Structural and Composition Evolution of Palladium Catalyst for CO Oxidation under Steady-State Reaction Conditions

Abstract

Understanding how the composition and structure of heterogeneous catalysts change during in situ reaction is of critical importance but greatly challenging because of the complexity of multiple time–space scale. Herein, we propose a self-adaptive simulation strategy driven by the first-principles microkinetic modeling and genetic-algorithm-based global structural search accelerated by machine learning that allows the interplay of surface reaction and catalyst evolution, and uncover the structural/compositional evolution of a Pd(111) single-crystal catalyst under the reaction conditions for CO oxidation, which is a classic open problem lasting for decades. The possible active phases at the kinetically steady state are identified and their common nature is unraveled. We show that, in the presence of CO/O2 reaction mixtures, a dynamically stable partially oxidized nonstoichiometric palladium oxide (PdO0.44 layer) grown on Pd(111) is formed with the O adatoms inserted into the Pd(111) sublayer that drives the formation of the surface oxide. Remarkably, the first-principles microkinetic analyses demonstrate that this self-evolved PdO0.44 surface at the steady state exhibits higher catalytic activity for CO oxidation as compared with Pd(111) and overoxidized PdO catalyst, which may explain the long-standing puzzle of the “PdO x ” active phase for Pd catalyst during CO oxidation.

General Rules of Coverage-Affected Activity Trends and a Widened Optimal Adsorption Energy Zone in Locating the Catalyst

“…Here, taking the more practical empirical logarithmic coverage effect as an example, we first simulated its catalytic activity trends under different values of the control parameter α 2 by applying a self-consistent microkinetic modeling method with coverage iteration proposed in our previous work. , As shown in Figure a, a variety of volcano-shaped curves can be obtained and several striking features can be highlighted. (i) Compared with no coverage effect (green curve in Figure a), the coverage-affected volcano curve is divided into three stages, i.e., the ascending stage, stable stage, and descending stage.…”

“…Here, taking the more practical empirical logarithmic coverage effect as an example, we first simulated its catalytic activity trends under different values of the control parameter α 2 by applying a self-consistent microkinetic modeling method with coverage iteration proposed in our previous work. 27,37 As Scheme 1. Energy Profile of the Two-Step Catalytic Model a a Notes for abbreviations throughout the paper: μ R ≠, °and μ P ≠, °represent the standard chemical potentials of the transition states of adsorption and desorption steps, respectively.…”

Early-Career and Emerging Researchers in Physical Chemistry Volume 2

“…The composition of the nominally metallic surface under reaction conditions has long been debated. Wang and co-workers introduce a composite computational strategy to address this question for CO oxidation. They first model the oxidation pathways and kinetics constraining the catalyst surface to be metallic.…”

Optimal Dynamic Regimes for CO Oxidation Discovered by Reinforcement Learning

“…The processes of adsorption, desorption, and reaction of CO and O 2 on the metal surface were described by the system of ODEs and model parameters were refined from experimental data. We choose the palladium (111) surface as a model for its high activity and wide range of industrial applications. , This numerical model was used as an environment for the algorithm, whose goal was the maximization of the mean reaction rate by varying the pressures of reactant gases. We have explored how the parameters of the model influence the algorithm’s behavior and found the ranges where the algorithm preferred nonstationary control.…”