Análise de requeijão por espectroscopia raman e ferramentas quimiométricas

Abstract

In this work, FT-Raman spectroscopy has been explored as a rapid technique to evaluate requeijão samples. Exploratory analysis using Raman data as well principal component analysis (PCA) has been able to separate the dairy samples according to their classifications: traditional or light food. Multivariate analysis has also been employed to classify requeijão samples which contain starch. To build the models two types of samples have been used: commercial (pure and adulterated in laboratory) and manufactured in local industries, which have given the best results when compared to the first samples. PCA model was able to allocate in groups different samples with and without lipid content reduction, as well samples with and without starch. In the partial least square discriminant analysis all samples were correctly classified in terms of starch content, being necessary the selection of other variables in some of the models. Multivariate regression has been done through partial least square method, and the results are satisfactory. However, the built models have shown high prevision errors for samples without starch, which can be indicative of a false positive result of fraud. At last, the results show clearly that Raman spectroscopy with chemometric models can be used as quality control tool for dairy products, in particular requeijão.