Siesta

Net Charge for Defect Formation Energy

Asked by Charith Ravana DeSilva

Dear all,

I am currently trying to calculate the defect formation energy of GaAs, specifically with a q=-3 Ga-vacancy.
To implement this charge defect, I am removing a Ga-atom from the pristine system and using the NetCharge function to add a -3 net charge to the system (i.e. NetCharge -3.0).
I am then doing a CG-relaxation of this defect-GaAs supercell to get the total Energy.

In the manual, there is a warning against using the NetCharge function for anything other than molecules/atoms. So my question is will this idea work for dealing with charged defects in supercells of GaAs or any other crystal system? Or is there
 a better way through Siesta to deal with charged defects.

Thank you for your help.

Question information

Language:
English Edit question
Status:
Solved
For:
Siesta Edit question
Assignee:
No assignee Edit question
Solved by:
Alberto Garcia
Solved:
Last query:
Last reply:
Revision history for this message
Alberto Garcia (albertog) said : 		#1

Hi,

The issue is non-trivial, and there are several methods available to deal with it. You can see the most recent attempt here:

Chagas da Silva, Mauricio, Michael Lorke, Bálint Aradi, Meisam Farzalipour Tabriz, Thomas Frauenheim, Angel Rubio, Dario Rocca, and Peter Deák. 2021. “Self-Consistent Potential Correction for Charged Periodic Systems.” Physical Review Letters 126 (7): 076401. https://doi.org/10.1103/PhysRevLett.126.076401.

That paper also contains references to previous works, which you should read for completeness.

In Siesta we are currently working on a scheme similar to that of Freysoldt et al, but the code is not ready yet.

Revision history for this message
Brian Leininger (bleininger) said : 		#2

Alberto,

I'm newer to Siesta, but I was wondering how Siesta differentiates between molecule/atom and periodic systems. If you are trying to calculate the total energy of a charged system with the goal of manually calculating defect formation energies (DFEs), then by setting a NetCharge = +- n, would there be any corrections to the energy that Siesta inherently applies?

Also, from my understanding, Freysoldt's method incorporates a correction to:

1. electrostatic interaction between charged defects and images
2. potential alignment correction for maintaining an overall neutral system.

(Gathered from Freysoldt and Broberg-PyCDT)

That said, from the way the Siesta manual reads it seems as if NetCharge along with the SimulateDoping = true would account for the item 2. Is that a correct way of thinking?

Thank you,

Brian Leininger

Revision history for this message
Best Alberto Garcia (albertog) said : 		#3

(I have provided an answer to this question in the other thread opened for it).

Revision history for this message
Charith Ravana DeSilva (charith44) said : 		#4

Thanks Alberto Garcia, that solved my question.