Speaker
Description
Ruijie Zhang, Takuji Oda Department of Nuclear Engineering, Seoul National University
Abstract:
Li₄SiO₄ is a promising solid tritium breeding material due to its high lithium density, thermal stability, and excellent tritium release performance. The Li-n reaction inevitably produces helium atoms, which are initially produced in bulk. Helium atoms have a closed-shell structure and small atomic radius, resulting in low solubility and a tendency to be trapped in defects such as dislocations, grain boundaries, and especially vacancies. Their accumulation causes crystal swelling and degrades mechanical properties. Hindering the diffusion of He atoms can prevent the coalescence and growth of He bubbles. Therefore, understanding the microscopic behavior of He atoms in crystals is essential.
Atomistic simulations such as density functional theory (DFT) calculations and molecular dynamics (MD) calculations help to uncover diffusion mechanisms and accumulate diffusion data. In this study, to realize accurate MD simulations, we construct a machine learning potential using DFT calculation data as the training set. The moment tensor potential (MTP) is adopted.
We study the He diffusivity in two cases: (1) the diffusion of a He atom in the perfect crystal as an interstitial atom (Dint); (2) the diffusion of a He atom interacting with a Li vacancy (DLi). At low temperatures (< 800 K), Dint exhibits anisotropy, with faster diffusion along the y-direction, while DLi remains nearly immobile due to strong He -Li vacancy binding. At high temperatures (>800 K), Dint converges to DLi because Li atoms start to diffuse significantly, and He is no longer deeply trapped in a Li vacancy.
