Pattern recognition methods in the study of thermal decomposition of α-amino acids

Abstract

There are many examples in the literature of a strict relation between the pathways of decomposition of a drug substance and chemical structure of its molecule. For this reason, a study has now been performed on the relation between thermal decomposition of α-amino acids and their chemical structure. To achieve this goal, a group of a dozen or so compounds was chosen at random, and the results obtained using the DTA, TG and DTG analyses of their thermal decomposition were interpreted by highly advanced multivariate methods, principal component analysis and cluster analysis. By this statistical analysis, the influence of specific functional groups on thermal decomposition of α-amino acids was determined. It has been found that first two principal components explain together more than 75 % of variance, and in an exceptional case, about 90 %. The third stage of decomposition was that at which the thermoanalytical data were best correlated with chemical constitution of a compound. It has also been recognized that a better discrimination among the analysed compounds was obtained for the DTA data set. The results can be useful for identification of a relation between the pathway of degradation of a drug substance and chemical structure of its molecule, and for predicting chemical stability of the compounds studied.