Chemistry Reference
In-Depth Information
of the experimental densities. The S%(Co) values are even smaller in magnitude
and positive at about 4% for the single molecule studied theoretically. Interestingly,
some of the atoms belonging to the other molecule in the asymmetric unit yield a
small but nonnegligible contribution to the experimental density at the Co-Co mp,
which may be one of the reasons for the observed S%(Co) difference respect to the
theoretical value (the interaction with the other molecule in the asymmetric unit and
in general the crystal field effects are clearly not taken into account in the gas-phase
ab initio computation). Conversely, a comparison between the SF values obtained
from the two experimental data sets would have enabled one to get a rough idea of
how stable are the SF values with respect to the change of collected reflections and
of the multipole model parameters refined thereof. Unfortunately, as for the scope of
the present review, such a comparison was not reported in the paper, whereas one
would argue that it could have been interesting in its own, in view of the assessment
of the intrinsic stability of the SF description against those changes. One of the main
reason that led Overgaard et al. [
80
] to analyze only the topological properties from
the synchrotron data set, starting from page 3839 onward of their paper, was the
closer match with theory observed for the density and Laplacian profiles along the
various C-O bonds of the molecule, and in particular close to the bcp. Whether an
assessment of the relative quality of two data sets against theory must be based on a
comparison among local rather than integral quantities like the S% values is clearly
debatable. Especially so, if one considers that local quantities are often too sensitive
to otherwise small changes in the model approaches or data. An analysis of the
different sensitivity of local and integral topological properties such as the SF
values to moderate changes in the data and/or in their multipole refinement is briefly
discussed in Sect.
4
, using synthetic data from ab initio calculations [
15
].
Overgaard et al. examined the profiles of the LS along the Co-Co internuclear
axis which showed for all densities and for both molecules a drop around the mp
similar to that observed for the double-bridged Co
2
(CO)
8
complex and compara-
tively missing in the M-M bonded unbridged form of the complex (see earlier).
Such a similarity of the LS profiles in systems which lack a Co-Co bcp was taken
by the authors as a clear corroborating sign of the absence of a direct Co-Co
interaction in their investigated complex. Although probably correct, an “internal”
comparison with a reference system having a Co-Co bcp is required to support this
belief. The observed analogy with the results of the study by Gatti and Lasi [
14
]is
also not stringent enough, since, as discussed earlier, the shapes of the profiles
around the LS singularity can be safely compared among each other only when
traced within the same distance from the singularity.
25
Another interesting example in this area is due to Farrugia et al. [
81
]. These
authors applied the SF, among other tools, to the study of a series of three binuclear
Cu(II) coordination complexes [Cu
2
(ap)
2
(L)
2
], (ap
¼
nitrite, nitrate, and formate), using both experimental and theoretical electron
densities. The complexes contain the same centrosymmetric alxoxy-bridged
¼
3-aminopropanolate, L
25
Information about the value of such a distance is not reported in [
80
].
