Semiquantitative model for chemometric detection of Brazilian cachaça adulterated with ethanol fuel: A portable near infrared-based approach.

This study presents the development and validation of a rapid and direct chemometric method for detecting the adulteration of cachaça, a typical Brazilian spirit made from fermented and distilled sugarcane juice, with ethanol fuel using a portable near-infrared (NIR) spectrometer. Chemometric models were built employing principal component analysis (PCA), soft independent modeling of class analogies (SIMCA) and partial least squares discriminant analysis (PLS-DA). A total of 124 cachaça samples of different origins were used to build the models, 62 authentic and 62 deliberately adulterated with ethanol fuel (in the range 1-40 % v/v, with a 40 % ethanol solution). While SIMCA model was more suitable to detect cachaça authenticity, PLS-DA was employed for detecting and characterizing the specific adulteration with ethanol fuel. These samples were split into 82 for the training set and 42 for the validation set for the PLS-DA model. This model achieved an accuracy of 100 %, showing no false positive or false negative predictions in both sets. It was spectrally interpreted by identifying the most discriminant molecular vibrations through informative vectors. The semiquantitative model was validated by estimating the detection capability (CCβ) at 0.03 % (v/v) and the decision limit (CCα) at 3.50 % (v/v), which indicated the method is reliable for detecting adulteration only above uncertainty region. Furthermore, the proposed model was validated with an external test set composed of 20 samples, 12 reference cachaças and 8 adulterated samples seized by the Brazilian police in a specific operation.