石墨炔原子催化剂的崭新道路:基于自验证机器学习方法的筛选策略

Although atomic catalysts(ACs) have attracted intensive attention in recent years, the current progress of this area is limited by the use of noble metal as well as single atomic catalysts(SACs). Here, we summarize the recent works in screening highly-efficient graphdiyne-ACs(GDY-ACs) with the utilization of density functional theory(DFT) calculations and machine learning(ML). Our studies showed that the Pd, Co, Pt and Hg could form stable zero-valence transition metal-GDY(TM-GDY), whereas the lanthanide-TM DAC(Ln-TM DAC) systems were also demonstrated as the promising electrocatalyst candidates because of their long-range site-to-site f-d orbital interactions. The further analysis revealed that the combination of main group elements with TM and Ln metals can achieve high stable GDY-DAC and preserve the high electroactivity due to the long-range p-orbital coupling, while the role of the s- and p-orbitals was studied via ML algorithm. In addition, the DFT calculation and ML techniques also showed great potential in screening possible GDY-based ACs with excellent hydrogen evolution reaction(HER) performances, and the potential of rare-earth-based GDY-ACs for HER has been predicted for the first time. This review has supplied an advanced strategy for future exploration of atomic catalyst.