Nonequilibrium molecular dynamics study on thermal conductivity of spinel AB2O4

• The development of a deep learning potential with high adaptability and accuracy for hexagonal spinel oxides. • A study of the effects of elements, temperature, grain boundaries, and antisite defects on thermal conductivity. • The finding offers valuable theoretical guidance for the development of advanced thermal barrier coating materials. Spinel oxides with the formula AB 2 O 4 are widely used in the design of materials for thermal barrier coatings, due to their low thermal expansion, thermal conductivity, and hydrogen diffusion. This study aims to identify spinel oxide ceramics that exhibit both low thermal conductivity and low thermal expansion by developing a machine learning potential for spinel oxides containing cation elements such as Cu, Al, Cr, Co, Ni, and Fe. Employ machine learning to fit DFT calculation data to obtain interatomic potential functions for molecular dynamics simulations. Non-equilibrium molecular dynamics simulations were conducted to investigate the thermal conductivity and thermal expansion of these materials, focusing on the influence of temperature, elemental composition, grain boundaries, and antisite defects. The results reveal that the high-entropy oxide (CuNiAlCrCoFe) 0.5 O 4 demonstrates a lower thermal conductivity and thermal expansion coefficient compared to other spinel oxides with different compositions, making it a promising candidate for thermal barrier coating applications. Moreover, in practical applications, the thermal conductivity of spinel oxides generally decreases as the temperature increases. For spinel oxides with the same composition, the thermal conductivity can be further reduced by the formation of grain boundaries and antisite defects during the material preparation process.