Programme

O.1 -- Oral, VCTP-48

Date: Monday, 31 July 2023

Time: 11:10 - 11:35

Development of machine learning-assisted semi-empirical model for twisted multi-layer materials

Huy-Viet Nguyen (1) and Cong-Huy Pham (2)

(1) Institute of Physics, Vietnam Academy of Science and Technology, Hanoi 10072, Vietnam (2) Physical and Life Sciences Directorate, Lawrence Livermore National Laboratory, Livermore, California 94550, United States

Twisted layer materials have recently attracted much interest due to its fascinating properties, such as the observation of unconventional superconductivity, the flat bands near Fermi energy, leading to strong electronic localization in bilayer graphene at `magic` twisted angle. From the computational standpoint, it is extremely challenging to perform direct quantum calculations for such systems, due to the large number of atoms in the unit-cell (can be 10-100 thousands or more). As a first step to achieve a viable description of twisted layer materials with the accuracy of first-principles quantum mechanical methods, we develop a machine learning-assisted semi-empirical model for twisted multi-layer graphene. We show that our model, while maintaining the accuracy of density-functional theory calculations, results in orders of magnitude improvement in computational cost. This would allow us to obtain accurate quantum data for very large systems which would not be possible with standard quantum approaches. We compare our calculation for structural properties and electronic band structure to previous published data. Our model also shows good transferability as it can describe reasonably well the structural properties of other carbon polymorphs at ambient conditions.

Presenter: Nguyễn Huy Việt