Comparison of time-dependent density-functional theory and coupled cluster theory for the calculation of the optical rotations of chiral molecules

A comparison of the abilities of time-dependent density-functional theory (TDDFT) and coupled cluster (CC) theory to reproduce experimental sodium D-line specific rotations for 13 conformationally rigid organic molecules is reported. The test set includes alkanes, alkenes, and ketones with known absolute configurations. TDDFT calculations make use of gauge-including atomic orbitals and give origin-independent specific rotations. CC rotations are computed using both the origin-independent dipole-velocity and origin-dependent dipole-length representations. The mean absolute deviations of calculated and experimental rotations are of comparable magnitudes for all three methods. The origin-independent DFT and CC methods give the same sign of [alpha]D for every molecule except norbornanone. For every large-rotation ketone and alkene for which DFT and CC yield the incorrect sign as compared to liquid-phase experimental data, the corresponding optical rotatory dispersion (ORD) curve is bisignate, suggesting that the two models cannot reliably reproduce the relative excitation energies and antagonistic rotational strengths of multiple competing electronic states that contribute to the total long-wavelength rotation. Several potential sources of error in the theoretical treatments are considered, including basis set incompleteness, vibrational and temperature effects, electron correlation, and solvent effects.