Qualitative and Quantitative Multivariate Analysis of Optical Density Spectra of Plastic Waste

Abstract

Five types of plastics were classified based on NIR optical density spectra using multivariate cluster analysis in principal component space. An accuracy above 0.96 was obtained by ranking spectral variables by decreasing variance of measured optical density. Changes in polycarbonate VIS-NIR spectra during thermal degradation were modeled using a partial least squares method with searching combination moving windows of optimal width. A quantitative calibration of an aging time up to 11.5 years was obtained with a root mean square error of the estimate around 19 days and a residual predictive deviation of more than 45.