comparison betweem EXP and Compuation
On Thu, 28 Aug 1997, Dr. Adel El-Azhary wrote:
> Dear CCL members:
>
> I sent an e-mail to ccl yesterday but it looks like that it was not
received.
> So my apologies if you receive it twice. My question is that I am doing
some
> calculation on some small molecules for which effective, substituted,
average
> or equilibrium experimental geometries are available. So which of these
is the
> best to be comapred to the calculated geometry.
This is a good question. Too often, we compare computed molecular
structure with experimental value without concerning whether they are
comparable.
First of all, WHAT IS A MOLECULE? (see, Gribov, L. A.
J. Mol. Struc. 1993,300,415). There is one quite accuarte definition:
it is the smallest particle of the substance made up of eletrons and
nuclei of a definit mass. This definition can be expressed mathematically
by the schrodinger equation. However, we must reduce to classical
expresions if we want to compare calculated structure with a real
experiment. We have to use classic models, different models for
different experiments.
Let us start from several experimental procedures obtaining
the information of molecular structure. Two kinds of operational bond
lengths R(0) and R(s) may be derived directly from rotational and
rotation-vibrational spectroscopy. The R(0) structure is derived from
the ground state rotational constants B(0) directly, or more usually with
assumptions about some of the structural parameters. When the rotational
constants for various isotopic species are observed, the substituted
structure R(s) can be determined by Kraitchman method. When vibrational
(both harmonic and anharmonic) properties are recorded, one may obtained
equilibrium structure R(e) with some approximation. Only is the R(e)
really comparable with the value from quantum mechanical calculations,
which give the distance between equilibrium nuclear position R(e).
X-ray or neutron diffraction experiments give the average nuclear
positions at thermal equilibrium, and from which the average nuclear
position
the internuclear distance R(alpha) is defined. When we refer to the
structure at 0 K, we get R(z) structure, which may derived from
rotational and rotation-vibrational spectroscopy.
Gas phase electron diffraction experiments give the thermal
average value of internuclear distance: R(g) structure.
The relationship and interconversion of those strcutures may be
formulated by classical approximation (see reference cited in the end
of this message). MM3(96) version will print out all those structures.
If you are using molecular mechanical method, MM3(96) is a good option.
you can compare your calculated results with comparable experimental
data. Other molecular mechanical packages, as I know, do not realized
the difference. If you are talking about QM calculation, of course,
experimental equilibrium structure is first choice. However, if it is
not available, R(s) structure is close to R(e). I have derived a formula
R(s) =[R(z) + R(e)]/2 . R(z) comes third, then R(alpha), and finally
R(g). Note that QM results are base sets and correlation dependent, and
convergence should be considered in the comparison.
Refernce:
a) Kuchitus, K. in Accurated Molecular Structures, Ed Domenicano, A et al.
Oxford Science Publication, 1992.
b) Ma, B.; et al. J. Physical chem. 1996, 100, 8763.
J. Am. Chem. Soc. 1997, 119, 2570.
J. Mol. Struc. In press ( the issure dedicated to Dr.
Kuchitsu).
=================================================================
Dr. Buyong Ma buyong -8 at 8- ibmnla.chem.uga.edu
Computational Center for Molecular Structure and Design
Department of Chemistry
University of Georgia
Athens, Georgia 30602 USA Voice (706) 542-2044
=================================================================