Machine Learning-Assisting DFT Simulations of Diffusion Behaviors in Doped Cr₂AlC

Cr₂AlC has garnered significant attention for high-temperature protective applications, but its diffusivity often requires tailoring for specific scenarios, especially in bond coats of thermal barrier coatings, where diffusion must be suppressed to prevent coating failure. In this study, the impact of doping with 11 typical alloying elements on diffusion activation energy (E_d) in doped Cr₂AlC was systematically investigated by density functional theory (DFT) simulations integrated with the statistical analysis and machine learning (ML) model. While most dopants reduce E_d and promote diffusion, substituting Cr in Cr₂AlC with Mo or W effectively suppresses diffusion. Combined with the bond stiffness model, statistical analysis confirms that diffusivity is fundamentally regulated by the local bonding environment, with stronger bonds leading to significantly higher E_d. Furthermore, ML analysis identifies atomic radius (R_at) and chemical potential (µ) as the dominant factors modulating diffusion, reflecting the synergistic impact of lattice distortion and thermodynamic stability. These findings provide a comprehensive microscopic understanding of doping-regulated diffusion and establish a critical theoretical foundation for the rational design of diffusion-resistant Cr₂AlC-based materials.