The ABINIT Users Mailing List ( CLOSED )

[Xavier Gonze <gonze@pcpm.ucl.ac.be>]

Dear Q.M. Hu,

> > Dear Abinitioners,
> > I'm very interested in the alchemical mixing of psps feature implemented
>
> in abinit 3.4.3. I have done some tests on it, and encontered some problems.
>
> > 1. For disordered Ti(50%)Zr(50%) and Ti(50%)Nb(50%), it gives reasonable
>
> equilibrium lattice constants, but the calculated heats of formation is
> considerablely large (about 9eV for TiZr and 10.2eV for TiNb, being
> endothermic). I have no idea about the reliability of the calculated heats
> of formation.
>
> > For disordered Ni(50%)Al(50%) and Ti(50%)Ni(50%), it gives unphysical
>
> equilibrium lattice constants and total energies. It may be reasonable
> because  disordered TiAl and TiNi do not exist at low temperature.
>
> > Therefore, can I conclude that the alchmeical mixing of psps is only
>
> applicable for the elements with similiar electronic structure, or say, for
> the elements that can form disordered solid solution?

The alchemical mixing of pseudopotentials is a "numerical trick"
to interpolate
between elements. When the two pseudopotentials are quite similar,
the results can be interpreted as giving a rough idea of the behaviour
of the corresponding alloy. Even in this case, it should be considered
as a starting point for studies using more accurate representations,
like disordered supercells, or a linear response approach, generating
an Ising model (in the spirit of de Gironcoli, Baroni and coworkers
a few years ago).

I would place little confidence in the simple mixing approach (that is,
not followed by a linear-response correction)
for mixing pseudopotentials with DIFFERENT ionic charge - zion -  (e.g : 
Ti-Nb ; Ni-Al, Ti-Ni pairs). Even for the Ti-Zr pair, one should make 
sure that the pseudopotentials are sufficiently similar.

Still, in the case Ti-Zr, supposing you have used
similar pseudopotentials for Ti and Zr, I wonder why
the heat of formation is so bad !

Examples of systems for which the alchemical approach has been successfully
applied are III-V semiconductors alloys (e.g. Gax Al(1-x) As)
and ferroelectrics alloys (e.g. Bax Sr(1-x) TiO3, see work
by Ghosez and coworkers), for which
the similarity of elements (and corresponding pseudopotentials)
has been checked.

Yours,
Xavier