We present a new open repository for chemical reactions on catalytic surfaces, available at https://www.catalysis-hub.org. The featured Surface Reactions database contains more than 100,000 chemical adsorption and reaction energies obtained from electronic structure calculations, and is continuously being updated with new datasets. In addition to providing quantum-mechanical results for a broad range of reactions and surfaces from different publications, the database features a systematic, large-scale study of chemical adsorption and hydrogenation on bimetallic alloy surfaces. The database contains reaction specific information, such as the surface composition and reaction energy for each reaction, as well as the atomic surface geometries used in the calculations together with the calculation parameters and output, which are essential for data reproducibility. Data can be accessed from the web-interface as well as from a Python API providing direct access from a local workstation. This enables researchers to efficiently use the data as a basis for further calculations and to generate surrogate models for accelerating the discovery of catalytic materials for sustainable energy applications.
In methanol-to-hydrocarbon chemistry, methanol and dimethyl ether (DME) can act as methylating agents. Therefore, we focus on the different reactivity of methanol and DME towards benzene methylation in H-ZSM-5 at operating conditions by combining first principles microkinetic modeling and experiments. Methylation reactions are known to follow either a concerted reaction path or a stepwise mechanism going through a framework-bound methoxide. By constructing a DFT based microkinetic model including the concerted and stepwise reactions, product formation rates can be calculated at conditions that closely mimic the experimentally applied conditions. Trends in measured rates are relatively well reproduced by our DFT based microkinetic model. We find that benzene methylation with DME is faster than with methanol but the difference decreases with increasing temperature. At low temperatures, the concerted mechanism dominates, however at higher temperatures and low pressures the mechanism shifts to the stepwise pathway. This transition occurs at lower temperatures for methanol than for DME, resulting in smaller reactivity differences between methanol and DME at high temperature. Our theory-experiment approach shows that the widely assumed rate law with zeroth and first order in oxygenate and hydrocarbon partial pressure is not generally applicable and depends on the applied temperature, pressure and feed composition.
Mean-field microkinetic models in combination with Brønsted-Evans-Polanyi like scaling relations have proven highly successful in identifying catalyst materials with good or promising reactivity and selectivity. Analysis of the microkinetic model by means of lattice kinetic Monte Carlo promises a faithful description of a range of atomistic features involving short-range ordering of species in the vicinity of an active site. In this paper, we use the "fruit fly" example reaction of CO oxidation on fcc(111) transition and coinage metals to motivate and develop a lattice kinetic Monte Carlo solver suitable for the numerically challenging case of vastly disparate rate constants. As a result, we show that for the case of infinitely fast diffusion and absence of adsorbate-adsorbate interaction it is, in fact, possible to match the prediction of the mean-field-theory method and the lattice kinetic Monte Carlo method. As a corollary, we conclude that lattice kinetic Monte Carlo simulations of surface chemical reactions are most likely to provide additional insight over mean-field simulations if diffusion limitations or adsorbate-adsorbate interactions have a significant influence on the mixing of the adsorbates.
Lattice kinetic Monte Carlo simulations have become a vital tool for predictive quality atomistic understanding of complex surface chemical reaction kinetics over a wide range of reaction conditions. In order to expand their practical value in terms of giving guidelines for the atomic level design of catalytic systems, it is very desirable to readily evaluate a sensitivity analysis for a given model. The result of such a sensitivity analysis quantitatively expresses the dependency of the turnover frequency, being the main output variable, on the rate constants entering the model. In the past, the application of sensitivity analysis, such as degree of rate control, has been hampered by its exuberant computational effort required to accurately sample numerical derivatives of a property that is obtained from a stochastic simulation method. In this study, we present an efficient and robust three-stage approach that is capable of reliably evaluating the sensitivity measures for stiff microkinetic models as we demonstrate using the CO oxidation on RuO2(110) as a prototypical reaction. In the first step, we utilize the Fisher information matrix for filtering out elementary processes which only yield negligible sensitivity. Then we employ an estimator based on the linear response theory for calculating the sensitivity measure for non-critical conditions which covers the majority of cases. Finally, we adapt a method for sampling coupled finite differences for evaluating the sensitivity measure for lattice based models. This allows for an efficient evaluation even in critical regions near a second order phase transition that are hitherto difficult to control. The combined approach leads to significant computational savings over straightforward numerical derivatives and should aid in accelerating the nano-scale design of heterogeneous catalysts.
A novel nanoparticulate catalyst of copper (Cu) and ruthenium (Ru) was designed for low-temperature ammonia oxidation at near-stoichiometric mixtures using a bottom-up approach. A synergistic effect of the two metals was found. An optimum CuRu catalyst presents a reaction rate threefold higher than that for Ru and forty-fold higher than that for Cu. X-ray absorption spectroscopy suggests that in the most active catalyst Cu forms one or two monolayer thick patches on Ru and the catalysts are less active once 3D Cu islands form. The good performance of the tuned Cu/Ru catalyst is attributed to changes in the electronic structure, and thus the altered adsorption properties of the surface Cu sites.
We present a framework for physically motivated models of adsorbate–adsorbate interaction between small molecules on transition and coinage metals based on modifications to the substrate electronic structure due to adsorption. We use this framework to develop one model for transition and one for coinage metal surfaces. The models for transition metals are based on the d-band center position, and the models for coinage metals are based on partial charges. The models require no empirical parameters, only two first-principles calculations per adsorbate as input, and therefore scale linearly with the number of reaction intermediates. By theory to theory comparison with explicit density functional theory calculations over a wide range of adsorbates and surfaces, we show that the root-mean-squared error for differential adsorption energies is less than 0.2 eV for up to 1 ML coverage.
Descriptor-based analysis is a powerful tool for understanding the trends across various catalysts. In general, the rate of a reaction over a given catalyst is a function of many parameters-reaction energies, activation barriers, thermodynamic conditions, etc. The high dimensionality of this problem makes it very difficult and expensive to solve completely, and even a full solution would not give much insight into the rational design of new catalysts. The descriptor-based approach seeks to determine a few "descriptors" upon which the other parameters are dependent. By doing this it is possible to reduce the dimensionality of the problem-preferably to 1 or 2 descriptors-thus greatly reducing computational efforts and simultaneously increasing the understanding of trends in catalysis. The "CatMAP" Python module seeks to standardize and automate many of the mathematical routines necessary to move from "descriptor space" to reaction rates for heterogeneous (electro) catalysts. The module is designed to be both flexible and powerful, and is available for free online. A "reaction model" can be fully defined by a configuration file, thus no new programming is necessary to change the complexity or assumptions of a model. Furthermore, various steps in the process of moving from descriptors to reaction rates have been abstracted into separate Python classes, making it easy to change the methods used or add new functionality. This work discusses the structure of the code and presents the underlying algorithms and mathematical expressions both generally and via an example for the CO oxidation reaction.
We present a multi-lattice kinetic Monte Carlo (kMC) approach that efficiently describes the atomistic dynamics of morphological transitions between commensurate structures at crystal surfaces. As an example we study the reduction of a $(\sqrt{5}\times \sqrt{5})R27^{\circ}$ PdO(101) overlayer on Pd(100) in a CO atmosphere. Extensive density-functional theory calculations are used to establish an atomistic pathway for the oxide reduction process. First-principles multi-lattice kMC simulations on the basis of this pathway fully reproduce the experimental temperature dependence of the reduction rate [Fernandes et al., Surf. Sci. 2014, 621, 31-39] and highlight the crucial role of elementary processes special to the boundary between oxide and metal domains.
We use multi-scale modeling to analyze laser-induced fluorescence (LIF) measurements of the CO oxidation reaction over Pd(100) at near-ambient reaction conditions. Integrating density-functional theory based kinetic Monte Carlo simulations of the active catalyst into fluid-dynamical simulations of the mass transport inside the reactor chamber we calculate the reaction product concentration directly above the catalyst surface. Comparing corresponding data calculated for different surface models against the measured LIF signals we can discriminate the one that predominantly actuates the experimentally measured catalytic activity. For the probed CO oxidation reaction conditions the experimental activity is due to pristine Pd(100), possibly coexisting with other (oxidic) domains on the surface.
Kinetic Monte Carlo (kMC) simulations have emerged as a key tool for microkinetic modeling in heterogeneous catalysis and other materials applications. Systems, where site-specificity of all elementary reactions allows a mapping onto a lattice of discrete active sites, can be addressed within the particularly efficient lattice kMC approach. To this end we describe the versatile kmos software package, which offers a most user-friendly implementation, execution, and evaluation of lattice kMC models of arbitrary complexity in one- to three-dimensional lattice systems, involving multiple active sites in periodic or aperiodic arrangements, as well as site-resolved pairwise and higher-order lateral interactions. Conceptually, kmos achieves a maximum runtime performance which is essentially independent of lattice size by generating code for the efficiency-determining local update of available events that is optimized for a defined kMC model. For this model definition and the control of all runtime and evaluation aspects kmos offers a high-level application programming interface. Usage proceeds interactively, via scripts, or a graphical user interface, which visualizes the model geometry, the lattice occupations and rates of selected elementary reactions, while allowing on-the-fly changes of simulation parameters. We demonstrate the performance and scaling of kmos with the application to kMC models for surface catalytic processes, where for given operation conditions (temperature and partial pressures of all reactants) central simulation outcomes are catalytic activity and selectivities, surface composition, and mechanistic insight into the occurrence of individual elementary processes in the reaction network.
We present first-principles kinetic Monte Carlo (1p-kMC) simulations addressing the CO oxidation reaction at Pd(100) for gas-phase conditions ranging from ultra-high vacuum to ambient pressures and elevated temperatures. For the latter technologically relevant regime there is a long-standing debate regarding the nature of the active surface. The pristine metallic surface, an ultra-thin \((\sqrt{5}\times \sqrt{5})R27^{\circ}\)PdO(101) surface oxide, and thicker oxide layers have each been suggested as the active state. We investigate these hypotheses with 1p-kMC simulations focusing on either the Pd(100) surface or the PdO(101) surface oxide and intriguingly obtain a range of (T, p)-conditions where both terminations appear metastable. The predicted bistability regime nicely ties in with oscillatory behavior reported experimentally by Hendriksen et al. (Catal Today 105:234, 2005). Within this regime we find that both surface terminations exhibit very similar intrinsic reactivity, which puts doubts on attempts to assign the catalytic function to just one active state.
We present high-pressure x-ray photoelectron spectroscopy (HP-XPS) and first-principles kinetic Monte Carlo study addressing the nature of the active surface in CO oxidation over Pd(100). Simultaneously measuring the chemical composition at the surface and in the near-surface gas phase, we reveal both O-covered pristine Pd(100) and a surface oxide as stable, highly active phases in the near-ambient regime accessible to HP-XPS. Surprisingly, no adsorbed CO can be detected during high CO(2) production rates, which can be explained by a combination of a remarkably short residence time of the CO molecule on the surface and mass-transfer limitations in the present setup.
We present first-principles kinetic Monte Carlo (1p-kMC) simulations addressing the CO oxidation reaction at Pd(100) for gas-phase conditions ranging from ultra-high vacuum (UHV) to ambient pressures and elevated temperatures. For the latter technologically relevant regime there is a long-standing debate regarding the nature of the active surface. The pristine metallic surface, an ultra-thin (sqrt5xsqrt5)R27^o PdO(101) surface oxide, and thicker oxide layers have each been suggested as the active state. We investigate these hypotheses with 1p-kMC simulations focusing on either the Pd(100) surface or the PdO(101) surface oxide and intriguingly obtain a range of (T, p)-conditions where both terminations appear metastable. The predicted bistability regime nicely ties in with oscillatory behavior reported experimentally by Hendriksen and coworkers [Catal. Today 105, 234 (2005)]. Within this regime we find that both surface terminations exhibit very similar intrinsic reactivity, which puts doubts on attempts to assign the catalytic function to just one active state.
We present first-principles kinetic Monte Carlo (1p-kMC) simulations addressing the CO oxidation reaction at Pd(100) for gas-phase conditions ranging from ultra-high vacuum to ambient pressures and elevated temperatures. For the latter technologically relevant regime there is a long-standing debate regarding the nature of the active surface. The pristine metallic surface, an ultra-thin (√(5)×√(5))R27^∘ PdO(101) surface oxide, and thicker oxide layers have each been suggested as the active state. We investigate these hypotheses with 1p-kMC simulations focusing on either the Pd(100) surface or the PdO(101) surface oxide and intriguingly obtain a range of ( T , p )-conditions where both terminations appear metastable. The predicted bistability regime nicely ties in with oscillatory behavior reported experimentally by Hendriksen et al. (Catal Today 105:234, 2005 ). Within this regime we find that both surface terminations exhibit very similar intrinsic reactivity, which puts doubts on attempts to assign the catalytic function to just one active state.
We present first-principles kinetic Monte Carlo (1p-kMC) simulations addressing the CO oxidation reaction at Pd(100) for gas-phase conditions ranging from ultra-high vacuum to ambient pressures and elevated temperatures. For the latter technologically relevant regime there is a long-standing debate regarding the nature of the active surface. The pristine metallic surface, an ultra-thin (5×5)R27∘\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$(\sqrt{5}\times \sqrt{5})R27^{\circ}$$\end{document}PdO(101) surface oxide, and thicker oxide layers have each been suggested as the active state. We investigate these hypotheses with 1p-kMC simulations focusing on either the Pd(100) surface or the PdO(101) surface oxide and intriguingly obtain a range of (T, p)-conditions where both terminations appear metastable. The predicted bistability regime nicely ties in with oscillatory behavior reported experimentally by Hendriksen et al. (Catal Today 105:234, 2005). Within this regime we find that both surface terminations exhibit very similar intrinsic reactivity, which puts doubts on attempts to assign the catalytic function to just one active state.