Characterization of the effects of different roasting conditions on coffee samples of different geographical origins by HPLC-DAD, NIR and chemometrics

Abstract

The effect of roasting conditions on the NIR and HPLC profiles of coffee samples was systematically evaluated by ANOVA-simultaneous component analysis (ASCA), also as a function of the varietal origin (Arabica or Robusta) of the beans. With respect to previously published paper, the use of ASCA allowed not to limit the characterization to individual analytes or specific family of compounds, by extending it to whole instrumental profiles, thus providing a holistic characterization of the roasting process, also in terms of how it changes depending on the species of the raw material. In particular, it was evidenced that both the species and the roasting time have a significant effect on the fingerprints and that, only in the case of NIR spectra, their interaction is significant as well. Moreover, it was possible to highlight what are the main variations in the experimental profiles ascribable to the effects of the controlled factors and in particular, in the case of roasting time, to discover a major longitudinal monotonic trend which is accompanied by a minor one that can be associated to the formation of intermediates.In a successive stage, based on the outcomes of ASCA, partial least squares-discriminant analysis (PLS-DA) and soft independent modeling of class analogies (SIMCA) have been used to build classification models to authenticate the varietal origin of coffee beans by NIR spectroscopy. PLS-DA resulted in about 98% correct classification rate on the test set (100% for Arabica and 95% for Robusta), while sensitivity and specificity values almost always above 90% were achieved with SIMCA (93% and 96%, respectively, for Arabica and 77% and 96% for Robusta).