Towards the best model for H atoms in experimental charge-density refinement.

The consequences of different treatments of H atoms in experimental charge-density studies are discussed. Geometric and topological parameters obtained after applying four different H-atom models in multipolar refinement on high-resolution X-ray data only were compared with the results obtained for a reference joint high-resolution X-ray/neutron refinement. The geometry and the topological critical point and integrated parameters closest to the reference values were obtained after a mixed refinement (high-order refinement of heavy atoms, low-angle refinement of H atoms and elongation of the X-H distance to the average neutron bond lengths) supplemented by an estimation of the anisotropic thermal motions of H atoms using the SHADE program. Such a procedure works very well even for strong hydrogen bonds. The worst fit to the reference results for both critical point and integrated parameters was obtained when only the standardization to the average neutron X-H distances was applied. The non-H-atom parameters are also systematically influenced by the H-atom modeling. In order to compare topological and integrated properties calculated for H and non-H atoms in multipolar refinement when there are no neutron data, the same treatment of H atoms (ideally the mixed refinement + estimated anisotropic atomic displacement parameters for H atoms) should be applied.