Design of (TiHfZr)(NiCoCu) High-Entropy Shape Memory Alloys: From Firstov's Experiments to Data-Driven Approach

Abstract

This paper deals with the design of (TiHfZr)(NiCoCu) high-entropy and high-temperature shape memory alloys (HE-HT-SMAs). It explains the chronology and the progress of this design starting from the experimental work of Georgi Firstov initiated in the 2015s until the advent of data-driven alloy approaches. A state-of-the-art (TiHfZr)(NiCoCu) HE-HT-SMA family is presented and enriched by a database used as input for a data-driven approach. The paper then focuses on the comparison of martensitic transformation temperatures provided by: (i) the experimental work of Firstov et al. started in 2015, (ii) other recent experimental studies and, (iii) those predicted by two numerical approaches. The first approach consists of a linear regression model proposed by Peltier et al., while the second one is proposed and enriched by Thiercelin et al. using a data-driven technique (random forest regression). The results from the data-driven approach yield accurate predictions that align with the experimental data from both the literature and previous studies. Thus demonstrating the importance of physics-informed, inspired techniques to optimize the design of future alloys, in particular HE-HT-SMAs.