ABINIT Discussion Forums

Hi everyone,

I have a problem with Bader charge integration in my PAW calculation.

I performed Bader charge analysis for simple system (rutile MgH2), using PAW datasets from ATOMPAW website, _PAWDEN valence charge density files obtained with prtopt .eq. 2 and site-supplied core-charge atomic files for AIM.

The calculated value for Mg Bader charge:

Different density contributions: Core (only spherical part) and the rest
9.95139224 8.86903873
Bader charge: 1 18.82043097

This is obviously overestimation. Typically, all-electron LAPW Bader analysis result is 10.36 e !!!
Also only 5 BCP's were found around Mg. Mg is surrounded with 6 hydrogen atoms in octahedral arrangement.

What is wrong here? Is there a need for core-charge files in Bader analysis when PAW method is used?

Best regards,

Nikola

So, I took a look again at the problem, and I managed to at least partially solve it...
It turned out that the reason for the large core contribution are really core density files, ill-suited for the most of PAW datasets available...

PAW dataset for Mg includes in valence panel all except 1s electrons, while supplied Mg core file contains 10 core electrons (all except 3s), which explains the AIM charge integration result of 9.951 for core contribution, and 8.86903873 for valence one.
So, I made new core charge file for Mg using FHI98pp with only 1s core electrons contribution, and now the Bader charge is 10.357 (core is ~2.00, as expected), in excellent agreement with FP-LAPW result (10.36).
However, AIM still manages to locate only 4 BCP's in vicinity of Mg, not 6, as expected...
Does anyone have an idea what else can be wrong?

Regards,

Nikola

input files:

aim.in
----------------
surf 1
crit 2
atom 1
gpsurf 1
irho 1

inpt 50
ntheta 32
nphi 48

atrad 1.7
radstp 0.03
ratmin 1.0
maxatd 8.0
maxcpd 5.0
lgrad2 1.0d-4
lstep2 1.0d-4

FHI98PP ini file for Mg:
---------------------------
12.00 1 4 8 1.80 : z nc nv iexc rnlc
1 0 2.00 : n l f
2 0 2.00
2 1 6.00
3 0 2.00
3 1 0.00
2 t : lmax s_pp_def