Anisotropic response under shock compression in refractory high-entropy alloy via machine-learning potential

• Trained a DP potential model that can be used to study the anisotropy of shock compression in TiZrNbTa. • TiZrNbTa exhibits strong orientation-dependent deformation during shock loading, unloading, and spallation. • Spall strength anisotropy is linked to void nucleation timing, with premature nucleation causing the weakest spall. The anisotropic shock response of BCC-structured refractory high-entropy alloys (RHEAs) single crystals fundamentally governs their dynamic deformation and failure. Given experimental challenges in observing orientation-dependent shock behavior, we employed molecular dynamics with a Deep Potential machine-learning potential to simulate TiZrNbTa RHEA under 800 m/s impact. The [1 1 0] orientation exhibited twinning-induced plasticity during compression and formed residual strain-induced distortion zones that triggered severe amorphization upon unloading. Premature void nucleation (32.2 ps) in amorphous regions before peak stress (35.2 ps) caused thermal softening and catastrophic void proliferation, yielding minimal spall strength (12.20 GPa). Conversely, [1 0 0] underwent elastic compression followed by BCC → FCC transformation during unloading, suppressing amorphization and achieving intermediate spall strength (15.12 GPa). The structurally stable [1 1 1] orientation showed minimal amorphization and maximal spall strength (15.25 GPa), revealing unloading-induced microstructural evolution as the dominant factor in anisotropic spall failure for impact-resistant RHEA design.