The use of centroids of localized orbitals as a method to derive oxidation states (OS) from first-principles is critically analyzed. We explore the performance of the closest-atom distance criterion to assign electrons for a number of challenging systems, including high-valent transition metal compounds, π-adducts, and transition metal (TM) carbenes. Here, we also introduce a mixed approach that combines the position of the centroids with Bader’s atomic basins as an alternative criterion for electron assignment. The closest-atom criterion performs reasonably well for the challenging systems, but wrongly considers O-H and N-H bonds as hydrides. The new criterion fixes this problem, but underperforms in the case of TM carbenes. Moreover, the OS assignment in dubious cases exhibit undesirable dependence on the particular choice for orbital localization.

Recent advances in serial femtosecond crystallography (SFX) of photosystem II (PSII), enabled by X-ray free electron lasers (XFEL), provided the first geometric models of distinct intermediates in the catalytic S-state cycle of the oxygen-evolving complex (OEC). These models are obtained by flash-advancing the OEC from the dark-stable state (S1) to more oxidized intermediates (S2 and S3), eventually cycling back to the most reduced S0. However, the interpretation of these models is controversial because geometric parameters within the Mn4CaO5 cluster of the OEC do not exactly match those expected from coordination chemistry for the spectroscopically verified manganese oxidation states of the distinct S-state intermediates. Here we focus on the first catalytic transition, S1 → S2, which represents a one-electron oxidation of the OEC. Combining geometric and electronic structure criteria, including a novel effective oxidation state approach, we analyze existing 1-flash (1F) SFX-XFEL crystallographic models that should depict the S2 state of the OEC. We show that the 1F/S2 equivalence is not obvious, because the Mn oxidation states and total unpaired electron counts encoded in these models are not fully consistent with those of a pure S2 state and with the nature of the S1 → S2 transition. Furthermore, the oxidation state definition in two-flashed (2F) structural models is practically impossible to elucidate. Our results advise caution in the extraction of electronic structure information solely from the literal interpretation of crystallographic models and call for re-evaluation of structural and mechanistic interpretations that presume exact correspondence of such models to specific catalytic intermediates of the OEC.

The prediction of isomer shifts in 57 Fe Mossbauer spectra is typically achieved by building calibration lines using the values of the density at the nuclear position. Using Slater-type orbital basis or large and specific Gaussian-type orbital basis has been thus far mandatory to achieve accurate predictions with density functional theory methods. In this work, we show that replacing the value of the density at the nucleus by the density integrated in a sphere of radius 0.06 au centered on the Fe nuclei yields excellent calibration lines ( r 2 = 0.976) with a high predictive power ( q 2 = 0.975, MAE = 0.055 mm·s –1 ) while using the conventional def2-TZVP basis set and X-ray geometrical parameters. Our data set comprises 69 57 Fe-containing compounds and 103 signals. We also find B3LYP performing significantly better than the PW91 functional.