Rapid Estimation of the Liquidus Curve for Unexplored Multicomponent Alloy Chemistries

Melting temperature is a critical parameter in alloy design, processing, manufacturing, and operational stability. Thermodynamic modeling tools exist for well-characterized alloy systems, but it is hard to predict melting temperatures for unexplored alloy systems. Here, we develop a model that can rapidly estimate the liquidus curve for alloy systems composed of the 70 most abundant metallic elements in the Earth’s crust. We digitize and extract 1,411 binary alloy phase diagrams from the Materials Platform for Data Science, (MPDS) and compare with phase energies obtained from Density Functional Theory (DFT). We then augment this dataset with high throughput liquidus measurements to improve accuracy for multi-component interactions. We discuss how Hume-Rothery style intuition can be applied to predict the melting behavior of multi-component alloys