This study presents a novel damage-programmable metamaterial design method inspired by nature's crack-resisting mechanisms. Using machine learning, the researchers developed a design engine to generate metamaterials with engineerable microfibers, enabling spatial control of micro-scale crack behavior. The resulting metamaterials exhibited enhanced toughening functionalities (crack bowing, deflection, shielding), increasing absorbed fracture energy by up to 1235% compared to conventional metamaterials. This approach holds significant implications for damage-tolerant materials and lightweight engineering systems.

Zhenyang Gao, Xiaolin Zhang, Yi Wu, Minh-Son Pham, Yang Lu, Cunjuan Xia, Haowei Wang, Hongze Wang