Analysis of deformation mechanism of dual-phase Mg–Li alloy, processed by high-pressure torsion, using experiments and machine learning

Abstract Dual-phase Mg–Li alloys display interesting mechanical properties, including an unusual increase in ductility at low strain rates due to grain refinement. The effects of grain size, testing temperature and strain rate on the flow stress of an Mg-8wt.% Li alloy processed by high-pressure torsion were investigated through mechanical tests at low strain rates and through machine learning algorithms trained on data in the literature. The data are also compared to the deformation mechanism of grain boundary sliding. The experimental results and creep analyses show a high strain rate sensitivity and ductility and a low activation energy in the material, suggesting the dominance of thermal activated deformation mechanisms. The mechanism of grain boundary sliding is able to predict the trends revealed by experiments, while machine learning models suggest the temperature as the most significant parameter in flow stress. The ability of the black-box and white-box machine learning algorithms to make predictions and errors of models is discussed.