peculiarities of d orbital electronic arrangements, in order to formulate and interpret
rational structures; when a structure isnot“rational”, because it is particularly
novel, background theoretical knowledge is even more valuable. Prosaic factual
knowledge of chemical properties does not hurt either. The elucidation of the
structure of ferrocene, (C 5 H 5 ) 2 Fe, provides a nice example of the role of factual
and theoretical knowledge in discovery. Ferrocene was initially assigned a conven-
tional C–Fe–C structure, but unlike known compounds with a metal-carbon sigma
bond it was very stable, and like benzene reacted by electrophilic substitution.
Theory led to the formulation of the correct and then-unprecedented sandwich
structure. The ferrocene saga, which initiated a revolution in transition metal
chemistry, has been summarized by Dagani [ 103 ] and Laszlo and Hoffmann [ 104 ].
In our short survey of the computational techniques available for investigating
TM compounds we first mention molecular mechanics (Chapter 3). It may seem
humble by the standards of the quantum mechanical ab initio, semiempirical and
DFT methods (Chapters 5, 6 and 8, respectively) but MM is useful for obtaining
input structures for submission to one of those calculations, may even provide in
itself useful information, and it is, of course, extremely fast. Indeed, a recent book
on the modelling of inorganic compounds, mainly TM species, is devoted very
largely to molecular mechanics and a program specially parameterized for TM
compounds, Momec3 [ 105 ].
Ab initio methods (unparameterized, or almost unparameterized, wavefunction
calculations) were at one time, in contrast to DFT, deprecated for the study of TM
compounds, but it now appears that this inferiority of ab initio is largely confined to
the first-row metals, titanium to copper [ 81 , 106 ]. DFTcansometimes be quite
inaccurate, and advanced correlated ab initio methods like CCSD(T) and even
CCSDTQ (Section 5.4.3), may be useful, although these are currently limited to
small systems [ 107 ]. Nevertheless, DFT calculations with pseudopotentials, com-
monly relativistic, are now the standard methods for performing calculations on TM
compounds [ 81 , 106 , 108 ]; for example Frenking, in a paper analyzing bonding in
such species, extols the virtues of DFT used with pseudopotentials [ 108 ]. The
suitability of various functionals for TM chemistry is commented on by Zhao and
Truhlar in a review which presents their new M0-class functionals (Sections 7.2.3.4.
and 7.3), and the most appropriate for this purpose are said to be M06 and,
especially, M06-L [ 109 ], but Tekarli et al. found that with the correlation-consistent
cc-p-VQZ basis the B97-1 functional can give formation enthalpies of first-row
transition metals within 4 kJ mol"^1 (1 kcal mol"^1 ) of high-level multistep ab initio
methods (cf.Section 5.5.2.2b) , G4(MP2) and ccCA-tm [ 110 ]. A DFT method
called SIESTA (Spanish Initiative for Electronic Simulation with Thousands of
Atoms), designed for big, extended systems like large metal clusters, has found use
in recent years [ 111 ].
Finally, TM compounds have been studied by semiempirical methods. One
thinks first offaux-ab initio-type methods like AM1 and PM3 (Chapter 6), since
these are surrogates for “full” quantum mechanical ab initio techniques. However,
the deepest insights into the nature of these compounds that have been afforded by a
semiempirical method have come from the uncomplicated and venerable extended
8.3 A Note on Heavy Atoms and Transition Metals 551