A Grain Boundary Embrittlement Genome for Substitutional Cubic Alloys

TL;DR

This paper develops a comprehensive computational database of grain boundary embrittlement across various alloys using machine learning, aiding alloy design by predicting embrittlement tendencies.

Contribution

It introduces a universal interatomic potential-based method to generate a large-scale grain boundary embrittlement genome for multiple alloy systems.

Findings

Created a database for 15 base metals with 75 solutes each

Predicted embrittlement tendencies across diverse alloy compositions

Provided a tool for alloy design and materials engineering

Abstract

Grain boundary chemistry plays a critical role for the properties of metals and alloys, yet there is a lack of consistent datasets for alloy design and development. With the advent of artificial intelligence and machine learning in materials science, open materials models and datasets can be used to overcome such challenges. Here, we use a universal interatomic potential to compute a grain boundary segregation and embrittlement genome for the {\Sigma}5[001](210) grain boundary for FCC and BCC binary alloys. The grain boundary database calculated here serves as a design tool for the embrittlement of high-angle grain boundaries for alloys across 15 base metals system of Ag, Al, Au, Cr, Cu, Fe (both BCC and FCC), Mo, Nb, Ni, Pd, Pt, Rh, Ta, V and W with 75 solute elements for each.