Insights into the GB effects on the self-diffusion, tensile deformation and pre-crack self-healing in UO2: Deep learning molecular dynamics simulations

HW Bao and Z Guo and DM Peng and ZP Sun and Y Li and Y Xin and F Ma, PROGRESS IN NATURAL SCIENCE-MATERIALS INTERNATIONAL, 35, 986-1000 (2025).

DOI: 10.1016/j.pnsc.2025.08.002

Understanding on the transportation and mechanical properties of polycrystalline UO2 is critical for the development and design of high- performance nuclear fuels. However, the influence of grain boundaries (GBs) on the basic properties of UO2 has not been fully understood. Herein, a high-accuracy deep learning (DP) potential is constructed for UO2 based on a comprehensive database obtained from density functional theory (DFT) calculations. The melting, self-diffusion and phase transition of single-crystal UO2 are studied by molecular dynamics (MD) simulations using DP potential, demonstrating the excellent predictive capabilities of the DP potential. The DP potential could also well describe the local atomic structure of symmetrical tilt grain boundaries (STGBs) in UO2. UO2 bi-crystals exhibit the pre-melting at GBs during heating process. The self-diffusion of O is enhanced by all the GBs, whereas the self-diffusion of U atoms is only enhanced by the dislocations in low-angle STGBs. Brittle fracture occurs in UO2 bi- crystals at the lower temperatures (<1200 K), while amorphization takes place around GBs at high temperatures (>1200 K). This leads to brittle- to-ductile transition in polycrystalline UO2 at high temperature. The self-healing of pre-cracks in UO2 bi-crystals is determined by the diffusion of O and U atoms at high temperatures. The developed DP potential can be further applied in the studies on the microstructure design of UO2 based nuclear fuels.

Return to Publications page