Origin of the DifferentTrends of Experimental Activityon Perovskite Catalysts between OER and ORR

Understanding the difference between the activities of catalysts in OER and ORR is crucial for designing a bifunctional catalyst for rechargeable fuel cells and metal–air batteries, which so far remains elusive. In this work, a wide range of 3d transition metal-based perovskite oxide catalysts were considered to uncover the difference in the trends between OER and ORR. By performing systematic symbolic regression on experimental data, we confirmed the previous descriptor (<i>d_B</i>, <i>n_B</i>) for OER activity and identified a new descriptor (<i>−d_B</i>, <i>–</i>|<i>e</i>_g – 0.8|) for ORR activity, where <i>d_B, n_B</i>, and <i>e</i>_g are the number of d-electrons, oxidation state, and <i>e</i>_g orbital occupancy of the transition-metal cation, respectively. To understand the descriptors, first-principles calculations based on multiple reaction mechanisms were performed. Results show that the dependence of activity on the B-site metal species of the ABO₃ perovskites exhibits a volcano-shaped relation in both OER and ORR. We found that the experimental activity descriptors can be explained by the computed results from multiple mechanisms (4e^– transfer). The difference between the experimental OER and ORR activity descriptors originates from that the volcano top of the OER performance against the 3d transition metal is located near the end of this period and that of the ORR is in the middle. For ORR, further calculations show that the 2e^– pathway was only important on Ni and Cu oxides on which the binding of *OOH is weak. These descriptors and insights can be helpful in guiding the design of perovskite catalysts for OER and ORR.