Deformation mechanisms of AlCoCrCuFeNi: A molecular dynamics and machine learning approach

Abstract

High-entropy alloys (HEAs) distinguish themselves from other multi-component alloys through their unique nanostructures and mechanical properties. This study employs molecular dynamics (MD) simulations and machine learning to investigate the deformation mechanisms of AlCoCuCrFeNi HEA under varying temperatures, strain rates, and average grain sizes. The modeling results show that interactions between partial dislocations in AlCoCrCuFeNi HEA during tension and compression deformation cause various lattice disorders. The effect of temperature, strain rates, and grain boundaries on lattice disorder, plastic deformation behavior, dislocation density, and von-Mises stress (VMS) is disclosed. This study offers new insights into the atomic-scale deformation mechanisms governing the mechanical behavior of AlCoCrCuFeNi HEAs. It also presents a comprehensive workflow for predicting the mechanical properties of this HEA using machine learning models. The proposed approach provides several advantages, including significantly reduced simulation time and robust model validation. By employing the machine learning model trained in Stage 1, the time needed to simulate mechanical properties in Stage 2 is significantly decreased. Additionally, the framework ensures that the machine learning model effectively captures and understands the underlying representations of the mechanical properties of HEAs, thereby enhancing both the efficiency and accuracy of the predictions.