Uncovering the eutectics design by machine learning in the Al–Co–Cr–Fe–Ni high entropy system

Abstract

Eutectics in high entropy alloys (HEAs) have shown excellent properties and promising applications. With empirical rules, various of eutectic high entropy alloys (EHEAs) have been proposed. The current design strategies shed light on the formation of eutectics in HEAs, but they are incapable of confirming multiple variables quantitatively in the selection of a specific system. In the present study, the eutectic formation in the multi-principal element systems is uncovered via data mining with machine learning (ML), where the critical elements and strongly associated elements were discovered. Taking the Al–Co–Cr–Fe–Ni system as an example, Al is confirmed to be the critical element for the eutectic formation and Cr is the strongly associated element with Al, Ni, Co, Fe and minor additions with comparably large solid solubility can be considered overall. With these understandings, a three-step approach can be summarized for designing EHEAs in a given system. Within the designed EHEAs, properties can be tested for optimization of application orientated design. The findings can not only accelerate the exploitation of EHEAs with better performance but also provide new ideas for designing compositionally complex alloys.