Abstract
The design of transition metal complexes has drawn much attention over the years because of their important applications as metallodrugs and functional materials. In this work, we present an extension of our recently reported approach, LigandDiff. The new model, which we call multi-LigandDiff, is more flexible and greatly outperforms its predecessor. This scaffold-based diffusion model allows de novo ligand design either with existing ligands or without any ligand. Moreover, it allows users to predefine the denticity of the generated ligand. Our results indicate that multi-LigandDiff can generate well-defined ligands and has great transferability with regard to transition metals and coordination geometries. In terms of its application, multi-LigandDiff successfully designs 338 Fe(II) SCO complexes from only 47 experimentally validated SCO complexes. And these generated complexes are configurationally diverse and experimentally reasonable. Overall, the results show that multi-LigandDiff is an ideal tool to design novel transition metal complexes from scratch.