Why the CHGCAR is negative?

[yiranwang]

Dear friends,
I want to do a research about a system of KI@SWNT. I calculate the KI crystal by PAW,and after the static calculation, I read the CHGCAR by Xcrysden, I find that the min grid isosurface value =-48, the max grid isosurface value=360. What's more, the CHGCAR has negative numbers. I thought the negtive value is too large, and this is not right. I haved tried all the PAW　pseudo potentials, such as LDA, GGA_pbe, GGA_91. The result are similar. After I use the USPP, there is not such problems, that is to say,the min grid isosurface value is near to ZERO. Why is it? Is the result of PAW dependable? The negative charge is around the center of I.

INCAR # for static
SYSTEM=K I
ISTART=0
ICHARG=2
PREC=Normal
ENCUT=350
ISMEAR=-5
SIGMA=0.2
#ISPIN=2
ISIF=2
EDIFF=1E-04
#EDIFFG=-0.02
IBRION=-1
NSW=0
#LORBIT=11
#MAGMOM=1.0
LWAVE=.F

POSCAR
K I
1.00000000000000
4.0929999999999999 0.0000000000000000 0.0000000000000000
0.0000000000000000 4.0929999999999999 0.0000000000000000
0.0000000000000000 0.0000000000000000 4.0929999999999999
1 1
Direct
0.5000000000000000 0.5000000000000000 0.5000000000000000
0.0000000000000000 0.0000000000000000 0.0000000000000000

0.00000000E+00 0.00000000E+00 0.00000000E+00
0.00000000E+00 0.00000000E+00 0.00000000E+00



<span class='smallblacktext'>[ Edited ]</span>

[Danny]

Short answer:
You are using PP, what you find in the center is the actual core. Since protons have the oposite sign of electrons you get a negative chargedensity for heavy elements.

Long answer:
Basically the same, but PP means you replace the core potential with samething that mimics the properties exactly, but is easier to calculate. If you do a single atom calculation, make an averaged radial distribution of the density, and then integrate you will find the correct number of electrons (given you take the consequences of Periodic boundaries into account and numerical accuracy) for neutral and positive ions, even though the radial distribution will most of the time not be stricktly declining, and the core value can be quite negative (Cl^+5 for example) ... I has a Z=53 then -48 does not look that unreasonable knowing that there are >40e in the "core"

Danny