|Computational chemistry is a standard tool to understand chemical phenomena at the
electronic and molecular levels, and is a rapidly growing branch of chemistry. In this thesis,
three distinct computational chemistry projects about dispersion interaction between small
molecules and models of nanotubes are investigated: (1) An accurate benchmark description
of the interactions between carbon dioxide and polyheterocyclic aromatic compounds containing
nitrogen; (2) Evaluation of DFT-D variants suitable for nanotube adhesion forces;
(3) Description of the interactions between carbon dioxide and polyheterocyclic aromatic
compounds containing nitrogen via local methods.
Chapter 1 presents a brief introduction to these subjects. Chapter 2 explains the
methodologies used in this work. Chapter 3 presents the performance of a large variety of
modern density functional theory approaches for the adsorption of carbon dioxide on molecular
models of pyridinic N-doped graphene. The benchmark interaction energies were established
at the complete-basis-set limit MP2 level plus a CCSD(T) coupled-cluster correction
in a moderate but carefully selected basis set. Using a set of 96 benchmark CCSD(T)-level
interaction energies as a reference, the performance of various DFT+D variants was examined.
It turns out that several schemes such as B2PLYP-D3 and M05-2X-D3 exhibit average
errors on the entire benchmark data set in the 5-10% range. The top DFT+D variants were
then used to investigate the energy profile for a carbon dioxide transition through model N-doped
graphene pores. The results obtained from these methods indicated that the largest,
N4H4 pore allows for a barrierless CO2 transition to the other side of a graphene sheet. In
Chapter 4, three sets of benchmark CCSD(T)/CBS data, all involving a coronene molecule
(flat at or curved away from the adsorbate), interacting with 1 methane, 2 carbon dioxide,
and 3 ethylene are combined to determine the performance of various DFT approaches for
describing interactions of solubilizer molecules with nanotubes. The combined data show
that the simple PBE-D3(BJ)refit variant emerges as an optimal combination of accuracy and
efficiency for weakly interacting complexes of this kind. Finally, Chapter 5 reports the results
for various local methods on investigation of the dispersion interaction between carbon
dioxide and polyheterocyclic aromatic compounds containing nitrogen. Unfortunately, so
far no single local method could yield benchmark-level accuracy for the models of interest.
Therefore, more detailed study is needed to generate acceptable results for local treatments.