Machine-Learning-Assisted Design of Mechanically Robust Room-Temperature Self-Healing Epoxy Resins

Epoxy resins are the most widely used thermosets, yet they typically lack the capability to self-heal at room temperature due to their molecular chains and networks being immobilized in a glassy state. Herein, machine learning identifies fractional free volume as a crucial factor for enabling self-healing in the glassy state. Guided by this insight, we designed an epoxy network incorporating dangling chains together with numerous hydrogen bonds and aromatic disulfide bonds. The dangling chains introduce large free volume, facilitating the reorganization of hydrogen bonds and the radical-mediated exchange of aromatic disulfide bonds, thereby imparting prominent self-healing capability at room temperature. Notably, the damaged epoxy not only can recover 81.2% of its tensile strength without intervention but also can autonomously and completely eliminate electrical tree damage and scratches at room temperature. Under mild compression, 100% healing occurs within tens of minutes in the glassy state. Additionally, the optimized epoxy exhibits high physicomechanical properties with a tensile strength of 42.1 MPa, a modulus of 2.9 GPa, and a glass transition temperature of 53.2 °C. Its ability to self-heal both electrical tree and mechanical damage at room temperature positions this epoxy as a promising material for advanced insulating and sealing applications.