Assessment of mean-field microkinetic models for CO methanation on stepped metal surfaces using accelerated kinetic Monte Carlo

Abstract

First-principles screening studies aimed at predicting the catalytic activity of transition metal (TM) catalysts have traditionally been based on mean-field (MF) microkinetic models, which neglect the effect of spatial correlations in the adsorbate layer. Here we critically assess the accuracy of such models for the specific case of CO methanation over stepped metals by comparing to spatially resolved kinetic Monte Carlo (kMC) simulations. We find that the typical low diffusion barriers offered by metal surfaces can be significantly increased at step sites, which results in persisting correlations in the adsorbate layer. As a consequence, MF models may overestimate the catalytic activity of TM catalysts by several orders of magnitude. The potential higher accuracy of kMC models comes at a higher computational cost, which can be especially challenging for surface reactions on metals due to a large disparity in the time scales of different processes. In order to overcome this issue, we implement and test a recently developed algorithm for achieving temporal acceleration of kMC simulations. While the algorithm overall performs quite well, we identify some challenging cases which may lead to a breakdown of acceleration algorithms and discuss possible directions for future algorithm development.

General concepts, assumptions, drawbacks, and misuses in kinetic Monte Carlo and microkinetic modeling simulations applied to computational heterogeneous catalysis

“…Typically, every kMC simulation will involve a huge number of steps (i.e., 10 8 –10 11 ) until the system achieves a steady‐state [i.e., temporal convergence of coverages (θ) and turnover frequencies (TOF), sometimes also called turnover rates (TOR)]. Temporal acceleration of kMC simulations to overcome the problem of the large differences in the time scales of surface processes can be carried out using more refined algorithms . ] Additional simple techniques can also be considered to reduce the kMC computational cost, as for instance, the use of scaling factors in reaction rates for the very fast processes (e.g., diffusion rates) or beginning the kMC simulation from a lattice with an initial coverage obtained from a MM solution.…”

“…Moreover, kMC and MM predicted TOFs and final average coverages are usually different . For instance, compared to similar kMC simulations, mean‐field models can overestimate the catalytic activity by several orders of magnitude as shown in the case of CO methanation on stepped transition metal surfaces . This is the case even when lateral interactions are neglected in both methods.…”

Beyond Scaling Relations for the Description of Catalytic Materials

“…Correspondingly, BEP relations are typically found to exhibit significantly lower errors than thermochemical scaling relations even for the pure metals. 28 We note that an alternative scaling-relationbased approach is to calculate all potential adsorption sites for the descriptors on a mixedmetal surface and then to consider only the most stable adsorption sites. 29 However, at concomitantly increased screening costs this still does not alleviate the problem since different adsorbates (e.g.…”

“…In addition, not only the most stable sites, but also metastable sites missed by this approach, can get populated at higher coverages and then play an important role in the catalytic pathway. 28…”

“…Once a selection of promising catalyst materials had been identified, a next step could then be the evaluation of a more thorough microkinetic model from e.g. a kinetic Monte Carlo simulation, 28 possibly taking into account also lateral interactions between the adsorbates through cluster expansion methods. 36 A full assessment of identified promising catalyst materials would ultimately also need to take into account other aspects such as bulk and surface segregation stability under realistic surface coverages for the chemical reaction and reaction conditions of interest, as well as stability against metal stripping in electrocatalysis applications.…”