Machine learning-aided design of two novel high entropy alloys with enhanced corrosion resistance for orthopedic implants

Abstract

This study showcases machine learning (ML) as a promising tool for developing new biomedical alloys with enhanced corrosion resistance. Specifically, two novel high entropy alloy (HEA) compositions, namely Ti34.8Ta17Nb21.4Zr14.2Mo12.6 and Ti35Ta23Nb20.8Zr14.2Mo7, were predicted as the optimum HEA compositions with enhanced corrosion resistance for orthopedic applications. For validation purposes, potentiodynamic polarization experiments were conducted on the predicted compositions in phosphate buffered saline (PBS) solution at human body temperature, demonstrating the new alloys' superior corrosion properties with the corrosion potential (Ecorr) of-0.40 V +/- 0.05 and-0.46 +/- 0.02 V, respectively. The samples were then subjected to static immersion experiments in PBS for 28 days to gather insight into ion release and for the sake of an initial assessment of the biocompatibility of the new HEAs. The results of this study demonstrate that by employing an optimal combination of feature selection and machine learning models, ML proves to be a powerful tool for predicting HEA compositions with superior corrosion resistance, as evidenced by the close correlation between experimental findings and predicted values reported herein.