Seamlessly joining length scales: From atomistic thermal graphs to anisotropic continuum conductivity

TL;DR

This paper introduces a new method to connect atomistic thermal transport data with continuum models, enabling accurate, anisotropic heat conduction simulations in complex solids by leveraging graph neural networks and adaptive finite element methods.

Contribution

The authors develop SCACS, a toolkit that predicts site-projected thermal conductivities from atomic structures and embeds them into continuum models, bridging atomistic and device-scale thermal simulations.

Findings

Accurately reproduces bulk and interfacial thermal conductivities in silicon nanostructures.

Captures anisotropic effects and defect-driven variations in thermal transport.

Predicts experimental conductance trends effectively.

Abstract

Thermal transport in complex solids is governed by local structure, defects, and anisotropy, yet most continuum models still rely on oversimplified, homogenized conductivities. Here, we bridge atomistic and continuum descriptions by building finite element (FE) models directly from the site-projected thermal conductivity (SPTC), an atomic-level decomposition of the Green-Kubo thermal conductivity. We introduce a new toolkit, the Simulator Collection for Atomic-to-Continuum Scales (SCACS), which uses a graph neural network to predict SPTC on large atomic structures, coarse-grain these fields into anisotropic conductivity tensors, and embeds them into the heat-flow FE equation with a customized, anisotropy-aware adaptive mesh refinement scheme. Applied to silicon nanostructures, the resulting FE models act as representative volume elements, reproduce bulk conductivities, and capture interfacial and defect-driven anisotropy while maintaining thermodynamic consistency. Additionally, SCACS predicts experimental conductance trends and fields. This innovation demonstrates a novel and general route for transferring atomistic transport information into device-scale thermal simulations with physics-based approximations.