Prediction and Explanation of Properties in Multicomponent Polyurethane Elastomers: Integrating Molecular Dynamics and Machine Learning

Abstract

Establishing quantitative connections among the chemical composition, molecular structure, and macroscopic properties of multicomponent polyurethane elastomers remains a challenging task. Molecular dynamics (MD) has been extensively utilized in the study of various materials and serves as a crucial tool for exploring the relationship between structure and properties. However, the intricate modeling process and lengthy computation times associated with the MD method complicate the attainment of complex combinatorial results for the various components of polyurethane elastomers. Machine learning (ML) offers a solution by integrating and analyzing existing data, along with the capability to predict new outcomes. Consequently, we combine MD and ML methods to conduct a comprehensive investigation of multicomponent polyurethane elastomers. MD simulations indicate the presence of various types of hydrogen bonds within the elastic matrix of polyurethane, and the strong hydrogen bonds formed in the hard segments significantly affect the tensile properties of material. While the incorporation of long molecular chains in the soft segments enhances the material’s flexibility, it simultaneously diminishes its tensile strength. Feature engineering techniques, including parametric representation and feature screening of the MD model, were employed to create a data set suitable for ML applications. The application of the interpretable ML method has demonstrated that the number of hydrogen bonds in the hard segment is regulated by the hydrogen bond donor and acceptor, while the rotatable bonds in the soft segment are the primary characteristics contributing to the material’s flexibility and are also key factors that regulate the number of free hydrogen bonds. This integration of MD and ML methods not only enhances predictive capabilities for novel polyurethane elastomers but also facilitates quantitative analysis of how microstructural characteristics affect macroscopic properties.