In transition metal compounds with spin states close in energy, the magnitude and sign of the energy splitting calculated with density functional theory depends strongly on the functional used. Therefore we must turn to additional criteria to assess the level of accuracy and reliability of predictions based on this level of theory. We report optimized geometries, total energies, and Mössbauer quadrupole splitting values for low-spin and high-spin, ferric and ferrous model hemes using a variety of gradient-corrected and hybrid functionals. In one model, the iron-porphyrin is axially ligated by two strong-field imidazole ligands [FeP(Im)2] and has a low-spin ground state. In the other model complex the axial ligands are two weak-field, water molecules [FeP(H2O)2], and have a high-spin ground state. Among all the functionals used (UHF, B3LYP, B3LYP*, BLYP, half-and-half, LSDA), the B3LYP hybrid functional most consistently reproduced the experimental geometry, Mössbauer, and spin state data for the two model hemes. Simply gradient-corrected functionals exhibit strong biases towards low spin states, while Hartree-Fock favours strongly high spin states. These findings suggest that for systems with similar characteristics of several accessible electronic spin configurations, it is imperative to include properties other than just the energy in the assessment of the DFT predictions.
ASJC Scopus subject areas
- Molecular Biology
- Condensed Matter Physics
- Physical and Theoretical Chemistry