Assessment of the Performance of Optimally Tuned Range-Separated Hybrid Functionals for Nuclear Magnetic Shielding Calculations

The performance of optimally tuned range-separated hybrid (OT-RSH) functional calculations in predicting accurate isotropic nuclear magnetic shielding (σ) and chemical shift values is examined. To that end, the results of OT-RSH and other approximate density functional theory calculations are assessed against recently published benchmark CCSD(T) calculations for a test set consisting of several molecules and bond types. It is found that for atoms in single bonds with a large paramagnetic contribution to σ, OT-RSH offers a significant improvement in prediction of shielding constants over popular semi-local and hybrid density functionals, yielding non-empirical results that are as accurate as those of semi-empirical density functionals often used for prediction of shielding constants. This success is attributed to the improved fundamental gap prediction of the OT-RSH approach. For atoms in multiple bonds, however, larger errors often persist. By comparing OT-RSH and recently reported double-hybrid functional results, the remaining difficulties are traced to significant non-local correlation.