adjust the values of force constants, reference bond lengths, etc. in simple non-
quantum-mechanical equations.
SE methods like PPP, CNDO, and AM1 are partly empirical and partly quantum-
mechanical: experimental (or nowadays, often high-levelab initioor DFT) para-
meters are used to simplify the evaluation of the integrals in the Fock matrix. In
contrast, there is no quantum-mechanical component to MM; it is not quite true,
however, that MM has no theoretical component, because the force constants and
reference geometric parameters are inserted into an ((albeit simple) classical-
physics ball-and-springs-model of a molecule (this model is augmented with energy
terms arising from dihedral angles, nonbonded interactions, and possibly other
factors).
Chapter 6, Harder Questions, Answers
Q3
What do you think are the advantages and disadvantages of parameterizing SE
methods with data fromab initio calculations rather than from experiment?
Could a SE method parameterized using ab initio calculations logically be called
semiempirical?
This question is similar to Chapter 3, Harder Question 3, for MM. For the first part
of the question I’ll just repeat the response to that question, tailored to be appropriate
to SE methods. Apart from a possible philosophical objection, which from a utilitarian
viewpoint can be dismissed, there is the question of the trustworthiness of the ab initio
or DFT results. For “normal” molecules, that is, species which are not in some way
exotic [1], these calculations deliver quite reliable results. The advantages they offer
over experimental acquisition of the required parameters is that these quantities
(1) can be obtained for a wide variety of compounds without regard to synthetic
difficulties or commercial availability, (2) are offered up transparently by the output of
the calculation, rather than being required to be extracted, perhaps somewhat tortu-
ously, from experiments, (3) are usually more quickly calculated than determined in
the lab, and (4) can be uniformly secured, that is, all parameters can be obtained from
calculations at the same level, say MP2/6-311G(df,p), in contrast to experiment,
where different methods must be used to obtain different parameters. This last point
may be more of an esthetic than a utilitarian advantage.
The advantage of parameterizing with experimental quantities is that, if the
experiment is reliable, then weknowthe values of the parameters; we need not
reflect on the reliability of the calculation. Of course, we might wish to ponder the
accuracy of the experiment.
Could a SE method parameterized using ab initio calculations logically be
called semiempirical? Literally, semiempiricalmeans semiexperimental. If we
parameterize with calculations we have not resorted to experiment (of course,
630 Answers