Rapid determination of main chemical components in tobacco leaf by NIR diffuse reflectance spectroscopy

FT-NIR was evaluated as a method to simultaneously analyze 17 chemical components in tobacco leaf. Spectra of 700 typical samples were collected to develop the calibration models, among which the raw spectra were pretreated with first derivative and partial least square. About 50 randomly selected validation samples were used to test the accuracy of each calibration model. The values of root mean square error of prediction (RMSEP) for total volatile acids, total volatile bases, total petroleum ether extracts, petroleum ether extracts neutral, polyphenol, starch, cellulose, sulfate, pH, ash, water soluble ash bases, total sugar, reductive sugar, total nitrogen, alkaloids, chlorine, and potassium were 0.020, 0.009, 0.402, 0.393, 0.578, 0.583, 0.932, 0.139, 0.117, 0.634, 0.235, 1.720, 1.407, 0.104, 0.173, 0.037, 0.300 respectively. The satisfactory results indicated that FT-NIR could be an alternative to the traditional wet chemical method to simultaneously analyze the above 17 chemical components in tobacco leaf. The utilization of FT-NIR as a quality control method will reduce costs and enhance efficiency for the tobacco industry.