Dissociating the phononic, magnetic and electronic contributions to thermal conductivity: a computational study in alpha-iron

Abstract

Computational tools to study thermodynamic properties of magnetic materials have, until recently, been limited to phenomenological modeling or to small domain sizes limiting our mechanistic understanding of thermal transport in ferromagnets. Herein, we study the interplay of phonon and magnetic spin contributions to the thermal conductivity in $$\alpha$$ α -iron utilizing non-equilibrium molecular dynamics simulations. It was observed that the magnetic spin contribution to the total thermal conductivity exceeds lattice transport for temperatures up to two-thirds of the Curie temperature after which only strongly coupled magnon-phonon modes become active heat carriers. Characterizations of the phonon and magnon spectra give a detailed insight into the coupling between these heat carriers, and the temperature sensitivity of these coupled systems. Comparisons to both experiments and ab initio data support our inferred electronic thermal conductivity, supporting the coupled molecular dynamics/spin dynamics framework as a viable method to extend the predictive capability for magnetic material properties.

Impact of electronic correlations on high-pressure iron: insights from time-dependent density functional theory

“…In recent studies, the total thermal conductivity, accounting, in addition, for the coupling between phonons and magnons, has been split into various contributions, as demonstrated by Nikolov et al [93]. Their work highlights the significant contribution of the electronic component (approximately 80%-90%) at ambient temperature [94,95].…”

FitSNAP: Atomistic machine learning with LAMMPS

“…This also allows performance improvements achieved in the LAMMPS production code to immediately speed up the training process, rather than having to replicate the code improvements in the training software. FitSNAP has already taken advantage of this intrinsic LAMMPS interface in a number of publications that performed large-scale simulations with innovative ML atomistic models (Cusentino et al, 2020(Cusentino et al, , 2021Nikolov et al, 2021Nikolov et al, , 2022Wood & Thompson, 2018). LAMMPS supports a rapidly growing and diverse set of descriptors and ML model forms (Zuo et al, 2020).…”

Recent advances in computational materials design: methods, applications, algorithms, and informatics

“…The article by Sose et al [8] investigates the structure and dynamics of confined water in hybrid layered materials. The article by Nikolov et al [9] demonstrates new capabilities to investigate the interplay between phonon and magnetic spin contributions to the thermal conductivity of a-iron using a new spectral neighbor analysis potential. The article by Mishra et al [10] demonstrates a new capability to characterize phase and twinning variants in atomistic microstructures using a new virtual texture analysis method.…”