(544ea) Evaluation of the Benefits of Kinetic Monte Carlo and Microkinetic Modeling for Catalyst Design Studies in the Presence of Lateral Interactions

Microkinetic models (MKM) are commonly used in catalyst, reactor and process design by solving ordinary differential equations within the mean-field approximation.¹ A more modern and powerful approach is provided by kinetic Monte Carlo (kMC) simulations of heterogeneously catalyzed reactions that integrate microscopic, mesoscopic and macroscopic levels into one multiscale simulation. A key advantage of kMC over traditional MKM is that it explicitly considers correlations, fluctuations, and spatial distributions of the adsorbed species on catalyst surface.²

To evaluate the practical advantages of both methods for descriptor-based catalyst design studies we consider the prototypical CO oxidation reaction on close-packed fcc(111) surfaces,³ and focus on the effect of lateral adsorbate-adsorbate interactions on the predicted activity trends.⁴ When the same elementary steps and rate constants are used in the absence of any lateral adsorbate-adsorbate interactions, both the kMC and MKM simulations result in identical CO oxidation mechanisms and activity trends. When lateral adsorbate-adsorbate interactions are explicitly accounted for, however, the simulation results for activity trends and surface coverages differ. By varying the rate constants for surface diffusion steps in the kMC model, we are able to eliminate surface diffusion as possible origin of the differences; all diffusion steps are quasi-equilibrated and rapid surface rearrangement is kinetically feasible. Instead, the repulsion between CO* and O* enlarges the O* and O₂* covered region, but the interaction effect becomes less relevant at low coverages, i.e., for weakly binding species. Thus, the region of highest CO oxidation activity is found near less reactive metals.

In summary, we have assessed the strengths and weaknesses of MKM and kMC approaches for catalytic activity evaluation. We find that MKM is generally preferred because of its easier implementation and computational efficiency, but it is unreliable for cases with lateral adsorbate-adsorbate interactions or heterogeneous surfaces with multiple active sites. For these cases, a kMC simulation is the preferred choice.

(1) Medford, A. J.; Shi, C.; Hoffmann, M. J.; Lausche, A. C.; Fitzgibbon, S. R.; et al. Lett. 145 (2015) 794-807

(2) Scheffler, K. Reuter, Phys. Rev. B, 73 (2006) 045433

(3) Falsig, H.; Hvolbæk, B.; Kristensen, J. S.; Jiang, T.; Bligaard, T.; et al. Chem. Int. Ed. 47 (2008) 4835-4839

(4) Grabow, L. C.; Hvolbæk, B.; Nørskov, J. K. Catal. 53 (2010) 298-310