Evaluation of a Tobacco Fractionation Procedure Using Pyrolysis Mass Spectrometry Combined with Multivariate Analysis

This paper presents an evaluation of a fractionation procedure for use with tobacco. Correlation coefficients calculated from the weights of the polymer fractions obtained and data obtained with classical wet-chemical methods show that these parameters have a low resolving power, which precludes any detailed distinction between tobaccos and tobacco-derived fractions. Pyrolysis mass spectrometry combined with multivariate analysis is presented as a promising approach for investigating the variability in the chemical composition of tobacco. Multivariate analysis performed on the pyrolysis mass-spectrometric fingerprints of all tobacco fractions indicates that the differences between the individual tobaccos are smaller than the differences induced by the fractionation procedure. Multivariate analysis of subfiles of the pyrolysis mass-spectrometric fingerprints of separate polymer fractions is an excellent method for classifying tobaccos. Leaf and stem material can easily be distinguished. A further differentiation is made between the flue-cured and Burley-type tobaccos. The strong clustering of tobacco samples after treatment with potassium hydroxide points to a uniform cell-wall skeleton. The greater variability in the ethanolized tobacco samples, and samples subjected to hot-water extraction or amylase digestion, is related to the cytosol characteristics. These fractions appear to reflect both the dissimilarities between the distinct tobacco types and phenotypic variations due to differences in cultural management. Pyrolysis mass spectrometry of the Klason lignin residues points to a large variety of chemical constituents unrelated to lignin. Pyrolysis gas-chromatography mass-spectrometric data on the Klason lignin residues of a Burley tobacco showed that the correlation made between the organic-nitrogen content and the protein content of this fraction might be misleading. As a consequence the maximum value for the lignin content of this tobacco is estimated incorrectly.