pseudopotential for charged atoms ?

[Valerio Bellini]

Dear all,

I am a REALLY beginner in the business of pseudopotential generation, so
forgive me if the question is trivial.
Suppose I want to create a GTH pseudo for an element (e.g. Ni), for which
I know that in the system I want to study it is found in a well defined
valence state (e.g. Ni2+ as in NiO)..
My question is: is it possible to create a pseudo that replicates the WFs of
a Ni2+ atom? I know that in principle pseudo might be transferable, but
it might be that checking the transferability error, as explained in the
README file of the program pseudo.x, the error are too big. Adding
semicore stated might resolve this situation or not, depending on the
elements and on the orbitals of interest.. I presume.
If it is possible, do I just need to modify the occupation in the atom.dat
file for the creation of the reference all-electron WFs, and start the
fitting of the parameters, as in a standard neutral atom or is there
something else I should take care?
I guess that the parameters in psp.par relative to a non-charged
calculation are not good a starting point, since if I plot the WFs for the
Ni2+ atom empty orbitals have WFs much more contracted towards the nucleus
due to the two holes (3d8 4s0 instead of 3d8 4s2), so that I will need to
start from scratch. Is is true?
Actually the Ni2+ is just an example, I am interested in several other
transition metals or even lanthanide atoms.

cheers,
Valerio