Investigation of the transport behavior of BP and BAs using the Boltzmann transport equation and machine learning methods

Abstract

It is important to predict the theoretical upper limit of lattice thermal conductivity and carrier mobility in the face of high uncertainty in experimental measurements of boron phosphide (BP) and boron arsenides (BAs). Using the Boltzmann transport equation (BTE) approach, we quantify the contributions of three-phonon and four-phonon scattering to thermal conductivity. The thermal conductivity of BP is determined by three-phonon scattering and is independent of four-phonon scattering. Due to the large bandgap between acoustic and optical phonons in BAs, three-phonon scattering is prevented, while four-phonon scattering is permitted. At room temperature, the predicted thermal conductivity with three- and four-phonon scattering is 505 W/m K (BP) and 1506 W/m K (BAs). After considering isotope scattering, the thermal conductivity is reduced to 468 and 1124 W/m K, respectively. Neuroevolution potential is trained using machine learning methods. Subsequently, the extrapolated lattice thermal conductivity through nonequilibrium molecular dynamics simulation is 304 W/m K (BP) and 1008 W/m K (BAs), respectively. The Heyd-Scuseria-Ernzerhof (HSE) hybrid functional and spin-orbit coupling (SOC) are employed to correct bandgap and band edge shape. Utilizing the Wannier function interpolation methodology, the ambipolar mobility predicted by the iterative BTE is μa=1.61×103cm2/Vs (BP) and μa=1.79×103cm2/Vs (BAs) at room temperature, which is in good agreement with the experimental measurements. By analyzing the characteristics of branch-dependent phonon scattering, the dominant scattering of charge carriers is attributed to longitudinal acoustic phonons.