[Pw_forum] comparing gaussian03 with pwscf

Eduardo Ariel Menendez Proupin eariel99 at gmail.com
Wed May 7 23:40:05 CEST 2008


Hi,
Bader anaylisis is well founded, but when I read of it I did not like two
things. One is that Bader atoms have strange shapes sometimes. Well, this is
subjective, and easier to understand than the collapse of the wavefunction.
The second that it is still a partition of the space, and it depends more on
the size of the atoms than on the charge transfer. As the valence charge
density changes really little, the total charge density changes
proportionally even less. I suspect that the atomic boundaries change little
during the self-consistent calculations, and if one performs a Bader
analysis with the self-consistent wave functions or with the charge density
obtained from the superposition of the the neutral atomic charge densities,
the assigned atomic charges are the almost the same. Hence an oxygen can
have a charge near -2 with and without self-consistency, but this is due to
the size of the atoms, and not its affinity to trap electrons or the
polarization of the bonds. I have not made Bader analysis because because it
is not implemented in the codes that I have worked with, and I may have
missed some important article. Voronoy assigned charge depend on the atomic
radius used: covalent, ionic. One can also obtain the proyected densities of
states and integrate them to the Fermi level to obtain the charges. I did
that once, and I chose whether to use covalent or atomic radii  guided by
the "principle" that oxygen must be negatively charged.

I would make some effort to study  this topic if I knew what is the real
importance of quantifiying the charge transfer.
Regards
EM

*Stefano de Gironcoli*
*Wed May 7 10:36:27 CEST 2008*
how to carve an atom from a molecule or a solid is a long standing problem
with no clear solution basically due to the fact that it is an ill defined
problem as .however I think that Bader analysis, based on defining atomic
volumesfrom topological properties of the charge density rather than
justnearest-neighbhor interatomic distances, give results that are more
consistent with chemical intuition.I remember for instance that in GaAs a
description based on Voronoy

polyhedra would assign higher charge to Ga than to As just because the
As is "fatter" than Ga and a fraction of the charge that one would
visually assign to it was in the Ga territory...  An analysis based on
Bader decomposition was instead consistent with the expectation.
Unfortunately Bader analysis is not presently implemented among QE
post-processing tools.
Other options to try to assign charges to atoms are Lowdin population
analysis (it however depends on the  choice of atomic functions used) or
construction of maximally localized Wannier functions that would help
splitting the charge density in contributions with well identified
origin...   both these two schemes are available using QE and/or
wannier90 code.

hope this helps,
stefano



2008/5/6 Eduardo Ariel Menendez Proupin <eariel99 at gmail.com>:

> If there is doubt on the approximations for lantanides, I suggest to make
> some tests with something more "classic", such as NaCl or MgO and then
> compare PWSCF vs GAUSSIAN with complete basis sets.
>
> One question? What is the useful information provided by the knowledge of
> the charge transfer? Can it be correlated to any measurable property
> (reactivity, entropy, adsorcion enrgy, ....) or with any measurable and
> difficult to calculate property?
>
> Some years ago I calculated and reported some charge transfer and said
> something about the ionicity. To know better, I calculated the charge
> transfer in NaCl. For this, I integrated the charge density in the Voronoi
> cells of Na and Cl, using the self consisten charge density and the
> superposition of atomic densities. The difference is the charge transfer or,
> more precisely, the charge that crosses the boundary of the Voronoi cells.
> To my surprise, the charge transfer in NaCl is 0.1 electron, and not 1
> electron as the textbooks suggest. I also tested it with MgO (0.2 instead of
> 2). So, where is hidden  the truth of the ionic bond?
>
>
>
> --
> Eduardo Menendez
>



-- 
Eduardo Menendez
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://www.democritos.it/pipermail/pw_forum/attachments/20080507/d87847a5/attachment-0001.htm 


More information about the Pw_forum mailing list