Spatial density neural network force fields with first-principles level accuracy and application to thermal transport

2020 ◽  
Vol 102 (3) ◽  
Author(s):  
Alejandro Rodriguez ◽  
Yinqiao Liu ◽  
Ming Hu
2021 ◽  
Author(s):  
Sandra Sæther ◽  
Merete Falck ◽  
Zhiliang Zhang ◽  
Anders Lervik ◽  
Jianying He

2020 ◽  
Author(s):  
Jianqin Tang ◽  
Jonathan Skelton

We present a systematic first-principles modelling study of the structural dynamics and thermal transport in the CoSb<sub>3</sub> skutterudites with a series of noble-gas filler atoms. A range of analysis techniques are proposed to estimate the filler rattling frequencies, to quantify the separate impacts of filling on the phonon group velocities and lifetimes, and to show how changes to the phonon spectra and interaction strengths lead to suppressed lifetimes. The fillers are found to reduce the thermal conductivity of the CoSb<sub>3</sub> framework by up to 15 % primarily by suppressing the group velocities of low-lying optic modes. Calculations show that the filler rattling frequencies are determined by a detailed balance of increasing atomic mass and stronger interactions with the framework, and are a good predictor of their impact on the heat transport. Lowering the rattling frequency below ~1.5 THz by selecting heavy fillers that interact weakly with the framework is predicted to produce a much larger suppression of the thermal transport, by inducing avoided crossings in the acoustic-mode dispersion and facilitating resonant scattering with a consequent large reduction in the lifetimes. Approximate rattling frequencies determined from the harmonic force constants may therefore provide a useful metric for selecting filler atoms to optimise the thermal transport in skutterudites and other cage compounds such as clathrates.


Author(s):  
Bohayra Mortazavi ◽  
Masoud Shahrokhi ◽  
Xiaoying Zhuang ◽  
Alexander V. Shapeev ◽  
Timon Rabczuk

In the latest experimental advances in the field of two-dimensional (2D) materials, penta-PdPS and penta-PdPSe layered materials have been fabricated. In this work, we conduct first-principles calculations to explore the...


2019 ◽  
Vol 21 (12) ◽  
pp. 6506-6516 ◽  
Author(s):  
Mário R. G. Marques ◽  
Jakob Wolff ◽  
Conrad Steigemann ◽  
Miguel A. L. Marques

We present a practical procedure to obtain reliable and unbiased neural network based force fields for solids.


Sign in / Sign up

Export Citation Format

Share Document