Revisiting thermal transport in CuCl : First-principles calculations and machine learning force fields

Ashis Kundu, Florian Knoop, Igor A. Abrikosov · 2025

Read Revisiting thermal transport in CuCl : First-principles calculations and machine learning force fields by Ashis Kundu, Florian Knoop, Igor A. Abrikosov (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2508-15525.txt.

Introduction

Ashis Kundu and co-authors revisit how heat flows through copper chloride, a crystal with famously low thermal conductivity. Its atoms vibrate in a strongly anharmonic way, so simple models miss the mark. The team recalculated the forces with temperature corrections and counted collisions among four vibrations at once, not just three.

The result matches experiments closely. Squeezing the crystal lowers its conductivity, an odd trend the team reproduces with a machine-learned force field. Stronger four-way scattering and slower sound modes explain the drop. The recipe should transfer to other anharmonic materials.

Catalog record

CREATORAshis Kundu, Florian Knoop, Igor A. Abrikosov
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE34 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2508-15525

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Revisiting thermal transport in CuCl : First-principles calculations and machine learning force fields Ashis Kundu, Florian Knoop, Igor A. Abrikosov Abstract Accurate prediction of lattice thermal conductivity ( κ l \kappa_{l} ) in strongly anharmonic materials requires renormalized interatomic force constants (IFCs) and appropriate incorporation of diagonal and off-diagonal contributions and higher-order scattering. We investigate CuCl , a highly anharmonic system with a simple zincblende structure and ultralow κ l \kappa_{l} . Our calculations, including IFC renormalization and four-phonon scattering, show excellent agreement with the experiment, underscoring the critical role of both effects in the accurate estimation of κ l \kappa_{l} . Furthermore, the unusual pressure dependence of κ l \kappa_{l} is explored using a rigorously validated machine-learned force field, with the predicted values showing good agreement with the experimentally observed trend of monotonic decrease. This behavior is primarily driven by a significant increase in four-phonon scattering and a reduction in the group velocity of transverse acoustic modes. Overall, this study establishes a robust framework for modeling thermal transport in strongly anharmonic materials. I Introduction Understanding and an accurate evaluation of lattice thermal conductivity in highly anharmonic materials is essential for thermoelectric Snyder and Toberer 2008 ; Sootsman et al. 2009 and thermal barrier coatings Padture et al. 2002 ; Clarke et al. 2012 , as these materials typically have low thermal conductivity. Binary semiconductors generally tend to have high thermal conductivity due to their simple crystal structures and phonon characteristics, which stem from large atomic mass differences that favour weak phonon scattering Ravichandran and Broido 2020 ; Xia et al. 2020 . However, some compounds exhibit strong anharmonicity despite their simple structures and significant mass difference, leading to ultralow thermal conductivity Xia et al. 2020 . One such example is CuCl , a metal halide with a zincblende crystal structure, which has been experimentally reported to exhibit a thermal conductivity of approximately 0.7 to 0.85 W m -1 K -1 at room temperature Slack and Andersson 1982 ; Haynes 2014 . Previous theoretical studies have attempted to explain this unusually low value and attributed it to extreme anharmonicity caused by strong orbital hybridization Togo et al. 2015 ; Mukhopadhyay et al. 2017 ; Yang et al. 2023 . However, the calculated thermal conductivity varies, with some studies overestimating Togo et al. 2015 ; Mukhopadhyay et al. 2017 while others underestimating Yang et al. 2023 the experimental result. This discrepancy highlights the need for a more accurate assessment of anharmonicity and the lattice thermal conductivity. Another peculiar aspect of the thermal conductivity of CuCl is its pressure dependence. Typically, thermal conductivity increases with pressure due to phonon hardening unless a structural phase transition occurs Zhou et al. 2022 . However, recent studies have reported anomalous behavior in some materials, where thermal conductivity initially increases and then decreases with pressure Lindsay et al. 2015 ; Ouyang and Hu 2015 ; Yuan et al. 2018 ; Ravichandran and Broido 2019 ; Ravichandran and Broido 2021 ; Kundu et al. 2024 . This behavior results from the interplay between harmonic and anharmonic interactions or competition among different anharmonic scattering mechanisms. In the case of CuCl , thermal conductivity continuously decreases with pressure before the phase transition Slack and Andersson 1982 . This open question requires further investigation. In this work, we study the lattice thermal conductivity of CuCl theoretically using Temperature Dependent Effective Potential (TDEP) method Hellman et al. 2011 ; Hellman and Abrikosov 2013 accurately incorporating anharmonic effects, which results in excellent agreement with the experiment. Our findings highlight the importance of both temperature-dependence of interatomic force constants (IFCs) and four-phonon interactions. Moreover, we study the significant role of selection of the exchange-correlation functional in the density functional theory (DFT) calculations of CuCl . To study pressure-dependent thermal conductivity, we employ a machine-learning force field (MLFF) to reduce computational costs while ensuring reliability. The accuracy of MLFF is verified at multiple levels, and our calculated thermal conductivity across temperatures and pressures matches well with experimental results. Thus, our work provides a robust framework for estimating thermal conductivity in highly anharmonic systems. II Methodology To compute the lattice thermal conductivity tensor, we employ the mode-coupling theory of anharmonic lattice dynamics Castellano et al. 2023 ; Castellano et al. 2025 , which provides a nonperturbative framework that incorporates phonon correlation functions and captures both collective and coherent contributions to thermal transport, as implemented in the TDEP package Knoop et al. 2024 . The total thermal conductivity ( \kappa ) is the sum of the diagonal and off-diagonal (coherent) contributions, derived from the respective heat flux operators, and is expressed as \kappa=\kappa^{\text{d}}+\kappa^{\text{nd}}. (1) The diagonal contribution to the thermal conductivity tensor is expressed as \kappa_{\alpha\beta}^{\mathrm{d}}=\frac{1}{V}\sum_{\lambda\lambda^{\prime}}v_{\lambda}^{\alpha}v_{\lambda}^{\beta}c_{\lambda}\mathbf{\Sigma}^{-1}(\lambda,\lambda^{\prime}), (2) where V is the volume of the primitive cell, and \alpha and \beta are Cartesian axes. The phonon mode \lambda\equiv(\mathbf{q},s) is labeled by the phonon wave vector \mathbf{q} and the branch index s . v_{\lambda}^{\alpha} is the group velocity of mode \lambda , and c_{\lambda}=n_{\lambda}(n_{\lambda}+1)\omega_{\lambda}^{2}/k_{\mathrm{B}}T^{2} is the heat capacity, where \omega_{\lambda} is the phonon frequency, n_{\lambda} is the Bose-Einstein occupation number, k_{\mathrm{B}} is the Boltzmann constant, and T is the temperature. The matrix \mathbf{\Sigma} denotes the phonon scattering matrix, whose diagonal elements correspond to the phonon scattering rates \Gamma_{\lambda} , that incorporates contributions from three-phonon, four-phonon, and phonon-isotope scattering. The off-diagonal elements of \mathbf{\Sigma} describe coupling between modes that gives collective phonon contributions to heat transport. The calculation of \kappa_{\alpha\beta}^{\mathrm{d}} in this form is computationally demanding, as it requires the diagonalization of the scattering matrix. This difficulty is alleviated using an iterative solution approach, as detailed in Ref. Castellano et al. 2025 . The off-diagonal (coherent) contribution to thermal conductivity tensor is expressed as Simoncelli et al. 2019 ; Isaeva et al. 2019 \kappa_{\alpha\beta}^{\mathrm{nd}}=\frac{1}{V}\sum_{\lambda\lambda^{\prime}}v_{\lambda\lambda^{\prime}}^{\alpha}v_{\lambda\lambda^{\prime}}^{\beta}\frac{c_{\lambda}+c_{\lambda^{\prime}}}{2}\Gamma_{\lambda\lambda^{\prime}} (3) where v_{\lambda\lambda^{\prime}}^{\alpha} is the off-diagonal group velocity. The off-diagonal scattering rate \Gamma_{\lambda\lambda^{\prime}} is defined as \Gamma_{\lambda\lambda^{\prime}}=\frac{\Gamma_{\lambda}+\Gamma_{\lambda^{\prime}}}{(\omega_{\lambda}-\omega_{\lambda^{\prime}})^{2}+(\Gamma_{\lambda}+\Gamma_{\lambda^{\prime}})^{2}} (4) where \Gamma_{\lambda} is the scattering rate of phonon mode \lambda and the detailed expression for different scattereing processes can be found in Ref. Castellano et al. 2025 . The key requirement to evaluate both the diagonal and off-diagonal contributions is the harmonic and anharmonic interatomic force constants (IFCs). To obtain these, we employ TDEP with efficient stochastic phase-space sampling scheme Shulumba et al. 2017 . This method iteratively refines the force constants by approximating atomic displacement distributions in the canonical ensemble and improving the approximation with actual forces. These forces are calculated using density functional theory (DFT) as implemented in the VASP package Kresse and Furthmüller 1996 ; Kresse and Joubert 1999 . The exchange-correlation functional used for the final results is PBESol Perdew et al. 2008 , with additional comparisons made using the LDA Kohn and Sham 1965 and PBE Perdew et al. 1996 functionals. All DFT calculations were performed using a plane-wave energy cutoff of 500 eV and \Gamma -point sampling for the Brillouin zone. The convergence criteria for the electronic self-consistent cycle was set to 10^{-8} eV. A 5\times 5\times 5 supercell containing 250 atoms was used to construct the second-, third-, and fourth-order IFCs. Cutoff radii of 6 Å and 4 Å were used for the third- and fourth-order IFCs, respectively. A detailed convergence analysis of the cutoff values for IFCs and the

Read the complete text