LASP 3.7 for Large-scale Atomic Simulation and the Application to Ethene Epoxidation on Silver

Back

Zhi-Pan Liu （刘智攀）

Department of Chemistry, Fudan University, Shanghai, China

This lecture introduces our recent progress in machine-learning atomic simulations for catalysis by using LASP software (version 3.7, www.lasphub.com) developed by our group[1] since 2018. Our methodology for bulk, surface structure search, and reaction sampling in the grand-canonical ensemble will be overviewed[2-3]. These methods were recently applied to resolve the active site for selective ethene epoxidation on silver, a long-standing problem in the field for more than 50 years. Ag-catalyzed ethene epoxidation is the only viable route for making ethene oxide (EO) in industry. Due to the lack of tools to probe the reaction at high temperatures and high pressures, the active site structure remains highly controversial. Here, with advanced machine-learning grand canonical global structure exploration and in-situ catalysis experiments, we identify a unique surface oxide phase, namely O5 phase, grown on Ag(100) under industrial catalytic conditions. This O5 phase features square-pyramidal subsurface O and strongly adsorbed ethene, which can selectively convert ethene to EO. The other Ag surface facets, Ag(111) and Ag(110), although also reconstructing to surface oxide phases, only produce CO2 due to the lack of subsurface O. The complex in-situ surface phases with distinct selectivity contribute to an overall medium (50%) selectivity of Ag catalyst to EO. Our further catalysis experiments with in-situ infrared spectrum confirm the theory-predicted IR-active C=C vibration (1583 cm-1) of adsorbed ethene on O5 phase and the microkinetics simulation results. The active phase structure and activity help to settle the long dispute on the nature of active oxygen in ethene epoxidation caused by the “pressure gap” and shed light on the design of better catalysts for olefin epoxidation.[4]

1. Pei-Lin Kang, Cheng Shang, Zhi-Pan Liu, Large-Scale Atomic Simulation via Machine Learning Potentials Constructed by Global Potential Energy Surface Exploration, Acc. Chem. Res .2020,  53, 2119 

2. Sicong Ma, Si-Da Huang, Zhi-Pan Liu , Dynamic Coordination of Cations and Catalytic Selectivity on Zn-Cr oxide Alloy during Syngas Conversion, Nature Catal.,  2019, 2, 671

3. Ye-Fei Li, Zhi-Pan Liu, Smallest stable Si/SiO2 interface that suppresses quantum tunneling   from machine-learning-based global search, Phys. Rev. Lett.   2022, 128, 226102 

4. Dongxiao Chen#, Lin Chen#, Qian-Cheng Zhao, Zheng-Xin Yang, Cheng Shang, Zhi-Pan Liu*, 'Square-pyramidal Subsurface Oxygen [Ag4OAg] Drives Selective Ethene Epoxidation on Silver', Nature Catal.,  2024, 7, 536