Negative and local-maximum charge in OUTCAR file for small and big RWIGS radius respectively

[ledssiul]

Dear Sirs,

I have calculated, for Co_1 (d8 s1), the s, p, d, and total charge as a function of Wigner-Seitz (RWIGS) radius using a simple INCAR, a big cubic box (3*18), GGA, and just Gamma point. I am using vasp.5.2.2

Here the INCAR file I am using.

............
ISYM = 0
ISPIN = 2
VOSKOWN = 1
ISMEAR = 0 (Gaussian)
SIGMA = 0.01

NPAR = 1
LORBIT = 0 (I tried four different values : 0,1,2,5)
RWIGS = (I tried the interval: 0.1, ... ,3.0 [angstroms])
...........

Looking in the OUTCAR files everything is well converged, also the occupation for each band is correct, and there is NO differences in all the OUTCAR files, except in the "total charge" section (and of course the "magnetization (x)" and "volume of typ" sections).

Please, find bellow the results for the "total charge" I got as a function of RWIGS radius. I am showing just two digits for simplicity. I am not showing here the section "magnetization (x)" because can be obtained directly from the "total charge" data.

========================================
========================================
========================================

total charge

# RWIGS s p d tot
----------------------------------------
0.1 -0.09 0.0 5.38 5.29 <-- 1)
0.2 -0.09 0.0 5.37 5.28
0.3 -0.08 0.0 5.38 5.29
0.4 -0.07 0.0 5.39 5.32
0.5 -0.06 0.0 5.45 5.38
0.6 -0.03 0.0 5.56 5.52
0.7 0.00 0.0 5.75 5.75 <-- 2)
0.8 0.03 0.0 6.02 6.05
0.9 0.09 0.0 6.31 6.40
1.0 0.15 0.0 6.59 6.75
1.1 0.23 0.0 6.83 7.06
1.2 0.31 0.0 7.02 7.33
1.3 0.39 0.0 7.16 7.56
1.4 0.48 0.0 7.26 7.75
1.5 0.57 0.0 7.35 7.92
1.6 0.65 0.0 7.41 8.06
1.7 0.72 0.0 7.46 8.19
1.8 0.79 0.0 7.51 8.30
1.9 0.86 0.0 7.54 8.40
2.0 0.91 0.0 7.56 8.48
2.1 0.96 0.0 7.58 8.54 <-- 3)
2.2 1.00 0.0 7.58 8.58
2.3 1.04 0.0 7.57 8.61
2.4 1.07 0.0 7.56 8.63
2.5 1.09 0.0 7.54 8.64 <-- 4)
2.6 1.11 0.0 7.51 8.63
2.7 1.13 0.0 7.48 8.62
2.8 1.15 0.0 7.44 8.59
2.9 1.16 0.0 7.39 8.56
3.0 1.18 0.0 7.34 8.52 <-- 5)
========================================
========================================
========================================

Here some comments about the unexpected results (<-- #) for different RWIGS radius:

[e_s == s-charge]
[e_d == d-charge]
[e_t == total charge]

<-- 1) RWIGS = 0.1 [angstroms]

a) e_s = -0.09 (e_s is NEGATIVE)
b) e_d = 5.38
c) e_t = 5.29 (e_t < e_d !!!)

<-- 2) RWIGS = 0.7 [angstroms]

a) e_s = 0.00 (the curve for e_s crosses the zero level)
b) e_d = 5.75
c) e_t = 5.75 (there is a crossing in the curves for e_d and e_t)

<-- 3) RWIGS = 2.1 [angstroms]

a) e_s = 0.96
b) e_d = 7.58 (here, there is a MAXIMUM in the curve for the e_d)
c) e_t = 8.54

<-- 4) RWIGS = 2.5 [angstroms]

a) e_s = 1.09
b) e_d = 7.54 (this value is LOWER than in RWIGS=2.1)
c) e_t = 8.64 (here, there is a MAXIMUM in the curve for the e_t)

<-- 5) RWIGS = 3.0 [angstroms]

a) e_s = 1.18
b) e_d = 7.34 (this value is even lower than before)
c) e_t = 8.52 (now I have LESS total charge)


I was expecting to have the d-charge --> 0 for RWIGS --> 0, and also, I was not expecting to have the local MAXIMUM values in the e_d and e_t curves since I am dealing just with ONE ATOM.

I tried Co_2, Co_43 and Co_bulk with similar results for small RWIGS radius.

The question is, am I misunderstanding the real meaning of RWIGS? Is there a reason to get such kind of estrange behavior? Maybe is too simple my INCAR file and I have to setup an extra flag?

Thanks in advance for any clarification.

Best regards,

Luis

PS. Last week I posted a similar thread but until now nobody have replayed. Maybe I was not enough clear. That is the reason why I am trying this new thread.

[ledssiul]

Please, could someone help me or point me in the right direction on how to correctly understand these strange behaviours? Is it clear enough my post?

[ledssiul]

I really would appreciate any explanation about the minimum and maximum behaviour explained in the original thread.

admin?

Is this a BUG in the code?

Thanks in advance!