The “free-electron” sea confers on the substance typical metallic properties
(Cotton FA, Wilkinson G, Gaus PL (1995) Basic inorganic chemistry, 3rd edn.
Wiley, New York, pp 249–251 and chapter 32).
So why are the transition elements all metals? A detailed answer would require a
discussion of concepts like band gaps and Fermi levels (Cotton FA,
Wilkinson G, Gaus PL (1995) Basic inorganic chemistry, 3rd edn. Wiley, New
York, chapter 32), but the beginning of an explanation emerges from consider-
ing, say, calcium, scandium and titanium (Z¼20, 21, 22). Calcium is a metal
because its nuclear charge is not high enough to prevent the two outer, 4s
electrons from merging into a common pool. The electrons that take us to
scandium and titanium get tucked into the 3d shell, still leaving, in the isolated
atom, the outermost 4s pair which in the bulk metal are pooled. Slight splitting of
the d levels by ligands confers typical transition metal properties, as touched on
inSection 8.3.4.- The simple crystal field analysis of the effect of ligands on transition metal
d-electron energies accords well with the “deeper” molecular orbital analysis (see
e.g. chapter 8, [99]). In what way(s), however, is the crystal field method unrealistic?
The crystal field method is a formalism. It perturbs the metal d orbitals with
point charges (Cotton FA, Wilkinson G, Gaus PL (1995) Basic inorganic
chemistry, 3rd edn. Wiley, New York, pp 503–509). It does not allow for the
role of other orbitals on the metal, nor does it invoke orbitals on the perturbing
charges. Thus it does not permit ligand electron donation to and electron
acceptance from the metal (Lewis basicity and Lewis acidity by the ligand;
the former is said to be essential, the latter desirable (chapter 8, [100])). - Suggest reasons why parameterizing molecular mechanics and PM3-type pro-
grams for transition metals presents special problems compared with parameter-
izing for standard organic compounds.
There are many more geometric structural possibilities with transition metal com-
pounds than with standard organic compounds. Carbon is normally tetrahedral and
tetracoordinate, trigonal and tricoordinate, or digonal and dicoordinate. This holds
for nitrogen too and the normal possibilities are even more restricted for other
common organic-compound atoms like hydrogen, oxygen and halogens. In con-
trast, a transition metal atom may have more stereochemical possibilities: square
planar, square pyramidal, tetrahedral, trigonal bipyramidal, and octahedral are the
common ones. The geometry of many transition metal molecules also poses a
problem for parameterization: consider ferrocene, for example, where iron(II) is
coordinated to two cyclopentadienyl anions. Should iron be parameterized to allow
for ten C–C bonds, or for two Fe-ring center bonds? This kind of conundrum arises
more for molecular mechanics parameterization, where bonds are taken literally,
than for PM3- or AM1-type parameterization, where the objective is to simplify the
ab initio molecular orbital method, which does not explicitly use bonds (although
the concept can be recovered from the wavefunction after a calculation). The
parameterization of molecular mechanics for transition metals is discussed in
[105] in connection with the Momec3 program (chapter 8, [105]).
Answers 653