Hundreds of millions of tonnes of plastic are produced each year, only a fraction of which is currently recycled. This enormous quantity of polymer waste represents both a significant environmental challenge as well as an opportunity to recover valuable chemicals or materials. Many of the polymers produced today have been optimized for properties relevant to their primary use, without regard to their ability to be depolymerized or recycled. Intrinsically circular polymers (ICPs) are a promising class of materials that can be recycled at mild conditions through chemical conversion to monomer, making them an attractive alternative to conventional plastics. Ceiling temperature (Tc), which is dependent on enthalpy and entropy of polymerization, is an experimentally observable property that is valuable to predict because it is directly linked to a polymer’s thermodynamic recyclability. Determining the Tc of a system requires precise knowledge of the polymerizing system’s thermodynamics, which is difficult to predict using molecular modeling methods. This complexity arises from the interplay between monomer structure and the reaction environment. To rein in the high dimensional nature of this problem, we have used machine learning (ML) methods to predict enthalpy and entropy of polymerization using features that describe system-wide interactions while staying within the constraints of the design space. We propose a framework in which model predictions can be contextualized with the goal of predicting novel ICPs that exhibit thermodynamic recyclability, contributing to plastic waste reduction.
However, it is important to note that having a reasonable TC is necessary but not sufficient to achieve recyclability. Circularity is tied closely to kinetics, as kinetics determine yield and selectivity for depolymerization. To address this, we have developed a mechanistic kinetic modeling framework for quantifying the depolymerization of various classes of polymers. Continuum modeling based on the method of moments allows the complexity of the chain length distribution to be taken into account while enabling mechanistic formulations. To obtain even more structural detail for hyperbranched polymers, kinetic Monte Carlo (kMC) was used to track chain-specific detail, including the full structure of each polymer chain. In both cases, kinetic parameter estimates are derived from density functional theory using transition state theory. Transport limitations are accounted for by using an effective rate constant that was a function of both the intrinsic kinetics rate constant and a chain-length dependent transport rate constant. Finally, the results of the detailed models are consolidated into a simple measure, the unzipping length, to allow for rapid screening of potential polymers towards circular polymer design.
