Removal of solvent effect from Raman and NIR spectra for real-time monitoring of dissolved polypropylene content across multi-solvent systems during dissolution-based recycling

A methodology is developed for real-time monitoring the dissolved polypropylene concentration during the polypropylene dissolution recycling process across several solvents. The selection of the most appropriate solvents is based on the thermodynamic dissolution properties of polypropylene and the solvents. In-situ Raman and near infrared (NIR) spectroscopy are employed for in-situ monitoring. First, partial least square (PLS) regression is used to develop several single-solvent models to predict the dissolved polymer concentration in the solvent. Then, based on the same data, a generalized multi-solvent model is developed to operate across several predefined solvents, mixture of solvents or new comparable solvents. To do so, an external parameter orthogonalization (EPO) approach is employed to eliminate the influence of the solvent on the spectra. The method allows to monitor in-situ the concentration of polypropylene dissolved in several types of solvents, not used during the model development.