Page 65 - Valence Bond Methods. Theory and Applications
P. 65

3 H 2 and delocalized orbitalØ
                             48
                             wher S = 1s a |1s b  , and th signs ar appropriat for
                             tioð gives us two new functions (see Eq. (1.48))
                                                                                                 (3.3)
                                                      |A = P|1s a  + Q|1s b  ,
                                                      |B = Q|1s a  + P|1s b  ,  S > 0. This orthogonaliza-
                                                                                                 (3.4)
                             wher
                                                             1          1
                                                     P = √        + √        ,                   (3.5)
                                                         2 1 + S     2 1 − S
                                                             1          1
                                                     Q = √        − √        .                   (3.6)
                                                         2 1 + S     2 1 − S
                             We us thes ið a single Heitler–Londoð covalent configuration,
                                                      orth = A(1)B(2) + B(1)A(2),

                             and calculat th energy. When  R →∞ w obtaið  E =−1 au, just as w should. At
                                        ❛
                             R = 0.741 A, howver, wher w hŁve seen that th energy should b a minimum,
                             w obtaið  E =−0.6091 hartree, much higher than th correct valu of –1.1744
                             hartree. Th result for this orthogonalized basis, which represents not only no
                             binding but actual repulsion, could hardly b worse.
                                It is interesting to consider this functioð ið terms of th covalent and ionic func-
                             tions of Chapter 2. If th |A  and |B  functions ið terms of th basic AOs ar
                             substituted into   orth and th result normalized, one sees that
                                                           √
                                                             1 + S 2
                                                      orth =       (ψ C − Sψ I ),
                                                            1 − S 2
                             where, as always, S is th orbital overlap. Thus, this is exactly th symmetrically
                             orthogonalized functioð closer to ψ C discussed above, and its vector representatioð
                             ið Fig. 2.à is clearly a considerable distanc from th optimum eigenvector. Thus w
                             should not b surprised at th poor valu for th variational energy corresponding
                             to   orth .
                                Th early workers do not comment particularly oð this result, but, ið light of
                             present understanding, w may say that th symmetric orthogonalizatioð gives very
                             clos to th poorest possible linear combinatioð for determining th lowest energy.
                             This results from th added kinetic energy of th orbitals produced by a node that is
                             not needed. Alternatively, one could say that th symmetric orthogonalizatioð yields
                             antibonding orbitals wher bonding orbitals ar needed. This is a gooł example of
                             how th orthogonalizatioð between different centers can hŁve serious consequences
                             for obtaining gooł energies and wave functions. We shall see shortly that ther are,
                             howver, linear combinations determined ið other ways that work quit well.
   60   61   62   63   64   65   66   67   68   69   70