NMR shielding calculations- Summary
Dear CCLers,
I've posted the following the question :
>I am performing NMR shielding calculations (of 7Li) but it seems to me
>that the results are very dependant of the molecular geometry and that a
>small change in the structure induces important changes for the chemical
>shift.
>
>Have you more informations about this assumption?
>More generally, what is the accuracy of such calculations?
>
Here are the messages I received on the topic :
----------------------------------------------------
A tutorial/review on this topic entitled "The Ab Initio Computation of
Nuclear
Magnetic Resonance Chemical Shielding" by D.B. Chesnut can be found in
Reviews
in Computational Chemistry, Volume 8, VCH Publishers, 1996, Chapter 5. It
addresses the issue you raised.
Kenny Lipkowitz <lipkowitz-: at :-chem.iupui.edu>
----------------------------------------------------
Pascal,
Since the experimental NMR shielding tensor as expressed as a chemical
shift is quite sensitive to geometries and even conformations (hence its
utility as a diagnostic) the calculated is also dependant upon these
influences. I expect vacume calculated NMR chemical shifts to be within 5
% of experimentallly obtained chemicals shifts from solid or liquid phase
experiments for normal basis sets. In many cases one obtains considerably
better agreement than this with values coming within 2 or 3 %. See the
references here for examples and the references therein for background
and further examples.
1. Journal of Chemical Physics. 104 (14), 8 April 1996.
2. International Journal of Quantum Chemistry, Vol. 57, 199-205 (1996)
Thank-you
Robert A. Kirby
BioChem Therapeutiques Inc.
275 Armand-Frappier Blvd.
Laval, Quebec,
H7V 4A7
#514-978-7914
fax 514-978-7777
kirbyr-: at :-biochem-pharma.com
----------------------------------------------------
Hi!
I saw your note on CCL, but wasn't quite sure if I had anything
significant to contribute with. However, I guess that although the
Bochum group has been investigating Li-compounds more than we have,
they haven't looked too much into the geometry dependence (apart from
a paper in which distinct conformers where distinguished by quantum
chemical means).
We did some work on C2H3Li a while ago (CPL 226, 1 (1994)), but there
our goal was both the shieldings and spin-spin couplings. In general:
As all shieldings also 7Li are dependent on geometry. The
problem with Li-compounds is that whereas you normally can use an MP2
geometry optimization as a starting point, this might not be
adequate for Li-compounds due the often very weak and long
bonds. Experimental geometries are often troubled by the very
large geometrical changes of Li-compounds in solvents (and it often
polymerizes as well), making such geometries difficult to use.
Thus, as I do not know the purpose of your investigation, it is
difficult to make any clearcut recommendations. If purely theoretical,
I would have done everything consistently, that is, optimize the
geometry at the same level of theory as you calculate the
shieldings. If comparison with experiment is the goal, use
experimental geometries where available, although I haven't, sitting
here thinking it through, decided quite what to do if the experiment
has been performed in solvent. I dielectric continuum model might not
be appropriate in such cases.
Being a purist, I would go for internal consistency in the
calculation (geometries and shieldings), but it is quite noteworthy
that it actually was the shieldings that destroyed our spectrum the
most compared to experiment in the CPL paper and not the spin-spin
couplings as you would normally expect.
It might be an idea to hear with Dr. Ulrich Fleischer or
Prof. W.Kutzelnigg as well for some recommendations for prefered
geometries.
Good luck!
Kenneth
_______________________________________________________________________________
Kenneth Ruud, Ph.d.-student in Chemical Physics at the Department of Chemistry,
University of Oslo, Norway. E-mail: kenneth-: at :-dalton.uio.no
I don't know what the computer language of the year 2000 will look like, but
I know it will be called FORTRAN.
_______________________________________________________________________________
----------------------------------------------------
Hi Pascal,
there was a very similar question recently in the CCL, about the use of
experimental vs. optimized geometries. So you might wish to search the CCL
archives.
About the sensitivity to the geometries: Yes, NMR shieldings are VERY
sensitive to the geometries. In fact, I think that is part of why
experimentalists
like the method. I have seen something on shielding gradients recently in
some review (the shielding gradient would be the geometry dependency taken
in a quantitative manner). I can't exactly remember wish review it was,
it might be in one of the annual reviews by C. Jameson, in Annual Reports
on NMR Spectroscopy. Jameson has another review in Chem Rev 1991, 91, 1375
that contains, among other things, the temperature dependence of the (measured)
shift/shielding. This temperature dependence is likely also due to the change
of equilibrium geometry with T, at least to a large degree.
> More generally, what is the accuracy of such calculations?
This is indeed a VERY general question. The answer depends on what you are
really
doing. E.g., if you have one of the highly correlated methods, e.g. MPn or
Coupled Cluster
(the latter by Gauss et al.), then you get quantitative results with large
enough
basis sets. Else you might miss some correlation effects. Further, as mentioned,
you need a good basis set, otherwise this gives you also an error. Finally,
what you are probably calculating is the zero-pressure, zero-density limit of a
gas-phase experiment. The experiment you are comparing to is likely done
under finite
pressures and temperatures, and in a condensed phase: all this can give
considerable
shifts.
See one of the various reviews for more details on all of this.
Yours, Georg
--
==============================================================================
Georg Schreckenbach Tel: (Canada)-403-220 8204
Department of Chemistry FAX: (Canada)-403-289 9488
University of Calgary Email: schrecke-: at
:-zinc.chem.ucalgary.ca
2500 University Drive N.W., Calgary, Alberta, Canada, T2N 1N4
==============================================================================
----------------------------------------------------
Yes shielding is dependent on bond length. Use an optimised structure for
gaseous molecules, experimental for solids. Take particular care if the bond
length you are using is less than the equilibrium value and the molecule does
not approach the combined atom smoothly. If there is a crossing of the ground
and a repulsive state, the shielding values can go to very large positive and
negative numbers.Shielding is also very dependent on basis set and gauge origin.
Chemical shifts or relative shieldings are generally less sensitive.
Cynthia Jameson and co-workers have done a lot of work on nmr shielding surfaces
that is the change of shielding with geometry.
Elaine A. Moore
e.a.moore-: at :-open.ac.uk
----------------------------------------------------
Many thanks to all.
Pascal
*****************************************************************************
Pascal HEBANT
Laboratoire d'Electrochimie et de Chimie Analytique
Ecole Nationale Superieure de Chimie de Paris
11 rue Pierre et Marie Curie
75005 Paris FRANCE
tel: 33 (1) 44 27 66 94 fax: 33 (1) 44 27 67 50
http://alcyone.enscp.jussieu.fr/Pages/LECA/Electrochimie.html
*****************************************************************************