TD-DFT benchmark for indigoid dyes

In this paper, we assess the efficiency of more than 20 functionals for reproducing the experimental absorption wavelength of a large set indigo derivatives (thioindigo, selenoindigo, indirubin, isoindigo,...). The pros and cons of each functional is analysed in terms of, on the one hand, the accuracy with respect to the experimental transition energies, and, on the other hand, the consistency of the predicted rankings. It turns out that global hybrids including between 20% and 25% of exact exchange yield the smallest mean absolute deviations. Functionals with a larger percentage of exact exchange (similar to 50%) or range-separated hybrids deliver larger absolute errors, but improve the linear correlation between experimental and theoretical values. (C) 2009 Elsevier B.V. All rights reserved.