Introduction

The use of ceramic vessels has a history of tens of millennia. These vessels were most likely used to store and transport water, various types of liquids, grain and other foods, and later wine (Twede, 2002). MacNeil (2015) mentions ceramic jars as one of the earliest vessels for transporting wine in the ancient world. Archaeological excavations in the Armenian highlands have discovered winemaking complexes (in Adablur, Karmir Blur, Areni and other sites), which describe the primary phases and patterns of wine production. Of particular importance are the hundreds of grape sprouts, barks, fruit stems, seeds, pressed fruits and pulp remains discovered in the Areni-1 cave in the Vayots Dzor Province of Armenia, as well as the oldest known wine-producing facility, which is at least 6000 years old (Hovhannisyan et al., 2017). Over the centuries, ceramic pots gradually gave up their place, both for transport or storage and for vinification, to other types of containers like oak barrels, concrete and stainless steel tanks and glass containers. However, ceramic pots remained a container of choice in countries such as Armenia and Georgia, but they are also used to some extent in Greece, Italy, France and Portugal (Harutyunyan et al., 2022; Hovhannisyan et al., 2017; McGovern et al., 2017). Nowadays, there is a growing interest in using ceramics for wine production due to the popular natural wine movement, which practices traditional, even ancient, winemaking techniques. Despite this interest, there are few analytical studies about the interaction between ceramic containers and wine, in particular concerning the migration of elements from the ceramic to the wine (Shackelford and Shackelford, 2020; Weltman, 2018).

Ceramic pots, like concrete tanks and wood (mainly oak) barrels, are known for their porous structure. It makes them favourable for oxygen transfer which plays an important role from the perspective of oxidative ageing (Work, 2014). Unlike stainless steel, wood barrels, ceramic pots and concrete tanks are thermally stable due to lower thermal conductivity (0.17 Wm^-1K^-1; 0.5 Wm^-1K^-1; 20.5 Wm^-1K^-1 for oak, concrete and stainless steel, respectively) and the thickness of their walls, which makes them stable to temperature fluctuations (Gil i Cortiella et al., 2021; Montalvo et al., 2021). Stainless steel and glass containers can claim to be the most flavour-neutral containers, being chemically stable, thus, preserving the flavour and aroma properties of wine, producing a more stable and homogenous wine (Gil i Cortiella et al., 2021; Montalvo et al., 2021). Opinions on flavour neutrality are not consensual for concrete tanks and ceramic pots, particularly if they are not coated (e.g., with epoxy paint for concrete (Montalvo et al., 2021), beeswax (Barisashvili, 2011; Johnson et al., 2013), or pine pitch for ceramic (Romanus et al., 2009). Wood barrels stand out for their influence on wine flavour. When interacting with wood, the wine is enriched with a remarkable diversity of phenolic and aromatic compounds from the wood, thereby influencing the organoleptic profile and increasing the antioxidant stability (oak barrels) of the wine (Gougeon et al., 2009; Martínez-Gil et al., 2022; Nikolantonaki et al., 2019).

Summarising the available information on the interactions between containers and wine, we can state that, in contrast to the influence of wood barrels on the physicochemical and organoleptic (texture and taste) properties of wine, which has been widely studied, the effects of alternative containers are poorly studied, and little is known about their impact on the chemical composition of the resulting wines (Cabrita et al., 2018; Gil i Cortiella et al., 2021; Guerrini et al., 2022). However, it can be said with confidence that the type of tank material plays an important role in directly impacting the quality, taste, and characteristics of wines. Wood and ceramic containers provide micro-oxygenation of wine thanks to their porous structure. Wood barrels are also a source of extra tertiary aromas. Stainless steel tanks allow for keeping the natural taste and aromas of the wine (Gil i Cortiella et al., 2021; Nevares and del Alamo-Sanza, 2021).

The main objective of this study is to characterise the interactions between Armenian clay-based ceramics and model wine in terms of wine composition and redox processes occurring during ageing. Quantification of the elements present in the model wine was determined by inductively coupled plasma atomic emission spectroscopy (ICP-AES). Iron redox processes were investigated by ¹H time-domain NMR (TD NMR).

Materials

1. Clay-based ceramics

The ceramic tablets (Figure 1) used for the experiments were prepared by mixing clay and sand (seeds smaller than 4 µm) in a ratio of 2:1, respectively. Armenian clay from Shahumyan village (chemical compositions of two samples of clay deposit and one sample of body clay (mixture of clay/sand (2:1) and water) reported in Figure 2 and detailed in Supplementary data, Table S1) was used as the starting material for the elaboration of ceramic tablets. The tablets were shaped cylindrically using a homemade plastic extruder. After shaping, the tablets were air-dried at room temperature for 3 days. The firing protocol is composed of three steps. The first step involves heating the samples for 4 hours from ambient temperature (20 °C) to 600 °C (2.4 °C/min). The second step is a fast temperature increase from 600 °C to 930 °C (20 °C/min) followed by firing at 930 °C for 2 hours. The final step is free cooling in the furnace. This firing protocol mimics the technique used in Armenia to produce such jars (arm. կարաս [karas]).

Figure 1. The tablets and their average dimensions before (a) and after (b) heat treatment

Save View full size Expand inline Collapse inline

Figure 2. Representative histogram of XRF analyses

Save View full size Expand inline Collapse inline

Methods

Results and Discussion

To study the interaction of ceramics with model wine, we choose to insert a piece of ceramic (Figure 1) into the model wine instead of studying the ageing of the model wine into a ceramic vessel. The principal reasons are (i) the scale of the experiments involved if real containers have to be used, which prevents in situ experiments, (ii) the impact of the stopper (e.g., ceramic plate sealed with beeswax) and (iii) the reproducibility issues, and the additional complexity of the data analysis (due to the impact of an external surface in contact with atmosphere or soil (if buried). Therefore, this study focuses on the interaction of the internal surface of a ceramic vessel in contact with the wine, removing the impact of the external surface (e.g., dioxygen migration through the ceramic). To compare with a real situation, besides the headspace volume, one critical dimension in this study is the hydraulic ratio of the ceramic that we consider by the ratio of the wine volume over the surface of the ceramic (R_V/S).

2. Elemental composition of model wine in contact with a ceramic determined by ICP-AES

2.1. Elements migration and pH evolution

When the model wine is in contact with the ceramic, significant migration of elements from the ceramic to the model wine is observed (Figure 3) (complete ICP-AES data are compiled in the Supplementary data, Table S3). It should be noted that the migration behaviour of elements changes over time, but if we try to rank the most abundant elements, a list could be ordered as follows: Ca>>Na>Al>K>Fe>Mg>S>Sr. Compared to the other elements, from the first instants, calcium stands out with an obviously high concentration (up to 193 mg/L, 4.8 mM, after 11 hours). Overall, the concentrations seem to evolve with at least three-time scales: hours, days and months. For most elements, there is a fast release during the first few hours, and then the rate decreases. From there, three rough trends of evolution can be suspected: (i) a decrease after 4 days (Al, Mg, Fe); (ii) a rather steady state situation at a constant level over 16 months (Na, S); (iii) an increase over a long time period (K, Cr).

Figure 3. Concentrations (log scale) of the six most abundant elements released from 1 g of ceramic tablet in contact with the model wine. The contact time varies from 1 h (1 hour) up to 16 m (16 months); on the x-axis, d stands for days. The concentrations of elements in the model wine were subtracted.

Save View full size Expand inline Collapse inline

Simple principles of liquid chemistry were used to approximate the evolution of the dominant element Ca, which concentration evolution is reported in Figure 4. The raw clay contains a lot of calcite (CaCO₃) (~15–20 % w/w in the clay and ~10 % in the clay/sand mixture, Figure 2 and Table S1 in Supplementary data). We assume that calcite is the main raw source of calcium that migrates into the model wine. Upon firing, calcite forms calcium oxide (CaO) (Nikulshina et al., 2007; Sanchez-Garmendia et al., 2021; Stanmore and Gilot, 2005), which is also expected to be significantly present in the ceramic.

CaCO3→~800 °CCaO+CO2. (1)

The calcium oxide reacts rapidly with liquid water (when exposed to wine and probably also, to a lesser extent, with atmospheric water during storage of the ceramic) to form calcium hydroxide (Ca(OH)₂) (Criado et al., 2014; Serris et al., 2011).

CaO+H2O→Ca(OH)2. (2)

This hydroxide is known to be slightly soluble in water (~1.65 g/L, 22 mM) (CRC, 2004; Miller and Witt, 1929) but is considered a strong base (pkb₁ = 2.43, pkb₂ = 1.41) (CRC, 2004). All measured Ca concentrations are below 5 mM (200 mg/L), which is below the solubility of Ca(OH)₂. It is difficult to estimate the dissolution kinetics, but it should be noted that the model wines analysed should be at equilibrium because, after separation from the ceramics, the model wine was kept in a flame-sealed tube at a controlled temperature for quite a very long period (from 16 months for the first samples to several weeks for the last ones). We can, therefore, assume that all available calcium hydroxide present in the model wine is dissolved and dissociated, in particular for the first experiments with small contact times (< 10 hours) where no precipitation or solid phase was observed. According to Eq. (3), the dissociation of calcium hydroxide should produce an increase in the pH of the model wine, as observed in Figure 4.

CaOH2s→ Ca2+aq+2OH-aq. (3)

Assuming a total dissociation of the base, for the first samples, the pH shift from 2.35 to 2.5 would correspond to a consumption of 1.3 mM of H₃O⁺, which results from the dissociation of 0.65 mM of Ca(OH)₂ (corresponding to 26 mg/L of Ca²⁺). When the pH reaches 7, the same calculation gives a dissociation of 2.2 mM (corresponding to 88 mg/L of Ca²⁺) of Ca(OH)₂. It is reassuring to observe that despite the rough approximations, the estimated values are of the same order as the concentrations measured by ICP-AES. Albeit the estimations for the first samples (26 mg/L) present the maximum discrepancy with the observed values (from 80 to 200 mM), it might be easier to discuss it since no precipitation or solid phase was observed for the first contact times, and the pH remained fairly stable. The simplest explanation for the underestimation of calcium concentration would be that more Ca is dissolved from other mineral sources. XRD revealed the presence of other calcium-containing-minerals in the ceramic: diopside (CaMgSi₂O₆), gehlenite (Ca₂Al(AlSi)O₇), but also plagioclase (labradorite (Ca_0.6Na_0.4(Al_1.6Si_2.4)O₈), bytownite (Ca_0.8Na_0.2(Al_1.6Si_2.4)O₈) and anorthite (CaAl₂Si₂O₈)).

Figure 4. Blue circles: evolution of the Calcium concentration in model wine measured by ICP-AES (in mg/L per g of ceramic) versus the contact time between the model wine and the ceramic (model wine subtracted).

Save View full size Expand inline Collapse inline

Red triangles: evolution of the pH. Due to the small volumes of model wine available, measurements were done by pouring a few drops of model wine onto pH indicator strips, which are subject to significant uncertainties estimated at ± 0.5 and reported by error bars. The dotted line corresponds to the pH of the starting model wine (2.35).

2.2. The dissolution of ceramic minerals in model wine

The dissolution of ceramics and minerals such as diopside, hematite, magnetite and anorthite has been studied by several researchers (Bennett et al., 1988; Knauss et al., 1993; Knauss and Wolery, 1986; Lüttge et al., 1999; Torrent, 1987; Welch and Ullman, 1993). Most of these studies have focused on mineral dissolution mechanisms as a function of pH and temperature. It is worth mentioning that results from the literature review may reveal noticeable discrepancies in dissolution constants depending on surface area or particle size (Bennett et al., 1988; Bragança and Bergmann, 2006). Some of these dissolution rates from the literature for a few minerals are given in Table 2. Figure 5 compares the elemental dissolution (Figure 3) with the simulation of the mineral dissolution according to Table 2. For the simulation, a hypothetical tablet with a mass of 679 mg made of 100 % of the mineral is considered. Figure 5 shows that the measured amount of dissolved element in the model wine is consistent with the dissolution constants reported in the literature. For example, a tablet of anorthite with the same surface area as the ceramic tablet can dissolve at 25 °C, pH 3, in one hour, up to 5.28 mg Ca. Compared to the amount of calcium (0.266 mg) measured after one hour of contact of the ceramic with the model wine, this suggests that if 5 % of the ceramic is anorthite (homogeneously distributed), this would be sufficient to produce the amount of calcium observed in the model wine (193 mg/L).

Figure 5. The mass of elements (µg) released from one tablet, measured in 2.8 mL of model wine by ICP-AES (full circles).

Save View full size Expand inline Collapse inline

Simulated dissolution from literature data of an equivalent tablet (same surface area, average mass of 679 mg) are reported with dotted lines. The upper horizontal dashed black line figures the level corresponding to a mass of 679 mg.

Table 2. Dissolution constant (Kd measured at 25 °C of some minerals identified in the ceramic. The reported values are generally obtained by a flow experiment. Quartz is given for comparison. l stands for low (not measured) pH. Dissolutions constants in bold were used for the calculations in Figure 5.

View popup Expand inline Collapse inline

Mineral	Formula	pH	Kd (mol cm-2 s-1)	Reference
Anorthite	CaAl2Si2O8	3	2.7 10-12	Lüttge et al. (1999)
Bytownite	Ca0.8Na0.2(Al1.6Si2.4)O8	2–3 7	10-12 10-15	Welch and Ullman (1993)
Diopside	CaMgSi₂O₆	2–3	2.9 10-14	Knauss et al. (1993)
Albite	NaAlSi3O8	4 6	4.7 10-17 4.6 10-17	Knauss and Wolery (1986)
Hematite	Fe2O3	l	5.8 10-14	Torrent (1987)
Iron (III) hydroxide	Fe(OH)3	4–6	10-13	Deng (1997)
Quartz	SiO2	2–3	3.6 10-17	Bennett et al. (1988)

In addition, it is worth noting that Qin et al. (2016), using a very similar material (ceramsites), have evidenced that in conjunction with calcium release from calcium hydroxide, minerals (like gehlenite and anorthite) are also important dissolution sources for calcium. The results of Qin et al. (2016) need to be further compared to our results, and a detailed analysis of the calcium dissolution during the first hours may be engaged. They have studied the release of Ca²⁺ and OH^- from ceramsites containing anorthite and gehlenite prepared from waste lime mud. Qin et al. (2016) have modelled (at initial pH 6.5) the dissolution of Ca²⁺ by an Avrami law (fixing the Avrami constant n to 0.9019):

CtCmax= x=1-e-k tn, (4)

where k is the kinetic constant, n is the characteristic constant of the solid (Avrami constant), t is the time, x is the fractional conversion, (C(t) is the concentration at time t and Cmax is the maximal concentration).

They obtained kinetic coefficients from 0.0077 to 0.0125 min^-n. Figure 6 reports the fits of the concentration evolution of some elements according to Eq. (4), and the numerical results for the most abundant nuclei are reported in Table 3. Most elements seem to have the same kinetic constants (around 0.017), indicating a homogenous dissolution. Potassium and, to a lesser extent, sodium (but the incertitude is relatively large) reveal a lower k value. This apparent decrease in the kinetic constant could be associated with the precipitation of potassium bitartrates occurring during more than 16 months of storage of the model wines with ceramic tablets (before ICP-AES analyses) (Coulter et al., 2015).

It is worth noting that in the experiments, the ratio of the model wine volume over the ceramic surface in contact with the wine (R_V/S) is less than 10 mm, which has to be compared with real containers. For example, R_V/S is 15.7 mm for a Bordeaux-type bottle (0.75 L), 128 mm for a Bordeaux barrel (225 L) and 186 mm for an antic Armenian pithos (660 L) (Hovhannisyan et al., 2017; p. 102, Fig. 138). For comparison between different containers, the dissolution rate should be scaled accordingly to the R_V/S ratio.

Figure 6. Refinement with an Avrami law:

Save View full size Expand inline Collapse inline

Table 3. Parameter of the kinetic Avrami law for the most abundant dissolved nuclei. Cmax is reported for 1 mg of ceramic. The number in the bracket is the refinement error on the precedent digit (0.248(9) means 0.248 ± 0.009).

View popup Expand inline Collapse inline

Element	Cmax (mg/L)/min.mgceram	K (t-n) n = 0.9019	Element	Cmax (mg/L)/min.mgceram	K (t-n) n = 0.9019
Ca	0.248(9)	0.016(3)	S	0.0047(2)	0.017(4)
Na	0.037(8)	0.010(7)	Sr	0.00072(2)	0.015(2)
K	0.021(2)	0.0049(7)	Mn	0.00028(2)	0.018(5)
Mg	0.0204(9)	0.018(4)	Zn	0.00024(7)	0.01(1)
Al	0.019(2)	0.017(7)	V	0.00011(1)	0.023(8)
Fe	0.010(2)	0.02(1)

3. In situ analysis of the model wine in contact with the ceramic by TD NMR

The dissolution of the minerals from the ceramics can impact the characteristics of the wine. It may be responsible for a pH increase because of the acidic attack of the wine on the ceramic, and the dissolution of metallic elements (Fe, Cu…) may also promote the oxidation of ethanol. In practice, to minimise these possible effects, the surface of the ceramic is usually covered by a glassing formed during the firing of the ceramic at a higher temperature (sometimes with the preparation of the surface) (Zhang and Zhang, 2014), or traditionally, by an organic layer applied after firing (beeswax, pine pitch) (Barisashvili, 2011; Romanus et al., 2009). The effect of a beeswax coating is illustrated by time-domain NMR measurement. Time domain NMR has the advantage of following the kinetic of dissolution in situ and, additionally, sheds light on the complex physicochemical evolution occurring in wine in the presence of a ceramic.

3.1. Contact of the ceramic with hydroalcoholic and tartaric solutions

The analysis of the proton spin–lattice relaxation in wine and hydroalcoholic solutions has been reported elsewhere (Bodart et al., 2020; Bodart et al., 2022), and Figure 7 shows the effect of the immersion of a ceramic in water, hydroalcoholic solution and tartaric acid solution. The physical act of immersion does not change the relaxation rate of the solution, as shown by the relaxation rate of the water protons from water and hydroalcoholic solution in contact with a ceramic. However, when the solution is acidic (tartaric acid solution), the relaxation rate increases rapidly as a result of an acidic attack of the ceramic, which releases some paramagnetic elements that enhance the relaxation rate.

Figure 7. The relaxation rate of water (green circles), hydroalcoholic solution (blue diamonds) 12 % v/v, and tartaric acid solution (purple triangles) before (up to 2 hours) and after contact (after 2 hours) with ceramic.

Save View full size Expand inline Collapse inline

3.2. Contributions of released paramagnetic elements to the proton relaxation rate of the model wine

ICP-AES has shown that the dissolution of the ceramic is susceptible to releasing all types of elements present in the ceramic. Amongst them, paramagnetic nuclei are of prime importance for the interpretation of TD-NMR experiments because they are responsible for the generally dominant dipolar paramagnetic relaxation mechanisms of the water protons of the solutions (Solomon, 1955). Interpretation of relaxograms can be difficult because of the variety of paramagnetic elements present in the model wine (Cd, Co, Cr, Cu, Fe, Mn, Mo, Ni, Ti and V) that have distinct relaxivity (i.e., the "power of relaxation" of paramagnetic nuclei on water protons, defined in solution as the ratio of the proton relaxation rate over the concentration of the paramagnetic element, Supplementary data, Table S3) (Bertini et al., 2017). However, in conjunction with the concentration of the element determined by ICP-AES, it is possible to estimate an effective contribution of the paramagnetic element present in the model wines to the ¹H relaxation rate of the water molecules of the model wine (Figure 8).

Figure 8. Ratios (in percent) of paramagnetic ions concentration with respect to the Fe concentration (extracted for ICP-AES data) versus ageing time, with ponderations relative to their relaxivity (Supplementary data, Table S2). Element concentrations from the model wine were subtracted. It is assumed that all elements are in the reported oxidation states. The horizontal dotted black line represents the calculated quantity for iron (100 %).

Save View full size Expand inline Collapse inline

Figure 8 reveals that under our experimental conditions, Fe³⁺ (dotted black line) is the main source of relaxation. The second contribution is Mn²⁺ (continuous black line) which has a maximum contribution (8.8 %) at t = 48 h. All other paramagnetic elements would contribute less than one-hundredth of the Fe³⁺ contribution. When the ratios become greater than 1, in particular for chromium and vanadium at long times, it is associated with a decrease in the iron concentration in the solution because of iron precipitation, whilst chromium and vanadium do not form solids.

Additionally, for the interpretation of the relaxogram, it is important to consider that the ¹H relaxation rate of water molecules in the model wine (R1) is linearly dependent on the concentration of the paramagnetic element:

R1 = R0 +rMM, (5)

where R0 is the relaxation rate of the solution without a paramagnetic element. rMandM are the relaxivity and the concentration of the paramagnetic element M, respectively. Figure S2 in Supplementary data reports the corresponding titration curve of Fe³⁺ in the model wine. The linear regression is numerically given by

R1 = 0.4783 + 0.0651Fe3+, (6)

where R1 is expressed in s^-1 and the concentration in mg/L (numbers in brackets are the incertitude given by the linear regression on the previous digit).

3.3. Iron dissolution

For comparison, Figure 9 reports the measured iron concentrations by ICP-AES and TD NMR techniques (a discussion of the relaxogram is reported in Section 3.4). For the first few hours, despite significant differences between the methods, the measured concentrations were fairly consistent. The equivalent Fe³⁺ concentrations measured by TD NMR are slightly higher than the concentrations measured by ICP-AES. Besides the possible variability between ceramic tablets, this overestimation could be attributed to the contribution of other paramagnetic ions to the relaxation rate of water molecules. For the comparison over long contact times, there are two major aspects to be considered: (i) unlike TD NMR, ICP-AES is not an in situ experiment and even if the model wine was in contact for 100 hours with the ceramics, for example, the ICP-AES measurements were performed more than 16 months later. The model wine, which was kept under controlled conditions in a sealed bulb, had a long period of time to evolve and reach an equilibrium state over the formation of various solid tartrates. (ii) with respect to the relaxogram, for long periods of time, the dissolved dioxygen concentration decreases (e.g., because of ethanol oxidation), leading to a reduction of ferric to ferrous ions, which are significantly less effective relaxation agents and giving low relaxations rate for the model wine.

Figure 9. (a) Comparison of the equivalent ferric ion concentration calculated from the relaxogram (blue) and the ICP measurement (green). Statistics performed on all the ceramic tablets used for the ICP experiment give an average mass of 680

Save View full size Expand inline Collapse inline

3.4. Relaxograms of model wine after coating the ceramic tablets with beeswax

Figure 10 compares the relaxogram of ceramic tablets uncovered or covered with one or two layers of beeswax. It displays two vertical axes, the measured relaxation rates on the right and its conversion to an equivalent Fe³⁺ concentration on the left, according to Eq. (6). First of all, it is necessary to discuss the relaxogram of the uncoated ceramic. Relaxograms of model wine in contact with a ceramic are rich in various information but may be complex to interpret; a detailed description is beyond the scope of this article and is reported to another one. Nevertheless, one can consider that the relaxation rate of water molecules depends on (i) the ratio of the volume of model wine over the surface of the ceramic, (ii) the amount of available dioxygen (dissolved and in the headspace), (iii) the pH of the solution. When the ceramic is immersed in the model wine, there is an immediate dissolution leading to the release (among many other elements) of Fe³⁺ from the dissolution of hematite, that is estimated to be 5 % of the ceramic. This induces a sharp increase in relaxation up to ~10 hours when the first maximum is reached. The dissolution of Fe³⁺ does not stop or decrease at 10 hours, but it continues and will decrease when the pH of the solution increases sufficiently. However, a decrease in the relaxation rate is observed with a minimum of 25 hours. A possible explanation could be attributed to the reduction of Fe³⁺ to iron Fe²⁺ (ferrous ions are significantly less effective as a relaxing agent than ferric ions, Supplementary data, Table S3, (Bodart et al., 2020)). This would result in a decrease in the dissolved dioxygen concentration in the solution because of oxygen consumption in the solution through the oxidation of ethanol (catalysed by iron and other metals). However, this decrease is transitory because there are competitive reactions between the dissolution of dioxygen from the headspace into the solution, the dissolution of dioxygen resulting in the dissolution of the minerals composing the ceramic and oxidation reactions. After 25 hours, the relaxation rate increases again because of the continuous dissolution of Fe³⁺ and because of a slowing down of the dioxygen consumption in the solution. However, a plateau is reached at ~ 40 hours, and the relaxation rate starts to continuously decrease down to a value slightly higher than the relaxation rate of pure model wine because of (i) the slowing down of the dissolution because of the pH increase, (ii) the appearance of solid phases that can precipitate iron and (iii) a significant decrease in dioxygen in the solution and a reduction of paramagnetic metals.

However, the main purpose of Figure 10 is to compare the evolution of the relaxation rate of the ceramic with or without a coating. It is clear that the coating of the ceramic with beeswax significantly decreases the amount of released Fe³⁺ (but also of all elements). At 30 hours, the relaxograms show that dissolution is decreased by at least a factor of 4 or 170 if the ceramic is coated with one or two beeswax layers, respectively. Over the course of one year of contact of the coated ceramics with model wines, changes were observed related to both the colour of the model wine and the texture of the beeswax (Supplementary data, Figure S3). After three months, the samples coated with one layer of beeswax turned yellow but remained transparent, and the beeswax texture was slightly damaged. It should also be noted that no yellow or white precipitates or crystals were formed, which is typically observed with uncoated samples. A year later, the model wine in contact with the one-layer coated ceramic became hazy, and the texture of the beeswax continued to deteriorate. In the samples coated with two layers of beeswax and beeswax alone in the model wine, none of these changes (crystallisation, precipitation, haze formation, change in colour) were observed. This allows one to conclude that the changes that occur in the sample with one layer of beeswax are associated with the dissolution of the ceramic. This means that, at least for the reported period of time, an appropriate coating (in our case, with two layers) can strongly limit the acidic attack of the model wine on the ceramic, the migration of elements and the increase in pH.

Figure 10. (a) Relaxograms of model wines in contact with beeswax-coated and non-coated ceramics. The black curve is the relaxogram of a non-coated ceramic (Vwine = 500 µL, RV/S = 5.2 mm, Vhs/Vwine ~ 17). The blue curve represents the relaxation rate of the model wine in contact with a ceramic coated by one layer of beeswax; the red curve is the relaxogram of the model wine in contact with a ceramic tablet coated with two layers of beeswax (Vwine = 700 µL, RV/S ~ 10 mm, Vhs/Vwine ~ 12). (b) is an expansion of (a) with the relaxogram of beeswax immersed in model wine (green triangles).

Save View full size Expand inline Collapse inline

Conclusion

ICP-AES analyses showed that a ceramic container could be easily deteriorated by the model wine it contains through an acidic attack, which dissolves the ceramic elements in the model wine. There is homogeneous dissolution of the minerals forming the ceramic consistent with the dissolution of the ceramic in acid water reported in the literature. This dissolution is concomitant with a significant increase in pH. Time domain NMR focused on the measurement of the ¹H relaxation time of water molecules in model wines, allowed to indirectly target the evolution of the Fe³⁺ concentration in the model wine in in situ experiments. It showed that the coating of the ceramic with beeswax could reduce the dissolution of the ceramic by at least a factor of 4 for one-layer coated or by a factor of 170 for two layers of coated ceramic.

Acknowledgements

Dr. Igor Bezverkhyy (Laboratoire Interdisciplinaire Carnot de Bourgogne, Université de Bourgogne) is thanked for performing BET measurements. Dr. Jean-Marc Dachicourt (École supérieure d'ingénieurs numérique et matériaux (ESIREM)) is thanked for the XRF measurements The authors thank the DIVVA platform (Institut Agro Dijon, Université de Bourgogne) for the access Bruker minispec. C.F. and S.V. acknowledge financial support from the European Union (ERDF) and Région Nouvelle Aquitaine. This work is part of the project VALVIGNE (2020–2022, SYNERGIE BG0027582) supported by the conseil régional Bourgogne Franche Comté and the European Union through the FEDER—FSE Bourgogne 2014–2020 program.

References

• Aceto, M., Abollino, O., Bruzzoniti, M. C., Mentasti, E., Sarzanini, C., & Malandrino, M. (2002). Determination of metals in wine with atomic spectroscopy (flame-AAS, GF-AAS and ICP-AES); a review. Food Additives and Contaminants, 19(2), 126–133. https://doi.org/10.1080/02652030110071336
• Barisashvili, G. (2011). Making Wine in Qvevri—A Unique Georgian Tradition. Www.Kvevri.Org. Retrieved February 23, 2023, from www.kvevri.org/wp-content/uploads/2018/03/Making-Wine-in-Qvevri-a-Unique-Georgian-Tradition.pdf
• Bennett, P. C., Melcer, M. E., Siegel, D. I., & Hassett, J. P. (1988). The dissolution of quartz in dilute aqueous solutions of organic acids at 25°C. Geochimica et Cosmochimica Acta, 52(6), 1521–1530. https://doi.org/10.1016/0016-7037(88)90222-0
• Bertini, I., Luchinat, C., Parigi, G., & Ravera, E. (2017). NMR of paramagnetic molecules: Applications to metallobiomolecules and models (Second edition). Elsevier. https://doi.org/10.1016/C2013-0-18690-4
• Bodart, P. R., Batlogg, A., Ferret, E., Rachocki, A., Knapkiewicz, M., Esoyan, S., Hovhannisyan, N., Karbowiak, T., & Gougeon, R. D. (2022). Analysis of the Proton Spin–Lattice Relaxation in Wine and Hydroalcoholic Solutions. Food Analytical Methods, 15(2), 266–275. https://doi.org/10.1007/s12161-021-02118-w
• Bodart, P. R., Rachocki, A., Tritt-Goc, J., Michalke, B., Schmitt-Kopplin, P., Karbowiak, T., & Gougeon, R. D. (2020). Quantification of manganous ions in wine by NMR relaxometry. Talanta, 209, 120561. https://doi.org/10.1016/j.talanta.2019.120561
• Bragança, S. R., & Bergmann, C. P. (2006). Effect of Quartz of Fine Particle Size on Porcelain Properties. Materials Science Forum, 530–531, 493–498. https://doi.org/10.4028/www.scientific.net/MSF.530-531.493
• Cabrita, M. J., Martins, N., Barrulas, P., Garcia, R., Dias, C. B., Pérez-Álvarez, E. P., Costa Freitas, A. M., & Garde-Cerdán, T. (2018). Multi-element composition of red, white and palhete amphora wines from Alentejo by ICPMS. Food Control, 92, 80–85. https://doi.org/10.1016/j.foodcont.2018.04.041
• Cacho, J., Castells, J. E., Esteban, A., Laguna, B., & Sagristá, N. (1995). Iron, Copper, and Manganese Influence on Wine Oxidation. American Journal of Enology and Viticulture, 46(3), 380–384. https://doi.org/10.5344/ajev.1995.46.3.380
• Coleman, R. E., Boulton, R. B., & Stuchebrukhov, A. A. (2020). Kinetics of autoxidation of tartaric acid in presence of iron. The Journal of Chemical Physics, 153(6), 064503. https://doi.org/10.1063/5.0013727
• Coulter, A. D., Holdstock, M. G., Cowey, G. D., Simos, C. A., Smith, P. A., & Wilkes, E. N. (2015). Potassium bitartrate crystallisation in wine and its inhibition: Cold stability. Australian Journal of Grape and Wine Research, 21, 627–641. https://doi.org/10.1111/ajgw.12194
• CRC (2004). CRC handbook of chemistry and physics: A ready-reference book of chemical and physical data (D. R. Lide, Ed.; 85. ed). CRC Press.
• Criado, Y. A., Alonso, M., & Abanades, J. C. (2014). Kinetics of the CaO/Ca(OH)2 Hydration/Dehydration Reaction for Thermochemical Energy Storage Applications. Industrial & Engineering Chemistry Research, 53(32), 12594–12601. https://doi.org/10.1021/ie404246p
• Danilewicz, J. C. (2016). Fe(II):Fe(III) Ratio and Redox Status of White Wines. American Journal of Enology and Viticulture, 67(2), 146–152. https://doi.org/10.5344/ajev.2015.15088
• Danilewicz, J. C. (2018). [Fe(III)]:[Fe(II)] Ratio and Redox Status of Red Wines: Relation to So-Called “Reduction Potential”. American Journal of Enology and Viticulture, 69(2), 141–147. https://doi.org/10.5344/ajev.2017.17081
• Deng, Y. (1997). Effect of pH on the Reductive Dissolution Rates of Iron(III) Hydroxide by Ascorbate. Langmuir, 13(6), 1835–1839. https://doi.org/10.1021/la9607013
• Elias, R. J., & Waterhouse, A. L. (2010). Controlling the Fenton Reaction in Wine. Journal of Agricultural and Food Chemistry, 58(3), 1699–1707. https://doi.org/10.1021/jf903127r
• Gil i Cortiella, M., Ubeda, C., Covarrubias, J. I., Laurie, V. F., & Peña-Neira, Á. (2021). Chemical and Physical Implications of the Use of Alternative Vessels to Oak Barrels during the Production of White Wines. Molecules, 26(3), 554. https://doi.org/10.3390/molecules26030554
• Gougeon, R. D., Lucio, M., Frommberger, M., Peyron, D., Chassagne, D., Alexandre, H., Feuillat, F., Voilley, A., Cayot, P., Gebefügi, I., Hertkorn, N., & Schmitt-Kopplin, P. (2009). The chemodiversity of wines can reveal a metabologeography expression of cooperage oak wood. Proceedings of the National Academy of Sciences, 106(23), 9174–9179. https://doi.org/10.1073/pnas.0901100106
• Guerrini, L., Maioli, F., Picchi, M., Zanoni, B., Parenti, A., & Canuti, V. (2022). Kinetic modeling of a Sangiovese wine’s chemical and physical parameters during one-year aging in different tank materials. European Food Research and Technology, 248(6), 1525–1539. https://doi.org/10.1007/s00217-022-03982-4
• Harutyunyan, M., Viana, R., Granja-Soares, J., Martins, M., Ribeiro, H., & Malfeito-Ferreira, M. (2022). Adaptation of Ancient Techniques to Recreate ‘Wines’ and ‘Beverages’ Using Withered Grapes of Muscat of Alexandria. Fermentation, 8(2), 85. https://doi.org/10.3390/fermentation8020085
• Hovhannisyan, N. A., Yesayan, A. A., Bobokhyan, A., Dallakyan, M. V., Hobosyan, S., & Gasparyan, B. Z. (2017). Armenian vine and wine.
• Johnson, K., Zhang, Y., & Trela, B. (2013). Evaluation of Qvevri in Winemaking. 64th ASEV National Conference, 128–129. Retrieved February 23, 2023, from http://www.asev.org/sites/main/files/file-attachments/2013technicalabstracts.pdf?1410302529
• Knauss, K. G., Nguyen, S. N., & Weed, H. C. (1993). Diopside dissolution kinetics as a function of pH, CO2, temperature, and time. Geochimica et Cosmochimica Acta, 57(2), 285–294. https://doi.org/10.1016/0016-7037(93)90431-U
• Knauss, K. G., & Wolery, T. J. (1986). Dependence of albite dissolution kinetics on ph and time at 25°c and 70°c. Geochimica et Cosmochimica Acta, 50(11), 2481–2497. https://doi.org/10.1016/0016-7037(86)90031-1
• Lüttge, A., Bolton, E. W., & Lasaga, A. C. (1999). An interferometric study of the dissolution kinetics of anorthite; the role of reactive surface area. American Journal of Science, 299(7–9), 652–678. https://doi.org/10.2475/ajs.299.7-9.652
• MacNeil, K. (2015). The wine bible (Revised Second Edition). Workman Publishing Company.
• Martínez-Gil, A., Del Alamo-Sanza, M., & Nevares, I. (2022). Evolution of red wine in oak barrels with different oxygen transmission rates. Phenolic compounds and colour. LWT, 158, 113133. https://doi.org/10.1016/j.lwt.2022.113133
• McGovern, P., Jalabadze, M., Batiuk, S., Callahan, M. P., Smith, K. E., Hall, G. R., Kvavadze, E., Maghradze, D., Rusishvili, N., Bouby, L., Failla, O., Cola, G., Mariani, L., Boaretto, E., Bacilieri, R., This, P., Wales, N., & Lordkipanidze, D. (2017). Early Neolithic wine of Georgia in the South Caucasus. Proceedings of the National Academy of Sciences, 114(48). https://doi.org/10.1073/pnas.1714728114
• Miller, L. B., & Witt, J. C. (1929). Solubility of Calcium Hydroxide. The Journal of Physical Chemistry, 33(2), 285–289. https://doi.org/10.1021/j150296a010
• Montalvo, F. F., García-Alcaraz, J. L., Cámara, E. M., Jiménez-Macías, E., & Blanco-Fernández, J. (2021). Environmental impact of wine fermentation in steel and concrete tanks. Journal of Cleaner Production, 278, 123602. https://doi.org/10.1016/j.jclepro.2020.123602
• Nevares, I., & del Alamo-Sanza, M. (2021). Characterisation of the Oxygen Transmission Rate of New-Ancient Natural Materials for Wine Maturation Containers. Foods, 10(1), 140. https://doi.org/10.3390/foods10010140
• Nikolantonaki, M., Daoud, S., Noret, L., Coelho, C., Badet-Murat, M.-L., Schmitt-Kopplin, P., & Gougeon, R. D. (2019). Impact of Oak Wood Barrel Tannin Potential and Toasting on White Wine Antioxidant Stability. Journal of Agricultural and Food Chemistry, 67(30), 8402–8410. https://doi.org/10.1021/acs.jafc.9b00517
• Nikulshina, V., Gálvez, M. E., & Steinfeld, A. (2007). Kinetic analysis of the carbonation reactions for the capture of CO2 from air via the Ca(OH)2–CaCO3–CaO solar thermochemical cycle. Chemical Engineering Journal, 129(1–3), 75–83. https://doi.org/10.1016/j.cej.2006.11.003
• Ohls, K., & Bogdain, B. (2016). History of inductively coupled plasma atomic emission spectral analysis: From the beginning up to its coupling with mass spectrometry. Journal of Analytical Atomic Spectrometry, 31(1), 22–31. https://doi.org/10.1039/C5JA90043C
• Qin, J., Yang, C., Cui, C., Huang, J., Hussain, A., & Ma, H. (2016). Ca2+ and OH− release of ceramsites containing anorthite and gehlenite prepared from waste lime mud. Journal of Environmental Sciences, 47, 91–99. https://doi.org/10.1016/j.jes.2016.03.013
• Romanus, K., Baeten, J., Poblome, J., Accardo, S., Degryse, P., Jacobs, P., De Vos, D., & Waelkens, M. (2009). Wine and olive oil permeation in pitched and non-pitched ceramics: Relation with results from archaeological amphorae from Sagalassos, Turkey. Journal of Archaeological Science, 36(3), 900–909. https://doi.org/10.1016/j.jas.2008.11.024
• Sanchez-Garmendia, U., Iñañez, J. G., & Arana, G. (2021). Alterations and Contaminations in Ceramics Deposited in Underwater Environments: An Experimental Approach. Minerals, 11(7), 766. https://doi.org/10.3390/min11070766
• Serris, E., Favergeon, L., Pijolat, M., Soustelle, M., Nortier, P., Gärtner, R. S., Chopin, T., & Habib, Z. (2011). Study of the hydration of CaO powder by gas–solid reaction. Cement and Concrete Research, 41(10), 1078–1084. https://doi.org/10.1016/j.cemconres.2011.06.014
• Shackelford, J., & Shackelford, P. (2020). Ceramics in the wine industry. International Journal of Ceramic Engineering & Science, 3. https://doi.org/10.1002/ces2.10073
• Solomon, I. (1955). Relaxation Processes in a System of Two Spins. Physical Review, 99(2), 559–565. https://doi.org/10.1103/PhysRev.99.559
• Stanmore, B. R., & Gilot, P. (2005). Review—Calcination and carbonation of limestone during thermal cycling for CO2 sequestration. Fuel Processing Technology, 86(16), 1707–1743. https://doi.org/10.1016/j.fuproc.2005.01.023
• Torrent, J. (1987). The Reductive Dissolution of Synthetic Goethite and Hematite in Dithionite. Clay Minerals - CLAY MINER, 22, 329–337. https://doi.org/10.1180/claymin.1987.022.3.07
• Twede, D. (2002). Commercial Amphoras: The Earliest Consumer Packages? Journal of Macromarketing, 22(1), 98–108. https://doi.org/10.1177/027467022001009
• van Duynhoven, J., Voda, A., Witek, M., & Van As, H. (2010). Time-Domain NMR Applied to Food Products. In Annual Reports on NMR Spectroscopy (Vol. 69, pp. 145–197). Elsevier. https://doi.org/10.1016/S0066-4103(10)69003-5
• Welch, S. A., & Ullman, W. J. (1993). The effect of organic acids on plagioclase dissolution rates and stoichiometry. Geochimica et Cosmochimica Acta, 57(12), 2725–2736. https://doi.org/10.1016/0016-7037(93)90386-B
• Weltman, P. (2018). Why an Ancient Winemaking Technique Is Making a Comeback. Daily.Sevenfifty.Com. Retrieved February 24, 2023, from https://daily.sevenfifty.com/why-an-ancient-technique-is-making-a-comeback/
• Work, H. (2014). Wood, whiskey and wine: A history of barrels. Reaktion Books.
• Zhang, X., & Zhang, H. (2014). Research on Meiping in the Application of Chinese Traditional Wine Packaging: International Conference on Education, Language, Art and Intercultural Communication (ICELAIC-14), Zhengzhou, Henan, China. https://doi.org/10.2991/icelaic-14.2014.120