A Deep Neural Network Potential to Study the Thermal Conductivity of MnBi2Te4 and Bi2Te3/MnBi2Te4 Superlattice

Magnetic topological insulators MnBi2Te4 and Bi2Te3/MnBi2Te4 have attracted extensive attention because of their novel physical properties and promising application potentials. In this work, based on the dataset from density functional calculations, a deep neural network potential is trained to study the thermal transport properties of MnBi2Te4 and Bi2Te3/MnBi2Te4. The thermal conductivity of MnBi2Te4 is first calculated using equilibrium molecular dynamics combined with this potential, and the results are in good agreement with the experimental values. The thermal conductivity of Bi2Te3/MnBi2Te4 is then predicted using the same approach. Compared with that of MnBi2Te4, the in-plane thermal conductivity of Bi2Te3/MnBi2Te4 does not change significantly, but the out-of-plane one is much lower. The low out-of-plane thermal conductivity of Bi2Te3/MnBi2Te4 is extremely desirable for high thermoelectric performance. By investigation of the corresponding phonon dispersion relations, it is found that the band gaps at the boundary of the Brillouin zone and the inhibition of phonon velocity lead to the large reduction of out-of-plane thermal conductivity in Bi2Te3/MnBi2Te4.