This thesis is focused on the characterization and the interpretation of the nonlinear optical (NLO) response of complex systems by quantum chemistry methods. The main goal is the reduction of the gap between real systems characterized experimentally and model systems usually employed in simulations and therefore the introduction of the structural complexity of the studied compounds. Computationally, the main targets are biological structures useful for second harmonic generation microscopy (SHIM), fluorescent proteins and collagen, of which the responses have different origins. The NLO response of fluorescent proteins originates mostly from a push-pull π-conjugated chromophore whereas, for collagen, small π-conjugated peptide bonds are at the origin of the NLO response. In the case of fluorescent proteins, theoretical simulations carried out at different levels of approximation are accompanied by experimental investigations in order to design proteins that fluoresce over the whole visible spectra and exhibit large second-order NLO responses. In particular, different applications of the ONIOM scheme have been assessed in comparison with experiment, demonstrating the need for an explicit description of the close surroundings of the chromophore and the inclusion of electron correlation. This study has substantiated the importance of the symmetry argument for the second-order nonlinear optical properties of fluorescent proteins and the sensitivity of this principle. In the investigation on collagen, a short peptide triple-helix model [(Pro-Pro-Gly)10]3 (PPG10) has been considered and its first hyperpolarizability has been calculated using the ONIOM method in combination with time-dependent Hartree-Fock and DFT calculations. Detailed analysis of the first hyperpolarizability tensors has unraveled that β originates from the individual peptide bonds and that the triple-helix structure gives the relative orientation of these bonds. So, the total response is mostly determined by summing the β tensors of each bond as well as by accounting for (hyper)polarization effects between the chains. The selection of quantum chemistry methods to describe the NLO responses of fluorescent proteins and collagen has been substantiated by investigations of the impact of electron correlation on the hyperpolarizabilities of small model systems, i) push-pull π-conjugated systems, which model the chromophore of fluorescent proteins and ii) open-shell systems such as the p-quinodimethane and two of its derivatives. For closed-shell systems, these studies validate the choice of the MP2 method for estimating the first hyperpolarizability and show the lack of reliability of DFT approaches with most exchange-correlation functionals. In parallel, several technical aspects associated with the evaluation of the hyperpolarizabilities have been addressed, i.e i) the automatization of the Romberg scheme to improve the numerical accuracy in the finite field method, ii) the evaluation of the frequency dispersion at correlated levels by approximated schemes, and iii) the deviations from Kleinman’s conditions.
|Date of Award||29 Apr 2014|
|Supervisor||Benoit CHAMPAGNE (Supervisor), Johan Wouters (President), Vincent Liegeois (Jury), Koen Clays (Jury), Tom Leyssens (Jury) & Michaël J. BEARPARK (Jury)|