文摘
Brazilian cachaças (115 samples; 73 samples derived from distillation in copper pot stills, 42 samplesderived from distillation in stainless steel columns), collected directly at the producers, were analyzedfor the contents of 34 constituents by chromatography, inductively coupled plasma optical emissionspectrometry, and atomic absorption spectrometry. The analytical data were subjected to principalcomponent analysis (PCA) and linear discriminant analysis (LDA). The PCA treatment led todiscrimination of the two groups of cachaças, explaining 65.0% of the database variance. Using LDAand ethanal, ethyl carbamate, dimethyl sulfide, isobutyl alcohol, n-propanal, copper, ethyl acetate,and phenylmethanal as chemical descriptors, a model was developed that presented 95.1% accuracyin predicting the type of distillation apparatus.