Prediction of energies of spin states for first row transition metal complexes: DFT vs. ligand field theory

DFT benchmarking on 127 base metal complexes revealed unreliable spin state predictions by B3LYP, PBE0, TPSSh and M06-L functionals, especially when energy gaps between the lowest energy spin states were small. PBE0/def2-TZVP offered the closest optimised geometries to crystal structures but failed to reliably predict spin-crossover (SCO) temperatures (TSCO) for 32 Fe(ii) complexes. Kestrel, a computational package using ligand field theory, achieved better TSCO predictions and demonstrated high sensitivity to ligand–metal angular arrangement. Application of Kestrel to 2 DFT-detected spin-crossover catalytic cycles confirmed agreement on ground-state spins but not SCO in key steps. These findings underscore that DFT-based TSCO predictions based on single complexes should not guide SCO assessments particularly in the solid state.