(147z) Harnessing Molecular Simulations and Machine Learning to Predict Macromolecular Properties and Drive Efficient Polymer Recycling

Through the integration of molecular dynamics simulations and machine learning, we have developed a computational framework that enables the prediction of polyolefin properties across a wide parameter space. In this work, we consider the effect of composition, chemistry, temperature and pressure in the transport and thermodynamic properties of polyolefin-like melts.

In the context of polyolefin deconstruction, we used this approach to predict the diffusion coefficients of binary mixtures of linear polyethylene chains with its oligomers as a function of molecular weight, concentration, pressure, and temperature. Our results allow us to obtain insight into the kinetics that affect polyolefin deconstruction during the evolution of the melt.

In the context of sustainable polymer design, this methodology allowed us to predict thermodynamic and transport properties of polyolefin-like materials featuring cleavable bonds within their structure. By leveraging the predictive capabilities of our framework, we can accelerate the design and development of innovative plastic materials with tailored properties to enhance circularity and at the same time meet industry requirements.

Our research presents an opportunity for industries to leverage the synergy between molecular simulations and machine learning, enabling accurate predictions and comprehensive exploration of macromolecular systems. By harnessing this integrated approach, companies can optimize processes and make significant strides in enhancing product performance.