Abstract

Chemma, an oral moist snuff widespread used in Algeria, is prepared from grounded tobacco and inorganic salts or other plant materials. In addition to its adverse health issues, Chemma contains unknown amounts of nicotine and other harmful constituents. Twenty seven samples of commercial products and tobacco leaves varieties were characterized in the aim to quantitatively determine total nicotine by attenuated total reflectance Fourier transform infrared (ATR-FTIR) spectroscopy coupled with two regression methods (univariate and multivariate). For this purpose, a highly specific-to-analyte extraction method was applied on calibration and test sets of standards for a wide concentration range. The optimal analytical models were evaluated by selecting the appropriate spectral measurement, spectral pre-processing strategies and critical figures of merit estimated according to the latest approaches. Then, the predictive ability of the method was demonstrated with the elliptical joint confidence region test and with comparing to results in literature. The linear baseline correction and the extended multiplicative scatter correction pre-processing of the combined reduced spectral regions 1333–1299 and 740–690 cm−1 followed by partial least squares regression (PLSR) offered the best analytical outcomes. Using this latter, total nicotine content in the manufactured smokeless tobacco (ST) products have been found ranging from 3.9 to 12.6 mg/g of dry weight; while in the ST leaves, it was between 5.9 and 45.0 mg/g, dry weight basis. The developed method, besides to the high predictive performance, provides advanced statistical tests that can qualitatively assess the target analyte in real sample spectra, allowing more reliable result. With the increasing number of new products containing complex agents and adulterants, the aforementioned method can be successfully adapted to routine analyses for rapid quality control of nicotine in ST and cigarettes.