Fullerene C70 characterization by 13C NMR and the importance of the solvent and dynamics in spectral simulations.

The nuclear magnetic resonance (NMR) spectroscopy combined with theoretical calculations is an important tool for fullerene identification. However, the accuracy of available theoretical methods is often not adequate. Therefore, in this work, different computational aspects needed to simulate realistically chemical shifts in the C70 molecule are investigated by density functional theory (DFT) calculations. The importance of the functional choice, basis set, solvent, and molecular motions was assessed. The solvent was simulated using the implicit conductor-like polarized continuum model. The molecular motions were included via anharmonic corrections and averaging of snapshots obtained from classical and first-principles molecular dynamics (MD) simulations. Comparison to experiment revealed that density functional calculations typically overestimate the (13)C NMR chemical shifts. Hybrid functionals, such as BHandH and BHandHLYP, and long-range corrected functionals, such as wB97xd and CAM-B3LYP, give the best results. While the solvent has a minor effect (chemical shift changes by ~1 ppm), the vibrational and dynamical effects are surprisingly large, causing changes up to 9 ppm. Consideration of the latter was also necessary to explain the observed temperature dependence. While the dynamical corrections for MD performed in vacuum were overestimated, inclusion of the solvent in simulations provided more realistic results. The study thus points out the importance of an appropriate solvent model and a complex approach to the modelling, balancing the static, dynamic and environmental factors.