What can we get from the Bader approach?

[chengdaojian]

We want to use ab initio in order to characterize the ionic component of bonds
between C and Ni atoms in the vicinity of a Ni surface. The purpose is to
incorporate the information into a classical molecular dynamics model in terms
of Coulomb interactions between fractional point charges at the centre of the
atoms. Can we get the appropriate information using the Bader approach to
charge distributions?

We can calculate the Bader charge of an isolated C atom (which is trivial) and
that associated with Ni atoms in pure Ni surfaces. We can also repeat the same
ab initio calculation when the C is adsorbed at the Ni surface. Expectedly,
when we do so, we get different numbers. What is the relationship between such
numbers and the point charge ionicity model we would like to use?

Charge transfer between C and Ni is estimated as the Bader charge of the
isolated C and Ni systems and the same systems when bound. However, by
putting C and Ni together, one necessarily modifies the topology of the "zero
flux" surfaces. So, is the charge difference an effect of changing the topology
(in which case Coulomb interaction should not be expected) or is it an effect
of transferring charge from Ni to C (in which case Coulomb forces should be
expected)?

All of your answers are appreciated.

Thank you very much.

With Regards
Cheng Daojian
chengdaojian@gmail.com
2009-11-21
Université Libre de Bruxelles, Belgium

[graeme]

I think that the Bader charges should be useful for point charge models. But there is, of course, no formal connection between the Bader charges and what should be optimal in an empirical model. Several years ago I did some work comparing DFT to point charge models for MgO diffusion on MgO(100). There, the Bader charges provided a qualitatively better potential. For details, see:
http://theory.cm.utexas.edu/henkelman/p ... 115437.pdf