We present a machine-learning-augmented chemisorption model that enables fast and accurate prediction of the surface reactivity of metal alloys within a broad chemical space. Specifically, we show that artificial neural networks, a family of biologically inspired learning algorithms, trained with a set of ab initio adsorption energies and electronic fingerprints of idealized bimetallic surfaces, can capture complex, nonlinear interactions of adsorbates (e.g., *CO) on multimetallics with ∼0.1 eV error, outperforming the two-level interaction model in prediction. By leveraging scaling relations between adsorption energies of similar adsorbates, we illustrate that this integrated approach greatly facilitates high-throughput catalyst screening and, as a specific case, suggests promising {100}-terminated multimetallic alloys with improved efficiency and selectivity for CO2 electrochemical reduction to C2 species. Statistical analysis of the network response to perturbations of input features underpins our fundamental understanding of chemical bonding on metal surfaces.

No abstract

We present the orbitalwise coordination number CN^{α} (α=s or d) as a reactivity descriptor for metal nanocatalysts. With the noble metal Au (5d^{10}6s^{1}) as a specific case, the CN^{s} computed using the two-center s-electron hopping integrals to neighboring atoms provides an accurate and robust description of the trends in CO and O adsorption energies on extended surfaces terminated with different facets and nanoparticles of varying size and shape, outperforming existing bond-counting methods. Importantly, the CN^{s} has a solid physiochemical basis via a direct connection to the moment characteristics of the projected density of states onto the s orbital of a Au adsorption site. Furthermore, the CN^{s} shows promise as a viable descriptor for predicting adsorption properties of Au alloy nanoparticles with size-dependent lattice strains and coinage metal ligands.

No abstract

Synthetically tuning the surface properties of many oxide catalysts to optimize their catalytic activity has been appreciably challenging given their complex crystal structure.Nickelate oxides (e.g., La2NiO4+δ) are among complex, layered oxides with great potential toward efficiently catalyzing chemical/electrochemical reactions involving oxygen (oxygen reduction, ammonia oxidation). Our theoretical calculations show that the surface structure of La2NiO4+δ plays a critical role in its activity, with (001)-Ni oxide terminated surface being the most active. This is demonstrated through the effect on the energetics associated with surface oxygen exchange -key process in reactions involving oxygen on these oxides. Using a reverse microemulsion method, we have synthesized La2NiO4+δ nanorod-structured catalysts highly populated by (001)-Ni oxide terminated surfaces. We show that these nanostructures exhibit superior catalytic activity toward oxygen exchange/reduction as compared to the traditional catalysts, while maintaining stability under reaction conditions. The findings reported here pave the way for engineering complex metal oxides with optimal activity.