A Study of Platinum Metal Surface as a Component of Fuel Cells

Abstract

Adsorption energies of small molecules carbon monoxide and hydrogen are studied using density functional theory and the neural network method of generating higher dimensional potential energy surfaces. Various lower index surfaces of platinum metal are considered, and the adsorption energies appear to increase according to: P t(211) > P t(100) > P t(110) > P t(111) with the adsorption of both molecules. The large adsorption energies mean only energetic interactions with the ions from air such as oxygen ion can detach the adsorbates from the surface thereby forming water and carbon monoxide as by-products. Exploring possible adsorption sites gives insight on better ways in which charge exchanges are maximized while at the same time forming of the by-products becomes efficient. The adsorption energies range between 0.85 eV to 2.08 eV with the adsorption of CO while the values are between 1.30 eV to 1.9 eV with the adsorption of hydrogen. Charge transfers give some insight into the study of electrification process in the system. These values are computed to be about 0.37e with the adsorption of hydrogen and upto 0.24e with the adsorption of carbon monoxide. Training of the system potential energies using the neural network method shows promising opportunity to study further complex problems in such systems.

Keywords: Neural networks, machine learning, density functional theory calculations, hydrogen, carbon monoxide, platinum, fuel cells