Chemistry Reference
In-Depth Information
Although many arguments for and against CP corrections have been
made over the years, this part of the chapter will briefly illustrate three impor-
tant concepts related to this issue.
The CP procedure can lead to unphysical descriptions of PESs that are not
easily corrected.
CP corrections often do not improve calculated interaction energies. In fact,
the procedure can even make results worse.
BSSE is a poor diagnostic for the quality of a computed
D
e
or
E
int
. The crucial
quantity is the basis set completeness error (BSCE).
While the CP procedure is useful when examining specific structures, it is
sometimes desirable to examine a path across the PES or reaction profile that
might include, for example, reactants, products, and a transition state (TS).
Generating a CP-corrected potential energy curve or reaction profile can intro-
duce a new set of problems in certain circumstances. As demonstrated by
Lendvay and Mayer, the Boys-Bernardi CP procedure (and variations thereof)
can actually produce discontinuities in the PES near the TS and even give dif-
ferent TS energies for the forward and reverse reactions.
196
Fortunately, this
unphysical behavior is normally limited to regions near the TS.
To illustrate the last two points, consider the
D
e
of (H
2
O)
2
and (H
2
O)
3
.
The MP2 CBS limits have been determined from explicitly correlated MP2-
R12 computations with the K2-
76 kJ mol
1
basis set to be 20
:
for the
C
s
12 kJ mol
1
global minimum of (H
2
O)
2
and 66
:
for the cyclic
C
1
global mini-
mum of (H
2
O)
3
.
118,119,197
basis set is constructed by removing
functions with the two highest angular momentum values from Klopper's
K2 basis set,
198
which corresponds to
f
and
g
functions for H, and
g
and
h
functions for O.) Figures 8 and 9 show how CP-corrected and uncorrected
MP2 calculations deviate from the corresponding CBS limit when using the
aug-cc-pV
X
Z and haug-cc-pV
X
Z (diffuse functions only added to O atoms)
families of basis sets. The height of a particular bar above the
x
axis indicates
the basis set completeness error (BSCE) for that basis set while the combined
height of a bar above and its CP-corrected counterpart below the
x
axis repre-
sents the magnitude of BSSE for that basis set. For both (H
2
O)
2
and (H
2
O)
3
,
the MP2 dissociation energies converge systematically to the CBS limit from
below when the CP procedure is applied and from above when it is not.
Closer examination of the data presented in Figures 8 and 9 reveals that
the errors associated with the CP-corrected dissociation energies are almost
always larger than those that are not corrected for BSSE (i.e., the bars below
the
x
axis are larger than those above it). Only for aug-cc-pVQZ, aug-cc-
pV5Z, and aug-cc-pV6Z
D
e
values for (H
2
O)
2
does the CP procedure offer
any improvement (and merely on the order of 0
(The K2-
1 kJmol
1
). In these particular
cases, the CP corrections are clearly not worth the substantial additional effort.
Although these trends have been observed elsewhere,
133,199
they do not
necessarily apply to all weakly bound complexes. Sinnokrot and Sherrill have
:
