A new data analysis approach for an AgNPs-modified impedimetric bioelectronic tongue for dairy analysis

Abstract

An alternative approach to analyze complex matrices, such as milk, lies on multivariate description of the chemical composition of samples, where fingerprints for each sample are generated using data mining methods (Principal Component Analysis, PCA, and Supported Vector Machine). Electronic tongues (ET) hold substantial potential for the dairy industry as analytical tools due to their fast analysis. The purpose of this work was to create an impedimetric ET with an array of microelectrode sensors for application in the dairy industry. A sensor array with enhanced sensitivity and selectivity silver nanoparticles and enzymes were developed. Higher sensitivity was evident in the responses of this array to milk components of interest like glucose, galactose, lactose, and urea. PCA of the signals obtained using the optimized ET has allowed the discrimination of milks with different characteristics. SVM was used to stablish correlations between the signals obtained from the ET and the physicochemical parameters. Electrochemical Impedance Spectroscopy have also shown that the ET in combination with the equivalent circuits approach could be a potential tool for its further application in dairy analysis. Moreover, the study of milk samples confirmed the high cross-selectivity of the system which allowed the development of classification models and to stablish correlations between the ET and physicochemical parameters.