(212b) Feature Engineering of Machine-Learning Chemisorption Models for Bifunctional Electrocatalyst Design

Transition metal alloys have shown great potential for catalyzing many chemical and electrochemical reactions. Attributed to the versatility of transition metals on the periodic table, the catalyst surface reactivity can be manipulated by varying its geometric characteristics, metal ligands, and extrinsic factors in the vicinity of active sites [1]. However, it is very time-consuming and costly to search for highly optimized alloy composition and geometry by high-throughput experiments and/or quantum-chemical calculations.

In this talk, we will introduce a new computational catalyst design framework, which integrates machine-learning algorithms with the descriptor-based design approach for rapid screening of transition-metal catalysts [2-3]. By engineering numerical representation of surface metal atoms using easily accessible features such as the local electronegativity and the effective coordination number that are dependent on the surroundings of an adsorption site, together with the intrinsic properties of active metal atoms including the electronegativity, ionic potential, and electron affinity, the machine-learning model optimized with ~500 ab initio adsorption energies on bimetallic alloys can capture complex, non-linear adsorbate/substrate interactions with the root mean squared errors (RMSE) ~0.12 eV. To validate the model, we applied the model to search for high-performance {111}-terminated bifunctional catalysts for electrocatalytic oxidation of methanol. The *CO and *OH adsorption energies represent important efficiency metrics to describe the electrocatalytic reactivity. We show that our machine-learning-augmented model exhibits outstanding prediction power in screening for the bifunctional, multimetallic surfaces with desired adsorption properties for both *CO and *OH on different adsorption sites. Compared with the traditional high-throughput computational and experimental trial-and-error approach, the machine-learning chemisorption models have great potential in accelerating the discovery of interesting catalytic materials.

[1] A. Vojvodic and J. K. Nørskov, â??New design paradigm for heterogeneous catalysts,â? Natl. Sci. Rev., p. nwv023, Apr. 2015.

[2] X. Ma, Z. Li, L. E. K. Achenie, and H. Xin, â??Machine-Learning-Augmented Chemisorption Model for CO2 Electroreduction Catalyst Screening,â? J. Phys. Chem. Lett., vol. 6, no. 18, pp. 3528â??3533, Sep. 2015.

[3] Z. Li, X. Ma, and H. Xin, â??Feature Engineering of Machine-Learning Chemisorption Models for Catalyst Design," Catalysis Today, 2015 (Accepted).