Mo-Re-W Alloys for High Temperature Applications: Phase Stability, Elasticity, and Thermal Property Insights via Multi-Cell Monte Carlo and Machine Learning

TL;DR

This paper combines multi-cell Monte Carlo simulations and machine learning to predict phase stability, elastic, and thermal properties of Mo-Re-W alloys, identifying promising compositions for high-temperature applications.

Contribution

It introduces a computational framework integrating Monte Carlo and machine learning to explore and optimize Mo-Re-W alloy properties at high temperatures.

Findings

Identified a (Mo,W)+Re alloy composition as optimal for high-temperature stability.

Machine learning models accurately predicted properties across the compositional domain.

Experimental validation confirmed computational predictions.

Abstract

The increasing demand for materials capable of withstanding high temperatures and harsh environments necessitates the discovery of advanced alloys. This study introduces a computational routine to predict solid-state phase stability and calculates elastic constants to determine high temperature viability. With it, machine learning models were trained on 1,014 Mo-Re-W structures to enable a large compilation of elastic and thermal properties over the complete Mo-Re-W compositional domain with extreme resolution. A series of heat maps spanning the full compositional domain were generated to visually present the impact of alloy constituents on the alloy properties. Our findings identified a balanced (Mo,W) + Re blend as a promising composition for high temperature applications, attributed to a strong and stable (Mo,W) matrix with high Re content and the formation of strengthening (W,Re) precipitates that enhanced mechanical performance at 1600oC. Several Mo-Re-W compositions were manufactured to experimentally validate the computational predictions. This approach provides an efficient and system-agnostic pathway for designing and optimizing alloys for high-temperature applications.