New developments in the calculation of metalloporphyrin Raman spectra via density functional theory

Abstract

The recent development of density functional theory (DFT) makes it possible to calculate accurately metalloporphyrin structures and potential surfaces. This is illustrated for nickel porphine, the vibrations of which are reliably assigned from extensive spectroscopic studies. With a minimal set of scaling factors, the DFT force field reproduces the experimental wavenumbers to higher accuracy than the best available empirical force field. Moreover, the calculated intensities are in good accord with experiment, including the surprisingly large off-resonant Raman intensities of non-totally symmetric (B1g) modes. DFT also predicts a slight ruffling distortion of the porphyrin, and accurately reproduces the IR intensity of a distortion-induced out-of-plane mode. In the case of iron porphine, DFT correctly predicts an intermediate-spin (3A2g) ground state, with short Fe—N bonds, and a high-spin (5A1g) excited state with a planar geometry but an expanded porphyrin core. When the Fe is displaced from the plane, the potential rises faster for the 3A2g than the 5A1g state, which becomes the ground state beyond a 0.4 Å displacement. The doming mode is predicted to be at 71 cm-1, close to the 75 cm-1 wavenumber determined from coherent reaction dynamics in myoglobin. The vibrational wavenumbers of CO bound to heme are correctly calculated, and the potential for distorting the CO away from the heme normal is found to be surprisingly soft. Also, the transition dipole for the CO stretching mode is calculated to lag significantly behind the CO bond vector, thereby resolving an apparent discrepancy between crystallography and polarized IR spectroscopy with regard to the CO geometry in its adduct with myoglobin. Finally, the DFT force field was used successfully in conjunction with INDO calculations of the excited states, to reproduce resonance Raman intensities for NiP, both for Soret and Q-band resonances. These results give promise for developing a quantitative modeling capability for heme protein vibrational spectra. Copyright © 1998 John Wiley & Sons, Ltd.