Thomas Bligaard

Solid–water interfaces are ubiquitous in nature and technology. In particular, technologies evolving in the green transition, such as electrocatalysis, heavily rely on the junction of an electrolyte and an electrode as a central part of the device. For the understanding of atomic-scale processes taking place at the electrolyte–electrode interface, density functional theory (DFT) has become the de facto standard. The validation of DFT’s ability to simulate the interfacial solid/water interaction is crucial, and ideal simulation setups need to be identified in order to prevent avoidable systematic errors. Here, we develop a rigorous sampling protocol for benchmarking the adsorption/desorption energetics of water on metallic surfaces against experimental temperature programmed desorption, single crystal adsorption calorimetry, and thermal energy atom scattering. We screened DFT’s quality on a series of transition metal surfaces, applying three of the most common exchange–correlation approximations: PBE-D3, RPBE-D3, and BEEF-vdW. We find that all three xc-functionals reflect the pseudo-zeroth order desorption of water rooted in the combination of attractive adsorbate–adsorbate interactions and their saturation at low and intermediate coverages, respectively. However, both RPBE-D3 and BEEF-vdW lead to more accurate water adsorption strengths, while PBE-D3 clearly overbinds near-surface water. We relate the variations in binding strength to specific variations in water–metal and water–water interactions, highlighting the structural consequences inherent in an uninformed choice of simulation parameters. Our study gives atomistic insight into water’s complex adsorption equilibrium. Furthermore, it represents a guideline for future DFT-based simulations of solvated solid interfaces by providing an assessment of systematic errors in specific setups.

Local optimization of adsorption systems inherently involves different scales: within the substrate, within the molecule, and between the molecule and the substrate. In this work, we show how the explicit modeling of different characteristics of the bonds in these systems improves the performance of machine learning methods for optimization. We introduce an anisotropic kernel in the Gaussian process regression framework that guides the search for the local minimum, and we show its overall good performance across different types of atomic systems. The method shows a speed-up of up to a factor of two compared with the fastest standard optimization methods on adsorption systems. Additionally, we show that a limited memory approach is not only beneficial in terms of overall computational resources but can also result in a further reduction of energy and force calculations.

Many breakthroughs have been achieved in rechargeable aluminum-ion battery technologies in recent years. Most recently, operando X-ray diffraction (XRD) combined with density functional theory (DFT) calculations was reported to study the chloroaluminate anion (AlCl₄ ⁻)-intercalated graphite cathode of the battery. However, there are quite a few discrepancies between the measured and simulated XRD patterns. This work is focused on the simulation of XRD patterns of graphite intercalation compounds (GICs) with DFT calculations. Our results reveal that both the geometry of AlCl₄ ⁻ in graphite and the gallery height of GICs are dependent on the intercalant density. At low intercalant density, the gallery height keeps constant, but at high intercalant densities, the gallery height is linearly related to the intercalant density. Our simulated XRD patterns are highly consistent with the measured operando XRD patterns. Not only do the angles of the peaks match very well, but also the relative intensities and the corresponding electrode capacities show reasonable agreement with the experimental results. The DFT simulation of the XRD pattern provides significant information on the stage index and the charge capacity of the GIC electrode.

A comprehensive database of chemical properties on a vast set of transition metal surfaces has the potential to accelerate the discovery of novel catalytic materials for energy and industrial applications. In this data descriptor, we present such an extensive study of chemisorption properties of important adsorbates - e.g., C, O, N, H, S, CH_x, OH, NH, and SH - on 2,035 bimetallic alloy surfaces in 5 different stoichiometric ratios, i.e., 0%, 25%, 50%, 75%, and 100%. To our knowledge, it is the first systematic study to compile the adsorption properties of such a well-defined, large chemical space of catalytic interest. We propose that a collection of catalytic properties of this magnitude can assist with the development of machine learning enabled surrogate models in theoretical catalysis research to design robust catalysts with high activity for challenging chemical transformations. This database is made publicly available through the platform www.Catalysis-hub.org for easy retrieval of the data for further scientific analysis.

We present a new open repository for chemical reactions on catalytic surfaces, available at https://www.catalysis-hub.org. The featured database for surface reactions contains more than 100,000 chemisorption 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 surface geometries and calculational parameters, essential for data reproducibility. By providing direct access via the web-interface as well as a Python API, we seek to accelerate the discovery of catalytic materials for sustainable energy applications by enabling researchers to efficiently use the data as a basis for new calculations and model generation.

We present the incorporation of a surrogate Gaussian process regression (GPR) atomistic model to greatly accelerate the rate of convergence of classical nudged elastic band (NEB) calculations. In our surrogate model approach, the cost of converging the elastic band no longer scales with the number of moving images on the path. This provides a far more efficient and robust transition state search. In contrast to a conventional NEB calculation, the algorithm presented here eliminates any need for manipulating the number of images to obtain a converged result. This is achieved by inventing a new convergence criteria that exploits the probabilistic nature of the GPR to use uncertainty estimates of all images in combination with the force in the saddle point in the target model potential. Our method is an order of magnitude faster in terms of function evaluations than the conventional NEB method with no accuracy loss for the converged energy barrier values.