Programme

O.35 -- Oral, VCTP-50

Date: Thursday, 7 August 2025

Time: 16:50 - 17:15

Machine learning enhanced global optimization and its application in elucidating hydrogen insertion in SrMO$_y$ (M = Fe and Co)

Thanh Ngoc Pham (1,3,4*), Huynh Thi Thanh Truc (3,4), Daisuke Kan (2), Yoshitada Morikawa (1)

(1) Department of Precision Engineering, Graduate School of Engineering, University of Osaka, Osaka 565-0871, Japan (2) Division of Applied Chemistry, Graduate School of Engineering, University of Osaka, Osaka, 565-0871, Japan (3) An Giang University (AGU), Long Xuyen City, An Giang Province, 880000,Vietnam. (4) Vietnam National University, Ho Chi Minh city, 700000, Vietnam

Hydrogen insertion/extraction can induce topotactic phase transitions, significantly altering the structural and electronic properties of Fe and Co-based-transition metal oxides, SrMO$_y$ (M = Fe, Co). These reversible transitions are of great interest for applications in resistive switching memory, and tunable electronic/ionic conductors for energy conversion and storage systems. Despite their promise, a comprehensive understanding of the phase transition behavior of SrMO$_y$ remains elusive, largely due to complex potential energy landscapes. Herein, we apply the machine learning-enhanced global optimization (GOFEE) method1,2 to investigate hydrogen insertion behavior in SrMO$_y$. By combining evolutionary algorithm (EA) search with a Gaussian Process Regression (GPR) surrogate model, we efficiently identify stable hydrogenated phases of SrMO$_y$ and elucidate thermodynamic stability of SrMO$_y$ under varying hydrogen and oxygen chemical potentials.3,4 First-principles calculations are used to analyze the structural transformations, energetics, and electronic properties of the predicted phases. To validate the computational models, we characterized protonated SrCoO$_2.5$ epitaxial films using photoelectron holography.5 Changes in holography patterns upon protonation, caused by cation displacements, were well reproduced by simulations based on GOFEE predicted structures. This confirms the effectiveness of combining advanced modeling and experimental characterization to elucidate H insertion to SrMOy. Acknowledgments: TNP and HTTT acknowledge the financial supports by AGU under grants No. 25.03.CM.HS and 25.04.CM.HS [1] M. K. Bisbo and B. Hammer, Phys. Rev. Lett. 124, 086102 (2020). [2] T. N. Pham, B. Andrea Choi Tan, Y. Hamamoto, K. Inagaki, I. Hamada, and Y. Morikawa, ACS Catal. 14, 1443 (2024). [3] Y. Isoda, T. N. Pham, R. Aso, S. Nakamizo, T. Majima, S. Hosokawa, K. Nitta, Y. Morikawa, Y. Shimakawa, and D. Kan, Nat. Commun. 16, 56 (2025). [4] Manuscript in preparation [5] D. Kan, Y. Hashimoto, T. N. Pham, L. Xie, Y. Isoda, T. Matsushita, Y. Morikawa, and Y. Shimakawa, J. Ceram. Soc. Jpn. 25062 (2025).

Presenter: Pham Ngoc Thanh