Thermophysical Property Prediction of Polyethylene-like Materials and Their Blends

The accurate determination of bulk thermophysical properties of polymers using only molecular structure and composition is a topic of critical academic and industrial interest. However, the ability to predict multi-molecule processes in polymer systems remains challenging for molecular modeling. In this context, polyethylene (PE) constitutes not only the major fraction of global plastic production but also an excellent system to develop structure-property and composition-property relationships. In this work, we propose a workflow to understand and predict thermophysical properties of PE-like materials and their blends using supervised Machine Learning (ML) techniques. We generate a data set from atomistic molecular dynamics [RA1] simulations and train a ML model in order to obtain accurate predictions. Here we present two important case studies in the context of chemical recycling where this workflow can be used to accelerate both circular polymer design and commercial polyolefin hydrocracking. For the former, we study ester-linked PE and its blends. For the latter, we modeled PE chains with smaller alkane chains in the presence of hydrogen. Our results show how quantitative predictive models can aid the design of sustainable solutions for plastic innovation.