Monitoring wine aging with Fourier transform infrared spectroscopy ( FTIR )

Oak wood has commonly been used in wine aging but recently other wood types such as Acacia and Chestnut, have attracted the interest of the researchers due to their possible positive contribution to wine quality. However, only the use of oak and chestnut woods is approved by the International Enological Codex of the International Organisation of Vine and Wine. In this study Fourier Transform (FT)-mid-infrared spectroscopy combined with Discriminant Analysis was used to differentiate wines aged in barrels made from French oak, American oak, Acacia and Chestnut and in tanks with oak chips, over a period of 12 months. Two red (Mandilaria, Kotsifali) and two white (Vilana, Dafni) native Greek grape varieties where used to produce four wines. The Fourier Transform Infrared (FT-IR) spectra of the samples were recorded on a Zinc Selenide (ZnSe) window after incubation at 40°C for 30 min. A complete differentiation of the samples according to both the type of wood used and the contact time was achieved based on their FT-IR spectra. The aim of this paper was to find a rapid, simple and economical method for the discrimination between wines aged in different types of containers and for different time periods using Fourier transform infrared spectroscopy (FT-IR). 2. Materials and methods 2.1. Wines and wood maturations The wines were 2013 vintage from the region of Heraklion in Crete, Greece. Two white wines from the grape varieties Vilana and Dafni and two red from Kotsifali and Mandilaria, all indigenous varieties of Crete, were used. Oenological parameters of resulting wines are shown in Table 1. French (Quercus robur) and American (Quercus alba) standard barrels (225 l) with a medium toasting were purchased from “Tonnellerie du Monde World Cooperage”, Acacia standard barrels with a light plus (L+) toasting were purchased from Tonellerie du sud ouest, and Chestnut standard barrels with a medium toasting were purchased from Tesias Metsovo. The oak chips (French oak) were from Seguin Moreau (Oenostick, V18). All wines completed fermentation in stainless steel tanks and were then transferred to the barrels and stainless steel tanks with oak chips for maturation. Control wines were transferred in stainless steel tanks. Samples for analyses were taken from all containers after 3, 6, 9 and 12 months of maturation. Chestnut barrels were used only for the red wines. 2.2. Sample pretreatment 100 μL of each sample were placed on a Zinc Selenide (ZnSe) window and were incubated at 40°C for 30 min. After incubation, the samples were ready for the spectroscopic analysis. BIO Web of Conferences 02016-p.2 2.3. Spectroscopic analysis Triplicate spectra of 48 samples for red varieties and 40 samples for white varieties (88 in total) were collected with a Thermo Nicolet 6700 FT-IR spectrophotometer (Thermo Electron Corporation, MA, USA) equipped with a deuterated triglycine sulfate (DTGS) detector. The FT-IR spectra were collected using a standard ZnSe window against a ZnSe background with a total of 100 scans (resolution 4cm) Each spectrum was smoothed using the Savitsky–Golay algorithm (5-point moving, second-degree polynomial) and its baseline was corrected (second-degree polynomial, 20 iterations). The average spectrum of each sample was calculated by means of the corresponding function of the software of the spectrometer using the software accompanying the spectrometer (OMNIC 7.3, Thermo Fisher Scientific Inc.).


Introduction
Wine ageing in wooden barrels is one of the most common methods in the winemaking process.The wood influences the organoleptic profile of the wine through important modifications that take place during barrel aging due to the slow and continuous diffusion of oxygen through the wood pores and the extraction of many substances (e.g., aromatic compounds and ellagitannins) which modulate its sensory attributes such as aroma, structure, astringency, bitterness and color [1].The main reasons that influence the release of wood extractable compounds is the ageing technology, involving the barrel characteristics, such as the wood botanical species [2,3], the toasting level [4] and the barrel size [5].The quantity of the barrel extractable compounds depends mainly on the geographical origin and the type of wood used [6] the age of the barrel and the contact time between wine and wood.The most frequently used type of wood is oak, and between the species of oak used in winemaking the most important are Quercus robur that originates from France and Quercus alba that originates from the United States, studies on the chemical composition of which have demonstrated that they are quantitatively different [7,8].Recently other types of wood, such as Acacia and Chestnut have been used for ageing, showing good potential for use in winemaking [9].Sophisticated methods and instrumentation have been developed in order to facilitate the accurate and precise measurement of wood extracted compounds.Infrared spectroscopy based methods are recently emerging because of their versatility and efficiency, their cost effectiveness, and their fast and non-invasive nature [10,11].Combining spectroscopic techniques with multivariate data analysis is proving to be very promising for the discrimination of wines of different varietal origin or wines from different countries, as was reported recently [12][13][14].
The aim of this paper was to find a rapid, simple and economical method for the discrimination between wines aged in different types of containers and for different time periods using Fourier transform infrared spectroscopy (FT-IR).

Wines and wood maturations
The wines were 2013 vintage from the region of Heraklion in Crete, Greece.Two white wines from the grape varieties Vilana and Dafni and two red from Kotsifali and Mandilaria, all indigenous varieties of Crete, were used.Oenological parameters of resulting wines are shown in Table 1.French (Quercus robur) and American (Quercus alba) standard barrels (225 l) with a medium toasting were purchased from "Tonnellerie du Monde World Cooperage", Acacia standard barrels with a light plus (L+) toasting were purchased from Tonellerie du sud ouest, and Chestnut standard barrels with a medium toasting were purchased from Tesias Metsovo.The oak chips (French oak) were from Seguin Moreau (Oenostick ® , V18).All wines completed fermentation in stainless steel tanks and were then transferred to the barrels and stainless steel tanks with oak chips for maturation.Control wines were transferred in stainless steel tanks.Samples for analyses were taken from all containers after 3, 6, 9 and 12 months of maturation.Chestnut barrels were used only for the red wines.

Sample pretreatment
100 μL of each sample were placed on a Zinc Selenide (ZnSe) window and were incubated at 40°C for 30 min.After incubation, the samples were ready for the spectroscopic analysis.
BIO Web of Conferences 02016-p.2

Spectroscopic analysis
Triplicate spectra of 48 samples for red varieties and 40 samples for white varieties (88 in total) were collected with a Thermo Nicolet 6700 FT-IR spectrophotometer (Thermo Electron Corporation, MA, USA) equipped with a deuterated triglycine sulfate (DTGS) detector.The FT-IR spectra were collected using a standard ZnSe window against a ZnSe background with a total of 100 scans (resolution 4cm -1 ) Each spectrum was smoothed using the Savitsky-Golay algorithm (5-point moving, second-degree polynomial) and its baseline was corrected (second-degree polynomial, 20 iterations).The average spectrum of each sample was calculated by means of the corresponding function of the software of the spectrometer using the software accompanying the spectrometer (OMNIC 7.3, Thermo Fisher Scientific Inc.).

Chemometrics
Discriminant analysis was performed using the JMP Statistical Discovery software, version 8. Spectral data were exported to JMP, and PCA was used to derive 30 principal components that ensure that 99% of the variability is considered by the analysis [20].Subsequently, using the stepwise selection and the linear method, discriminant analysis was performed.

Spectroscopic analysis
FT-IR spectra were obtained for the 88 samples analysed; a typical spectrum for the whole spectral range can be shown in Fig. 1 (white wines) and Fig. 2 (red wines).
The band in the 3702-3012 cm -1 region of the spectra originates from compounds with an -OH group, such as water [15].Various IR bands occur in the region from 1800 to 900 cm -1 , which is part of the fingerprint region, such as those corresponding to the vibration of the C-O, C-C, C-H and C-N bonds [16].The peak centered at 1718 cm -1 is due to the stretching of carbonyl group (C=O), while the second peak at 1613 cm -1 is due to C=C stretching (typical for aromatic molecules) [17][18][19].The absorption around 1449 cm -1 corresponds to antisymmetric in-plane bending of CH 3 and to deformation of -CH 2 -groups [18,19].The peak at 1350 cm -1 is associated with CH bending and CH 2 waging, while peaks at 1285 cm -1 correspond to in-plane bending of O-H [16][17][18].The peaks at 1204, 1114 and 1066 cm -1 correspond to the stretching vibration of C-O [17][18][19].

Discriminant analysis
Figure 3 shows the results obtained regarding the contact time period.The results indicate a complete discrimination of 100%.According to Table 2, no samples were misclassified.
Regarding the container type, Fig. 4 shows an almost complete discrimination of 95.4%.In this case, there were found 4 misclassified samples (Table 3).Namely, one sample of American Oak was classified as French Oak, one sample of French Oak was classified as Chestnut, one sample of Tank was classified as Tank

Conclusion
The objective of the research proposed was to explore the suitability of FT-IR spectroscopy for the discrimination of wines aged in different containers for different time periods.The spectroscopic analysis of the samples combined with the discriminant analysis resulted in a statistically significant discrimination of both the type of container and the time spent in it.The proposed method is promising since it is simple, rapid, economical without the requirement of chemical reagents and sample pretreatment steps.

Figure 3 .
Figure 3. Discrimination results regarding time of contact.

Table 1 .
Oenological parameters of wine samples.

Table 2 .
JMP discrimination results regarding time of contact.
th World Congress of Vine and Wine 02016-p.3 with Chips and one sample from Tank with Chips was classified as French Oak.