Beyond Scaling Relations for the Description of Catalytic Materials

Abstract

Computational screening for new and improved catalyst materials relies on accurate and lowcost predictions of key parameters such as adsorption energies. Here, we use recently developed compressed sensing methods to identify descriptors whose predictive power extends over a wide range of adsorbates, multi-metallic transition metal surfaces and facets. The descriptors are expressed as non-linear functions of intrinsic properties of the clean catalyst surface, e.g. coordination numbers, d-band moments and density of states at the Fermi level. From a single density-functional theory calculation of these properties, we predict adsorption energies at all potential surface sites, and thereby also the most stable geometry. Compared to previous approaches such as scaling relations, we find our approach to be both more general and more accurate for the prediction of adsorption energies on alloys with mixed-metal surfaces, already when based on training data including only pure metals. This accuracy can be systematically improved by adding also alloy adsorption energies to the training data. arXiv:1902.07495v1 [cond-mat.mtrl-sci]

Strategies to Break the Scaling Relation toward Enhanced Oxygen Electrocatalysis

“…112,139 Other factors should also be reconsidered to optimize the adsorption energies of all intermediates as close as possible to the ideal situation. 140 Recently, Govindaragan et al reported a new descriptor, that is, the electrochemical-step symmetry index (ESSI), and the decrease of ESSI could effectively reduce the calculated overpotentials. 102 With that in mind, more novel strategies to break adsorption-energy scaling relations and achieve high activity should be explored, with inspiration from other catalytic fields.…”

Determining the adsorption energies of small molecules with the intrinsic properties of adsorbates and substrates

“…Another breakthrough is the discovery of linear relations of adsorption energies, by Nørskov et al 11 , for atoms and their partially hydrogenated species on TM surfaces. These linear scaling relationships (LSRs), which have been successfully extended to intermetallics 29,30 , nanoparticles (NPs) 31 , TM compounds 32 , etc. 33 , not only allow the ascertainment of the trend of catalytic activity but also impose a thermodynamic limitation on some catalytic reactions [12][13][14][15][16][17][18][19][20][21]34 .…”

Progress in Computational and Machine‐Learning Methods for Heterogeneous Small‐Molecule Activation

“…For a couple of examples, Andersen et al implemented the SISSO feature selection algorithm to correlate various known surface descriptors with the binding energy. [129] Though the model is not necessarily the most accurate, the SISSO algorithm explores various functional forms with descriptors, thus the explicit relationship between the descriptors in the interest of computing binding energy could be found. In addition, García-Muelas et al performed PCAbased algorithms to binding energies of C 1 -C 2 species on 12 metals and found that thermodynamics of only a handful species is necessary to predict the binding energy of the rest of the molecules.…”