Improving the accuracy of solid-state nuclear magnetic resonance chemical shift prediction with a simple molecular correction.

A fast, straightforward method for computing NMR chemical shieldings of crystalline solids is proposed. The method combines the advantages of both conventional approaches: periodic calculations using plane-wave basis sets and molecular computational approaches. The periodic calculations capture the periodic nature of crystalline solids, but the computational level of the electronic structure calculation is limited to general-gradient-approximation (GGA) density functionals. It is demonstrated that a correction to the GGA result calculated on an isolated molecule at a higher level of theory significantly improves the correlations between experimental and calculated chemical shifts while adding almost no additional computational cost. Corrections calculated with a hybrid density functional improved the accuracy of C-13, N-15 and O-17 chemical shift predictions significantly and allowed identifying errors in previously published experimental data. Applications of the approach to crystalline isocytosine, methacrylamide, and testosterone are presented.