilst changes during reductive storage depend on the
Fractionation extent of protein stabilisation of the wine.
Bentonite 2017 Elsevier Ltd. All rights reserved.
Aging
Oxidation
Protein

1. Introduction
Metal ions are known to impact the colloidal, oxidative and
reductive stability of wine, and recent research has centered on
the role of copper (Cu) and iron (Fe) in the latter two processes
(Clark, Dias, Smith, Ghiggino, & Scollary, 2011; Clark, Prenzler, &
Scollary, 2003; Clark, Wilkes, & Scollary, 2015; Danilewicz, 2007;
Danilewicz, 2014; Li, Guo, & Wang, 2008; Lukton & Josyln, 1956;
Ribreau-Gayon, Glories, Maujean, & Dubourdieu, 2006; Viviers,
Smith, Wilkes, & Smith, 2013). Mn has also been the subject of a
recent study on its ability to influence wine oxidation in the pres-
ence of high Fe and Cu concentrations (Danilewicz, 2016a). How-
ever, many of these studies were conducted in model wine
systems, and the complexity of the wine matrix with its multitude
of potential binding agents, or components able to interact with
the metals, complicates the assessment of metal activity in wine.
The metal ions Cu and Fe, in their respective oxidation states
found in wine, that is Cu(II), Fe(III) and Fe(II), are predominantly
Corresponding author at: National Wine and Grape Industry Centre, Mambarra
Drive, Wagga Wagga, NSW 2678, Australia.
E-mail addresses: nkodoudakis@csu.edu.au (N. Kontoudakis), guoanque@
nwsuaf.edu.cn (A. Guo), scollary@unimelb.edu.au (G.R. Scollary), aclark@csu.edu.
au (A.C. Clark).
bound to organic acids in model wine systems but can still interact
with phenolic and thiol compounds resulting in redox reactions
(Clark, Kontoudakis, Barril, Schmidtke, & Scollary, 2016; Clark &
Scollary, 2006; Clark et al., 2015; Danilewicz, 2014). In addition,
hydrogen sulfide can bind Cu in wine conditions to such an extent
that it can form an insoluble dispersed Cu(I) complex that does not
tend to aggregate and settle in wine conditions (Clark, Grant-
Preece, Cleghorn, & Scollary, 2015; Kreitman, Danilewicz, Jeffrey,
& Elias, 2016). This form of Cu(I) sulfide has been shown to be suf-
ficiently bound to the extent that the Cu is not able to be detected
on analysis of the wine by the electrochemical technique of strip-
ping potentiometry (Clark et al., 2016). However, the formation of
the compound is reversible and the Cu can be released from the
sulfide via volatilization of hydrogen sulfide (Clark et al., 2015,
2016), heating (Clark & Scollary, 2006), dilution with salt followed
by heating (Clark & Scollary, 2006; Franco-Luesma & Ferreira,
2014), or with sufficient oxidation (Kreitman et al., 2016). Macro-
molecules present in wine also have the potential to interact
and/or bind metals, and may include tannins, polysaccharides, pro-
teins, and complex/aggregate combinations of these macro-
molecules. Past work has shown the ability of Cu to interact with
proteins and induce protein-hazes in wines, particularly in wines
with elevated Cu concentrations, low oxygen conditions and with

http://dx.doi.org/10.1016/j.foodchem.2017.02.065
0308-8146/ 2017 Elsevier Ltd. All rights reserved.
320 N. Kontoudakis et al. / Food Chemistry 229 (2017) 319328
residual protein present (Peterson, Josyln, & Durbin, 1958). Alter-
natively, Fe has been linked to the development of hazes through
interactions with phenolic compounds or phosphates when it is
at high concentrations in oxidative conditions (Joslyn & Lukton,
1953).
The fractionation of wine with solid phase extraction (SPE) car-
tridges is an approach taken to provide the broad categories of
metals present within a wine (Pohl & Prusisz, 2009; Pohl &
Sergiel, 2009; Rousseva, Kontoudakis, Schmidtke, Scollary, &
Clark, 2016). The advantage of this approach is the ability to ana-
lyze a range of metals simultaneously at low concentrations if an
inductively coupled plasma (ICP) technique (OES or MS) is utilised
for metal quantification in the subsequent fractions. The disadvan-
tage may be the ability of the solid phase extraction cartridges to
perturb the normal metal binding/interactions during the extrac-
tion process. However, with clearly defined methodology, the frac-
tionation can be performed reproducibly to provide an
operationally defined measure of metal fractionation in wine.
A previous study has used this approach to identify a fraction of
metal that provided the best correlation with the metals ability to
induce loss of oxygen in wines containing ascorbic acid (Rousseva
et al., 2016). The wines were passed sequentially through C18 and
cationic exchange SPE cartridges to provide measures of three
metal fractions: hydrophobic, residual and cationic. The hydropho-
bic fraction was that retained on the C18 cartridge, the residual
fraction passed through both cartridges, and the cationic fraction
was retained on the cation exchange cartridge. In assessing the
oxygen consumption in the wines, by utilizing ascorbic acid at an
elevated concentration (200 mg/L), the study was able to offset
the inherent variability of oxygen consumption between different
wines by ensuring not only similar pH and temperature conditions,
but also eliminating the influence of the phenolic profile of the
wines. The lower redox potential of ascorbic acid in wine favors
its reaction over the phenolic compounds during oxygen consump-
tion (Danilewicz, 2003). Consequently, it was only the variable
metal concentration in the wine, and their different forms, that
were able to influence the rate of oxygen consumption. The results
(Rousseva et al., 2016) showed, similar to other studies
(Danilewicz, 2007), that total Cu concentration correlated better
with oxygen consumption than total Fe concentration. Further-
more, it was the residual fraction of Cu that provided the best cor-
relation with oxygen consumption followed by the cationic
fraction, whilst the hydrophobic fraction had poor correlation with
the rate of oxygen consumption by the wine (Rousseva et al.,
2016). The binding agents responsible for classification of the met-
als to specific fractionation categories were not ascertained in this
study, other than the tartrate complexes of Cu(II) and Fe(II) con-
tributing to their respective cationic fractions. It was speculated
that the cationic fraction may also include protein-metal com-
plexes, whilst the residual fraction may include metal complexes
of acidic-polysaccharides, phenolic acids or citric acid, with an
overall negative charge on the complex (Pohl & Prusisz, 2009;
Pohl & Sergiel, 2009; Rousseva et al., 2016). However, it must be
noted that the SPE fractionation technique does not always result
in a specific metal ligand combination being measured in a single
fraction, but instead they can be distributed amongst several frac-
tions (i.e., residual and cationic) (Pohl & Prusisz, 2009). In this case,
it is likely that the equilibria between the different ligand com-
plexes of Fe(II) (i.e., complexes with different ratios of ligand to
metal) and their speed of re-equilibration during passage through
the SPE cartridges, influences the overall partitioning of Fe.
The general change in metal species during the oxidative and
reductive aging of wine is not well understood. Danilewicz
(2016b) showed that in model wine systems, oxygen consumption
occurred most rapidly until a redox equilibrium between Fe(II) and
Fe(III) forms was reached, and that the favored proportion of each
at equilibrium was wine dependent. Under accelerated oxidation
conditions (in darkness, 45 C), Clark and Scollary (2006) demon-
strated an increase in electrochemically measured labile Cu con-
centration in four white wines with the rate of increase being
dependent on the specific wine. It was proposed that the labile
Cu would be the more reactive in terms of catalyzing oxidative
reactions than the non-labile Cu. Related to the release of labile
Cu is the earlier work by Singleton (1987) that showed whites
wines browned at a progressively faster rate as they consumed
oxygen, and coined the term that browning begets browning,
implying that wines were more easily oxidised as they become
oxidised.
This present study was undertaken to assess how the fraction-
ation of metals may change with bottle aging (1 year) of wine
under low and high oxygen conditions. The study was conducted
with the Chardonnay wine from Rousseva et al. (2016) prepared
from juice of different metal concentrations, and with various ben-
tonite treatments and consequently different protein concentra-
tions (Table 1). The fractionation of these wines at bottling was
described in a previous work (Rousseva et al., 2016). The results
after 1-year of bottling were examined with emphasis on the
changes in the residual and cationic fractions of Cu that are known
to be correlated well with oxygen decay in wine. Furthermore, the
fraction of Cu corresponding to Cu(I) sulfide was also to be
established.
2. Materials and methods
2.1. Reagents and general instrumental analysis
All glassware and plastic-ware were soaked for at least 16 h in
10%(v/v) nitric acid (BDH, AnalaR, Poole, UK) and then rinsed with
copious amounts of water (18.2 MX). Solutions and dilutions were
prepared using 18.2 MX water. Metal salts were obtained from
BDH (Poole, UK) (copper(II) sulfate pentahydrate (AnalaR)), AJAX
(Taren Point, Australia) (iron(II) sulfate heptahydrate (98%)) and
Sigma-Aldrich (iron(III) sulfate nonahydrate (97%)).
Cu and Fe concentrations in Chardonnay wine were determined
with a Varian 710 ICP-OES (California, USA) as described by
Rousseva et al. (2016). To assess the efficacy of the dilution
approach in allowing the total Fe and Cu concentrations to be
assessed, a subset of two Chardonnay wine samples described in
this study were quantified after pre-treatment of the wine with
4-fold dilution with 5%(v/v) nitric acid as well as by digestion (with
concentrated nitric acid). The results showed no significant differ-
ence between the determined total concentrations by these two
strategies and hence pretreatment of wines via dilution was
adopted. The concentrations of total and free sulfur dioxide (SO2)
was measured with a flow injection analyser (FIAstarTM 5000, FOSS,
Denmark), connected to an auto-sampler (5027, FOSS, Denmark).
2.2. Wines and bentonite treatment
The Chardonnay wines utilised were those prepared in a previ-
ous study (Rousseva et al., 2016) and the experimental design is
shown in Supplementary Fig. 1. Chardonnay juice was used for
six different metal ion treatments with each treatment prepared
in triplicate (6  3  12.0 L demijohns). The treatments included
no addition (T1), 2.9 mg/L Cu (T2), 5.8 mg/L Cu (T3), 6.3 mg/L Fe
(T4), 12.5 mg/L Fe (T5), 5.8 mg/L Cu and 12.5 mg/L Fe (T6). Table 1
shows the metal concentrations in the juice and those in the bot-
tled wine. Cu was added as copper(II) sulfate pentahydrate and
Fe was added as iron(II) sulfate heptahydrate. After fermentation
(see details within Rousseva et al. (2016)) the wines were racked
twice.
 N. Kontoudakis et al. / Food Chemistry 229 (2017) 319328 321

Table 1
The Cu and Fe concentrations, as reported in Rousseva et al. (2016), in the juice and corresponding wine samples, and the first order oxygen decay rates determined for each of the
wines. The oxygen decay rates were assessed after addition of 216 mg/L ascorbic acid and 232 mg/L sulfur dioxide to the wines and incubation at 25 C.
1
Juice Cu total (mg/L) Juice Fe total (mg/L) Wine Cu total (mg/L) Wine Fe total (mg/L) Wine oxygen decay (min )
a,b a a a a
T1 0.77 0.03 0.37 0.01 0.083 0.008 0.423 0.006 0.00020 0.00003
T2 3.61 0.09c 0.38 0.02a 0.233 0.005c 0.425 0.009a 0.00045 0.00002b,c
T3 6.35 0.03d 0.31 0.02a 0.293 0.006d 0.421 0.006a 0.00054 0.00007b,e
T4 0.69 0.02a 6.0 0.1b 0.068 0.014ab 6.20 0.05b 0.00038 0.00006c
T5 0.80 0.03b 11.1 0.1c 0.052 0.011b,e 11.7 0.1c 0.00038 0.00005c
T6 6.52 0.06d 11.3 0.1c 0.288 0.007d 11.8 0.1c 0.00072 0.00003d
A 0.77 0.03a,b 0.37 0.01a 0.076 0.001a 0.400 0.001d 0.00023 0.00003a
A + Bent 0.77 0.03a,b 0.37 0.01a 0.034 0.002e 1.39 0.02e 0.00043 0.00005b,c
B 6.35 0.03d 0.31 0.02a 0.290 0.002d 0.41 0.02a,d 0.00046 0.00001b
B + Bent 6.35 0.03d 0.31 0.02a 0.046 0.002b 1.55 0.02f 0.00054 0.00005b
C 6.52 0.06d 11.3 0.1c 0.35 0.02f 12.3 0.1g 0.00064 0.00002e
C + Bent 6.52 0.06d 11.3 0.1c 0.092 0.003a 12.7 0.2h 0.00061 0.00002e

The uncertainty indicated is the standard deviation. Concentrations of a given metal with the same superscript letters in a given column are not significantly different at
p = 0.05.

A heat stability test (Iland, Bruer, Edwards, Weeks, & Wilkes,
2004) was conducted on the wines and all failed, indicating the
presence of proteins within these wines capable of inducing haze.
A bentonite trial (Iland, Bruer, Ewart, Markides, & Sitters, 2004)
using SIHA ActivBentonit G (calcium bentonite by EnolTech (Grif-
fith, Australia)) was conducted and established that 1.6 g/L of ben-
tonite was required to afford sufficient removal of protein from the
wines to allow them to pass the heat stability test. Consequently,
before bottling, the volume of one replicate of T1, T3 and T6 was
halved and one half of each treated with an equal amount (1.6 g/
L) of the SIHA ActivBentonit G bentonite and the wines were left
to settle for 5 days at 4 C before racking and bottling. The samples
without bentonite addition were designated wines A, B and C (Sup-
plementary Fig. 1), respectively, whilst the equivalent wines with
bentonite treatment as +bent (i.e., wine A + bent). Wines were
bottled, without filtration, under screw cap in 375 mL capacity bot-
tles with inert gas coverage for the wines with standard low oxy-
gen conditions. Wines A, A + bent, B + bent, and C + bent were also
bottled with 190 mL of wine in a 375 mL and the ullage left with
air before sealing under screw cap to afford high oxygen condi-
tions. Wines were stored in cellar conditions (17 C) for 1-year
prior to analysis.
The following are the combined compositional details of all
wines at bottling (average stdev.): ethanol 12.8 0.2%(v/v), pH
3.13 0.01, titratable acidity 6.3 0.01 g/L (tartaric acid equiva-
lents to pH 8.2), free sulfur dioxide 39 2 mg/L and total sulfur
dioxide 169 6 mg/L.
2.3. Preparation of Cu(II), Cu(I) sulfide, Fe(II) and Fe(III) in model wine
Cu(I) sulfide was generated in the model wine using similar
conditions as that described by Kreitman et al. (2016). A standard
of 491 mg/L sodium sulfide was prepared in chilled water (4 C),
and 0.100 mL was added to 100 mL of model wine (12% aqueous
ethanol, 0.011 M potassium hydrogen tartrate, 0.007 M tartaric
acid, pH 3.25) to afford a concentration of hydrogen sulfide of
6.30 lM (or 0.214 mg/L). Then 0.40 mL of a 50.0 mg/L Cu(II) stan-
dard in model wine (as copper(II) sulfate pentahydrate) was added
to the 100 mL to afford a concentration of 3.15 lM (or 0.200 mg/L)
and providing a mole ratio of 2:1 for hydrogen sulfide to Cu(II). The
Cu(I) sulfide solution was prepared in triplicate and analysed
within 10 min of preparation. A Cu(II) standard in the model wine
was prepared in an identical manner but without addition of the
sodium sulfide.
A stock concentration (100 mg/L) of both Fe(II) and Fe(III) were
prepared in model wine, utilising Fe(II)SO4
7H2O and
Fe(III)2(SO4)3
9H2O, respectively. The stock solutions were utilised
immediately after preparation to prepare 1.0 mg/L or 10.0 mg/L
samples, and once prepared the samples were immediately frac-
tionated by the SPE method outlined in Section 2.4. The 100 mg/L
stock solutions of Cu(II), Fe(II) and Fe(III) were utilised to generate
calibration standards in model wine solution. The calibration
standards were diluted 4-fold with 5%(v/v) nitric acid prior to
analysis. During automated flow to the ICP-OES, all samples and
standards were mixed (1:1), via a t-piece, with a stream of
10 mg/L Lu in 5%(v/v) nitric acid. The calibration graph for Cu
(324.754 nm) and Fe (259.940 nm) utilised Lu (261.541 nm) as
an internal standard.
2.4. Solid phase extraction scheme
The fractionation of the different metal species was performed
as described previously (Rousseva et al., 2016) (Supplementary
Fig. 2). Strata C18-E (non-polar extraction) and Strata SCX (cation
exchange) cartridges were used in series (both supplied by Phe-
nomenex, Torrance, USA). The Strata C18-E cartridge (1 g/6 mL)
was conditioned with 20 mL methanol, followed by 40 mL water,
10 mL 1.0 M HCl, and finally with 20 mL of a model wine solution,
consisting of 12% aqueous ethanol, 0.011 M potassium hydrogen
tartrate, and 0.007 M tartaric acid (pH 3.25). The cartridge was
then dried under vacuum prior to usage. The second cartridge
(Strata SCX, 1 g/6 mL) was conditioned with the following solu-
tions in sequence: 40 mL of water, 10 mL of 1.0 M HCl, 20 mL of
water, 10 mL of 1.0 M NaOH and finally with 20 mL water. The car-
tridge was then dried under vacuum.
The outlet of the conditioned cartridge was attached to a peri-
staltic pump via 0.76 mm (internal diameter) polyvinylchloride
tubing (Gilson, Middleton, USA) and 30 mL of each filtered wine
sample (0.45 lm polyethersulfone membrane) was passed through
the Strata C18-E cartridge at a rate of 1 mL/min. A 10 mL aliquot of
the eluent was used for measurement of the metal ions not
retained by the C18 cartridge, allowing calculation of the hy-
drophobic metal fraction, being equivalent to the total metal con-
centration in the wine minus the metal not retained by the C18
cartridge. The remaining 20 mL aliquot of the solution eluted
through the C18 column was then passed through the second col-
umn (Strata SCX) at a rate of 1 mL/min. The eluent from this second
column was measured for the residual fraction of metal ions. To
obtain the third and final fraction of metals found in the sample,
10 mL of 2.0 M HCl was passed through the Strata SCX column at
a rate of 1 mL/min. This 10 mL fraction was collected and quanti-
fied for metal ions (termed cationic fraction). Thus, three metal
fractions were obtained hydrophobic, residual and cationic. It
must be reinforced that the cationic fraction does not imply
322 N. Kontoudakis et al. / Food Chemistry 229 (2017) 319328
measurement of only positively charged metal ions and/or com-
plexes from a sample, but rather that it is any metal species
extracted by a cationic exchange cartridge during passage of a
sample through the cartridge. Blank samples were performed using
a model wine system (12% aqueous ethanol, 0.011 M potassium
hydrogen tartrate, 0.007 M tartaric acid, pH 3.2) which identified
background Fe contamination stemming from the cation exchange
cartridge. The background contamination was identified as arising
from the 10 mL of 1.0 M NaOH utilised to wash the cartridge, and
was able to be accounted for in all analyses.
Prior to ICP-OES analysis, the primary eluents from the C18 and
cationic columns were diluted 4-fold with 5.0%(v/v) nitric acid. The
acidic secondary eluent from the cation exchange column was
diluted 8-fold, allowing for the 2-fold concentration in the SPE car-
tridge, as an 1.25 mL aliquot was treated with 1.00 mL of 1.0 M
sodium hydroxide, 0.25 mL of water and 7.5 mL of 5.0%(v/v) nitric
acid. Lu was introduced into the sample flow during ICP-OES anal-
ysis and all samples were quantified using Lu as the internal stan-
dard. All uncertainty quoted is the standard deviation (n = 3) and
stated significant differences are at the 95% confidence limit and
calculated according to the Student-t test.
2.5. GC analysis of sulfur compounds
Analysis was carried out with an Agilent 7890B gas chromatog-
raphy system (Santa Clara, CA, USA) equipped with a CombiPal
auto-sampler (CTC Analytics, Zwingen, Switzerland) and coupled
to an Agilent 355 SCD Sulfur Chemiluminescence Detector. The col-
umn was an Agilent J&W DB-sulfur SCD 60 m  0.32 mm  4.2 lm.
GC conditions were adapted from the work of Siebert, Solomon,
Pollnitz, and Jeffery (2010). Helium (BOC, NSW, Australia-ultra
high purity) was used as a carrier gas at a linear velocity of
34.8 cm/s and a constant flow of 2.8 ml/min. The initial oven tem-
perature was 35 C and held for 5 min, followed by an increase to
150 C at a rate of 5 C/min and holding at this temperature for
10 min. SCD parameters were adjusted based on recommendations
by Agilent. SCD burner temperature was 800 C, burner base tem-
perature was 150 C, hydrogen (BOC-ultra high purity) flow rate
was 44.5 sccm and oxidizer flow low rate was 63.0 sccm (instru-
mental grade air-BOC). The ozone regulator pressure was 6 psig,
the SCD pressure was 6 Torr while the dual plasma controller pres-
sure was 378 Torr.
The total and free concentrations of both hydrogen sulfide and
methanethiol were determined. The free forms were analysed
adapting the protocol of Franco-Luesma and Ferreira (2014).
10 ml of sample was added into a 20 ml standard headspace vial
(CrossLab, Agilent). The vial was tightly sealed with a PTFE/silicone
lined screw cap and immediately put in the sampler tray at 10 C.
The vial was incubated at 45 C for 5 min with agitation at
400 rpm. Then 800 ll of the headspace was taken, with a 1 ml
heated headspace syringe (MSH 01-00B, Shimadzu, Kyoto, Japan)
thermostated at 60 C, and further injected into the injector (at
150 C under a 1:8 split ratio). Quantification was performed using
an external calibration graph.
Determination of total concentrations of hydrogen sulfide and
methanethiol was adapted from Siebert et al. (2010), whereby
10 ml of sample and 50 ll of EMS internal standard were trans-
ferred to a vial containing 3 g of NaCl. The vial was incubated at
45 C for 30 min and injected as described for the free forms. Stock
solutions of both Na2S and NaSCH3 were prepared in chilled water
and used to make external calibration standards in model wine for
determination of free concentrations, and to prepare internal stan-
dard calibration standards in model wine for determination of total
concentrations.
3. Results and discussion
3.1. Fractionation of Cu(II), Cu(I) sulfide, Fe(II) and Fe(III) in model
wine
In a model wine system buffered to pH 3.25 with tartaric acid
(3 g/L) and with the addition of 0.2 mg/L Cu(II), fractionation
and Cu measurement in each fraction by SPE ICP-OES showed that
the Cu(II) was fractionated predominantly (>95% of total measured
Cu) in the cationic fraction. The remaining Cu portion (<5%) was
found in the hydrophobic fraction, most likely as a consequence
of incomplete recovery of the entire sample volume from the SPE
C18 cartridge. Given that ion selective electrode studies have pre-
viously showed that the majority of Cu in this model wine system
is in the form of Cu(II) tartrate (Clark et al., 2015), it is evident that
Cu tartrate is measured in the cationic fraction of Cu by SPE ICP-
OES. This does not imply that Cu tartrate is positively charged,
but rather that the Cu is predominantly dissociated from tartrate
and sequestered by the cation exchange material before it can pass
through the cation exchange cartridge and be measured as residual
Cu. Using a related SPE methodology, Pohl and Sergiel (2009)
showed that the proportion of Cu measured in the cationic fraction
decreased as the pH in a 0.5 g/L tartaric acid solution increased
from 3.5 to 5.0, while the residual fraction increased. The pH
3.25 was adopted for the results in our current study to ensure that
the results would be consistent with the wines analysed in Sec-
tions 3.2 and 3.3, and it is evident that under these conditions Cu
tartrate is measured in the cationic fraction.
Given that Cu(I) sulfide is known to be the predominant form of
Cu when Cu(II) interacts with hydrogen sulfide in a model wine
system (Kreitman et al., 2016), it was desirable to identify the frac-
tionation category for this Cu complex. After the addition of Cu(II)
to a model wine containing hydrogen sulfide, at a mole ratio of 2:1
for hydrogen sulfide to Cu, the model wine system was fraction-
ated. The results showed that the Cu(I) sulfide formed was pre-
dominantly fractionated as hydrophobic Cu (98%). This
association with the hydrophobic fraction meant that copper(I)
sulfide was most likely retained on the C18 SPE cartridge due to
a filtration effect, with the capture of dispersed Cu(I) sulfide parti-
cles, but also possibly through non-polar interactions.
When the model wine was prepared with 1.0 mg/L of either Fe
(III) or Fe(II), the subsequent fractionation provided no significant
differences with 60% (of measured total Fe) contributing to the
cationic Fe fraction and the remaining 40% to the residual Fe frac-
tion, irrespective of the initial oxidation state of the Fe. There was a
negligible amount of hydrophobic Fe. However, when the concen-
tration of Fe was increased to 10 mg/L, the cationic fractions
became 74 10% for Fe(III) versus 94 1% for Fe(II), and the resid-
ual fractions became 30 10 and 5.4 0.8%, respectively. This
demonstrated a higher fraction of residual Fe for Fe(III) in the
model wine compared to Fe(II), but the difference was only signif-
icant at the high concentration of Fe. It also shows that, unlike the
case for Cu tartrate at pH 3.25, Fe(III) tartrate and Fe(II) tartrate can
presumably both be distributed amongst two different fractions
rather than being measured within one fractionated category. Sim-
ilar results were observed by Pohl and Prusisz (2009) for Fe(III),
where the distribution of 10 mg/L Fe(III) in 0.5 g/L citrate solution
at pH 3.5 and 3.0 was 16% and 27% in the cationic Fe fraction,
respectively, with the remaining Fe measured as residual (84 and
73%, respectively). The residual Fe(III) was attributed by Pohl and
Prusisz (2009) to be due to the presence of neutral and anionic
citrate complexes of Fe that were able to induce more Fe to elute
through the cationic exchange material. However, it is more likely
that the overall equilibrium between the different complexes of Fe
(II) and tartrate, or Fe(III) and tartrate, and their speed of
 N. Kontoudakis et al. / Food Chemistry 229 (2017) 319328 323

re-equilibration during passage through the cation exchange car- concentrations of Cu resulted in the largest proportional decreases
tridge, influences the overall partitioning of the Fe between the in total Cu during the 1-year storage period (see T2, T3 and T6,
residual and cationic fractions. In terms of our increased residual Fig. 1a). Presumably this Cu was encapsulated into the protein
Fe concentrations for Fe(III) compared to Fe(II) at 10 mg/L, the deposit.
complexes formed between Fe(III) and tartrate are known to be It is also evident that the wines maintained a similar rank order
stronger than those between Fe(II) and tartrate (Danilewicz, with respect to Cu concentration as was the case at bottling
2014; Timberlake, 1964). At pH 3.25, a dimeric form of the com- (Fig. 1a) and also as the Cu concentration in the juice prior to fer-
plex (2:2 complex for Fe(III):tartrate) dominates but the mono- mentation (Table 1). Although the magnitude of concentration of
meric form (1:1 complex) is also present (Timberlake, 1964). the Cu in the juice is around 10 times higher than in the wines,
Hence, the relative stability of the Fe complexes is consistent with and much of the Cu was lost during fermentation, the juices with
the stronger Fe(III) tartrate complexes surviving the cation higher initial Cu led to wines with higher Cu concentrations and
exchange column and inducing more residual Fe than the Fe(II) consequently to aged wines (at 1-year) with higher Cu
tartrate complexes. This would be particularly the case at higher concentrations.
Fe(III) tartrate concentrations when the ratio of cation exchange In terms of the fractionated forms of metal, the concentrations
sites to Fe(III) tartrate complexes would be lower. Indeed, Fe(II) of cationic and residual fractions of Cu decreased during the 1-
and Fe(III) have been separated in a pH 2.0 tartrate medium using year aging period (Supplementary Fig. 3), while the hydrophobic
an anion exchange column (Morie & Sweet, 1964), and the tech- Cu concentration remained constant (T1-T5) or increased for T6
nique makes use of the different stability constants of tartrate with (high Cu and high Fe). Consistent with the total Cu concentrations,
Fe(III) or Fe(II). Also, in acidic aqueous solutions (0.2 M H2SO4), a the highest proportional losses in cationic and residual fractions
cation exchange column has been used to retain both redox forms were evident for the samples with elevated Cu concentrations
of Fe and then oxalic acid used to elute only Fe(III) from the column (T2, T3 and T6). In terms of proportions of fractionated Cu (Fig. 1b),
(Zagorchev & Balushev, 1960). the general trend was towards an increase in the hydrophobic frac-
tion of Cu and decrease in the proportion of residual and cationic
3.2. Measured sulfur dioxide and calculated oxygen consumed in wines fractions of Cu. The increase for hydrophobic Cu was only signifi-
after 1-year cant for T5 and T6. Such a result is consistent with the formation
of hydrogen sulfide during the aging of wine in low oxygen condi-
The decrease in total sulfur dioxide for the wines in the 1-year tions (Ugliano et al., 2011, 2012; Viviers et al., 2013). Such hydro-
storage period was used to provide an estimate of the oxygen con- gen sulfide formed would be able to bind with Cu(II) tartrate (i.e.,
centration consumed over the same period and thereby define the
low and high oxygen storage conditions. The wines stored in low
oxygen bottling conditions resulted in an average consumption of
total sulfur dioxide of 29 5 mg/L (stdev, n = 9). On the assumption
that two molecules of sulfur dioxide are consumed for each mole-
cule of oxygen, as proposed in white wine (Danilewicz, Seccombe,
& Whelan, 2008), then this equates to a consumption of 7 1 mg/L
molecular oxygen. Although recent research suggests this stoi-
chiometry of sulfur dioxide consumed to oxygen consumed is wine
dependent (Waterhouse et al., 2016) and often smaller than 2:1,
this ratio still provides an approximation of the amount of oxygen
consumed. This oxygen would be oxygen dissolved in the wine at
bottling and oxygen trapped within the head-space of the wine
bottle (between the closure and the surface of the wine). It would
also include oxygen that permeated through the screw cap closure
although this is usually quite low for this type of closure (0.3
0.9 lg/day (Lopes, Saucier, Teissedre, & Glories, 2006)). For the
wines stored in high oxygen conditions the average consumption
of total sulfur dioxide was 118 12 mg/L, and hence a calculated
oxygen consumption of 30 2 mg/L.

3.3. The aging of wines without bentonite addition in low oxygen

conditions

Cu. The wines prepared without bentonite treatment were all

found to fail the heat stability test (Iland et al., 2004), indicating
the presence of residual protein. A bentonite fining trial estab-
lished that the concentration of protein in the wines was such that
1.6 g/L bentonite would have been required for its removal. For all
the wines bottled with residual protein (i.e., no bentonite treat-
ment), a deposit was evident at the bottom of the bottle after 1-
year of aging. The amount of deposit was visually assessed to be
similar regardless of the sample treatment. Such deposit is well
known to occur in wines with residual protein (Waters et al.,
2005; Wilkes, Hart, Clark, Blackman, & Scollary, 2013) and is attrib-
uted to the gradual denaturing, aggregation and settling of the Fig. 1. The total Cu concentration (a) and proportion of fractionated forms (b) in
grape-derived protein over time (Joslyn & Lukton, 1953). In terms wines T1-T6 at bottling and after 1 year in low oxygen conditions. The uncertainty
of total Cu, it was observed that the samples bottled with higher shown corresponds to the standard deviation (n = 3).
324 N. Kontoudakis et al. / Food Chemistry 229 (2017) 319328

cationic Cu fraction), to increase the Cu(I) sulfide concentration
and hence the hydrophobic fraction of Cu. The production of
hydrogen sulfide has been observed to occur in non-protein sta-
bilised wine (Wilkes et al., 2013), and would be described as de
novo formation of hydrogen sulfide by Franco-Luesma and
Ferreira (2016). It must be noted that the T5 sample, which has
the largest deviation between summed Cu fractions compared to
the measured total Cu (80% versus 100%, Fig. 1b), was the sample
with the lowest Cu concentration (<0.050 mg/L). It is likely that
at such low Cu concentrations the measurement uncertainty
would have contributed significantly to the deviation. A similar
deviation (125% versus 100%) is observed for sample WB + Bent
in Fig. 3b, and the total Cu concentration in this sample is less than
0.03 mg/L.
The total, bound and free concentrations for hydrogen sulfide
and methanethiol for samples T1-T6 after 1-year of bottle aging
are shown in Table 2. The bound form of hydrogen sulfide is pre-
dominantly attributed to the formation of Cu(I) sulfide (Kreitman
et al., 2016), although other metal ions can contribute to the bind-
ing of hydrogen sulfide as well (Franco-Luesma & Ferreira, 2014).
The concentrations of the sulfur compounds were measured after
the wine was aged 1-year in bottle (Table 2), but analysis was
not performed at bottling (Rousseva et al., 2016). This means that
it is not possible to directly distinguish between the amounts of
sulfur compounds at bottling versus those generated during bottle
aging. From the six protein-containing samples (T1-T6), there is a
trend towards higher total and bound hydrogen sulfide and lower
total methanethiol concentrations with increased total Cu concen-
tration (Table 2). These relationships are evident in the correlations
for total Cu and the sulfur compounds (r2 = 0.750.99) shown in
Table 2. The correlations are higher when the total Cu concentra-
tions utilised were those measured at bottling rather than those
measured a year later. Hence the total Cu that was initially present
in the wine at bottling, rather the total Cu remaining in the wine
after the gradual protein-induced loss, correlates best with total
and bound hydrogen sulfide (positively) and total methanethiol
(negatively) in the wine after 1 year.
The results (Table 2) also show that at bottling (year 0), it is the
cationic and residual fractions of the total Cu that correlate best
with total hydrogen sulfide concentration (positive correlation)
and methanethiol concentration (negative correlation) after 1 year
of aging. It was these same fractions of Cu that also correlated best
with oxygen consumption in these wines (Rousseva et al., 2016),
opening the possibility that they may be reactive in terms of both
oxidative and reductive reactions (Bekker, Smith, Smith, & Wilkes,
2016). The cationic fraction of Cu includes the Cu(II) tartrate com-
plexes, whilst the residual fraction of Cu has been observed to
increase with concentrations of organic acid Cu complexes that
may have a negative charge at wine pH (Pohl & Sergiel, 2009)
(i.e. citrate complexes). Correlation of the total hydrogen sulfide
after a year with these two different fractions of Cu at bottling sug-
gests they may be active in the production of hydrogen sulfide in
the protein-containing wine during bottle aging. A similar outcome
is evident for methanethiol and the cationic and residual fractions
of Cu, although in this case the correlations are negative instead of
positive. Consequently, it is the cationic and residual fractions of
Cu that have an inverse relationship with the concentration of
methanethiol.
When measured after one year of bottle aging (year 1, Table 2),
the cationic fraction of Cu correlates best with total hydrogen sul-
fide, while the residual fraction has poorer correlation than when
measured at bottling. For residual Cu, the poorer correlation after
a year of aging is most likely due to the significantly decrease of
residual Cu during bottle aging to concentrations that are similar
in all the samples (Supplementary Fig. 3).
The results for bound hydrogen sulfide concentration at 1 year
versus Cu fractions are similar to that of total hydrogen sulfide,
which is consistent with the fact that the majority of hydrogen

Table 2
Concentration of total, free and bound forms of hydrogen sulfide and total and free methanethiol after 1 year of wine storage, and their correlation with total Cu and fractionated
Cu when measured at bottling (0 year) or 1 year after bottling.

Sample Total H2S Bound H2S Free H2S % Free H2S Total H3CSH % Free H3CSH
T1 3.4 0.6a,b,c 2.1 1.3 0.3a,b 38 10a,b,f 4.3 0.3a,c 71 12a,b
T2 3.8 0.1a 3.4 0.5 0.2a,c 12 4a,c,d 2.2 0.3b,d 59 14a,b,c
T3 4.9 0.9a,b,c 4.2 0.7 0.3a,b,c 14 6a,c,d 2.0 0.1b 55 15a,b,c
T4 3.2 0.5a,b,c 2.0 1.1 0.1b 36 5b 3.9 0.6a,c,d 66 2a
T5 2.7 0.3b 1.6 1.17 0.05b 43 2b 4.3 0.1a 60 6a,b
T6 4.4 0.2c 4.1 0.32 0.04c 6 1c 1.9 0.2b 51 3b
WA 3.4 0.6a,b,c 2.1 1.3 0.3a,b 38 10a,b,f 4.3 0.3a,c 71 12a,b
WB 4.9 0.9a,b,c 4.2 0.7 0.3a,b,c 14 6a,c,d 2.0 0.1b 55 15a,b,c
WC 4.4 0.2c 4.1 0.3 0.1c 6 1c 1.9 0.2b 51 3b
WA + bent 2.6 0.4b 2.2 0.4 0.2a,c 16 9a,c,d 2.8 0.5c,b 54 7a,b,c
WB + bent 2.6 0.4b 2.3 0.32 0.04c 12 2d 1.4 0.2b,e 38 3c
WC + bent 3.6 0.9a,b,c 3.3 0.3 0.1c 7 3c,d 1.8 0.3b 47 4b,c
WA ox 0.94 0.04d 0.3 0.66 0.02a 70 2e 1.06 0.03e 42 1c
WA + bent ox 1.13 0.01e 0.4 0.76 0.05a 68 5e 0.94 0.05e,f 42 10b,c,d
WB + bent ox 0.63 0.02f 0.6 0.32 0.02c 51 2f 0.71 0.02g 34 16a,b,c,d
WC + bent ox 0.55 0.04f 0.3 0.31 0.02c 57 4f 0.79 0.04f,g 20 4d

Correlation of sulfur compound concentration in T1-T6 after 1 year with total Cu and fractionated Cu after 0 or 1 years
Year 0
Total Cu 0.9013 0.9863 ( ) 0.8374 ( ) 0.9446 ( ) 0.9566 ( ) 0.6440
Hyd Cu 0.5981 0.5204 ( ) 0.6826 ( ) 0.5997 ( ) 0.4914 ( ) 0.5136
Res Cu 0.9343 0.9512 ( ) 0.6947 ( ) 0.8851 ( ) 0.8606 ( ) 0.0287
Cat Cu 0.9108 0.9719 ( ) 0.8041 ( ) 0.9029 ( ) 0.9135 ( ) 0.6877
Year 1
Total Cu Y1 0.7506 0.8760 ( ) 0.8393 ( ) 0.8632 ( ) 0.8088 ( ) 0.6945
Hyd Cu Y1 0.3654 0.6311 ( ) 0.6162 ( ) 0.6474 ( ) 0.6881 ( ) 0.2106
Res Cu Y1 0.2899 0.3529 ( ) 0.3639 ( ) 0.4517 ( ) 0.3223 ( ) 0.3899
Cat Cu Y1 0.9500 0.9799 ( ) 0.7175 ( ) 0.8437 ( ) 0.8783 ( ) 0.6666

The uncertainty indicated is the standard deviation. Concentrations of a given metal with the same superscript letters in a given column are not significantly different at
p = 0.05.
 N. Kontoudakis et al. / Food Chemistry 229 (2017) 319328
sulfide in the wine is in the bound form (Table 2). The total, resid-
ual and cationic fractions of Cu at bottling showing the highest cor-
relation, followed by the total and cationic Cu after 1 year aging.
The bound hydrogen sulfide measured after 1-year correlates
poorly with the hydrophobic fraction of Cu at bottling
(r2 = 0.5204), but the correlation is improved against the
hydrophobic fraction of Cu at 1 year (r2 = 0.6311). However, the
magnitude of this latter correlation (being less than the other Cu
fractions) suggests that hydrogen sulfide may be bound to other
wine components (i.e., metals (Franco-Luesma & Ferreira, 2014)),
or that the smaller CuS particles in wine (Clark et al., 2015), com-
pared to the large CuS particles formed in model wine system (see
Section 3.1), may be partly measured in the cationic fraction of Cu.
The influence of the CuS particle size on the SPE fractionation war-
rants further investigation in future studies. The correlation
between the total Cu concentration and the free hydrogen sulfide
concentration (in both lg/L and percentage of total) was negative
as expected by the ability of Cu to convert free hydrogen sulfide
to bound hydrogen sulfide (Franco-Luesma & Ferreira, 2014). The
cationic fraction of total Cu had the highest magnitude of the neg-
ative correlation inferring the more of this fraction of Cu the less
free hydrogen sulfide.
Fe. In contrast to Cu, no loss in total Fe concentration was evi-
dent after the wine (without bentonite treatment) was aged for
1 year in bottle (Fig. 2a). This is consistent with Fe being associated
more with oxidative colloidal instability in wine rather than reduc-
tive colloidal instability. Furthermore, the concentration of Fe in
the wines are similar to the concentration at bottling and also
the concentrations in the juice (Fig. 2a and Table 1), as fermenta-
Fig. 2. The total Fe concentration (a) and proportion of fractionated forms (b) in
wines T1-T6 at bottling and after 1 year in low oxygen conditions. The uncertainty
shown corresponds to the standard deviation (n = 3).
325
tion is known to have little effect on the concentration of Fe, unlike
the outcome for Cu (Hsia, Planck, & Nagel, 1975).
The cationic fraction of Fe dominated the wines after 1 year as
was the case at bottling (Fig. 2b, Supplementary Fig. 4), and this
was true for both absolute concentrations and proportion. As noted
above (Section 3.1), the cationic fraction is the major fraction of Fe
(II) that was found for the model wine system containing added Fe
and supports the observation that Fe is mainly bound by organic
acids in wine conditions. A decrease was also evident in the pro-
portion and concentration of the residual Fe present in the wines,
significant for T1-T4, which accounts for the increase in the catio-
nic fraction of Fe (Fig. 2b). The presence of some Fe(III) at bottling,
induced by the total packaged oxygen, and then the gradual con-
version of Fe(III) to Fe(II) as the oxygen became depleted during
bottle aging, may have contributed to the decrease in residual frac-
tion of Fe. As noted in Section 3.1, the Fe(III) organic acid com-
plexes contributed more to residual Fe than Fe(II) organic acid
complexes, albeit at high total concentrations of Fe. A small
increase was evident in the hydrophobic fraction of Fe concentra-
tion and proportion in the samples with high Fe concentration
(T5-T6) (Fig. 2b, Supplementary Fig. 4).
Fe at bottling or after 1 year, and regardless of the fractionated
form, showed negligible correlation with total hydrogen sulfide,
methanethiol or any of the free and bound forms (data not shown).
For example, the correlation of total Fe at bottling with total hydro-
gen sulfide after a year in bottle was r2 = 0.0924.
3.4. The aging of wines with bentonite addition in low oxygen
conditions
Cu. The wines treated with bentonite exhibited significantly
lower Cu concentrations at bottling compared to those wines not
treated with bentonite as already reported (Rousseva et al., 2016)
(Fig. 3a). Bentonite would appear to be particularly efficient in
removing the Cu that survives the fermentation process, when uti-
lised with grape-derived protein still present in the wine. For the
wines with bentonite added, after 1 year aging in bottle, there were
slight (but not significant) decreases in total Cu concentrations for
wines A and B, whilst wine C had a significant loss albeit of low
magnitude (<0.04 mg/L) (Fig. 3a). The lower Cu concentration in
the wines treated with bentonite (Fig. 3a), in combination with
the lack of protein present in the wine, was most likely responsible
for the minimal Cu lost during bottle aging. Although a minor
amount of deposit was observed after 1-year in the bentonite-
treated wines, it was much less than observed for the wines bottled
without bentonite treatment.
After 1 year of aging there were decreased concentrations and
proportions of hydrophobic and cationic fractions of Cu in the
wines with bentonite treatment, while the concentration of resid-
ual Cu remained constant with a consequent increase in its propor-
tion (Fig. 3b, Supplementary Fig. 5). This residual fraction of Cu was
previously shown to have the strongest positive correlation with
oxygen consumption in the wine (Rousseva et al., 2016), and the
previous section (3.3) showed this it was correlated with hydrogen
sulfide production in the protein-rich wine. The largest loss in
hydrophobic fraction of Cu was for the bentonite treated wine with
the highest total Cu concentration (WC + bent, Fig. 3b, Supplemen-
tary Fig. 5). After bentonite treatment it must be noted that the
rank order of Cu concentrations in the wines after 1 year, no longer
reflects the concentration present in the wines at bottling, prior to
bentonite treatment or in the original juice (Table 1 vs Supplemen-
tary Fig. 5). Such an outcome is consistent with the ability of ben-
tonite to remove both Cu and proteins, the latter being a potential
source of de novo hydrogen sulfide formation, and hence limit for-
mation of Cu(I) sulfide that contributes to the hydrophobic fraction
of Cu. Without a source of hydrogen sulfide, the initial Cu(I) sulfide
326 N. Kontoudakis et al. / Food Chemistry 229 (2017) 319328
Fig. 3. The total Cu concentration (a) and proportion of fractionated forms (b) in
wines A-C, with and without bentonite treatment, at bottling and after 1 year in low
and high oxygen conditions. The uncertainty shown corresponds to the standard
deviation (n = 3).
present at bottling would gradually diminish presumably as oxy-
gen consumption occurs within the wine. Indeed, Table 2 shows
a trend to lower concentrations of total hydrogen sulfide and
methanethiol in the wines that underwent bentonite treatment.
Fe. The treatment of wine with bentonite is well known to
increase the concentration of Fe in the wines (Catarino et al.,
2008), and this was evident in this wine (Fig. 4a) as reported by
Rousseva et al. (2016). The higher Fe concentrations as induced
by bentonite fining were maintained in the wines after 1 year of
aging and unlike Cu, no loss of Fe was evident. Similar to the wines
without bentonite treatment, after 1 year of aging, the bentonite
treated wines showed slight decreases in the concentration of
the residual and hydrophobic fractions and increases in the catio-
nic fraction (Supplementary Fig. 6), the latter being the dominant
fraction in the wine. This result is particularly evident in the pro-
portional representation of the data (Fig. 4b).
3.5. The aging of wines in high oxygen conditions
Wines A-C, without bentonite treatment, and wine A, with ben-
tonite treatment, were aged in high oxygen conditions for 1 year.
Under these conditions, only wine C, with bentonite treatment,
and wine A, without bentonite treatment, showed significant
amounts of deposit (as assessed visually) in the bottle after 1 year.
Cu. In high oxygen conditions (Fig. 3a) no loss of Cu during aging
was evident, supporting the requirement of low oxygen conditions,
in combination with protein, to afford the loss of Cu in the form of
a protein haze. Of particular note was the near absence of the
hydrophobic fraction of Cu in the oxidised wines, and the domi-
Fig. 4. The total Fe concentration (a) and proportion of fractionated forms (b) in
wines A-C, with and without bentonite treatment, at bottling and after 1 year in low
and high oxygen conditions. The uncertainty shown corresponds to the standard
deviation (n = 3).
nance of the residual fraction of Cu both in terms of concentration
and proportion (Fig. 3b, Supplementary Fig. 5), apart from wine C
+ bent(oxidised), which had equal amounts of the cationic and
residual fractions of Cu. After 1 year aging in bottle, the concentra-
tion of the residual fraction of Cu was higher in the oxidised wines
than that for the wines aged in low oxygen conditions. The increase
in the residual fraction of Cu, the form with the strongest correla-
tion to oxygen consumption rates (Rousseva et al., 2016), is consis-
tent with the report that oxidation of wines can progressively
cause them to oxidise more rapidly (Singleton, 1987). In this pre-
sent study, it is possible that the loss of the hydrophobic fraction
of Cu during oxidation may be a consequence of Cu(I) sulfide being
gradually converted to Cu(II) and oxidised products of sulfide (such
as elemental sulfur, sulfate or polysulfides (Kreitman et al., 2016)),
and the released Cu(II) then binding with other wine components
to form the residual fraction of Cu. At this stage it is not certain
which components of wine contribute to Cu forming the residual
fraction of Cu and this is the subject of an on-going study. One pos-
sibility is that the oxidation of wine can generate degradation
products of organic acids (i.e., oxalic acid (Clark, Prenzler, &
Scollary, 2007), which subsequently may bind Cu(II) sufficiently
to allow its elution through the cation exchange column
(Zagorchev & Balushev, 1960). Consequently, future work may
investigate whether the proportion of residual Cu fraction in a
wine may provide an indication for the extent of oxidation of a
wine. The oxidised samples all have lower total hydrogen sulfide
and methanethiol concentrations than the corresponding non-
oxidised samples (Table 2).
Fe. For the wine aged in high oxygen conditions, the total Fe
concentration remained unchanged from their initial level at
 N. Kontoudakis et al. / Food Chemistry 229 (2017) 319328
bottling apart from the sample with highest Fe concentration,
which decreased dramatically (by 60%) (Fig. 4a). This is consis-
tent with the ability of a high Fe concentration in wine to induce
haze formation in oxidative conditions, which favors the conver-
sion of Fe(II) to Fe(III) and its interaction with phosphate in white
wines to afford a deposit (Joslyn & Lukton, 1953). In the oxidised
wines, the concentration and proportion of the residual fraction
of Fe dominated at the cost of the cationic fraction, and the
hydrophobic fraction was largely absent (Fig. 4b, Supplementary
Fig. 6). This result suggests that oxidative conditions favors the
conversion of Fe into its residual fraction, regardless of any precip-
itation of Fe in the form of a haze. As observed in Section 3.1, the
residual fraction of Fe can be linked to an increase in the oxidation
state of Fe (Fe(III) vs Fe(II)), however this was only at the high con-
centration of Fe and not the lower concentration. Fig. 4b shows
more Fe in the residual fraction in the oxidised samples regardless
of the difference in the total Fe concentration of the samples
(Fig. 4a). It is possible that the additional components of wine in
comparison to the model wine, including phenolic compounds
and organic acids other than tartaric acid, may enhance the mea-
surement of Fe(III) as the residual fraction.
The results present a novel approach to follow changes in broad
metal classification, based on a SPE fractionation process, during
oxidative and reductive aging of wine. It allows insight into the
impact of the initial concentration of metals in the juice prior to
fermentation, and also wine protein, on metal fractionation during
wine aging. The experimental strategy developed in this study has
shown that the initial juice concentrations of metals, bentonite
treatment and extent of oxygen supply to wines during bottle
aging can influence both the total and the fractionated metal con-
centrations after 1 year. Low oxygen conditions in wines with pro-
tein present induced a significant loss of Cu particularly for those at
elevated concentrations and a general increase of the hydrophobic
fraction, consistent with the formation of Cu(I) sulfide during
aging. The concentration of residual and cationic Cu fractions at
bottling correlated well with the total hydrogen sulfide in the wine
after 1 year. Alternatively, no loss of total Cu or increase in the
hydrophobic fraction of Cu was evident in bentonite fined wines
but instead an increase in the residual fraction of Cu was favored.
Oxidative storage of the wines exhibited no loss of total Cu and
caused a significant increase in the residual fraction and essentially
eliminated the hydrophobic fraction.
In contrast to Cu, Fe was lost in oxidative conditions when at
high concentrations rather than in low oxygen conditions. The
cationic fraction of Fe dominated when aging occurring in low oxy-
gen conditions, regardless of the presence of bentonite, and in high
oxygen conditions the residual fraction of Fe was favored, and
linked to the formation of Fe(III).
Acknowledgements
This project (NWG1401) was funded by Wine Australia with
funds from Australian grape growers and winemakers with match-
ing funds from the Commonwealth of Australia. The authors
declare that there are no conflicts of interest. The small lot wine-
making was conducted by Ms Michaela Rousseva under the super-
vision of NWGIC experimental winemaker Campbell Meeks.
Appendix A. Supplementary data
Supplementary data associated with this article can be found, in
the online version, at http://dx.doi.org/10.1016/j.foodchem.2017.
02.065.
327
References
Bekker, M. Z., Smith, M. E., Smith, P. A., & Wilkes, E. N. (2016). Formation of
hydrogen sulfide in wine: Interactions between copper and sulfur dioxide.
Molecules, 21(9), 1214.
Catarino, S., Madeira, M., Monteiro, F., Rocha, F., Curvelo-Garcia, A. S., & Bruno De
Sousa, R. (2008). Effect of bentonite characteristics on the elemental
composition of wine. Journal of Agricultural and Food Chemistry, 56, 158165.
Clark, A. C., Dias, D. A., Smith, T. A., Ghiggino, K. P., & Scollary, G. R. (2011). Iron(III)
tartrate as a potential precursor of light-induced oxidative degradation of white
wine: Studies in a model wine system. Journal of Agricultural and Food Chemistry,
59, 35753581.
Clark, A. C., Grant-Preece, P., Cleghorn, N., & Scollary, G. R. (2015). Copper(II)
addition to white wines containing hydrogen sulfide: Residual copper
concentration and activity. Australian Journal of Grape and Wine Research, 21,
3039.
Clark, A. C., Kontoudakis, N., Barril, C., Schmidtke, L., & Scollary, G. R. (2016).
Measurement of labile copper in wine by medium exchange stripping
potentiometry utilising screen printed carbon electrodes. Talanta, 154, 431437.
Clark, A. C., Prenzler, P. D., & Scollary, G. R. (2003). The role of copper(II) in the
bridging reactions of (+)-catechin by glyoxylic acid in a model white wine.
Journal of Agricultural and Food Chemistry, 51, 62046210.
Clark, A. C., Prenzler, P. D., & Scollary, G. R. (2007). Impact of the condition of storage
of tartaric acid solutions on the production and stability of glyoxylic acid. Food
Chemistry, 102, 905916.
Clark, A. C., & Scollary, G. R. (2006). Medium exchange stripping potentiometry for
the measurement of labile copper in white wine. Electroanalysis, 18,
17931799.
Clark, A. C., Wilkes, E. N., & Scollary, G. R. (2015). Chemistry of copper in white wine:
A review. Australian Journal of Grape and Wine Research, 21, 339350.
Danilewicz, J. C. (2003). Review of reaction mechanisms of oxygen and proposed
intermediate reduction products in wine: Central role of iron and copper.
American Journal of Enology and Viticulture, 54(2), 7385.
Danilewicz, J. C. (2007). Interaction of sulfur dioxide, polyphenols, and oxygen in a
wine-model system: Central role of iron and copper. American Journal of Enology
and Viticulture, 58, 5360.
Danilewicz, J. C. (2014). Role of tartaric and malic acids in wine oxidation. Journal of
Agricultural and Food Chemistry, 62, 51495155.
Danilewicz, J. C. (2016a). Chemistry of manganese and interaction with iron and
copper in wine. American Journal of Enology and Vitculture. http://dx.doi.org/
10.5344/ajev.2016.16033.
Danilewicz, J. C. (2016b). Fe(II):Fe(III) ratio and redox status of white wines.
American Journal of Enology and Vitculture, 67, 146152.
Danilewicz, J. C., Seccombe, J. T., & Whelan, J. (2008). Mechanism of interaction of
polyphenols, oxygen, and sulfur dioxide in model wine and wine. American
Journal of Enology and Vitculture, 59, 128136.
Franco-Luesma, E., & Ferreira, V. (2014). Quantitative analysis of free and bonded
forms of volatile sulfur compounds in wine. Basic methodologies and evidences
showing the existence of reversible cation-complexed forms. Journal of
Chromatography A, 1359, 815.
Franco-Luesma, E., & Ferreira, V. (2016). Formation and release of H2S,
methanethiol, and dimethysulfide during the anoxic storage of wines at room
temperature. Journal of Agricultural and Food Chemistry, 64, 63176326.
Hsia, C. L., Planck, R. W., & Nagel, C. W. (1975). Influence of must processing on iron
and copper contents of experimental wines. American Journal of Enology and
Viticulture, 26(2), 5761.
Iland, P., Bruer, N., Edwards, G., Weeks, S., & Wilkes, E. (2004). Chemical analysis of
grapes and wines: Techniques and concepts. Campbelltown (Australia): Patrick
Iland Wine Promotions PTY LTD.
Iland, P., Bruer, N., Ewart, A., Markides, A., & Sitters, J. (2004). Monitoring the
winemaking process from ggrapes to wine techniques and concepts. Campbelltown
(Australia): Patrick Iland Wine Promotions PTY LTD.
Joslyn, M. A., & Lukton, A. (1953). Prevention of copper and iron turbidities in wine.
Hilgardia, 22(14), 451533.
Kreitman, G. Y., Danilewicz, J. C., Jeffrey, D. W., & Elias, R. J. (2016). Reaction
mechanisms of metals with hydrogen sulfide and thiols in model wine. Part 1:
Copper-catalysed oxidation. Journal of Agricultural and Food Chemistry, 64,
40954104.
Li, H., Guo, A., & Wang, H. (2008). Mechanisms of oxidative browning of wine. Food
Chemistry, 108, 113.
Lopes, P., Saucier, C., Teissedre, P.-L., & Glories, Y. (2006). Impact of storage position
on oxygen ingress through different closures into wine bottles. Journal of
Agricultural and Food Chemistry, 54, 67416746.
Lukton, A., & Josyln, M. A. (1956). Mechanism of copper casse formation in white
table wine. II. Turbidimetric and other physico-chemical considerations. Journal
of Food Science, 21, 456476.
Morie, G. P., & Sweet, T. R. (1964). Anion exchange separations of iron(II) and iron
(III) in tartrate medium. Analytical Chemistry, 36, 140142.
Peterson, R. G., Josyln, M. A., & Durbin, P. W. (1958). Mechanism of copper casse
formation. III. Source of the sulfur in the sediment. Journal of Food Science, 23,
518524.
Pohl, P., & Prusisz, B. (2009). Application of tandem column solid phase extraction
and flame atomic absorption spectrometry for the determination of inorganic
and organically bound forms of iron in wine. Talanta, 77, 17321738.
328 N. Kontoudakis et al. / Food Chemistry 229 (2017) 319328
Pohl, P., & Sergiel, I. (2009). Evaluation of the total content and the operationally
defined species of copper in beers and wines. Journal of Agricultural and Food
Chemistry, 57, 93789384.
Ribreau-Gayon, J., Glories, Y., Maujean, A., & Dubourdieu, D. (2006). Handbook of
enology: The chemistry of wine stabilisation and treatments (Vol. 2) Chichester,
England: Wiley.
Rousseva, M., Kontoudakis, N., Schmidtke, L., Scollary, G. R., & Clark, A. C. (2016).
Impact of wine production on the fractionation of copper and iron in
Chardonnay wine: Implications for oxygen consumption. Food Chemistry, 203,
440447.
Siebert, T. E., Solomon, M. R., Pollnitz, A. P., & Jeffery, D. W. (2010). Selective
determination of volatile sulfur compounds in wine by gas chromatography
with sulfur chemiluminescence detection. Journal of Agricultural and Food
Chemistry, 58, 94549462.
Singleton, V. L. (1987). Oxygen with phenols and related reactions in musts, wines,
and model systems: Observations and practical implications. American Journal
of Enology and Viticulture, 38(1), 6977.
Timberlake, C. F. (1964). Iron-tartrate complexes. Journal of the Chemical Society,
12291240.
Ugliano, M., Dieval, J.-B., Siebert, T., Kwiatkowski, M., Aagaard, O., Vidal, S., et al.
(2012). Oxygen consumption and development of volatile sulfur compounds
during bottle aging of two Shiraz wines. Influence of pre- and postbottling
controlled oxygen exposure. Journal of Agricultural and Food Chemistry, 60,
85618570.
Ugliano, M., Kwiatkowski, M., Vidal, S., Capone, D., Siebert, T., Dieval, J.-B., et al.
(2011). Evolution of 3-mercaptohexanol, hydrogen sulfide, and methyl
mercaptan during bottle storage of Sauvignon blanc wines. Effect of
glutathione, copper, oxygen exposure, and closure-derived oxygen. Journal of
Agricultural and Food Chemistry, 59(6), 25642572.
Viviers, M. Z., Smith, M., Wilkes, E., & Smith, P. A. (2013). Effects of five metals on the
evolution of hydrogen sulfide, methanethiol and dimethyl sulfide during
anaerobic storage of Chardonnay and Shiraz wine. Journal of Agricultural and
Food Chemistry, 61, 1238512396.
Waterhouse, A. L., Frost, S., Ugliano, M., Cantu, A. R., Currie, B. L., Anderson, M., et al.
(2016). Sulfur-dioxide-oxygen consumption ratio reveals differences in bottled
wine oxidation. American Journal of Enology and Vitculture, 67, 449459.
Waters, E. J., Alexander, G., Muhlack, R., Pocock, K. F., Colby, C., ONeill, B. K., et al.
(2005). Preventing protein haze in bottled white wine. Australian Journal of
Grape and Wine Research, 11, 215225.
Wilkes, E. N., Hart, A., Clark, A. C., Blackman, J. W., & Scollary, G. R. (2013). The
compositional and sensory impact of copper fining on bottle aging of Riesling.
In P. Jeandet (Ed.), Proceedings of the 8th Symposium: In Vino Analytica Scientia,
(pp. p. 143 (http://www.csu.edu.au/nwgic/about-us/our-people/profiles/
research-staff/andrew-clark)). Reims, France: The University of Reims.
Zagorchev, B., & Balushev, B. (1960). Chromatographic separation of iron(II) from
iron(III). I. Oxalic acid as complexing agent. Godishnik na Khimiko-
Tekhnologicheskiya Institut, 7, 2535.