Basis set dependence of coupled cluster optical rotation computations.

Specific rotations for five notoriously difficult molecules, (S)-methyloxirane, (S)-methythiirane, (S)-2-chloropropionitrile, (1S,4S)-norbornenone, and (1R,5R)-β-pinene, have been computed using coupled cluster (CC) and density functional theory (DFT). The performance of the recently developed LPol basis sets compared to the correlation-consistent sets of Dunning and co-workers has been examined at four wavelengths: 355, 436, 589, and 633 nm. We find that the LPol basis sets are an efficient choice, often outperforming the more commonly used correlation-consistent basis sets of comparable size. The smallest of the four, LPol-ds, performs nearly as well as the rest of the series and often yields results closer to the basis set limit than appreciably larger basis sets. While the performance of the LPol bases is admirable, they still do not alleviate the need for high levels of electron correlation, vibrational corrections, and the inclusion of solvent effects to accurately reproduce experimental rotations. In particular in the case of β-pinene we find that they do not produce agreement between DFT and experiment as was previously suggested.