Combined Neural Network Potential and Density Functional Theory Study of TiAl₂O₅ Surface Morphology and Oxygen Reduction Reaction Overpotentials

Titanium alloys, such as Ti-6Al-4V, are used in a variety of applications due to their high strength-to-weight ratio and corrosion resistance. Despite resisting corrosion, Ti-6Al-4V facilitates the galvanic corrosion of less noble metals when they are in contact. Atmospheric galvanic corrosion is limited by the rate of cathodic reduction reactions, such as the oxygen reduction reaction (ORR). To better understand the factors that make a material a poor ORR catalyst in these conditions, we use an <i>in silico</i> procedure to predict how the ORR overpotentials of TiAl₂O₅ (a possible oxide present on the Ti-6Al-4V surface) surface sites are impacted by surface morphology and the presence of metal dopants. We trained Behler–Parrinello neural networks to reproduce the Kohn–Sham density functional theory energy and forces of TiAl₂O₅ structures and used these neural networks to create a variety of defective and amorphous surface models. We calculated and compared the ORR overpotentials of these TiAl₂O₅ surfaces with density functional theory. Our calculations show that ORR activity can be modulated by the presence of metal dopants in the oxide. Some dopants are consistently poor ORR catalyst sites (Si⁴⁺, Ga³⁺, and Sn⁴⁺), while others depend on the surface and the magnitude of solvation (Co²⁺, Nb⁵⁺, and Mn²⁺). This modulation may reduce the ORR activity on the oxide surface and therefore improve the corrosion resistance of a material in atmospheric conditions.