The University of Western Ontario

Published purpose

Biomimicry, adapting and implementing nature's designs, provides a first-order solution to achieving superior mechanical performance. NRC’s recent study, showed laser engraved architectured ceramics to have 50 times more quasi-static energy absorption and 70% more impact resistance compared to plain ceramics. However, the design space for further enhancements is too vast even using biomimetic designs as prototypes. The main objective of this research is to propose a new approach to design architectured ceramics using machine learning, trained with a database of hundreds of structures from finite element analysis, together with a self-learning algorithm for discovering high-performing ceramics where inferior designs are phased out for superior candidates. Preliminary results, validated through additive/subtractive manufacturing and experiments, show that this approach can create microstructural patterns that lead to tougher and stronger ceramics.