Analysis of car shredder polymer waste with Raman mapping and chemometrics

A novel evaluation method was developed for Raman microscopic quantitative characterization of polymer waste. Car shredder polymer waste was divided into different density fractions by magnetic density separation (MDS) technique, and each fraction was investigated by Raman mapping, which is capable of detecting the components being present even in low concentration. The only method available for evaluation of the mapping results was earlier to assign each pixel to a component visually and to count the number of different polymers on the Raman map. An automated method is proposed here for pixel classification, which helps to detect the different polymers present and enables rapid assignment of each pixel to the appropriate polymer. Six chemometric methods were tested to provide a basis for the pixel classification, among which multivariate curve resolution-alternating least squares (MCR-ALS) provided the best results. The MCR-ALS based pixel identification method was then used for the quantitative characterization of each waste density fraction, where it was found that the automated method yields accurate results in a very short time, as opposed to manual pixel counting method which may take hours of human work per dataset