Analysis of Stability and Mechanical Properties of Al-Cr-Fe-Co-Ni-Cu HEA Based on Machine Learning

A machine learning (ML) interatomic potential was developed to predict the behavior of complex high-entropy alloys (HEAs) containing Al, Cr, Fe, Co, Ni, and Cu. This ML potential accurately reproduces DFT calculations, allowing for extensive molecular dynamics simulations. We analyzed the mechanical properties of face-centered cubic (FCC), body-centered cubic (BCC), and polycrystalline HEA phases at different element concentrations and temperatures. The ML potential accurately predicts key system parameters such as lattice constants and elastic moduli and describes the stress–strain material behavior. Strain analysis shows a complex phase transformation and dislocation formation in the FCC phase under tension. High Co and Ni concentrations improve the mechanical properties of the FCC phase. Polycrystalline structures form BCC-HCP mixtures, highlighting the dependence of their stability on the element concentration. Temperature-dependent simulations show a transition from the FCC phase to the amorphous phase at 1200 K. Interestingly, Cu-rich clusters play a crucial role in stabilizing the FCC phase at lower temperatures. This study demonstrates the efficiency of the ML potential for studying HEAs and unveils the complex relations between element composition and the resulting mechanical properties.