Chemistry Reference
In-Depth Information
[
124
]). The bcp properties of CO, for either the isolated molecule or the molecule
acting as a ligand, depend even more strongly on the chosen computational scheme
or MM refinement of structure factors [
2
,
6
,
125
]. The
2
r
b
value for CO is indeed
extremely sensitive to the location of the bcp, which most often lies close to the
nodal
r
2
r
surface separating the core-shell charge depletion region of the carbon
from the single, merged VSCC of CO. Despite numerous evidences of a shared
interaction character of the C-O bond [
125
],
r
2
r
b
is found often positive, but even
very small environmental or modeling changes, capable to slightly shift the bcp
toward the oxygen atom, may lead to negative
r
2
r
b
values [
2
,
66
]. The three
PD models listed in Table
10
for CO differ as for their minimum energy geometry
and dipole moment estimate. The corresponding S% results are rather stable, with
differences in values among the three models not exceeding 3 points, whereas
r
b
and in particular
r
r
2
r
b
show a much larger variability. The electron density values
at the C-O bcp differ among each other by a maximum of about 10%, but the largest
jr
2
value is 14 times higher in magnitude and with sign reversed than the lowest
one. When PDs data are compared with the corresponding MMPD results, one
observes that the bias due to the multipole model refinement leads to acceptably
limited changes on the
r
b
and S% values, the largest ones being about 7.9% and
5 points, respectively. Conversely, the
r
b
j
2
r
b
values are not at all reproduced, with
changes even as large as 530% in one case and with a value qualitatively similar in
magnitude, but reversed in sign in another case. Table
10
reports MMPD data for
the MM with refined
k
and
k
0
screening parameters [
10
]. Adoption of standard
unitary screening parameters, though leading to quite different numerical values for
r
r
2
r
b
, does not qualitatively affect the picture [
15
]. It is worth noting that, despite
the large variations observed for the
2
r
b
magnitudes with changes in the PD
models or following the MM refinement of their corresponding structure factors,
the S% values remain reasonably stable for such shared and partially polar bond.
This holds true even when the
r
2
r
b
sign changes. The large and positive QTAIM
net charge of C and the location of the bcp closer to the C than to the O nucleus
easily explain why the S%(O) always exceeds that from C.
The results obtained for C
4
F
6
and CO seem to indicate that the S% values are
carrying a more robust and thus more chemically meaningful information than do
the
r
b
and, in particular, the
r
2
r
b
local descriptors. Indeed, being the result of
integration over an atomic basin, the SF or the S% contribution averages out the
local bias introduced by the MM refinement or by the PD model change on the LS
integrand (which is defined through the Laplacian). One also observes that since the
SF contributions simply reconstruct the electron density, PD and MMPD SF results
should roughly differ as their electron densities do. However, rather then being
related to the
r
b
values, the S% contributions express how
r
b
is comparatively
shared between the two linked atoms. Because of the dependence on the relative
contribution to
r
b
rather than to the
r
b
magnitude itself, the S% descriptor turns out
to be generally more stable than it is
r
b
(and a fortiori
r
2
r
b
) against both the MM
bias and the changes due to different theoretical approaches or multipole model
treatments.
r
