OUTCAR: what is "total charge" and "magnetization (x)" ?

[lahaye]

Hi,

I have a system of 11 atoms, which contains 45 electrons.
At the end of the OUTCAR file I get;

Code: Select all

 total charge     
 
# of ion     s       p       d       tot
----------------------------------------
  1        0.841   1.729   0.000   2.569
  2        0.844   1.734   0.000   2.578
  3        0.827   1.698   0.000   2.525
  4        0.864   1.760   0.000   2.623
  5        0.869   1.782   0.000   2.651
  6        0.864   1.765   0.000   2.629
  7        0.861   1.768   0.000   2.628
  8        0.862   1.744   0.000   2.606
  9        1.624   3.425   0.000   5.049
 10        1.598   3.549   0.000   5.147
 11        0.622   0.113   0.000   0.735
------------------------------------------------
tot       10.675  21.067   0.000  31.741
 


 magnetization (x)
 
# of ion     s       p       d       tot
----------------------------------------
  1       -0.004  -0.021   0.000  -0.025
  2        0.004   0.032   0.000   0.037
  3       -0.003  -0.014   0.000  -0.017
  4        0.013   0.213   0.000   0.226
  5       -0.005  -0.049   0.000  -0.054
  6        0.010   0.164   0.000   0.174
  7       -0.007  -0.057   0.000  -0.063
  8        0.010   0.152   0.000   0.162
  9        0.002   0.088   0.000   0.090
 10        0.002   0.059   0.000   0.062
 11       -0.001   0.001   0.000  -0.001
------------------------------------------------
tot        0.022   0.569   0.000   0.591

In general, what does this information mean?
More specific:
1) How come the total charge (bottom right of 'total charge' table)
is 31.741, though I have 45 electrons in the system? I would
expect it to be 45....!?!

2) The OSZICAR file tells me that "mag= 1.0000".
The total in the 'magnetization (x)' table is 0.591.
Why are these numbers different?

3) I need to know the direction of the magnetic moment.
Can I get this from the magnetization table?

Thanks,
Rob.

[lahaye]

Hmm, the "total charge" and "magnetization" tables are crippled by
the Forum software (the preview was OK though).
For uncrippled tables, please have a look here:
http://surfion.snu.ac.kr/~lahaye/vasp/OUTCAR

[admin]

these 'total charges' and magnetizations are obtained form a summation over the local quantities (integration of the charge and spin densities in the 'atomic spheres' of radii r=RWIGS. As the volumes of the spheres do not sum up to the volume of the unit cell, these charges also do not sum up to the total number of valence electrons (the contribution of the "interstitial" space is missing)

[bandy]

I believe that
rho(r)=Sum_i^Nocc (psi(r)^* psi(r))
How Nocc is determined in CHGCAR file construction such that in some systems I find nonintegral charge? Is Nocc determined by the number of electrons? In case of fractional occupancy of bands, it could be a problem. I am going through the relevant portion of the code and will need some time to find out. Meanwhile if any user/head admin can comment about this, then it will be useful to forum.
<span class='smallblacktext'>[ Edited Sun Nov 19 2006, 08:42PM ]</span>

[lahaye]

admin wrote:
these 'total charges' and magnetizations are obtained
from a summation over the local quantities (integration of
the charge and spin densities in the 'atomic spheres' of
radii r=RWIGS. As the volumes of the spheres do not sum
up to the volume of the unit cell, these charges also do not
sum up to the total number of valence electrons (the
contribution of the "interstitial" space is missing)


For a structure with C, O and H atoms, I get from
"grep RWIGS OUTCAR" this:
RWIGS = 1.630; RWIGS = .863 wigner-seitz radius (au A)
RWIGS = 1.550; RWIGS = .820 wigner-seitz radius (au A)
RWIGS = .700; RWIGS = .370 wigner-seitz radius (au A)
RWIGS = -1.00 -1.00 -1.00

There is no RWIGS line in my INCAR file, because I use
"LORBIT = 12" with PAW potentials.
Do RWIGS default to -1 in that case?
What RWIGS values are then used for determining the total charges
on the atoms?

Also, is following statement from the VASP Guide still valid:

Code: Select all

With the current implementation, it is for instance
hardly possible to determine charge transfer. 

See:
http://cms.mpi.univie.ac.at/vasp/vasp/node112.html

Thanks.
Rob.

[admin]

1) RWIGS is taken from POTCAR in that case, -1 indicates that it has not been read from INCAR explicitely
2) this holds for all PW and mixed basis set calculations, because "local charges" are somewhat arbitrary in these methods, depending on the choice of RWIGS for each atomic species.
3) the directions of magnetic moments: please use the LNONCOLLINEAR=True feature of VASP to obtain them.

[lahaye]

admin wrote:
3) the directions of magnetic moments: please use the
LNONCOLLINEAR=True feature of VASP to obtain them.

Thank you for the quick answer.

To obtain the direction of the mag. moment with respect to the
structure, should I then also include LSORBIT=.TRUE. ?

The VASP Guide says for LNONCOLLINEAR=True:

Code: Select all

THIS FEATURE IS IN LATE BETA STAGE (BUGS ARE POSSIBLE)

and on LSORBIT = .TRUE., it says:

Code: Select all

...this feature...is still in a late beta stage....
No promise, that your results will be usefull!!!

Are these warnings still valid, or has this part of the code become
as mature as the rest of the VASP code?

Thanks,
Rob.