(693i) Integrating Molecular Simulation with Machine Learning to Investigate Microstructure-Property Relationships in Copolymers

Copolymer materials exhibit various arrangements of monomers in the chain, leading to a broad spectrum of mechanical, thermal, electronic and chemical properties.¹ However, the complex interplay between the microstructural characteristics of different monomers and resultant mechanical and thermal properties hinders the development of advanced copolymer materials.² This work introduces a comprehensive methodology designed to unravel these intricate relationships. We commence by adapting the kinetic Monte Carlo method to fine-tune critical features of a Poly(methyl methacrylate) and Poly(butyl acrylate) copolymer system (PMMA-PBA), including monomer ratio, sequence distribution, dispersity, and chain length. Subsequently, we employ coarse-grained molecular dynamics simulations (CGMD) to evaluate critical properties, including mechanical and thermal properties, such as the glass transition temperature (), tensile strength, and Young's modulus. Machine learning algorithms are then applied to elucidate the latent microstructure-property correlations. Preliminary findings indicate that the machine learning model adeptly captures the nonlinear associations between microstructure and properties, with monomer ratio and chain length emerging as the most critical factors influencing these mechanical performances. Our integrated method not only sheds light on the synergistic effects of various microstructures on copolymer properties, but also paves the way for exploring a broader chemical space to discover materials with enhanced functionalities. This holistic approach establishes a potential framework for the design and synthesis of next-generation copolymers with optimized performance.