Collaborative neural networks-accelerated prediction of transition state energy barriers for CO catalytic oxidation

Tang，Tian1,2; Xue，Jingyu1,2; Shen，Xiaoqiang1,2; Chen，Jinfei1,2; Rac，Vladislav3; Rakić，Vesna3; Li，Xinbao4; Zhou，Bo5; Chen，Yanrong1,2; Song，Hao6; Du，Xuesen1,2

2024-10-15

10.1016/j.jallcom.2024.175546

Journal of Alloys and Compounds   影响因子和分区

0925-8388

We present a machine learning (ML) framework to explore the predictability of energy barriers of transition state (activation energy) for CO catalytic oxidation from non-collected ∼300 single-point energy data. Radial basis function (RBF) and back propagation (BP) neural networks for sensitivity analysis and small sample prediction training are used to fuse descriptor data to predict transition state energy barriers. Although the trained ML model may underfit, it is capable of predicting energy barriers of transition state with NH adsorption energy (E NH*), O adsorption energy (E O*), CO adsorption energy (E CO*), fermi energy (E) and d-band center, and yields prediction precision with relative error of 9.88 % (RMSE = 0.02 eV). This model avoids the extensive iterative process of high-dimensional spatial extreme value search and numerical optimization, and ∼5-fold reduction of computation time compared to the traditional nudged elastic band (NEB) method. This study represents a significant step towards transition state energy barriers prediction of surface catalytic reactions as a way to improve the efficiency of high-throughput screening catalysts.

CeO2-based catalyst CO oxidation DFT+U Machine learning

SCI ; EI

英语

其他

20242916701446

Activation energy ; Adsorption ; Ammonia ; Backpropagation ; Catalysis ; Catalysts ; Catalytic oxidation ; Forecasting ; Iterative methods ; Kinetic theory ; Numerical methods ; Optimization ; Radial basis function networks ; Sensitivity analysis ; Surface reactions

Air Pollution Control:451.2 ; Artificial Intelligence:723.4 ; Chemical Reactions:802.2 ; Chemical Operations:802.3 ; Chemical Agents and Basic Industrial Chemicals:803 ; Chemical Products Generally:804 ; Inorganic Compounds:804.2 ; Mathematics:921 ; Optimization Techniques:921.5 ; Numerical Methods:921.6

MATERIALS SCIENCE

2-s2.0-85198264681

Scopus

1.Key Laboratory of Low-Grade Energy Utilization Technologies and Systems,Ministry of Education of PRC,Chongqing University,Chongqing,400044,China
2.School of Energy and Power Engineering,Chongqing University,Chongqing,400044,China
3.Faculty of Agriculture,University of Belgrade,Zemun-Belgrade,Nemanjina 6,11080,Serbia
4.College of Metrology and Measurement Engineering,China Jiliang University,Hangzhou,310018,China
5.Department of Mechanics and Aerospace Engineering,Southern University of Science and Technology,Shenzhen,518055,China
6.State Key Lab of Clean Energy Utilization,State Environmental Protection Engineering Center for Coal-Fired Air Pollution Control,Zhejiang University,Hangzhou,Zhejiang,310027,China

Tang，Tian,Xue，Jingyu,Shen，Xiaoqiang,et al. Collaborative neural networks-accelerated prediction of transition state energy barriers for CO catalytic oxidation[J]. Journal of Alloys and Compounds,2024,1002.

Tang，Tian.,Xue，Jingyu.,Shen，Xiaoqiang.,Chen，Jinfei.,Rac，Vladislav.,...&Du，Xuesen.(2024).Collaborative neural networks-accelerated prediction of transition state energy barriers for CO catalytic oxidation.Journal of Alloys and Compounds,1002.

Tang，Tian,et al."Collaborative neural networks-accelerated prediction of transition state energy barriers for CO catalytic oxidation".Journal of Alloys and Compounds 1002(2024).