I have calculated the Bader charges for a H2S/ZnO structure using VASP and USPP. I have run the calculation with the default NG(X,Y,Z)F values as well as with 2x and 3x the values (ie. increasing the NG(X,Y,Z)F values). However some of the charges are still not converged using 3x the values. The difference between some of the charges is as large as 1.36e.
Do you have any suggestions why the charges are not converging?
My INCAR file for the 3x is:
NPAR = 8
IALGO = 38
NELM = 400
ISPIN = 2
PREC = ACCURATE
ENCUT = 400
ISMEAR = -5
VOSKOWN = 1
NGXF = 1152
NGYF = 1152
NGZF = 216
VASP No convergence with increasing NG(X,Y,Z)F values
Moderator: moderators
Re: VASP No convergence with increasing NG(X,Y,Z)F values
A difference of over 1 e is way too much. Have you included core charges?
Re: VASP No convergence with increasing NG(X,Y,Z)F values
As I am using ultrasoft pseudopotential I have not included core charges
Re: VASP No convergence with increasing NG(X,Y,Z)F values
The core charges are essential; see: http://theory.cm.utexas.edu/bader/vasp.php
Re: VASP No convergence with increasing NG(X,Y,Z)F values
Thank you for the extra information. I had read this previously but I didn't think the core charges could be written when using USPP?
Re: VASP No convergence with increasing NG(X,Y,Z)F values
I think you are correct, this information is only written in the PAW potentials.
Re: VASP No convergence with increasing NG(X,Y,Z)F values
sir, what is the meaning of 2x and 3x values ?
Re: VASP No convergence with increasing NG(X,Y,Z)F values
double and triple the grid point density