Graph neural network-driven prediction of high-performance CO_(2)reduction catalysts based on Cu-based high-entropy alloys  

基于图神经网络的高性能二氧化碳还原高熵合金催化剂预测

作　　者：Zihao Jiao Chengyi Zhang Ya Liu Liejin Guo Ziyun Wang 焦自浩;张成翼;刘亚;郭烈锦;王子运(动力工程多相流国家重点实验室,陕西西安710049;奥克兰大学化学科学学院,新西兰奥克兰1010)

机构地区：[1]International Research Center for Renewable Energy,State Key Laboratory of Multiphase Flow in Power Engineering,Xi’an Jiaotong University,Shaanxi 710049,China [2]School of Chemical Sciences,University of Auckland,Auckland 1010,New Zealand

出　　处：《Chinese Journal of Catalysis》2025年第4期197-207,共11页催化学报(英文)

基　　金：马尔斯登基金委员会(21-UOA-237);种子计划通用资助(22-UOA-031-CGS).

摘　　要：High-entropy alloy(HEA)offer tunable composition and surface structures,enabling the creation of novel active sites that enhance catalytic performance in renewable energy application.However,the inherent surface complexity and tendency for elemental segregation,which results in discrepancies between bulk and surface compositions,pose challenges for direct investigation via density functional theory.To address this,Monte Carlo simulations combined with molecular dynamics were employed to model surface segregation across a broad range of elements,including Cu,Ag,Au,Pt,Pd,and Al.The analysis revealed a trend in surface segregation propensity following the order Ag>Au>Al>Cu>Pd>Pt.To capture the correlation between surface site characteristics and the free energy of multi-dentate CO_(2)reduction intermediates,a graph neural network was designed,where adsorbates were transformed into pseudo-atoms at their centers of mass.This model achieved mean absolute errors of 0.08–0.15 eV for the free energies of C_(2)intermediates,enabling precise site activity quantification.Results indicated that increasing the concentration of Cu,Ag,and Al significantly boosts activity for CO and C_(2)formation,whereas Au,Pd,and Pt exhibit negative effects.By screening stable composition space,promising HEA bulk compositions for CO,HCOOH,and C_(2)products were predicted,offering superior catalytic activity compared to pure Cu catalysts.本文针对铜基高熵合金(HEA)在C0_(2)还原反应中的催化性能,探讨了其可调组成和表面结构对催化活性的提升作用.由于HEA材料的表面复杂性及元素偏析倾向,导致体相与表面组成不一致,这为通过密度泛函理论进行直接研究带来了挑战.本研究旨在克服这些困难,探索如何通过模拟和机器学习方法优化HEA催化剂的设计.为了模拟表面元素的分离行为,本文采用了蒙特卡罗模拟与分子动力学相结合的方式,涵盖了包括Cu,Ag,Au,Pt,Pd和Al在内的多个元素.通过这一方法,揭示了表面偏析倾向的规律:Ag>Au>Al>Cu>Pd>Pt.为了解析表面位点特征与多齿C0_(2)还原中间体自由能之间的关系,设计了一个基于图神经网络的模型,其中吸附物被转化为位于其质心的伪原子.该模型在C_(2)中间体自由能预测中取得了0.08–0.15 eV的平均绝对误差,能够精准量化表面位点活性.结果表明,增加Cu,Ag和Al的浓度显著提高了C_(0)和C_(2)的生成活性,而Au,Pd和Pt则表现出抑制作用.对于产氢(HER)反应,Al和Au是关键调节因子:提高Au含量或降低Al含量有助于提升HER活性;而Pd含量的增加或Au含量的减少则促进了甲酸的生成.此外,通过筛选稳定的组成空间,研究还预测了比纯Cu催化剂更具催化活性的HEA体相组成,适用于C_(0),HC0OH和C_(2)产品的生成.综上,HEA通过其可调的组成和表面结构能够有效促进C0_(2)还原反应,尽管表面组成和元素分离带来了挑战,蒙特卡罗模拟与图神经网络的结合提供了一个高效的催化剂设计框架.本文不仅为HEA催化剂的理性设计提供了理论依据,还为高通量筛选具有优异催化性能的HEA电催化剂奠定了基础.

关 键 词：Density functional theory Machine learning CO_(2)reduction High entropy alloys Graph neural network 

分 类 号：O643.36[理学—物理化学]