This paper introduces a novel approach to designing ultrahigh-entropy alloys (HEAs) using text mining (TM) on a corpus of 6.4 million scientific papers. The method, based on the concept of "context similarity," identifies chemical elements frequently co-occurring in the literature, reflecting researchers' experiences in materials design. This overcomes limitations of traditional TM by discovering HEAs not explicitly present in the data. The approach successfully identifies known HEAs like Cantor and Senkov alloys and screens for promising six- and seven-component lightweight HEAs, yielding nearly 500 candidates from 2.6 million possibilities. The method integrates with Integrated Computational Materials Engineering (ICME) approaches for further refinement.

Zongrui Pei, Junqi Yin, Peter K. Liaw, Dierk Raabe