Correlating Near-Infrared Spectra to Bulk Properties in Polyolefins

The industry standard for sorting plastic wastes is near-infrared (NIR) spectroscopy, which offers rapid and nondestructive identification of various plastics. However, NIR does not provide insights into the chain composition, conformation, and topology of polyolefins. Molar mass, branching distribution, thermal properties, and comonomer content are important variables that affect final recyclate properties and compatibility with virgin resins. Heterogeneous mixtures arise through sorting errors, multicomponent materials, or limits on differentiation of polyolefin subclasses leading to poor thermal and mechanical properties. Classic polymer measurement methods can quantify physical properties, which would enable better sorting; however, they are generally too slow for application in commercial recycling facilities. Herein, we leverage the limited chemistry of polyolefins and correlate the structural information from slower measurement methods to NIR spectra through machine learning models. We discuss the success of NIR-property correlations to delineate between polyolefins based on topology.