Poncé S, Gillet Y, Laflamme Janssen J, Marini A, Verstraete M, Gonze X
European Theoretical Spectroscopy Facility and Institute of Condensed Matter and Nanosciences, Université catholique de Louvain, Chemin des étoiles 8, bte L07.03.01, B-1348 Louvain-la-neuve, Belgium.
Consiglio Nazionale delle Ricerche (CNR), Via Salaria Km 29.3, CP 10, 00016 Monterotondo Stazione, Italy.
J Chem Phys. 2015 Sep 14;143(10):102813. doi: 10.1063/1.4927081.
The renormalization of electronic eigenenergies due to electron-phonon coupling (temperature dependence and zero-point motion effect) is sizable in many materials with light atoms. This effect, often neglected in ab initio calculations, can be computed using the perturbation-based Allen-Heine-Cardona theory in the adiabatic or non-adiabatic harmonic approximation. After a short description of the recent progresses in this field and a brief overview of the theory, we focus on the issue of phonon wavevector sampling convergence, until now poorly understood. Indeed, the renormalization is obtained numerically through a slowly converging q-point integration. For non-zero Born effective charges, we show that a divergence appears in the electron-phonon matrix elements at q → Γ, leading to a divergence of the adiabatic renormalization at band extrema. This problem is exacerbated by the slow convergence of Born effective charges with electronic wavevector sampling, which leaves residual Born effective charges in ab initio calculations on materials that are physically devoid of such charges. Here, we propose a solution that improves this convergence. However, for materials where Born effective charges are physically non-zero, the divergence of the renormalization indicates a breakdown of the adiabatic harmonic approximation, which we assess here by switching to the non-adiabatic harmonic approximation. Also, we study the convergence behavior of the renormalization and develop reliable extrapolation schemes to obtain the converged results. Finally, the adiabatic and non-adiabatic theories, with corrections for the slow Born effective charge convergence problem (and the associated divergence) are applied to the study of five semiconductors and insulators: α-AlN, β-AlN, BN, diamond, and silicon. For these five materials, we present the zero-point renormalization, temperature dependence, phonon-induced lifetime broadening, and the renormalized electronic band structure.
在许多含有轻原子的材料中,由于电子 - 声子耦合(温度依赖性和零点运动效应)导致的电子本征能量重整化相当可观。这种效应在从头算计算中常常被忽略,可以使用基于微扰的艾伦 - 海涅 - 卡尔多纳理论在绝热或非绝热谐波近似下进行计算。在简要描述该领域的最新进展并对理论进行简要概述之后,我们将重点关注声子波矢采样收敛问题,到目前为止该问题还鲜为人知。实际上,重整化是通过缓慢收敛的q点积分以数值方式获得的。对于非零的玻恩有效电荷,我们表明在q → Γ时电子 - 声子矩阵元中会出现发散,导致能带极值处绝热重整化的发散。玻恩有效电荷随电子波矢采样的缓慢收敛加剧了这个问题,这使得在对物理上不存在此类电荷的材料进行从头算时会留下残余的玻恩有效电荷。在此,我们提出一种改进这种收敛的解决方案。然而,对于玻恩有效电荷在物理上不为零的材料,重整化的发散表明绝热谐波近似失效,我们在此通过切换到非绝热谐波近似来评估这一点。此外,我们研究重整化的收敛行为并开发可靠的外推方案以获得收敛结果。最后,将考虑了缓慢玻恩有效电荷收敛问题(以及相关发散)修正的绝热和非绝热理论应用于五种半导体和绝缘体的研究:α - 氮化铝、β - 氮化铝、氮化硼、金刚石和硅。对于这五种材料,我们给出了零点重整化、温度依赖性、声子诱导的寿命展宽以及重整化的电子能带结构。