Assessment of wool chlorination by infrared spectroscopy: II. The chemometric approach

Abstract

In this paper we have investigated the use of the chemometrics methods partial least squares (PLS), principal components regression (PCR) and principal component analysis (PCA) to develop both quantitative and qualitative correlations between mid-infrared attenuated total reflectance spectra and variation in chlorination treatments in wool samples. Quantitative correlations were obtained using both PLS and PRC that were significantly improved in terms of both standard errors of prediction and calibration correlation coefficient values over the calibrations obtained by classical least squares regression. These results reflect the variability and complexity of the samples produced by the industrial chlorination process. The chemometric models based on spectral regions that were truncated to include the frequencies where only the primary and intermediate oxidation products are active were found to be significantly better than those based on the full low wavenumber spectral region. PLS calibrations were used successfully to predict chlorination levels of samples obtained from commercial mills. These predictions were found to be in much better agreement with stain test results than those obtained using the least squares approach. The use of PCA to develop qualitative correlations between the chlorinated samples was investigated. It was demonstrated that PCA could be used as a tool to discriminate between samples according to their chemical treatment.