Design biomedical β-Ti alloys with exceptional strength-ductility balance via domain knowledge-based machine learning

Shiyu He, Fei Xiao, Leiji Li, Yang Liu, Yi Zeng, Mingyu Gong, Ying Zhou, Jing Han, Jiannan Liu, Xuejun Jin,
Design biomedical β-Ti alloys with exceptional strength-ductility balance via domain knowledge-based machine learning,
Acta Materialia,
Volume 301,
2025,
121550,
ISSN 1359-6454,
https://doi.org/10.1016/j.actamat.2025.121550.
(https://www.sciencedirect.com/science/article/pii/S1359645425008365)
Abstract: The optimization of strength and ductility in biomedical titanium alloys is critical for improving the performance of biomedical implants. This study presents a novel domain knowledge-based machine learning approach to design a Ti-15Zr-15Nb-1Fe biomedical β-Ti alloy, achieving an exceptional balance of 35 % elongation and 700 MPa yield strength. The phase constitution and microstructure were characterized using X-ray diffractometry, electron backscatter diffraction, and transmission electron microscopy. The study also explores the internal mechanisms of the machine learning model and investigates the relationship between slip systems and kink band formation. Results reveal that the evolution and interaction of multi-slip/kinking mechanisms promote uniform deformation and dynamically enhance the strain-hardening rate, leading to a synergistic improvement in strength and ductility. These findings underscore the potential of machine learning in accelerating the development of advanced biomaterials and provide mechanistic insights into deformation behavior, offering a pathway for designing next-generation biomedical implants.
Keywords: β-Ti alloys; Machine learning; Ductility; Kink band