Entropy-Driven Microstructure Evolution Predicted with the Steepest-Entropy-Ascent Quantum Thermodynamic Framework

TL;DR

This paper introduces a thermodynamic framework that predicts microstructural evolution in crystalline solids by using an energy landscape and entropy principles, aligning well with experimental data and reducing computational costs.

Contribution

It develops a novel approach combining the steepest-entropy-ascent quantum thermodynamics framework with energy landscape modeling to predict microstructure kinetics without detailed mechanistic data.

Findings

Accurately predicts sintering, coarsening, and grain growth kinetics.

Aligns well with experimental results for ZrO2, Al3Li, and Pd.

Reduces computational effort compared to traditional Monte Carlo simulations.

Abstract

A Potts model and the Replica Exchange Wang-Landau algorithm are used to construct an energy landscape for a crystalline solid containing surfaces and grain boundaries. The energy landscape is applied to an equation of motion from the steepest-entropy-ascent quantum thermodynamic (SEAQT) framework to explore the kinetics of three distinct kinds of microstructural evolution: polycrystalline sintering, precipitate coarsening, and grain growth. The steepest entropy ascent postulate predicts unique kinetic paths for these non-equilibrium processes without needing any detailed information about the underlying physical mechanisms of the processes. A method is also proposed for associating the kinetic path in state space to a set of smoothly evolving microstructural descriptors. The SEAQT-predicted kinetics agree well with available experimental kinetics for ZrO2 sintering, Al3Li precipitate coarsening, and grain growth in nanocrystalline Pd. The computational cost associated with calculating the energy landscape needed by the approach is comparable to a Monte Carlo simulation. However, the subsequent kinetic calculations from the SEAQT equation of motion are quite modest and save considerable computational resources by obviating the need for averaging multiple kinetic Monte Carlo runs.