localization of sigma molecular orbitals: towards a better theoretical characterization of the excited electronic states of big conjugated molecules

Precise characterization of the first excited electronic states of conjugated molecules requires the circumvention of two technical difficulties inherent to the calculation of the interaction of configurations: on the one hand, the calculation and storage of the bi-electronic integrals (their number varies like the power four of the number of orbitals) and on the other hand, the selection of the configurations important for the description of the excited states. There are very precise ab initio methods (CASPT2 for example). However, they cannot yet deal with molecules much larger than naphthalene. In order to study the excited states of larger conjugated molecules, always taking into account the reorganization of the σ system, additional approximations are indispensable. First of all, we have chosen to place ourselves within the framework of the semi-empirical CS-INDO method. Moreover, we did not use the canonical molecular orbital set. Indeed, the delocalization of these orbitals over the whole molecule does not make it possible to deduce from the topological criterion which eliminates the bi-electronic integrals between atomic orbitals a simple criterion to select the large bi-electronic integrals between molecular orbitals. To do this, we have located each orbital σ between two atoms. In this way, it is possible to neglect a priori a large number of bi-electronic integrals built on two σ OMs when these two orbitals are not located on the same bond. The same topological criterion also makes it possible to considerably restrict the number of excited configurations. We have verified on the first excited states of small aromatic molecules that these approximations do not induce any significant error on the calculation of the excitation energies even when the excited state has a strong charge transfer character (error less than 500 cm^-1 with respect to the calculation made with all integrals and all di-excitations). The development of this method has allowed us to characterize charge transfer states in molecules of the DMABN family and in bi-anthryl. Our approximations have also allowed the characterization of excited states in super systems built on several chromophores. As long as the chromophores are not too conjugated, we have thus validated the application of the excitonic model to predict and understand the nature of the excited states of the super system. (author)