Accessing complex reconstructed material structures with hybrid global optimization accelerated via on-the-fly machine learning - Journal Article

Aug 2, 2026

Rational design of highly efficient materials depends on an atom-level structure–performance understanding, yet real working conditions trigger unpredictable structural reconstructions. This work presents a hybrid evolutionary algorithm combining differential evolution and genetic algorithms within a multi-tribe framework, using an on-the-fly machine learning calculator to rapidly identify low-lying structures. The method is validated by optimizing oxidized Pt/Pd/Cu surfaces under (4×4) periodicity, yielding energetically lower structures consistent with experimental results.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages9

Reading Time23 min

AccessFree to download

Introduction

Global optimization and reconstructed surfaces

Challenges in exploring complex PES and reducing computational cost

UCLA UCLA Previously Published Works Title Accessing complex reconstructed material structures with hybrid global optimization accelerated via on-the-fly machine learning Permalink [https://escholarship.org/uc/item/6887z4tz](https://escholarship.org/uc/item/6887z4tz) Journal Chemical Science, 14(33) ISSN 2041-6520 Authors Shi, Xiangcheng Cheng, Dongfang Zhao, Ran et al. Publication Date 2023-08-23 DOI 10.1039/d3sc02974c Copyright Information This work is made available under the terms of a Creative Commons Attribution License, available at [https://creativecommons.org/licenses/by/4.0/](https://creativecommons.org/licenses/by/4.0/) Peer reviewed [eScholarship.org](eScholarship.org) Powered by the California Digital Library University of California Chemical Science EDGE ARTICLE Cite this: Chem. Sci., 2023, 14, 8777 All publication charges for this article have been paid for by the Royal Society of Chemistry Received 12th June 2023 Accepted 13th July 2023 DOI: 10.1039/d3sc02974c[rsc.li/chemical-science](rsc.li/chemical-science) Accessing complex reconstructed material structures with hybrid global optimization accelerated via on-the-ﬂy machine learning† Xiangcheng Shi,abdf Dongfang Cheng,abd Ran Zhao,abd Gong Zhang, abd Shican Wu,abd Shiyu Zhen,abd Zhi-Jian Zhao *abcd and Jinlong Gong *abcde The complex reconstructed structure of materials can be revealed by global optimization. This paper describes a hybrid evolutionary algorithm (HEA) that combines diﬀerential evolution and genetic algorithms with a multi-tribe framework. An on-the-ﬂy machine learning calculator is adopted to expedite the identiﬁcation of low-lying structures. With a superior performance to other well-established methods, we further demonstrate its eﬃcacy by optimizing the complex oxidized surface of Pt/Pd/Cu with diﬀerent facets under (4 × 4) periodicity. The obtained structures are consistent with experimental results and are energetically lower than the previously presented model. Introduction The rational design of highly eﬃcient materials requires an atom-level understanding of their structure–performance relationship.1–5 However, under working conditions, most materials undergo a structural reconstruction accompanied by an unpredictable performance.6–9 For example, some bimetallic catalysts like Au–Ag alloys can exhibit dynamic geometrical and compositional reconstruction during the reaction, which generates active sites to boost performance.6,10–14 In contrast, under high voltage, metal catalysts can be partially oxidized, which results in destabilization and higher dissolution of the active species.15,16 Surface reconstruction sensitively varies with the nature of unreconstructed surfaces and the compositions/ concentrations of adsorbent.17,18 Due to the diﬃculty to model a reconstructed surface, it is a generally adopted, but improper, practice to oversimplify a complex surface when modeling. For example, to model a partially oxidized surface, some studies place adsorbed oxygen,19 or only a layer of metal oxides,20 upon aSchool of Chemical Engineering and Technology, Key Laboratory for Green Chemical Technology of Ministry of Education, Tianjin University, Tianjin 300072, China. [E-mail: zjzhao@tju.edu.cn](E-mail: zjzhao@tju.edu.cn);[jlgong@tju.edu.cn](jlgong@tju.edu.cn) bCollaborative Innovation Center of Chemical Science and Engineering (Tianjin), Tianjin 300072, China cHaihe Laboratory of Sustainable Chemical Transformations, Tianjin 300192, China dNational Industry-Education Platform of Energy Storage, Tianjin University, 135 Yaguan Road, Tianjin 300350, China eJoint School of National University of Singapore and Tianjin University, International Campus of Tianjin University, Binhai New City, Fuzhou 350207, Fujian, China fDepartment of Chemistry, National University of Singapore, 3 Science Drive 3, Singapore, 117543, Republic of Singapore † Electronic supplementary informat

FAQs

What is the main purpose of the hybrid global optimization approach in the paper?

To access complex reconstructed material structures by improving global optimization efficiency and accelerating the identification of low-lying structures.

Which algorithmic components are combined in the proposed method?

The method combines differential evolution and genetic algorithms under a multi-tribe framework.

How is the method validated, and what system is optimized?

It is benchmarked by optimizing the complex oxidized surface of Pt/Pd/Cu with different facets under (4×4) periodicity, producing structures consistent with experiments and energetically lower than earlier models.

Aug 2, 2026

Upload by Emma Wilson