Unlocking the efficiency of nonaqueous Li–air batteries through the synergistic effect of dual metal site catalysts: an interpretable machine learning approach.

The recent growing attention towards non-aqueous Li–air batteries (LABs) stems from their high energy density, positioning them as a key solution to the surging demand for electrical energy driven by portable electronics. Despite the potential of LABs, sluggish cathode kinetics and large overpotentials, coupled with storage challenges from insoluble discharge products, impede their commercialization. Herein, we have proposed an interpretable machine learning (ML) approach to screen the synergistic effect of transition metal pairs in dual metal site catalysts (DMSCs) as a new frontier cathode for efficient LABs. The extreme gradient boosting regression-based ML algorithm systematically explored different transition metal combinations and predicted efficient cathode electrocatalysts for LABs. Our findings revealed highly active DMSCs for LAB reactions, which have the potential to exhibit promising ORR/OER activity, surpassing that of the novel platinum cathode. The outcome of the ML algorithm unveils the significant role of metal d-electrons in predicting the catalytic performance. Additionally, the determined Coulomb interaction energy for active DMSCs provides valuable insights, which hold promise for modeling efficient LABs and advancing their practical applications in the future. [ABSTRACT FROM AUTHOR]