The ABINIT Users Mailing List ( CLOSED )

[m_mikami@hf.rim.or.jp]

Dear David,

> I'm attempting to simulate the dielectric properties of Fe doped TiO2. I'm
> using the Teter "extended norm conserving" psps for Ti and O. Undoped this
> gives reasonable results. However, when doped (replacing a fraction of Ti
> with Fe in a supercell), response functions will not converge for any choice
> of available Fe pseudopotential. Geometry relaxation does work. As an
> extended norm conserving psp does not exist for Fe, are there any general
> rules for mixing different kinds of psps? Should I look into creating a
> custom psp?

I have a different opinion, rather than the issue on the psps.

For the computation on dielectric properties, 
the model should be semiconductor/insulator. 
I did not understand your model exactly (lack of the input...), 
but still I could suppose your model (Fe-doped TiO2) would be 
"metal," i.e. little/no energy gap near the Fermi level... 
(You might want to draw the DOS to check this.)

So, I am afraid that your computation on Fe-doped TiO2
for dielectric properties might not produce anything...

Regards,
Masayoshi