ML_Sluggish_Diffusion

Single-phase concentrated solid solution alloys (CSAs), including medium- and high-entropy alloys (HEAs), show outstanding functional and structural properties, which are thought to be related to their sluggish diffusion properties. Despite decades of efforts, the origin of sluggish diffusion and even whether it exists at all, are still under intense debate. Generally, sluggish diffusion is understood as a result of the rough potential energy landscape (PEL) in CSAs due to the intrinsic chemically disordered states. Nonetheless, how the rugged PEL affects diffusion properties remains elusive due to the complications induced by the enormous atomic environments experienced by diffusing atoms. Here, we uncover the links between the PEL and self-diffusion in CSAs by combining machine learning (ML) and kinetic Monte Carlo (kMC).