Group velocity matters for accurate prediction of phonon-limited carrier mobility* *Project supported by the National Natural Science Foundation of China (Grant Nos. 11925407 and 61927901) and the Key Research Program of Frontier Sciences, Chinese Academy of Sciences (Grant No. ZDBS-LY-JSC019).

Abstract

First-principles approaches have recently been developed to replace the phenomenological modeling approaches with adjustable parameters for calculating carrier mobilities in semiconductors. However, in addition to the high computational cost, it is still a challenge to obtain accurate mobility for carriers with a complex band structure, e.g., hole mobility in common semiconductors. Here, we present a computationally efficient approach using isotropic and parabolic bands to approximate the anisotropy valence bands for evaluating group velocities in the first-principles calculations. This treatment greatly reduces the computational cost in two ways: relieves the requirement of an extremely dense k mesh to obtain a smooth change in group velocity, and reduces the 5-dimensional integral to 3-dimensional integral. Taking Si and SiC as two examples, we find that this simplified approach reproduces the full first-principles calculation for mobility. If we use experimental effective masses to evaluate the group velocity, we can obtain hole mobility in excellent agreement with experimental data over a wide temperature range. These findings shed light on how to improve the first-principles calculations towards predictive carrier mobility in high accuracy.