Comparison of basis set effects and the performance of ab initio and DFT methods for probing equilibrium fluctuations

The electronic absorption and emission spectra of large molecules reflect the extent and timescale of electron-vibration coupling and therefore the extent and timescale of relaxation/reorganization in response to a perturbation. In this paper, we present a comparison of the calculated absorption and emission spectra of NADH in liver alcohol dehydrogenase (LADH), using quantum mechanical/molecular mechanical methods, in which we vary the QM component. Specifically, we have looked at the influence of basis set (STO-3G, 3-21G*, 6-31G*, CC-pVDZ, and 6-311G**), as well as the influence of applying the DFT TD-B3LYP and ab initio TD-HF and CIS methods to the calculation of absorption/emission spectra and the reorganization energy (Stokes shift). The ab initio TD-HF and CIS methods reproduce the experimentally determined Stokes shift and spectral profiles to a high level of agreement, while the TD-B3LYP method significantly underestimates the Stokes shift, by 45%. We comment on the origin of this problem and suggest that ab initio methods may be naturally more suited to predicting molecular behavior away from equilibrium geometries.