High entropy alloy (HEA) is a new group of material which consists of five or more principal elements. Due to its distinct design concept, these alloys often exhibit unusual properties and have attracted significant interest of research, leading to an emerging yet exciting new field in recent years. Within various topics on HEAs, the stability of solid solution by mixing several elements, is one essential issue and the pre-condition of various application.
This talk will introduce our study on the stability and relevant properties of transition metal HEAs using the first-principles method combining machine learning. This research was started from two quinary high entropy alloys (HEAs), Cantor alloy FeCoNiCrMn, and FeNiCoCrPd which is synthesised by intentionally substituting Mn in Cantor alloy with Pd, and was reported to achieve 2.5 times higher strength than that of FeNiCoCrMn [1].
In order to understand the mechanism of Pd’s role, we investigated the stability and structural properties of these two HEAs based on first-principles calculation combining the density functional theory (DFT) and finite temperature effects with taking special quasi-random structures (SQS) as model of solid solution. It has been revealed that the inhomogeneous feature of Pd increases the root mean squared displacement (RMSD) of the alloy, which consequently enhances the mechanical properties, and also keeps the stability [2]. Along this work we have accumulated more than 1,000 DFT SQS data of all sub-systems of binary, ternary, quaternary in all equiatomic compositions and typical non-equiatomic compositions of FeCoNiCrMn/Pd, for fcc. bcc and hcp structures.
Based on this data set, systematic predictions are conducted by machine learning (ML). The elemental convolution graph neural networks (ECNet) [3] in cooperating with transfer learning attempted to predict the stability and properties of the higher compositional systems mainly based on the data of binaries and ternaries, then three new compositions of (FeCoNiCrMn)1-xPdx with superior values of RMSD than known quinary HEAs were explored [4.5]. Furthermore, the mesh searching [6] for virtual systems of Fe, Co, Ni, Cr, Mn, Pd +a (a =all 3d-, 4d-elements, Mg, Al, Si, etc.) gave a general picture of solid solution stability of the transition metal ternaries, quaternaries at zero K and finite temperature.
The study has been further extended to 6-element HEA with introducing Al into FeCoNiCrMn/Pd since the experiments reported interesting stability behavior of fcc-bcc dual phase and fcc to bcc transformation with varying Al concentration. The calculations found that the partial disordering which is the sublattices inequivalent element distribution relates very much to the phase stability of (FeNiCoCrMn /Pd) Alx. The analysis on the atomic pair interactions in the HEAs further provide the origin of the stability [7]. These results enriched the physics of those high entropy alloys.
[1] Q. Ding, Y. Zhang, X. Chen., et. al. Nature 574, 223-227 (2019).
[2] Nguyen-Dung Tran, Ying Chen, et al., J. Phase Equilibria and Diffusion, 42, 606 (2021).
[3] Shuming Zeng, Jun Ni, et al., npj Computational Materials, 5, 84 (2019).
[4] Xinming Wang, Nguyen-Dung Tran, Ying Chen, Jun Ni, et. al., npj Computational Materials, 8, 253 (2022)
[5] Nguyen-Dung Tran, Theresa Davey, Ying Chen, J. Applied Physics, 133, 045101 (2023).
[6] Hironao Yamada, Chang Liu, Ryo Yoshida, et. al., ACS Central Science 5, 1717 (2019)
[7] under preparation
Short biography:
Professor Ying Chen majored in solid state physics, and her main research field is the computational materials science using the integrated approaches of first-principles, statistical physics and thermodynamic modelling combining materials informatics. She received her PhD from The University of Tokyo in 1996.
After working at the Japan Science and Technology Agency (JST) for 6 years, she became an Associate Professor in The University of Tokyo in 2002; moved to Tohoku University in 2010, and was promoted to Full Professor in 2013. She has been actively involved in large national research projects to conduct computational research on wide range of materials such as intermetallics, alloys, steel, nuclear materials, magnets and high entropy alloys. She also has experience in data science.
During her work at JST, and as a principal scientist of MPDS in 2009, she was one of the main members of the international project of development of the large materials database.
Peer review papers ~110.
