Microscopic Mechanisms of Self-Healing in Polymers Revealed by Molecular Simulations and Machine Learning

At present, there is still a lack of in-depth research into the relationship between structural changes during the healing process and self-healing efficiency of self-healing polymers. In the current work, molecular dynamics (MD) simulations and machine learning methods were applied to investigate the microscopic mechanisms of self-healing in polymers. Based on MD simulations, it was found that chain segments and the entire chains diffuse into the crack region during healing, together with the reformation of reversible interactions. Segment diffusion is closely related to the reconstruction of reversible interactions, and chain diffusion can lead to the interpenetration of chains. Based on machine learning methods, it was demonstrated that chain diffusion (the interpenetration of chains) plays the most crucial role in influencing the self-healing efficiency of polymers. Furthermore, based on the Random Forest method, a quantitative relationship between the structural changes in the healing process and the self-healing efficiency can be established.