c20 JWBS043-Rogers September 13, 2010 11:29 Printer Name: Yet to Come
324 QUANTUM MOLECULAR MODELING
(NDO) andneglect of diatomic differential overlap(NDDO) approximations led to
a series of semiempirical programs denoted CNDO (complete neglect of differential
overlap), INDO, MINDO, and so on. Initially, NDDO programs were parameterized
to reproduceab initiovalues for simple molecules but later semiempirical programs
such as the AM1 programs of Dewar et al. and the PM3 method of Stewart are
parameterized against experimental thermochemical results so as to calculate energies
and enthalpies. Presently, they are both in wide use for this purpose. Along with
thermochemical data, dipole moments, geometries, and isomerization potentials are
also calculated by modern semiempirical programs.
20.5 AB INITIO METHODS
An exact solution of the Schrodinger equation would employ no empirical parameters ̈
beyond mass and charge of the subatomic particles, and it would be an absolute
solution to the problem that could never be changed or revised. In practice, the
absolute properties of a molecule are never found, though they may be approached
by ever-improving approximate methods. Today, what we call “ab initioprocedures”
contain small empirical “corrections,” but a very serious effort has been put forth
to minimize these aspects of the procedure. One condition we wish for the set of
functions{χμ}we choose as a basis set is that they be as nearly complete as possible,
that is, we hope that the set willspanthe vector space.
20.6 THE GAUSSIAN BASIS SET
We saw in Chapter 17 that the single Gaussian functionφ(r)=Ce−αr
2
, whereC=
1 .0 andα= 0 .2829, does not give a very good approximation to the energy of the
hydrogen atom. The result, HF=0.4244Eh, is 84.9% of the true (defined) energy
of the hydrogen atom. The Gaussian, withr^2 in the exponent, drops off faster than
the 1sorbital, which hasrin the exponent (Fig. 20.4). The Gaussian is too “thin” at
larger distances from the nucleus.
031 2
0.5
1
1
0
(r)
(r)
0 r 3
FIGURE 20.4 The 1sSTO (solid line) and a Gaussian approximation (dotted line).