Data-inspired atomic environment-dependence of vacancy formation energy in high-entropy alloys

FS Tan and X Liu and XF Liang and YN Cui, INTERNATIONAL JOURNAL OF PLASTICITY, 195, 104545 (2025).

DOI: 10.1016/j.ijplas.2025.104545

Vacancy properties in High-entropy alloys (HEAs) play a critical role in governing hightemperature microstructural stability, yet the fundamental relationship between Vacancy Formation Energy (VFE) and heterogeneous Local Atomic Environments (LAE) in HEAs remains far from well understood, owing to the complex and heterogeneous nature of LAE. To address this, we developed an interpretable machine learning framework integrating high-throughput molecular dynamics simulations and physics- informed features. Using CoNiCrFeMn as model system, our approach achieves exceptional prediction accuracy (R2 = 0.98) for VFE. It is found that the LAE within the first-nearest-neighbor shell around vacancy dominates VFE variations, and the local atomic spatial ordering exerts influence on VFE comparable in magnitude to local chemical composition. Based on the designated LAE descriptor, namely multilevel element pair probability, and feature analysis-guided physics interpretation, we identify for the first time the physical origin of LAE-mediated VFE as the synergistic strong/weak-bond elements competition and lattice distortion effects. Specifically, coexisting strong-bond (e.g., Ni) and weak-bond (e.g., Mn) atoms in 1NN shell around central vacancy drive offsetting displacements through lattice distortion, dynamically tailoring VFE. The mechanism explains anomalously high lattice distortion and elevated vacancy concentrations observed in Mn-containing CoNiCrFeMn HEAs, and further enables a strategy for enhancing vacancy stability via annealing-induced elemental aggregation. These results establish a theoretical framework for defect engineering in the design of complex solid-solution alloys.

Return to Publications page