186
M. K. Bilonda and L. Mammino
2 Computational Details
Calculations were performed using the same levels of theory as in previous works
on NIQ alkaloids [3–6], to enable meaningful comparisons. These are Hartree-Fock
(HF) with the 6-31G(d,p) basis set, and Density Functional Theory (DFT) with the
B3LYP functional [9, 10] and the 6-31+G(d,p) basis set. The reasons for selecting these two levels of theory have been explained in [3–6]. The results obtained
in [3–6], as well as studies on other molecules [11, 12] showed that HF can successfully handle intramolecular H-bonding at a comparatively low cost and yields
HOMO-LUMO energy gaps approaching those of experiments. DFT calculations
are important because DFT takes into account part of the electron correlation. Utilising at least two calculation methods of different natures provides more complete
information and validates trends-verification. Although the molecule is symmetric,
no symmetry conditions were imposed, to verify how closely the symmetry present
in the input is maintained on optimization.
Vibrational frequencies (harmonic approximation) were calculated in the gas
phase at the HF/6-31G(d,p) level to verify that the stationary points from optimization results correspond to true minima, to obtain the zero-point energy (ZPE) corrections, and to evaluate the red shifts caused by IHBs, which, in turn, enable an
approximate comparison of the IHB strengths. The frequency values were scaled by
0.9024, as recommended for the HF/6-31G(d,p) level [13]. Frequency calculations
at the DFT/B3LYP/6-31+G(d,p) level proved unaffordable (did not complete), likely
because of the high number of atoms in the molecule.
Calculations in solution considered the same three solvents (chloroform, acetonitrile and water) as in the previous studies of NIQ alkaloids [3–6] and utilised the same
computational approach. They were performed as single point (SP) calculations on
the in-vacuo-optimised geometries, at the HF/6-31G(d,p) level and using the Polarizable Continuum Model (PCM [14–19]). In this model, the solvent is considered
infinite and is modelled by a continuous isotropic dielectric, and the solute molecule
is considered inserted in a cavity within the continuum solvent. The geometry of the
cavity follows the geometry of the solvent accessible surface of the solute molecule.
The default settings of Gaussian03 [20] for PCM were used, namely, the Integral
Equation Formalism model (IEF [15–19]) and Gepol model [21–23]. The SCFVAC
option was selected to obtain more thermodynamic data. These calculations were
performed as SP because the size of the SHA molecule makes re-optimisation in
solution computationally exceedingly expensive. While SP calculations cannot provide information on the geometry changes caused by the solvent (including changes
in the IHB parameters), they can provide reasonable information on the energetics,
such as the conformers’ relative energies in solution and the free energy of solvation,
and also information about the changes in the molecule’s dipole moment caused by
the solvent’s polarizing effects. PCM calculations at the DFT/B3LYP/6-31+G(d,p)
level exhibited frequent convergence problems, which prevented the obtainment of
enough results to identify patterns.
M. K. Bilonda and L. Mammino
2 Computational Details
Calculations were performed using the same levels of theory as in previous works
on NIQ alkaloids [3–6], to enable meaningful comparisons. These are Hartree-Fock
(HF) with the 6-31G(d,p) basis set, and Density Functional Theory (DFT) with the
B3LYP functional [9, 10] and the 6-31+G(d,p) basis set. The reasons for selecting these two levels of theory have been explained in [3–6]. The results obtained
in [3–6], as well as studies on other molecules [11, 12] showed that HF can successfully handle intramolecular H-bonding at a comparatively low cost and yields
HOMO-LUMO energy gaps approaching those of experiments. DFT calculations
are important because DFT takes into account part of the electron correlation. Utilising at least two calculation methods of different natures provides more complete
information and validates trends-verification. Although the molecule is symmetric,
no symmetry conditions were imposed, to verify how closely the symmetry present
in the input is maintained on optimization.
Vibrational frequencies (harmonic approximation) were calculated in the gas
phase at the HF/6-31G(d,p) level to verify that the stationary points from optimization results correspond to true minima, to obtain the zero-point energy (ZPE) corrections, and to evaluate the red shifts caused by IHBs, which, in turn, enable an
approximate comparison of the IHB strengths. The frequency values were scaled by
0.9024, as recommended for the HF/6-31G(d,p) level [13]. Frequency calculations
at the DFT/B3LYP/6-31+G(d,p) level proved unaffordable (did not complete), likely
because of the high number of atoms in the molecule.
Calculations in solution considered the same three solvents (chloroform, acetonitrile and water) as in the previous studies of NIQ alkaloids [3–6] and utilised the same
computational approach. They were performed as single point (SP) calculations on
the in-vacuo-optimised geometries, at the HF/6-31G(d,p) level and using the Polarizable Continuum Model (PCM [14–19]). In this model, the solvent is considered
infinite and is modelled by a continuous isotropic dielectric, and the solute molecule
is considered inserted in a cavity within the continuum solvent. The geometry of the
cavity follows the geometry of the solvent accessible surface of the solute molecule.
The default settings of Gaussian03 [20] for PCM were used, namely, the Integral
Equation Formalism model (IEF [15–19]) and Gepol model [21–23]. The SCFVAC
option was selected to obtain more thermodynamic data. These calculations were
performed as SP because the size of the SHA molecule makes re-optimisation in
solution computationally exceedingly expensive. While SP calculations cannot provide information on the geometry changes caused by the solvent (including changes
in the IHB parameters), they can provide reasonable information on the energetics,
such as the conformers’ relative energies in solution and the free energy of solvation,
and also information about the changes in the molecule’s dipole moment caused by
the solvent’s polarizing effects. PCM calculations at the DFT/B3LYP/6-31+G(d,p)
level exhibited frequent convergence problems, which prevented the obtainment of
enough results to identify patterns.