Benchmark accuracy of predicted NMR observables for quadrupolar ¹⁴N and ¹⁷O nuclei in molecular crystals

Abstract

Nuclear quadrupole resonances for 14 N and 17 O nuclei are exquisitely sensitive to interactions with surrounding atoms. As a result, nitrogen and oxygen solidstate nuclear magnetic resonance (ssNMR) provides a powerful tool for investigating structure and dynamics in complex systems. First-principles calculations are increasingly used to facilitate spectral assignment and to evaluate and adjust crystal structures. Recent work combining the strengths of planewave density functional theory (DFT) calculations with a single molecule correction obtained using a higher level of theory has proven successful in improving the accuracy of predicted chemical shielding (CS) tensors and 17 O quadrupolar coupling constants (C q ). Here we extend this work by examining the accuracy of predicted 14 N and 17 O electric field gradient (EFG) tensor components obtained using alternative planewave-corrections involving cluster and two-body fragment-based calculations. We benchmark the accuracy of CS and EFG tensor predictions for both nitrogen and oxygen using planewave, two-body fragment, and enhanced planewave-corrected techniques. Combining planewave and two-body fragment calculations reduces the error in predicted 17 O C q values by 35% relative to traditional planewave calculations. These enhanced planewave-correction methods improve the accuracy of 17 O and 14 N EFG tensor components by 15% relative to planewave DFT but yield minimal improvement relative to a simple molecular correction. However, in structural environments involving either high symmetry or strong intermolecular interactions, enhanced planewave-corrected methods provide a distinct advantage.

Do Models beyond Hybrid Density Functionals Increase the Agreement with Experiment for Predicted NMR Chemical Shifts or Electric Field Gradient Tensors in Organic Solids?

“…Similar behavior has been found previously for isotropic 17 O chemical shifts, and it has been attributed to the high sensitivity of 17 O magnetic properties to the electrostatic environment. 4,26 As previously observed for 17 O quadrupolar coupling constants, 7,26 a monomer correction performed at the PBE0 level modestly improves the accuracy of 17 O EFG tensor components. However, the 15% rms error reduction obtained for the full set of tensor components by applying the PBE0 monomer correction is about half the improvement seen previously for the 17 O quadrupolar coupling constants alone.…”

“…32 Monomer-correction methods for GIPAW which employ hybrid density functionals have been shown to reduce the errors in predicted 17 O quadrupolar coupling constants by ∼30% relative to uncorrected GIPAW calculations. 7,26 Here we extend these earlier studies by examining the accuracy of predicted 17 O and 14 N EFG tensor components obtained from monomer corrections performed using MP2 or double-hybrid density functionals. The complete list of EFG experimental tensor values and reported uncertainties are provided in the Supporting Information Section 2.3.…”

“…26,32 Benchmark SCRMP fragment calculations 5 of molecular crystal chemical shifts using the PBE functional perform about as well as GIPAW PBE for 1 H, 13 C, and 15 N. They give errors that are modestly larger for 17 O shifts, which are especially sensitive to the description of the electrostatic environment. 26 Switching to the hybrid PBE0 functional instead of PBE improves the accuracy of 13 C and 15 N chemical shifts by ∼30−50% when using the HMBI SCRMP method. For example, the 13 C rootmean-square (rms) errors improved from 2.1 ppm with PBE to 1.1 ppm with PBE0.…”

“…Second, both finite molecular cluster models ,− and GIPAW calculations combined with a local, nonperiodic correction ,,, have been shown to predict chemical shifts and electric field gradient (EFG) tensors in organic crystals with accuracy on par with or higher than standard GIPAW PBE results. A ∼10–15 molecule cluster is often sufficient to reproduce chemical shifts in organic crystals, , especially when electrostatic embedding is employed to mimic the longer-range lattice contributions.…”

“…In other words, the SCRMP fragment approach replaces the calculation of the chemical shieldings in an infinite periodic crystal or in a large cluster with an embedded monomer and ∼1–2 dozen embedded dimer calculations. A comparable many-body expansion approach has proved effective for EFG tensors as well. , …”