GIPAW Pseudopotentials of d Elements for Solid-State NMR

Computational methods are increasingly used to support interpreting, assigning and predicting the solid-state nuclear resonance magnetic spectra of materials. Currently, density functional theory is seen to achieve a good balance between efficiency and accuracy in solid-state chemistry. To be specific, density functional theory allows the assignment of signals in nuclear resonance magnetic spectra to specific sites and can help identify overlapped or missing signals from experimental nuclear resonance magnetic spectra. To avoid the difficulties correlated to all-electron calculations, a gauge including the projected augmented wave method was introduced to calculate nuclear resonance magnetic parameters with great success in organic crystals in the last decades. Thus, we developed a gauge including projected augmented pseudopotentials of 21 d elements and tested them on, respectively, oxides or nitrides (semiconductors), calculating chemical shift and quadrupolar coupling constant. This work can be considered the first step to improving the ab initio prediction of nuclear magnetic resonance parameters, and leaves open the possibility for inorganic compounds to constitute an alternative standard compound, with respect to tetramethylsilane, to calculate the chemical shift. Furthermore, this work represents the possibility to obtain results from first-principles calculations, to train a machine-learning model to solve or refine structures using predicted nuclear magnetic resonance spectra.

Automated summary

Computational methods play an increasingly important role in interpreting, assigning and predicting solid-state nuclear resonance magnetic spectra. Density functional theory is considered to achieve a good balance between efficiency and accuracy. By introducing gauge and pseudopotentials, successful calculations of nuclear resonance magnetic parameters have been achieved, providing the possibility to improve the ab initio prediction of nuclear magnetic resonance parameters and train machine learning models to solve or refine structures.