Study of Non-Covalent Interactions in Onium Ions

Abstract

Non-covalent interactions are a key factor in the stabilization of a structure or conformer. Even if they have been discovered and studied for decades, there are still gaps in the theoretical and experimental studies of these forces. A better understanding should, ultimately, translate into more accurate predictions of structures and processes, such as reactions. In an effort to consolidate the current knowledge, this thesis reports findings about hydrogen bonding, London dispersion interactions and the treatment of solvation. The bending of intramolecular hydrogen bonds was studied experimentally using a test system of 14 protonated bipyridines. The main finding was that the bending of the hydrogen bond correlates linearly with the proton chemical shift. The localization/delocalization of the hydrogen between the nitrogen was studied in dichloromethane by isotopic perturbation. The observations were coherent with previous results and assumptions, but the results were not unambiguous. Through the study, it became evident that some compounds have multiple conformers in equilibrium. Even if a hydrogen bond is stabilizing, it was in some cases broken to favor a NH-$\pi$ interaction or to minimize steric repulsion. Non-hydrogen bonded conformers were, in a few cases, visible by X-ray crystallography. Gas phase DFT calculations appeared to underestimate steric repulsion and systematically predicted the hydrogen bonded structure as the energetical minimum. While DFT is generally considered to be accurate in the gas phase, DFT with implicit solvation is much less reliable. To get some answers on the source of the failure of implicit solvent models to reproduce experimental results, an in-depth computational study was undertaken. This study is based on results, obtained by previous group members, about the dissociation of proton-bound pyridine or quinoline dimers in gas phase and in dichloromethane. Ultimately, the explicit treatment of solvent molecules was required to reproduce experimental result. To overcome the limitations of classical force fields, more specifically the inability to simulate chemical reactions, the use of thermodynamic integration was investigated as a versatile method to calculate the solvation free energy. Unfortunately, the force field was not accurate enough in the gas phase to yield reasonable results. The effect of the parameterization of non-bonded interactions on free energy calculations was crudely studied but was not sufficient to find a method to calculate solvation free energy adequately.