Electronic nose for wine aging detection

This paper reports a novel application of an electronic nose (e-nose) for recognition and detection of wine ageing. Two different measurements are performed with the following samples: first, in an experimental cellar the same wine is aged in different type of oak barrel (French and American oak) and during different time (0, 3, 6 and 12 months) and second, several wines made with the same grape variety and from different wine cellars aged in French and American oak. This identification has a great importance for origin denominations for control of frauds. The e-nose is home-developed and home-fabricated for this purpose: a tin oxide multisensor prepared with RF sputtering onto an alumina substrate and doped with chromium and indium is used. The sampling method employed is static headspace followed by a dynamic injection. Linear techniques like principal component analysis (PCA) and nonlinear ones like probabilistic neural networks (PNN) are used for pattern recognition. A classification success rate (correct predicted number over total number of measurements) of 97% and 84% is achieved in detection of the different ageing process experimented by the wines tested.