Self-Validated Machine Learning Study of Graphdiyne-Based Dual Atomic Catalyst

摘要

Although the atomic catalyst has attracted intensive attention in the past few years, the current progress of this field is still limited to a single atomic catalyst (SAC). With very few successful cases of dual atomic catalysts (DACs), the most challenging part of experimental synthesis still lies in two main directions: the thermodynamic stability of the synthesis and the optimal combination of metals. To address such challenges, comprehensive theoretical investigations on graphdiyne (GDY)-based DAC are proposed by considering both, the formation stability and the d-band center modifications. Unexpectedly, it is proven that the introduction of selected lanthanide metals to the transition metals contributes to the optimized stability and electroactivity. With further verification by machine learning, the potential f-d orbital coupling is unraveled as the pivotal factor in modulating the d-band center with enhanced stability by less orbital repulsive forces. These findings supply the delicate explanations of the atomic interactions and screen out the most promising DAC to surpass the limitations of conventional trial and error synthesis. This work has supplied an insightful understanding of DAC, which opens up a brand new direction to advance the research in atomic catalysts for broad applications.