Classification of polymeric materials by evolving factor analysis and principal component analysis of thermochromatographic data

Thermal decomposition of different polymeric materials was investigated by thermochromatography (ThGC), a temperature programmed pyrolysis chromatogra phic method. ThGC produces two-dimensional results; the co-ordinates of whi ch are the retention time and the pyrolysis temperature at the time of samp ling. Therefore, principal component analysis (PCA), on results from evolvi ng factor analysis (EFA) successfully applied would decompose the complete data of each run into two parts: 'thermograms' and 'chromatograms'. Factor analysis at this stage compresses the data, making it more convenient for f urther analysis of the data structure composed of a few dozen of samples. T he aim of this stage of the data analysis process is to extract 'real therm ograms' as close as possible to the corresponding 'thermograms' - answering the question "which products are evolved at each temperature." Combination of 'chromatograms' and related 'thermograms' obtained on the fi rst stage were used as characteristic vectors in the further analysis. Sets of significant 'thermograms'-'chromatograms' were subjected to PCA. Mappin g of the polymeric samples onto planes defined by factors allows one to ide ntify clusters as related to different classes of polymers, as well as diff erent mechanisms of their thermal decomposition. The data was proven to giv e a very good basis for characterization of the samples by their polymer co ntent. (C) 2000 Elsevier Science B.V. All rights reserved.