(Leach pgs. 228-231; Grant and Richards pgs 41-43)
·
Some methods for deriving atomic charges are mentioned above.
·
Other parameters (e.g. force constants and equilibrium values) can
be chosen by trial and error, or by fitting (e.g. by a least-squares fit), to
experimental data or ab initio calculation results.
·
Bond lengths and angles can be taken from gas-phase structural
studies on small molecules (e.g. microwave spectra, or electron diffraction).
Force constants can be found by fitting to experimental or calculated harmonic
vibrational frequencies.
·
Barriers to rotation around torsion angles can be calculated at the
ab initio level can be found from ab initio calculations on small molcules
(recall that Hartree-Fock calculations can give useful results for
conformational energy changes).
·
When fitting molecular mechanics torsion terms, it should be
remembered that MM 1-4 non-bonded interactions will also contribute.
·
Non-bonded parameters (van der Waals and electrostatic) can be
found from ab initio calculations, with the caveat that Hartree-Fock
calculations do not include dispersion effects, so better calculations (e.g.
MP2) would be required for a good fit. However, typically the van der Waals
terms have little influence on structure, so are optimizeded in the final
stages of fitting to improve the performance of the model.