Tribological behavior of Sr and Yb modified AZ91 magnesium alloy using machine learning prediction and multi-objective optimization for biomedical applications

Magnesium (Mg) alloys are attractive biodegradable implant materials, but inadequate wear resistance and tribological instability in physiological environments remain major limitations. To address this challenge, this study investigates the effect of dual Sr–Yb modification on the tribological performance and surface stability of AZ91 Mg alloy. Specifically, the wear, friction, and surface degradation behavior of the Sr–Yb–modified AZ91 alloy were evaluated under simulated physiological conditions representative of implant-relevant biomechanical and biochemical environments. The alloy was synthesized via ultrasonic-assisted stir casting to ensure homogeneous dispersion and grain refinement. A D-optimal experimental design involving six factors, normal load (N), sliding speed (m/s), sliding distance (m), simulated body fluid (SBF), soaking time (hrs), and temperature (°C), was employed to evaluate the tribological response systematically. A stacked ensemble learning model was developed to predict specific wear rate (SWR), average coefficient of friction (COF), and surface roughness (R a ), achieving high R 2 values (>0.89). Multi-objective optimization using a surrogate-assisted Tree-Seed Algorithm (MOTSA) identified near-optimal process conditions yielding SWR ≈ 0.11991 mm 3 /N-m, COF ≈ 0.344, and Ra ≈0.185 μm. Experimental validation showed deviations within 5% of predicted values. Time-resolved COF analysis revealed transitional friction behavior, while post-wear profilometry and field emission scanning electron microscope (FESEM) imaging identified hybrid wear mechanisms, including delamination, oxidative smearing, and debris-induced microcracking. The integrated experimental-modeling framework demonstrates the potential of Sr and Yb-modified AZ91 alloy as a bio-tribological material, offering high wear resistance, stable frictional performance, and structural integrity under simulated body conditions.