Realistic Modeling of Catalyst Dynamics at Interfaces

Poths Group

Welcome to the “Realistic Modeling of Catalyst Dynamics at Interfaces” group in the Theory department of the Fritz Haber Institute.

Our goal is to explore the role of active site restructuring of catalytic interfaces under realistic conditions. Catalysts are known to undergo dramatic restructuring going from UHV conditions to actual operando conditions; this is, however, difficult to address computationally. Here, our goal is to bridge the complexity gap between surface science and heterogeneous catalysis, where we begin to directly address the kinetics of how active sites can evolve as part of a catalytic reaction coordinate.

We combine first-principles calculations with machine-learned interatomic potentials (MLIPs), automatic process exploration, and multiscale modeling, to span the length- and time-scales from the atomistic to what is measurable experimentally. Catalysis is inherently limited by so-called rare events, or the rate-determining step of a reaction mechanism. If we want to explore catalytic behavior, we must be able to generate complex reaction networks that incorporate both active site dynamics and catalytic reaction steps. To do this, we use the Automatic Process Explorer (APE) approach, developed by the Multiscale Modeling Group in the department. APE, when coupled with MLIPs as a surrogate model for the system potential energy surface (PES), can rapidly and automatically identify reaction barriers for both surface restructuring and reactivity. Using the vast library of APE-identified reaction barriers, also known as processes, we can construct models for dynamics simulations. In this group we explore two main systems, which might be expected to have different dynamic contributions:

Fluxional Sub-nanoscale Clusters. Cluster catalysts are known to have unique size-dependent physical and catalytic properties; and they are small enough that a full treatment of their dynamics is computationally feasible. In this group, one extreme of catalyst dynamics that we explore is that of cluster “fluxionality”, or the ability to interconvert between and populate multiple structures of similar energies. Here, we are exploring (a) the factors which impact the kinetic fluxionality of sub-nanoclusters and (b) the role which fluxionality plays in determining the reaction mechanism of cluster catalysts. Utilizing MLIP-APE, we construct reaction networks of cluster fluxional behavior. We are additionally developing methodologies to treat the state-to-state fluxional dynamics of clusters, using methods from Markov state modeling.

Extended Surfaces. Extended surfaces have been experimentally observed to be dynamic, though they might be expected to be less so than fluxional clusters. We use the MLIP-APE approach to create process libraries for surface restructuring under reaction conditions. To explore the real-time dynamic restructuring of the surfaces, we are developing approaches to map complex restructuring processes onto a lattice, which will be used as the basis for kinetic Monte Carlo (kMC) simulations that combine catalyst restructuring and reactivity into a single microkinetic model. This will enable us to explore the extent to which restructuring plays a role in the reaction mechanism for catalytic oxidation processes.

Ultimately, the catalyst dynamics that we explore across multiple systems can be linked to experimental observables from spectroscopy and microscopy. This will help validate our approach, and give deeper “under-the-hood” insight into the atomistic behavior of catalyst active site evolution under reaction conditions.

Selected Publications

Selected recent publications are listed below (a full list can be found under Publications):

Automatic Process Exploration through Machine Learning Assisted Transition State Searches, K.C. Lai, P. Poths, S. Matera, C. Scheurer, and K. Reuter, Phys. Rev. Lett. (2025)

ML-Accelerated Automatic Process Exploration Reveals Facile O-Induced Pd Step-Edge Restructuring on Catalytic Time Scales, P. Poths et al., ACS Catal. (2024)

Thermodynamic Equilibrium versus Kinetic Trapping: Thermalization of Cluster Catalyst Ensembles Can Extend Beyond Reaction Time Scales, P. Poths, S. Vargas, P. Sautet, and A.N. Alexandrova, ACS Catal. (2024)

Promoter–Poison Partnership Protects Platinum Performance in Coked Cluster Catalysts, P. Poths et al., The Journal of Physical Chemistry C (2023)