Quantification of cocaine in ternary mixtures using partial least squares regression applied to Raman and Fourier transform infrared spectroscopy

Abstract

AbstractThe use of chemicals is often related to suicide attempts and acute poisoning, which account for a significant number of hospital admissions. The differential diagnosis of patients exposed to poisoning is intricate and varies according to the substance used. The identification of drugs and drug abuse in cases of poisoning often requires time‐consuming and complex techniques, such as chromatography. The use of vibrational spectroscopy such as Raman spectroscopy and Fourier transform infrared spectroscopy (FTIR) enables the quick identification of toxic substances using no reagents or markers. This study aims to apply these techniques in the long term to identify and quantify substances found at the locations of suicide attempts, poisonings, and even drug consumption sites, including crack samples containing a cocaine base and adulterants commonly seized in drug busts, such as caffeine, lidocaine, and sodium carbonate. In this study, Raman and FTIR spectra of ternary mixtures of crack with caffeine and sodium carbonate and crack with lidocaine and sodium carbonate were obtained. These spectra were used for building multivariate models based on partial least squares (PLS) for determining the composition of the samples quantitatively. High correlation coefficients (r > 0.98) and small cross‐validation prediction errors (< 6%) were obtained for both Raman and FTIR spectral models. The results showed that PLS regression enables crack quantification in ternary mixtures using both Raman and FTIR spectroscopy, allowing for quick medical intervention or determination of the cause of death in cases of cocaine toxicity.