Session: 17-01-01: Research Posters

Paper Number: 150521

150521 - Sensitivity of Higher-Order Interatomic Force Constants and Thermal Conductivity to the Energy Surface Roughness of Exchange-Correlation Functionals 

Accurately predicting lattice thermal conductivity holds immense significance in a wide range of applications such as thermal management, thermal barrier coatings, and thermoelectric devices. Over the past years, first principles-based phonon theories have emerged as the most reliable and widely accepted approaches for the lattice thermal conductivity prediction. In these methods, the harmonic (second-order) interatomic force constants (IFCs), or 2^nd-IFCs, are used to calculate the phonon frequencies, velocities, and specific heat. The third- and fourth-order IFCs, or 3^rd- and 4^th-IFCs, are used to calculate the three- and four-phonon scattering rates, respectively. These quantities are implemented in the exact solution to the linearized Boltzmann transport equation (BTE) to predict the thermal conductivity. To determine the IFCs, the most widely used method is the finite difference method (FDM) based on the density functional theory (DFT) calculations. Thus, the precision of DFT calculations determines the accuracy of the predictions of IFCs and thermal conductivity. While DFT is an ab initio theory, which does not require any prior knowledge of the system or fitting parameters, its accuracy can be affected by the choice of exchange-correlation (XC) functionals. While many studies have investigated the impact of XC functionals on three-phonon scattering, its impact on higher-order IFCs or scattering rates remains unclear.

In this work, we meticulously investigate the sensitivity of 4^th-IFCs and 4ph thermal conductivity to the energy surface roughness of XC functionals, i.e., LDA, PBE, and PBEsol, on silicon and boron arsenide.

We find that while 2^nd- and 3^rd-IFCs exhibit minimal sensitivity to the choice of δ, the finite displacement utilized in the FDM, 4^th-IFCs demonstrate a pronounced susceptibility, indicating the substantial energy roughness of XC functionals in terms of higher-order. This roughness varies for different functionals in different materials and can cause misprediction of thermal conductivity by several times of magnitude. As a result, when calculating the 4^th-IFCs using the FDM, the atomic displacement needs to be taken large enough to overcome the energy surface roughness, in order to accurately predict phonon lifetime and thermal conductivity. In the end, we provide the general guidance on how to select the optimal δ value for IFCs extraction using FDM.

This work delves into the energy surface roughness of XC functionals and its impact on 4^th-IFCs and 4ph thermal conductivity. We anticipate that our finding will provide valuable insights and guidance for future phonon scattering and thermal transport studies.

We believe this work is well-suited to the scope of Applied Physics Letters. We really appreciate your consideration.

Presenting Author: Hao Zhou University of Utah

Presenting Author Biography: Hao Zhou is a Ph.D. candidate at the University of Utah. His research focuses on first principles, molecular dynamics, and machine learning-based thermal transport simulations.

Authors:

Hao Zhou University of Utah
Shuxiang Zhou Idaho National Laboratory
Zilong Hua Idaho National Laboratory
Kaustubh Bawane Idaho National Laboratory
Tianli Feng University of Utah