Physics-Guided Neural Networks for Transferable Prediction of Polymer Properties

Abstract

The architectural, compositional, and chemical complexities of polymers are fundamentally important to their properties; however, these same factors obfuscate effective predictions. Machine learning offers a promising approach for predicting polymer properties, but model transferability remains a major challenge, particularly when data is insufficient due to high acquisition costs and practical limitations. We explore the integration of polymer physics theory with machine learning architectures to enhance the predictive capabilities of polymer properties. Using a dataset of 18,450 polymers with diverse architectures, molecular weights, compositions, and chemical patterns, we focus on transferability tasks for predicting moments of the distribution of squared radius of gyration. Our tandem model, GC-GNN, which combines a graph neural network with a fittable model based on ideal Gaussian chain theory, surpasses both standalone polymer-physics and graph neural network models in predictive accuracy and transferability. We also demonstrate that predictive transferability varies with polymer architecture due to deviations from the ideal Gaussian chain assumption. This study highlights the potential of combining polymer physics with data-driven models to improve predictive transferability across diverse conditions and also pathways for improvement.