My Community

hi,
I am calculating electron charge density difference for a system comprising of a metallic surface and an adsorbate. To do so, I am calculating the electron charge density of the adsorbate+surface, the surface, and the adsorbate systems using the same supercell size, INCAR, and KPOINTS. While the CHGCAR for the adsorbate+surface and surface only have the same dimensions (NGX, NGY, and NGX), the CHGCAR for the adsorbate only has slightly different number of grid points. Why is that? Although I do recognize that we can fix NGx,y,z in the INCAR file for all three runs, I don't understand why the program does not generate the same grid for the three cases given the same INCARs and KPOINTS.

Also, to perform this kind of analysis (i.e. the electron charge density difference) do we have to work with more kpoints just like in the case of generating accurate DOS?

I would appreciate if someone would comment.
Thanks.

The answer to the second question is no you don't have to add more kpoints.

I am not sure of the answer to the first question.

The size of the grid is IMHO determined from the combination of the PREC, ENCUT and ENAUG tags according to the table in the manual
http://cms.mpi.univie.ac.at/vasp/vasp/PREC_tag.html

If you use older versions of PREC (Low, Med, High) then the grid is related to the cut-off for the augmentation charges which is read from POTCAR. If your substrate and adsorbate have different ENAUGs in POTCAR (which they most probably do), this will give you different grids (even when you set ENCUT in INCAR).

If you set PREC to Normal, Single or Accurate, then the grid should differ only if you don't specify ENCUT in INCAR since the grid is directly controlled by ENCUT. If you don't have ENCUT in INCAR (sounds like a bad thing to do), ENCUT is determined from POTCAR, specifically from the ENMAX values (which will again differ for the adsorbate and substrate).

The point is that even the same INCAR can lead to different grids when older PREC settings are used or if ENCUT is not set at all.

Thank you for your response. I found it very helpful.