Ferramentas de análise multivariada associadas as espectroscopias NIR e 1H-NMR na avaliação de cafés comerciais

Abstract

This dissertation presents the use of different chemometric methods for analysis of roasted and ground coffee. In summary, alternative methods for analyzing quality parameters and sensory attributes of commercial coffee by Near Infrared (NIR) Spectroscopy and by the Partial Least Squares (PLS) were reported. Additionally, a discrimination model for different coffees quality was successfully proposed using NIR spectroscopy combined with Partial Least Squares with Discriminant Analysis (PLS-DA). In both cases, a validation by the determination of the parameters of merit showed that all prediction models presented satisfactory performance. In addition, a metabolomic analysis using Nuclear Magnetic Resonance (1H-NMR) spectroscopy was proposed in two studies. In the first one, the information obtained by the 1H-NMR were correlated with the quality parameters, the sensory properties including the quality of the coffees and the spectral information of the NIR using the method ComDim. The results showed that there is a strong correlation between the information 1H-NMR spectroscopy with the sensory properties and the absorptions in the NIR and a lesser correlation with quality parameters. In all the applications of this dissertation, the combination of NIR and 1H-NMR spectroscopy techniques in conjunction with the different chemometric methods used, presented as interesting alternatives for analysis of roasted and ground coffee.