Chemistry Reference
In-Depth Information
R2
R
R1
R
A
B
C
Figure 1
Three most commonly studied configurations of the benzene dimer: the
sandwich (A), T-shaped (B), and parallel-displaced (C) configurations.
CHALLENGES FOR COMPUTING
p
INTERACTIONS
Many chemical problems can be addressed easily and reliably using
Hartree-Fock molecular orbital theory or Kohn-Sham density functional
theory with modest-sized basis sets. Unfortunately,
interactions, and non-
covalent interactions in general, are not among them. In this section we con-
sider the electron correlation and basis set requirements for computations of
p
p
interactions.
To illustrate the difficulties in finding a suitable theoretical method for
the reliable computation of
p
interactions, let us consider the simplest possible
prototype of aromatic
interactions, the benzene dimer. Despite a large
number of theoretical
24-28
and experimental studies,
26,29-36
a clear picture
of the the geometrical preferences and binding energy of the benzene dimer
was not available until high-level theoretical studies were conducted by Tsu-
zuki's group
37
and our group
38
in 2002 using a combination of coupled-
cluster theory and large-basis second-order Møller-Plesset (MP2) perturbation
theory computations. Figure 1 shows the three most commonly studied geome-
trical configurations of the benzene dimer.
p
-
p
Electron Correlation Problem
One of the primary challenges for computing
interactions is that differ-
ent theoretical methods can give quite different results. This is illustrated in
Figure 2, which shows potential energy curves for the sandwich benzene dimer
when the distance between the rings is systematically varied; the monomers are
frozen at the recommended geometry of Gauss and Stanton.
39
Figure 2 com-
pares the results from restricted Hartree-Fock (RHF), the very popular B3LYP
hybrid density functional method,
40,41
MP2 perturbation theory [also referred
to as many-body perturbation theory through second order, or MBPT(2)],
coupled-cluster theory with single and double substitutions (CCSD),
42
and
coupled-cluster through perturbative triple substitutions, CCSD(T),
43
all using
p
