Beyond single-crystal surfaces: The GAL21 water/metal force field.

Solvent effects are notoriously difficult to describe for metallic nanoparticles (NPs). Here, we introduce GAL21 which is the first pairwise additive force field that is specifically designed to modulate the near chemisorption energy of water as a function of the coordination numbers of the metallic atoms. We find a quadratic dependence to be most suitable for capturing the dependence of the adsorption energy of water on the generalized coordination number (GCN) of the metal atoms. GAL21 has been fitted against DFT adsorption energies for Cu, Ag, Au, Ni, Pd, Pt, and Co on 500 configurations and validated on about 3000 configurations for each metal, constructed on five surfaces with GCNs varying from 2.5 to 11.25. Depending on the metals, the root mean square deviation is found between 0.7 kcal mol-1 (Au) to 1.6 kcal mol-1 (Ni). Using GAL21, as implemented in the open-source code CP2K, we then evaluate the solvation energy of Au55 and Pt55 NPs in water using thermodynamic integration. The solvation free energy is found to be larger for Pt than for Au and systematically larger than 200 kcal mol-1, demonstrating the large impact of solvent on the surface energetics of NPs. Still, given that the amorphous NPs are both, the most stable and the most solvated ones, we do not predict a change in the preferred morphology between the gas-phase and in water. Finally, based on a linear regression on three sizes of NPs (from 38 to 147), the solvation energy for Au and Pt surface atoms is found to be -5.2 and -9.9 kcal mol-1, respectively.

Transferable Gaussian Attractive Potentials for Organic/Oxide Interfaces.

Organic/oxide interfaces play an important role in many areas of chemistry and in particular for lubrication and corrosion. Molecular dynamics simulations are the method of choice for providing complementary insight to experiments. However, the force fields used to simulate the interaction between molecules and oxide surfaces tend to capture only weak physisorption interactions, discarding the stabilizing Lewis acid/base interactions. We here propose a simple complement to the straightforward molecular mechanics description based on "out-of-the-box" Lennard-Jones potentials and electrostatic interactions: the addition of an attractive Gaussian potential between reactive sites of the surface and heteroatoms of adsorbed organic molecules, leading to the Gaussian Lennard-Jones (GLJ) potential. The interactions of four oxygenated and four amine molecules with the typical and widespread hematite and γ-alumina surfaces are investigated. The root mean square deviation (RMSD) for all probed molecules is only 5.7 kcal/mol, which corresponds to an error of 23% over hematite. On γ-alumina, the RMSD is 11.2 kcal/mol using a single parameter for all five chemically inequivalent surface aluminum atoms. Applying GLJ to the simulation of organic films on oxide surfaces demonstrates that the mobility of the surfactants is overestimated by the simplistic LJ potential, while GLJ and other qualitatively correct potentials show a strong structuration and slow dynamics of the surface films, as could be expected from the first-principles adsorption energies for model head groups.

Solvation Free Energies and Adsorption Energies at the Metal/Water Interface from Hybrid Quantum-Mechanical/Molecular Mechanics Simulations.

Modeling adsorption at metal/water interfaces is a cornerstone toward an improved understanding in a variety of fields from heterogeneous catalysis to corrosion. We propose and validate a hybrid scheme that combines the adsorption free energies obtained in the gas phase at the density functional theory level with the variation in solvation from the bulk phase to the interface evaluated using a MM-based alchemical transformation, denoted MMsolv. Using the GAL17 force field for the platinum/water interaction, we retrieve a qualitatively correct interaction energy of the water solvent at the interface. This interaction is of near chemisorption character and thus challenging, both for the alchemical transformation and also for the fixed point-charge electrostatics. Our scheme passes through a state characterized by a well-behaved physisorption potential for the Pt(111)/H2O interaction to converge the free energy difference. The workflow is implemented in the freely available SolvHybrid package. We first assess the adsorption of a water molecule at the Pt/water interface, which turns out to be a stringent test. The intrinsic error of our quantum-mechanical/molecular mechanics (QM/MM) hybrid scheme is limited to 6 kcal mol-1 through the introduction of a correction term to attenuate the electrostatic interaction between near-chemisorbed water molecules and the underlying Pt atoms. Next, we show that the MMsolv solvation free energy of Pt (-0.46 J m-2) is in good agreement with the experimental estimate (-0.32 J m-2). Furthermore, we show that the entropy contribution at room temperature is roughly of equal magnitude as the free energy but with an opposite sign. Finally, we compute the adsorption energy of benzene and phenol at the Pt(111)/water interface, one of the rare systems for which experimental data are available. In qualitative agreement with the experiment, but in stark contrast with a standard implicit solvent model, the adsorption of these aromatic molecules is strongly reduced (i.e., less exothermic by ∼30 and 40 kcal mol-1 for our QM/MM hybrid scheme and experiment, respectively, but ∼0 with the implicit solvent) at the solid/liquid interface compared to the solid/gas interface. This reduction occurs mainly because of the competition between the organic adsorbate and the solvent for adsorption on the metallic surface. The semiquantitative agreement with experimental estimates for the adsorption energy of aromatic molecules thus validates the soundness of our hybrid QM/MM scheme.

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.

Ten Facets, One Force Field: The GAL19 Force Field for Water-Noble Metal Interfaces.

Understanding the structure of the water/metal interfaces plays an important role in many areas ranging from surface chemistry to environmental processes. The size, required phase-space sampling, and the slow diffusion of molecules at the water/metal interfaces motivate the development of accurate force fields. We develop and parametrize GAL19, a novel force field, to describe the interaction of water with two facets (111 and 100) of five metals (Pt, Pd, Au, Ag, Cu). To increase transferability compared to its predecessor GAL17, the water-metal interaction is described as a sum of pairwise terms. The interaction energy has three contributions: (i) physisorption is described via a Tang and Toennies potential, (ii) chemisorption and surface corrugation rely on an attractive Gaussian term, and (iii) the angular dependence is explicitly included as a truncated Fourier series. Thirteen parameters are used for each metal surface and were fitted on 250 water adsorption energies computed at the PBE+dDsC level. The performance of GAL19 was evaluated on a set of more than 600 DFT adsorption energies for each surface, leading to an average root-mean-square deviation of only 1 kcal/mol, correctly reproducing the adsorption trends: strong on Pt and Pd but weaker on Ag, Au, and Cu. This force field was then used to simulate the water/metal interface for all ten surfaces for 1 ns. Structural analyses reveal similar tendencies for all surfaces: a first, dense water layer that is mostly adsorbed on the metal top sites and a second layer up to around 6 Å, which is less structured. On Pt and Pd, the first layer is strongly organized with water lying flat on the surface. The pairwise additive functional form allows one to simulate the water adsorption on alloys, which is demonstrated at the example of Ag/Cu and Au/Pt alloys. The water/Ag-Cu interface is predicted to be disordered with water mostly adsorbed on Cu which should exacerbate the Ag reactivity. On the contrary, incorporating Pt into Au materials leads to a structuring of the water interface. Our promising results make GAL19 an ideal candidate to get representative sampling of complex metal/water interfaces as a first step toward accurate estimation of free energies of reactions in solution at the metal interface.