Impedimetric electronic tongue based on nanocomposites for the analysis of red wines. Improving the variable selection method

An impedimetric electronic tongue to discriminate red wines with similar characteristics (same region, vintage and ageing method) is developed. The excellent performance achieved is due to the remarkable sensing properties of the nanocomposites forming the array and to the methodology developed to extract the input variables used in the statistical analysis. The multisensor system is formed by one PEDOT:PSS sensor and two nanocomposites formed by layers of PEDOT:PSS and gold nanoparticles (PEDOT:PSS/AuNP) or layers of PEDOT:PSS and lutetium bisphthalocyanine (PEDOT:PSS/LuPc2). The well-known electrocatalytic activity of the sensing materials is promoted in the layered composites and enhances the cross-selectivity of the sensors. Besides, a feature extraction method that uses an equivalent electric circuit model to obtain the input variables used in the statistical analysis has been developed. By using the statistical procedure of Principal Component Analysis, the improved electronic tongue is able to discriminate three wines from the variety Tempranillo, one wine of the variety Tinta de Toro (a clon of Tempranillo) and one wine elaborated with a coupage of 90% Tempranillo and 10% Garnacha variety. The combination of layered nanomaterials and the new feature extraction method also enhances the capability of the impedimetric electronic tongue to predict chemical variables using Partial Least Squares regression.