RESUMO
In this study, samples coming from large seizures of cocaine which took place in Italian Customs areas during 2011 and 2012 were examined. Minor alkaloids and residual solvents, analyzed by gas chromatography-mass spectrometry (GC-MS) and head space (HS)-GC-MS, respectively, were processed by principal component analysis (PCA), highlighting groupings of samples according to their chemical similarity. A hypothesis about the geographical origin of samples was also provided: most of them were compatible with Colombia as country of origin. Results of these analyses were used as starting point for the development of a "fast profiling" method, based on Fourier transform infrared spectroscopy (FTIR) and chemometric tools. Two models were developed and compared: KBr-FTIR and attenuated total reflection (ATR)-FTIR for comparative analysis of pure samples. Linear discriminant analysis (LDA) was applied to the model based on ATR-FTIR spectroscopy, obtaining a classification and a prediction ability both of 97.56% for pure samples. Finally, "cut" samples were tested as an external test set, and the assignment class provided by LDA was compared with results obtained by the analyses of alkaloids and residual solvents: in the case of samples added with only one substance, prediction errors began to occur for percentages of cocaine lower than 50%.
