Problem about zero bader charge!

[zhantian]

Dear All:

I have met the zero charge problem when I calculate the boron-nitride with VASP package. Based on previous topics, admins suggest to add the extra charge with the modified VASP. However, can I add some extra charge with a "crude" scheme, i.e., to add point charge (or a charged ball with smooth cutoff) to the CHGCAR? If it is ok, is there something need to be taken care?

[graeme]

I don't think this can be fixed in a simple way that does not have information about the correct shape of the core charge. If there is no Bader maximum at the atomic center, the location of the minimum between atoms has been lost. Whatever core charge you add will be what determines the location of the Bader dividing surface. It could make sense to add the core from an all electron calculation, but the best solution is to add the frozen core that was used to construct the PAW potentials. It may be possible to get this in the current version of vasp, but if not, they say it will be in vasp 4.5, which is due early next year.

Anyways, we recognize that the core charge is the most important limitation for using the bader analysis with vasp, and we'll try to get something working as soon as possible.

[zhantian]

Thanks for admin. Today, I have recompiled the VASP code (ver 4.6.28) based on the instruction on http://theory.cm.utexas.edu/vtsttools/bader/vasp.php

But I find that no any PARTIAL CORE CHARGE has been added to CHGCAR for the normal POTCAR (PAW). If I adopt the PAW_h potential (with a high cutoff energy of 700eV!), about 3.92191 e can be obtained for CHGCAR. Therefore, I obtained a charge transfer about 2.0 e for bulk c-BN, significant larger than results from DMOL3 or CASTEP package. Therefore, is there something wrong with my work?