Journal of Chemical Information and Modeling, Год журнала: 2025, Номер unknown
Опубликована: Июнь 3, 2025
Chemical space exploration motivates the development of data-driven models that bypass explicit computation or experiment. Cost-efficient strategies include concept additivity via many-body expansion treats a molecule as sum its parts. In context transition metal chemistry, ligand-wise has been established powerful tool to infer properties heteroleptic complexes (TMCs) from homoleptic TMCs excellent accuracy, including spin-splitting, orbital energies, and reaction energies. Nevertheless, this framework is incompatible with anionic ligands because stable homoleptic, thus polyanionic, parent complex cannot be simulated readily. Here, I explore alternative approaches, first identifying limits stability when successive Cl- anions are added in representative formed neutral H2O CO ligands. establish expected linear relationships preserved, albeit not strongly propose data-efficient interpolation extrapolation schemes for achieve root-mean-square errors low 0.15-0.36 eV on HOMO/LUMO levels gaps ionization potentials electron affinities 4 kcal/mol adiabatic spin-splitting energies Fe(II) complexes. show approach generalizes well across 14 other 3d, 4d, 5d metals. Finally, extend predict thousands binary ternary Zn(II) involving single ligand up two unique by leveraging handful calculations. how interpolated can used valid discover novel properties.
Язык: Английский
