Food Chemistry 150 (2014) 489–493

Contents lists available at ScienceDirect

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

Analytical Methods

Application of FTIR-ATR to Moscatel dessert wines for prediction of total phenolic and flavonoid contents and antioxidant capacity Sandra D. Silva a,b,⇑, Rodrigo P. Feliciano a, Luís V. Boas a,b, Maria R. Bronze a,b,c a

Instituto de Biologia Experimental e Tecnológica, Apartado 12, 2780 – 901 Oeiras, Portugal Instituto de Tecnologia Química e Biológica, Av. da República, Estação Agronómica Nacional, 2780-157 Oeiras, Portugal c Faculdade de Farmácia de Lisboa, Av. das Forças Armadas, 1649 – 019 Lisboa, Portugal b

a r t i c l e

i n f o

Article history: Received 10 August 2012 Received in revised form 12 March 2013 Accepted 3 November 2013 Available online 15 November 2013 Keywords: Moscatel wine Fourier-transform infrared spectroscopy Phenolic compounds Antioxidant capacity

a b s t r a c t Fourier transform infrared spectroscopy (FTIR) attenuated total reflectance (ATR) was applied for the determination of total phenolic and flavonoid contents and antioxidant capacity (DPPH and FRAP assays) in Moscatel dessert wines (n = 56). Prediction models were developed for the referred parameters using Partial Least Squares (PLS) considering the spectral region 1800–900 cm1. The determination coefficients (r2) values in the calibration models ranged from 0.670 to 0.870. Cross validation (leaveone-out technique) was applied to the data. Root mean square errors of calibration (RMSEC) and cross validation (RMSECV) as well as the relative errors of prediction (REP) were calculated. Minimum errors of prediction were obtained for total flavonoid content (0.2%) and maximum values (22%) for antioxidant capacity measured by FRAP. The proposed method may be used for rapid screening of total phenolic and flavonoid contents in Moscatel dessert wines. The implemented methodologies may also be used to get rough estimates for DPPH and FRAP antioxidant capacities. Ó 2013 Elsevier Ltd. All rights reserved.

1. Introduction Wines are known as important sources of phenolic compounds. ‘‘Moscatel de Setúbal’’ dessert wines are produced in a Denomination of Controlled Origin (DOC) region Setúbal in Portugal, using at least 85% of the white grape variety Moscatel de Setúbal also known as Moscatel Graúdo. This type of wine can also be prepared from red grape varieties (Moscatel Roxo). After crushing the grapes, the fermentation time is short and ends with the addition of spirit or wine alcohol to the grapes must. The non-fermented sugars (90–100 g/L of residual sugar) give the required sweetness characteristic of this type of wine. The resulting product stays in contact with grape skins (maceration period) and later is separated from the solid materials, which are pressed and the liquid obtained added to the wine collected previously (Feliciano et al., 2009). These fortified wines are stored for at least 18 months in Setúbal DOC region (e.g.: in wood casks). Due to the type of processing, maceration with grape skins and ageing, these wines usually present a higher content of phenolic compounds than white wines (Feliciano et al., 2009). In fact, the phenolic content of white wines varies from 165 to 1425 mg/L, and higher contents are observed in red wines ranging between 1018 and 4059 mg/L expressed as gallic acid equivalents (GAE) (Bauer et al., 2008; Katalinic, Milos, ⇑ Corresponding author at: Instituto de Biologia Experimental e Tecnológica, Apartado 12, 2780 – 901 Oeiras, Portugal. Tel.: +351 21 4469781. E-mail address: [email protected] (S.D. Silva). 0308-8146/$ - see front matter Ó 2013 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.foodchem.2013.11.028

Modun, Music, & Boban, 2004; Versari, Giuseppina, Scazzina, & Del Rio, 2010). Data reported in the literature (Feliciano et al., 2009) compared the total phenolic content (GAE) of different fortified wines such as Moscatel (400–1700 mg/L), Port (735–1116 mg/ L), Madeira (282–1936 mg/L) and Sherry wines (201–446 mg/L). For the analysis of individual phenolic compounds, reversed phase high performance liquid chromatography (HPLC) may be combined with several detectors (UV–Vis, electrochemical, fluorescence, mass spectrometry) and is commonly used. The Folin– Ciocalteu colorimetric method has been used for a long time to evaluate the total phenolic content (TPC) in different types of samples. This method is based on a chemical reduction of a reagent, a mixture of tungsten and molybdenum oxides yielding a bluish compound that is measured at 750 nm. Total flavonoid content (TFC) can also be measured with a spectrophotometric method, which is based on a reaction between aluminium chloride and flavonoids at alkaline pH yielding a reddish compound, which is measured at 510 nm (Feliciano et al., 2009). These methods are time consuming and often require sample pre-treatment. For antioxidant capacity determination, indirect measurements such as ferric reducing antioxidant power (FRAP) and 1,1-diphenyl-2-picrylhydrazyl (DPPH) assays have been used because of their capacity to scavenge reactive oxygen species and produce species that absorb at 593 and 517 nm, respectively (Feliciano et al., 2009). Infrared spectroscopy (IR) is based on the principle that functional groups within a sample will vibrate on exposure to

490

S.D. Silva et al. / Food Chemistry 150 (2014) 489–493

infrared radiation and these methods present some advantages such as versatility, efficiency and speed (Tarantilis, Troianou, Pappas, Kotseridis, & Polissiou, 2008). Fourier transform infrared spectroscopy is a widespread technique in the analysis of food components (Cozzolino, Cynkar, Shah, & Smith, 2011; Roussel, Bellon-Maurel, Roger, & Grenier, 2003; Silva, Rosa, Ferreira, Boas, & Bronze, 2009) that has already been used in the quantification of ethanol, glucose, fructose (Boulet, Williams, & Doco, 2007; Cozzolino et al., 2011) and organic acids (Mato, Suarez-Luque, & Huidobro, 2005) in wines with acceptable accuracy. The applicability of mid infrared technology has also been explored to detect problems with early-stage wine fermentation (Urtubia, PerezCorrea, Pizarro, & Agosin, 2008), to develop a rapid and simple method for wine authenticity (Bevin, Fergusson, Perry, Janik, & Cozzolino, 2006) and as a tool for rapid evaluation of procyanidin composition of white and red grape seeds varieties (Passos, Cardoso, Barros, Silva, & Coimbra, 2010). Fourier transform infrared spectroscopy (FTIR) attenuated total reflectance (ATR) was also used in an attempt to differentiate red wines (Edelmann, Diewok, Schuster, & Lendl, 2001; Tarantilis et al., 2008) of different varietal origin (Austrian and Greek wines). The development of sampling accessories, such as ATR cells, has led to major improvements in routine IR analysis, by simplifying sample handling and avoiding measurement problems often found with transmission cells (Roussel et al., 2003). In FTIR analysis, a large number of data are produced and different treatments have been developed to analyse these data. Versari et al. (2010) used Partial Least Squares (PLS) regression to analyse data from FTIR spectra in order to predict red wine total antioxidant capacity. PLS statistical analysis is a robust chemometric method for data treatment of complex mixtures if a suitable number of calibration samples with the appropriate variation in composition across the range for the unknown samples is available (Silva et al., 2009). The technique is based on the so-called bilinear projection, meaning that two sets of variables (x, y) are linked to each other by means of linear projection models. Instead of multiple linear regression (MLR) being applied to the full set of regressors, the information carried by the original x variables is projected onto a smaller set of latent, uncorrelated variables called PLS components (Bauer et al., 2008). The aim of this study was to develop a FTIR methodology for screening total phenolic content and total flavonoid content as well as total antioxidant capacity in Moscatel dessert wines using infrared spectral data, without sample pre-treatment. PLS models were developed for the prediction of these parameters (TPC, TFC and TAC). This method could be important for the rapid screening of wine phenolic composition and evaluation of age and antioxidant capacity.

2. Materials and methods 2.1. Reagents Folin–Ciocalteu reagent (Fluka, Switzerland) and sodium carbonate (Panreac, Spain) were used for the determination of total phenolic content using gallic acid P98% (Fluka, Germany) as a standard. For total flavonoids analysis, (+)-catechin hydrate 98% (Sigma, Germany) was used as a reference compound and aluminium chloride and sodium nitrite (Sigma, Germany) as reagents. DPPH 90% (Sigma, Germany) and methanol (Fischer, UK) were used for the DPPH assay. For FRAP assay, 6-hydroxy-2,5,7,8-tetramethylchroman-2-carboxylic acid (Trolox) 97% (Sigma, Germany) was used as the standard, and the following reagents were employed: sodium acetate buffer (Merck, Germany), formic acid 98% (Panreac, Spain), 2,4,6-Tris(2-pyridyl)-s-triazine (TPTZ) 98%

(Sigma, Germany), iron (III) chloride hexahydrate (Sigma, Germany), and hydrochloric acid 37% (Fisher, UK). For ORAC, 20 ,20 -azobis (2-amidinopropane) dihydrochloride (AAPH) was obtained from Fluka (Sigma Aldrich, Portugal) and disodium fluorescein was from TCI Europe, Belgium. Water was purified by means of Milli-Q from Millipore (Bedford, MA, USA). 2.2. Moscatel wine samples Samples from 13 different producers were supplied by the Setúbal Regional Wine Comission (Comissão Vitivínicola Regional da Península de Setúbal – CVRPS) from producing years 1980 and 1995–2001 (n = 52). Other samples from Setúbal (two) and Douro (two) regions were bought in a local supermarket. A total of 56 samples were analysed. 2.3. Total phenolic content assay (TPC) A standard solution of gallic acid (100 mg/L) was prepared in methanol, and calibration solutions between 1 and 7 mg/L were prepared by dilution with deionized water. Briefly, 0.050 mL of gallic acid solution or sample diluted 1:5 with water was mixed with 2.850 mL of deionized water, 2 mL of 2% sodium carbonate solution and finally, 0.100 mL of Folin–Ciocalteu reagent. After 60 min of incubation at room temperature in the dark, the absorbance was read against deionized water at 750 nm on a Beckmann DU-70 (Fullerton, USA) spectrophotometer (Feliciano et al., 2009). The total phenolic content was determined in triplicate and reported as mg GAE/L. 2.4. Total flavonoid content assay (TFC) A standard solution of catechin (5000 mg/L) was prepared in methanol, and calibration solutions between 10 and 150 mg/L were prepared by dilution with water. Briefly, 0.500 mL of catechin standard solution or sample diluted 1:2 with water was mixed with 2 mL of deionized water and with 0.150 mL of sodium nitrite 5%. After 5 min at room temperature, 0.150 mL of aluminium chloride 10% was added, and after 6 min, 1 mL of sodium hydroxide 1 M was also added. Deionized water was used to adjust the total volume to 5 mL and absorbance was read at 510 nm. The total flavonoid content was determined in triplicate and reported as mg catechin equivalents (CE)/L (Feliciano et al., 2009). 2.5. 1,1-diphenyl-2-picrylhydrazyl (DPPH) assays A standard solution of DPPH 24 mg/L was prepared in methanol. Undiluted sample (0.050 mL) was mixed with 2 mL of DPPH solution, and the absorbance was measured immediately at 517 nm against a methanol blank. After 16 min at room temperature, the absorbance was read again. The % inhibition of DPPH radical caused by a wine sample was determined according to the following formula: (AC(0)  AA(t))/AC(0)  100, where AC(0) is the absorbance of the sample at t = 0 min and AA(t) is the absorbance of sample at t = 16 min. All samples were analysed as triplicates (Feliciano et al., 2009). 2.6. Ferric reducing antioxidant power (FRAP) A standard solution of Trolox 100 mg/L was prepared in methanol, and working solutions between 0.5 and 4.0 mg/L were used. Briefly, 0.100 mL of sample diluted 1:20 with water was mixed with 3 mL of FRAP reagent (25 mL of acetate buffer 300 mmol/L at pH = 3.6 (corrected with formic acid) + 2.5 mL of Fe (II)-TPTZ 10 mmol/L in HCl 40 mmol/L + 2.5 mL of FeCl3.6H2O, 20 mmol/L). FRAP reagent was used as the blank, and final absorbance was read

S.D. Silva et al. / Food Chemistry 150 (2014) 489–493

at 593 nm after 10 min at room temperature. The ferric reducing antioxidant power of the samples was determined in triplicate and reported as mg of Trolox equivalents (mg TEAC/L) (Feliciano et al., 2009). 2.7. FTIR-ATR spectroscopy A Thermo Scientific FTIR Spectrometer (San Jose, USA), Class 1 Laser Product Nicolet 6100, was used. The equipment included an accessory with a ZnSe ATR crystal. The crystal provided an angle of incidence of 45°. The software used for FTIR data collection was Omnic version 7.3 (Thermo Electron Corporation). Before analysis the instrument was purged with nitrogen for 15 min. As reference, the background spectrum of air was collected before the acquisition of the sample spectrum. After each sample, the crystal was rinsed with acetone and then dried with a soft tissue. To record spectra, 0.600 mL of the sample was poured on the ATR crystal. Spectra were recorded with a resolution of 4 cm1, and 32 scans were averaged for each spectrum (scan 4000–650 cm1). 2.8. Data treatment PLS was used to predict the phenolic and flavonoid contents as well as antioxidant capacities (as FRAP and DPPH) in 56 Moscatel wines analysed. The values obtained by the reference methods were used as variables using the software programe TQ Analyst, version 7.2 (Thermo Electron Corporation). Cross-validation (leave-one-out technique) was used in all cases for calibration models evaluation. Determination coefficient (r2), root mean square error of calibration (RMSEC) and calibration accuracy as root mean square error of cross validation (RMSECV) and relative error of prediction (REP) were the statistical parameters used to evaluate the significance of the calculations. 3. Results and discussion 3.1. Phenolic content and total antioxidant capacities in Moscatel wines Average and standard deviation values obtained for total phenolic and flavonoid contents and antioxidant capacities of 56 samples are presented in Table 1. Results for all the parameters measured show there are important differences among the samples studied, probably due to differences in the winemaking techniques and composition of the raw materials: the total phenolic content in the samples ranged from 483 to 2263 GAE mg/L, with an average value of 1090 ± 409 GAE mg/L. Total flavonoid content had also a wide range of variation from 49 to 511 mg CE/L, with an average value of 227 ± 123 mg CE/L. For the evaluation of the total antioxidant capacity of samples, DPPH and FRAP assays were used. In the DPPH assay, a higher percentage inhibition corresponds to a higher antioxidant capacity for this radical. In the present study, DPPH inhibition ranged from 40%

Table 1 Parameters evaluated in Moscatel wines. Parameter

Mean ± SD (n = 56)

Total phenolic content (mg GAE/L) Total flavonoid content (mg CE/L) DPPH inhibition (%) FRAP (mg TEAC/L)

1090 ± 409 227 ± 123 72 ± 11 2957 ± 1141

Abbreviation: GAE – gallic acid equivalents; CE – catechin equivalents; TEAC – trolox equivalent antioxidant capacity; DPPH – 1,1-diphenyl-2-picrylhydrazyl assay as antioxidant capacity determination; FRAP – ferric reducing antioxidant power.

491

to 88% with a mean value of 72 ± 11%. FRAP assays measure the capacity of reducing metals, namely, ferric to ferrous iron. FRAP values varied between 827 and 5664 mg TEAC/L with an average value of 2957 ± 1141 mg TEAC/L. Results show Moscatel wines have a significant antioxidant capacity due to their chemical composition and values obtained are comparable with those reported for red wines (Katalinic et al., 2004). 3.2. FTIR analysis and wavenumber selection FTIR spectra were obtained for the 56 samples analysed and a typical spectrum obtained for the whole spectral range (4000–650 cm1) is presented in Fig. 1. The intense band detected in the 3627–2971 cm1 region originated from compounds with –OH groups such as water and ethanol, which are major compounds in these samples, was not useful in this work. Region 1800–900 cm1 was selected for working range since the RSD (relative standard deviation) between absorption values for the samples were high in this region as shown in Fig. 1, where an overlay of all samples analysed is presented. These wavelengths are part of the fingerprint region and include infrared typical absorption of phenolic molecules such as the stretching band of carbonyl (C@O) groups (1712–1704 cm1) and C@C stretching bands (1609–1608 and 1519–1516 cm1), which are typical of aromatic molecules. Moreover, signal from the phenols can be found in the region 1680–900 cm1. The absorption at 1448–1444 cm1 corresponds to antisymmetric in-plane bending of –CH3. Furthermore, in the same spectral region corresponding to the phenyl nuclei (C@C bonds), there are also bands of deformation of –CH2– groups. The peak at 1376–1373 cm1 is associated with symmetric in-plane bending of –CH3. The absorption at 1340–1339 cm1 has been assigned to CH bending and CH2 wagging. The peak at 1281–1278 cm1 corresponds to in-plane bending of O–H. The bands at 1207, 1110–1107, 1068–1062 cm1 correspond to the stretching vibration of C–O. These assignments are based on previous work on phenolic compounds in wine (Tarantilis et al., 2008). 3.3. FTIR calibration The results obtained from the reference methods and those calculated from infrared spectra of the samples were used to establish the calibration ranges. PLS was used to build calibration models to predict total phenolic and flavonoid content and total antioxidant capacities (DPPH and FRAP). The statistical PLS approach used in this work relies on the linear combination of FTIR spectral variables, so-called factors. A factor is a set of components that contains spectral and concentration information and is used to describe the variation in a PLS method model. The software used automatically selects the number of components/factors for the quantitative analysis. The number suggested for each parameter is related to the minimum level of predicted residual error sum of squares (PRESS). The factors used were: ten for phenolic and flavonoid contents and DPPH antioxidant capacity and seven for FRAP antioxidant capacity. Increasing the number of factors for FRAP, up to ten, introduced a bias in the model and, therefore, seven factors – as recommended by the software – were retained. Using these conditions, the accumulated variances accounted for the different sets of factors were more than 73%, which is an acceptable variability in these methods. Calibration evaluation was done using the RMSEC and determination coefficients. The differences between the calculated FTIR and the reference data were considered as the residual value used for RMSEC calculation. Values of r2 closer to 1 mean a higher probability that the FTIR predicted value (y-axis) is related to the measured parameters (x-axis). Calibrations with determination

492

S.D. Silva et al. / Food Chemistry 150 (2014) 489–493

Fig. 1. Typical FTIR spectrum of a sample of Moscatel wine: zoom of the selected wavenumber range with an overlay of all samples analysed in that range.

coefficients higher than 0.8 were considered acceptable (Feliciano et al., 2009; Versari et al., 2010). Another useful parameter for calibration evaluation is the relative error of prediction (REP), which shows the predictive ability of the model (Galtier et al., 2007). Low values of REP mean an acceptable performance of the model. Table 2 summarises the results obtained for the referred parameters. The determination coefficients obtained were greater than 0.8 for phenolic and flavonoid contents. However, the values obtained for DPPH and FRAP parameters were lower than 0.8 suggesting low correlation of these parameters with the calibration models implemented, especially for FRAP. The FRAP prediction range and high REP value obtained, presented in Table 2, also support these observations. However, the same statement cannot be extended to DPPH with a determination coefficient less than 0.8 but a good prediction of the FTIR range when compared with the calibration range. This effect could be explained with the range of calibration values, which were not broad enough to ensure a good dispersion of the data. REP values calculated for each of the studied parameters enabled us to conclude that the calibration models for total flavonoid content and DPPH were the ones with best results (REP less than 10%), showing that these predictions methods are adequate.

sample. This process was repeated until all calibration samples had been removed once. The total error obtained was expressed as RMSECV. Values obtained for RMSECV and the corresponding correlation coefficients are presented in Table 3. Results of cross validation showed higher RMSECV values than RMSEC obtained during calibration. The best results were obtained for the total phenolic and total flavonoid contents. This observation can be related to the lack of specificity for the DPPH and FRAP methods used and also to the contribution to antioxidant capacity of other compounds in wine apart from phenolic compounds (e.g.: polysaccharides) (Versari et al., 2010). Moreover, the DPPH radical does not react with flavonoids that lack hydroxyl groups in B-rings (e.g. kaempferol) or with monohydroxylated aromatic acids, such as ferulic and p-coumaric acids, and FRAP has the drawback that compounds with low redox potentials can reduce the Fe3+, even though they do not behave as antioxidants in vivo (Feliciano et al., 2009).

Table 3 Validation results for the implemented FTIR models obtained using the leave one out technique.

3.4. FTIR validation Validation of the implemented models was done using cross validation with the leave-one-out technique: one sample was removed from the calibration model and the remaining samples were then used to predict the value corresponding to the removed

Parameter

RMSECV

Correlation coefficient (r)

Total phenolic content (mg GAE/L) Total flavonoid content (mg CE/L) DPPH inhibition (%) FRAP (mg TEAC/L)

265 72 9 929

0.763 0.811 0.606 0.619

Abbreviation: RMSECV – root mean square error of cross validation; GAE, CE, TEAC, DPPH and FRAP as in Table 1.

Table 2 Calibration results for the implemented FTIR models. Parameter

Calibration range

Correlation coefficient (r)

Determination coefficient (r2)

RMSEC

Prediction range

REP %

Total phenolic content (mg GAE/L) Total flavonoid content (mg CE/L) DPPH inhibition (%) FRAP (mg TEAC/L)

483–2263 49–511 40–88 828–5665

0.933 0.948 0.868 0.818

0.870 0.899 0.753 0.670

146 39 6 650

468–2049 27–522 54–78 1344–4512

13 0.2 8 22

Abbreviation: RMSEC – root mean square error of calibration; REP – relative error of prediction; GAE, CE, TEAC, DPPH and FRAP as in Table 1.

S.D. Silva et al. / Food Chemistry 150 (2014) 489–493

4. Conclusions These preliminary results show that FTIR-ATR can be a useful tool for rapid screening of total phenolic and flavonoids contents in Moscatel dessert wines. The implemented methodologies may also be used to get rough estimates for DPPH and FRAP antioxidant capacities. Further validation of the implemented models should include a larger number of Moscatel wines samples from different regions. Acknowledgements The authors thank the Regional Wine Commission from Setúbal Region (CVRPS) for providing Moscatel wine samples. References Bauer, E. R., Nieuwoudt, H., Bauer, F. F., Kossmann, J., Koch, K. R., & Esbensen, K. H. (2008). FTIR spectroscopy for grape and wine analysis. Analytical Chemistry, 80, 1371–1379. Bevin, C. J., Fergusson, A. J., Perry, W. B., Janik, L. J., & Cozzolino, D. (2006). Development of a rapid ‘‘fingerprinting’’ system for wine authenticity by midinfrared spectroscopy. Journal of Agricultural and Food Chemistry, 54, 9713–9718. Boulet, J. C., Williams, P., & Doco, T. (2007). A Fourier transform infrared spectroscopy study of wine polysaccharides. Carbohydrate Polymers, 69, 79–85. Cozzolino, D., Cynkar, W., Shah, N., & Smith, P. (2011). Feasibility study on the use of attenuated total reflectance mid-infrared for analysis of compositional parameters in wine. Food Research International, 44, 181–186. Edelmann, A., Diewok, J., Schuster, K. C., & Lendl, B. (2001). Rapid method for the discrimination of red wine cultivars based on mid-infrared spectroscopy of

493

phenolic wine extracts. Journal of Agricultural and Food Chemistry, 49, 1139–1145. Feliciano, R. P., Bravo, M. N., Pires, M. M., Serra, A. T., Duarte, C. M., Boas, L. V., & Bronze, M. R. (2009). Phenolic content and antioxidant activity of Moscatel dessert wines from the Setúbal region in Portugal. Food Analytical Methods, 2, 149–161. Galtier, O., Dupuy, N., Le-Dréau, Y., Ollivier, D., Pinatel, C., Kister, J., & Artaud, J. (2007). Geographic origins and compositions of virgin olive oils determinated by chemometric analysis of NIR spectra. Analytica Chimica Acta, 595, 136–144. Katalinic, B. V., Milos, M., Modun, D., Music, I., & Boban, M. (2004). Antioxidant effectiveness of selected wines in comparison with (+) catechin. Food Chemistry, 86, 593–600. Mato, I., Suarez-Luque, S., & Huidobro, J. F. (2005). A review of the analytical methods to determine organic acids in grape juices and wines. Food Research International, 38, 1175–1188. Passos, C. P., Cardoso, S. M., Barros, A. S., Silva, C. M., & Coimbra, M. A. (2010). Application of Fourier transform infrared spectroscopy and orthogonal projections to latent structures/partial least squares regression for estimation of procyanidins average degree of polymerization. Analytica Chimica Acta, 661, 143–149. Roussel, S., Bellon-Maurel, V., Roger, J. M., & Grenier, P. (2003). Authenticating white grape must variety with classification models based on aroma sensors, FT-IR and UV spectrometry. Journal of Food Engineering, 60, 407–419. Silva, S. D., Rosa, N. F., Ferreira, A. E., Boas, L. V., & Bronze, M. R. (2009). Rapid determination of a-tocopherol in vegetable oils by Fourier transform infrared spectroscopy. Food Analytical Methods, 2, 120–127. Tarantilis, P. A., Troianou, V. E., Pappas, C. S., Kotseridis, Y. S., & Polissiou, M. G. (2008). Differentiation of Greek red wines on the basis of grape variety using attenuated total reflectance Fourier transform infrared spectroscopy. Food Chemistry, 111, 192–196. Urtubia, A., Perez-Correa, J. R., Pizarro, F., & Agosin, E. (2008). Exploring the applicability of MIR spectroscopy to detect early indications of wine fermentation problems. Food Control, 19, 382–388. Versari, A., Giuseppina, P. P., Scazzina, F., & Del Rio, D. (2010). Prediction of total antioxidant capacity of red wine by Fourier transform infrared spectroscopy. Food Control, 21, 786–789.