Machine Learning Interatomic Potential for Finite-Size Graphene
Zhang Yan, Wang Lifeng, Zheng Zhuoqun
Abstract: Empirical potentials are insufficient for describing the edge effect of graphene. First-principle calculations based on density functional theory can correctly simulate the boundary effect, but the calculation cost is high. On the basis of deep neural networks, a machine learning interatomic potential suitable for simulating finite-size graphene is developed. Using first-principles data as a reference, the machine learning interatomic potential and empirical potential are evaluated. Compared with the empirical potential, the machine learning interatomic potential can predict the stability, static mechanical properties and dynamic behaviours of graphene nanoribbons well. The effect of hydrogen atom adsorption on the edge of graphene on the structure and stability can also be described. This machine learning interatomic potential can facilitate future research on graphene nanoribbons and help the development of resonators based on graphene materials.
文章链接：https://www.sciencedirect.com/science/article/pii/S0927025625005993
