ABINIT Discussion Forums

Dear All,

I want to study the effect of doping in the electronic structure of a certain compound. Until now, I have used PAW pseudopotentials because the ecut needed for them (and for my system) is acceptable. However, I read that doping via mixalch can only be performed with Norm Conserving (NC) pseudopotentials. The problem is that with NC pseudopotentials my system requires around ecut=1400 eV to ensure a convergence of 0.1 eV in the energy obtained in SCF calculations. I think that I need a much better accuracy. What do you recommend?

Thank you very much.

Nahasamapetilah

Dear Nahasamapetilah,
Indeed VCA does not work with PAW in Abinit, you have to use NC pseudos.
Which NC pseudos did you use? If you are not using them, I advise you to use the ones from the recent Pseudodojo table (v0.4):
http://www.pseudo-dojo.org/
And the related paper:
https://www.sciencedirect.com/science/article/pii/S0010465518300250?via%3Dihub
These pseudos usually require lower ecut depending on what you want to calculate (30-35 Ha).
Best wishes,
Eric

Dear ebousquet,

I am using the NC pseudopotentials of the first table of https://www.abinit.org/psp-tables, which is the official page of Abinit. Are these the same than the ones you recommended me (http://www.pseudo-dojo.org)?

Thank you for your help. I am looking forward to your answer.

Nahasamapetilah

Dear Nahasamapetilah,
The pseudos from Abinit and Pseudodojo websites are supposed to be the same, you have more exchange correlation functionals in Pseudodojo (as of today).
Your system might be tricky and you need large ecut? Be careful that looking at the total energy convergence only might induce you larger ecut than the one you need for the physical properties you want to look at (which are usually related to energy differences).
Best wishes,
Eric

Dear ebousquet,

I tried with the pseudopotentials that you recommended me and it worked well. Finally, I reached a good precision after the convergence studies. Thank you.