Bayesian optimization of laser-aided fracture toughness specimen preparation

Reliable lifetime prediction of ceramic components requires fracture toughness measurements supported by sharp, reproducible artificial defects. Conventional notching techniques often lack consistency, are time-consuming, and are particularly limited for fine-grained ceramics and geometrically complex specimens. This study investigates a modified femtosecond-laser vaporization approach that targets an ablation area rather than a single line. To efficiently explore the complex, material-dependent parameter space, Bayesian optimization was employed for zirconia and both conventionally and additively manufactured alumina. The method rapidly identified suitable parameters for zirconia, achieving sharp, smooth notches within only a few experimental iterations. In contrast, alumina exhibited unsuitable fracture behavior, with failure initiating at notch flanks rather than roots. Overall, laser notching demonstrated clear advantages in speed, reproducibility and geometric flexibility, including applicability to complex designs such as the CharAM specimen. Further research is required to ensure consistent root-initiated fracture and to expand understanding across a wider range of materials.