Challenges for Kinetics Predictions via Neural Network Potentials: A Wilkinson's Catalyst Case.

Ab initio kinetic studies are important to understand and design novel chemical reactions. While the Artificial Force Induced Reaction (AFIR) method provides a convenient and efficient framework for kinetic studies, accurate explorations of reaction path networks incur high computational costs. In this article, we are investigating the applicability of Neural Network Potentials (NNP) to accelerate such studies. For this purpose, we are reporting a novel theoretical study of ethylene hydrogenation with a transition metal complex inspired by Wilkinson's catalyst, using the AFIR method. The resulting reaction path network was analyzed by the Generative Topographic Mapping method. The network's geometries were then used to train a state-of-the-art NNP model, to replace expensive ab initio calculations with fast NNP predictions during the search. This procedure was applied to run the first NNP-powered reaction path network exploration using the AFIR method. We discovered that such explorations are particularly challenging for general purpose NNP models, and we identified the underlying limitations. In addition, we are proposing to overcome these challenges by complementing NNP models with fast semiempirical predictions. The proposed solution offers a generally applicable framework, laying the foundations to further accelerate ab initio kinetic studies with Machine Learning Force Fields, and ultimately explore larger systems that are currently inaccessible.

DockOnSurf: A Python Code for the High-Throughput Screening of Flexible Molecules Adsorbed on Surfaces.

We present the open-source python package DockOnSurf which automates the generation and optimization of low-energy adsorption configurations of molecules on extended surfaces and nanoparticles. DockOnSurf is especially geared toward handling polyfunctional flexible adsorbates. The use of this high-throughput workflow allows us to carry out the screening of adsorbate-surface configurations in a systematic, customizable, and traceable way, while keeping the focus on the chemically relevant structures. The screening strategy consists in splitting the exploration of the adsorbate-surface configurational space into chemically meaningful domains, that is, by choosing among different conformers to adsorb, surface adsorption sites, adsorbate anchoring points, and orientations and allowing dissociation of (acidic) protons. We demonstrate the performance of the main features based on varying examples, ranging from CO adsorption on a gold nanoparticle to sorbitol adsorption on hematite. Through the use of the presented program, we aim to foster efficiency, traceability, and ease of use in research within tribology, catalysis, nanoscience, and surface science in general.

Water adlayers on noble metal surfaces: Insights from energy decomposition analysis.

Water molecules adsorbed on noble metal surfaces are of fundamental interest in surface science, in heterogeneous catalysis, and as a model for the metal/water interface. Herein, we analyze 28 water structures adsorbed on five noble metal surfaces (Cu, Ag, Au, Pd, and Pt) via density functional theory and energy decomposition analysis based on the block localized wave function technique. Structures, ranging from monomers to ice adlayers, reveal that the charge transfer from water to the surface is nearly independent from the charge transfer between the water molecules, while the polarization energies are cooperative. Dense water-water networks with small surface dipoles, such as the 39×39 unit cell [experimentally observed on Pt(111)], are favored compared to the highly ordered and popular Hup and Hdown phases. The second main result of our study is that the many-body interactions, which stabilize the water assemblies on the metal surfaces, are dominated by the polarization energies, with the charge transfer scaling with the polarization energies. Hence, if an empirical model could be found that reproduces the polarization energies, the charge transfer could be predicted as well, opening exciting perspectives for force field development.

Parameter-free coordination numbers for solutions and interfaces.

Coordination numbers are among the central quantities to describe the local environment of atoms and are thus used in various applications such as structure analysis, fingerprints, and parameters. Yet, there is no consensus regarding a practical algorithm, and many proposed methods are designed for specific systems. In this work, we propose a scale-free and parameter-free algorithm for nearest neighbor identification. This algorithm extends the powerful Solid-Angle based Nearest-Neighbor (SANN) framework to explicitly include local anisotropy. As such, our Anisotropically corrected SANN (ASANN) algorithm provides with a fast, robust, and adaptive method for computing coordination numbers. The ASANN algorithm is applied to flat and corrugated metallic surfaces to demonstrate that the expected coordination numbers are retrieved without the need for any system-specific adjustments. The same applies to the description of the coordination numbers of metal atoms in AuCu nanoparticles, and we show that ASANN based coordination numbers are well adapted for automatically counting neighbors and the establishment of cluster expansions. Analysis of classical molecular dynamics simulations of an electrified graphite electrode reveals a strong link between the coordination number of Cs+ ions and their position within the double layer, a relation that is absent for Na+, which keeps its first solvation shell even close to the electrode.

Energy Decomposition Analysis for Metal Surface-Adsorbate Interactions by Block Localized Wave Functions.

The energy decomposition analysis based on block localized wave functions (BLW-EDA) allows one to gain physical insight into the nature of chemical bonding, decomposing the interaction energy in (1) a "frozen" term, accounting for the attraction due to electrostatic and dispersion interactions, modulated by Pauli repulsion, (2) the variationally assessed polarization energy, and (3) the charge transfer. This method has so far been applied to gas- and condensed-phase molecular systems. However, its standard version is not compatible with fractionally occupied orbitals (i.e., electronic smearing) and, as a consequence, cannot be applied to metallic surfaces. In this work, we propose a simple and practical extension of BLW-EDA to fractionally occupied orbitals, termed Ensemble BLW-EDA. As illustrative examples, we have applied the developed method to analyze the nature of the interaction of various adsorbates on Pt(111), ranging from physisorbed water to strongly chemisorbed ethylene. Our results show that polarization and charge transfer both contribute significantly at the adsorption minimum for all studied systems. The energy decomposition analysis provides details with respect to competing adsorption sites (e.g., CO on atop vs hollow sites) and elucidates the respective importance of polarization and charge transfer for the increased adsorption energy of H2S compared to H2O. Our development will enable a deeper understanding of the impact of charge transfer on catalytic processes in general.