Optimizing the Catalytic Activity of Pd-Based Multinary Alloys toward Oxygen Reduction Reaction

The development of cost-effective catalysts for oxygen reduction reaction (ORR) has an enormous impact on fuel cells toward highly efficient low emission energy conversion. Recently, a Pt-free multinary PdAuAgTi alloy was discovered with excellent ORR activity and low overpotential close to that of Pt. To rationalize the experimental results, a model based on first-principles methods accelerated with deep learning is developed to rapidly compute and with high fidelity the *OH adsorption energy on the alloyed surface. The ensemble-average *OH adsorption energy is shown to explain the experimentally reported OER activities of PdAuAgTi and further is utilized to provide precise maps of the catalytic activity in the total composition space. Notably, the ORR activity of PdAuAgTi is found to be optimum in a narrow region of the composition space with 8-12 at. % Ti, which agrees with the experimental finding for enhanced ORR activity at 11-13 at. % Ti. In addition, replacing Au and Ag with the more cost-effective elements Cu and Zn is also shown to yield optimum catalysts for ORR. The current study shows that first-principles methods in conjunction with machine learning approaches are an effective tool for discovering multinary alloy systems for catalytic applications.

Automated summary

The discovery of a low-cost PdAuAgTi alloy as a catalyst for oxygen reduction reaction (ORR) with excellent activity and low overpotential is reported in this study. A model based on first-principles methods accelerated with deep learning is developed to calculate the adsorption energy on the alloy surface, which provides precise maps of the catalytic activity distribution. The study also reveals that the ORR activity of the PdAuAgTi alloy is optimum in a specific composition range.