Achieving dynamic strength-plasticity synergy in refractory high-entropy alloys via machine learning

Dynamic strength-plasticity synergy in refractory high-entropy alloys (RHEAs) is realized via a machine-learning-driven design strategy in this study. High-precision prediction models for dynamic compressive yield strength and fracture strain are constructed and validated, and the Ti 18 V 24 Nb 12 Ta 20 W 26 alloy designed through Pareto frontier multi-objective optimization is prepared. At a strain rate of ∼3000 s −1 , the alloy exhibits a dynamic compressive strength of 1905 MPa and a fracture strain of 30%, presenting simultaneous enhancement in strength and plasticity relative to conventional TiVNbTaW-based RHEAs. Machine learning identifies a non-equiatomic Ti-V-Nb-Ta-W composition that balances W-driven strengthening and deformation-mode activation. The designed alloy avoids premature adiabatic shear failure by activating cross-slip-assisted dislocation glide, {112} twinning, twin intersection, and local BCC-to-ω phase transformation, thereby achieving improved dynamic strength-plasticity synergy.