Determinants of the FeXO (X = C, N, O) vibrational frequencies in heme adducts from experiment and density functional theory

Vibrational spectra of CO, NO, and O-2 adducts of heme proteins contain information on interactions of the heme and its bound ligands with the surrounding protein matrix that may help in elucidating the mechanism of small-molecule activation. Whereas the heme-CO system is well studied and a framework exists for the interpretation of such interactions, heme-NO and -O-2 complexes have not been systematically investigated. Here we examine resonance Raman spectra of all three classes of adducts, combining literature values with new data for (FeNO)-N-II porphyrins having both electron-donating and electron-withdrawing substituents. Negative linear correlations are observed for all three adducts between their Fe-XO and X-O stretching frequencies. The slopes of these correlation lines are -0.4 for five-coordinate FeCO and FeNO porphyrins and -0.8 for five-coordinate FeO2 adducts. Thus, Fe-NO and Fe-O-2 bonds are equally or even more sensitive than FeCO bonds to electronic influences that affect metal-to-ligand pi back-bonding. However, the responses of the NO and O-2 adducts to trans ligand binding are very different from those for CO complexes. Ligands trans to CO displace the plot to steeper slopes and lower Fe-CO frequencies, reflecting competition of the ligand lone pairs for the a acceptor orbital, d(z)2. However, no displacement of the line is observed for six-coordinate FeNO and FeO2 adducts, but only a shift to higher positions on the line, indicating greater back-bonding. We infer that trans ligand competition for the d(z)2 orbital is not as effective for NO and O-2 as for CO, reflecting the lower energy of the N and O orbitals relative to that of the C orbitals. These results are discussed with the aid of a simple bonding model involving FeXO valence isomers. To examine this model, we applied density functional theory to five- and six-coordinate XO adducts of Fe(II) porphine. Geometries were in good agreement with experiment, as were vibrational frequencies for CO adducts. However, DFT overestimated the Fe-NO bond extension on binding a trans ligand and predicted a decrease in the Fe-NO stretching frequency, whereas an increase is observed. The predicted frequency change was likewise in the wrong direction for Fe-O stretching in six- vs five-coordinate FeO2 adducts. The results suggest that DFT captures the essential features of back-bonding, but not of the a competition with the trans ligand, in the cases of NO and O-2.