Exploring high-performance environmental barrier coatings for rare earth silicates: A combined approach of first principles calculations and machine learning

Abstract

RE₂Si₂O₇ is promising materials for environmental barrier coating (EBC), but the vast phase space poses challenges for the screening of RE₂Si₂O_7. It follows that a combined approach of first principles calculations and machine learning is proposed for this problem, with establishing a comprehensive database comprising β-, γ- and δ-RE₂Si₂O₇ (RE = La–Lu, Y, Sc) and correlating their mechanical/thermal properties on structural characteristics. It is revealed the [O₃SiOSiO₃] structure and polyhedron distortion affect mechanical properties of RE₂Si₂O₇, while criteria for selecting RE₂Si₂O₇ with low thermal conductivity are identified, including complex crystal structures, chemical bond inhomogeneity, and strong non-harmonic lattice vibrations. Also, the machine learning model accurately predicts the coefficient of thermal expansion (CTE) and minimum thermal conductivity (λ_min) of RE₂Si₂O₇, with volume and mass variations identified as critical factors, respectively. This integrated approach efficiently screens RE₂Si₂O₇ for EBC application and enables rapid assessments of their thermal properties.