Machine learning driven discovery of low modulus biomedical titanium alloys for additive manufacturing

Designing alloys for additive manufacturing opens opportunities for next-generation biomedical orthopaedic implants. However, most biomedical alloys currently in use are legacy compositions that cannot fully harness the potential of additive manufacturing. Here, we present a machine-learning-driven computational framework for discovering additive-manufacturing-specific β-titanium alloys with low modulus, optimised printability, mechanical and corrosion properties, and biocompatibility. Using this framework, an additive-manufacturing-specific Ti-Nb-Ta-Zr-Sn alloy was designed and validated via laser powder bed fusion. The alloy shows good printability and reduced sensitivity to keyhole pore formation compared with commercial Ti-6Al-4V. Under optimised process parameters, the as-built alloy combines a low Young's modulus (~42.7 GPa) with high ductility (~30.9%), attributed to its metastable β-phase microstructure, cubic <001> texture, and reduced dislocation density. In this work, we show that this machine-learning-driven approach can facilitate the efficient design of biomedical β-titanium alloys, offering promising opportunities for advancing additive manufacturing in medical and related applications.