Refining crystal structures with experimental 13C NMR shift tensors and lattice-including electronic structure methods

Large differences are found in the quality of crystal structures obtained from different diffraction methods. The most accurate studies identify all atomic sites, including hydrogens, while others lack even the resolution needed to locate individual atoms. The gauge including projector augmented wave method (GIPAW) provides a technique to further refine any of these structures under lattice constraints. Here, the sensitivity of solid-state NMR C-13 shift tensor principal value data to GIPAW refinement is investigated. The refinement is shown to improve X-ray powder, X-ray single crystal and even neutron single crystal diffraction data. Convergence to a single structure is observed in most cases. Surprisingly, the final refined structures usually diverge from the original neutron diffraction coordinates - data typically viewed as the most accurate. To ensure that the structural changes represent improvements, three metrics are monitored comprising fit to C-13 shift tensors, forces upon the atoms and changes in atomic positions relative to a reference structure. In all cases these parameters improve upon refinement suggesting that GIPAW creates structures surpassing the accuracy of single crystal neutron diffraction data. However, the influence of thermal motions remains unknown. Improvements are seen most strongly in forces and NMR fits and least in atom positions. This study evaluates reasonably accurate model structures to quantify improvements. However, structures obtained from lower resolution methods (e. g. electron diffraction) will benefit most from GIPAW refinement. In such structures the refinement has the potential to convert structures with questionable atom positions into coordinates rivaling neutron diffraction single crystal data.