Is my Bader result reasonable?

[andyfan]

Hi, all

I have recently performed the Bader charge analysis on a p(3*3) four-layer Ni(111) slab With VASP code. After I obtained the CHGCAR and CHGCAR_sum (the sum of AECCAR0 and AECCAR2), I typed "bader CHGCAR -ref CHGCAR_sum". The result in the ACF.dat file is listed as follows.
# X Y Z CHARGE MIN DIST ATOMIC VOL
--------------------------------------------------------------------------------
1 0.0000 0.0000 0.0000 10.0192 1.1738 37.9654
2 0.0000 2.4975 0.0000 10.0253 1.1767 38.0529
3 0.0000 4.9950 0.0000 10.0368 1.1806 38.2234
4 2.1629 -1.2487 0.0000 10.0254 1.1806 38.0417
5 2.1629 1.2487 0.0000 10.0374 1.1741 38.2369
6 2.1629 3.7463 0.0000 10.0180 1.1768 37.9488
7 4.3258 -2.4975 0.0000 10.0368 1.1767 38.2343
8 4.3258 0.0000 0.0000 10.0173 1.1811 37.9610
9 4.3258 2.4975 0.0000 10.0255 1.1741 38.0467
10 0.0000 0.0000 6.0426 10.0186 1.1764 37.9966
11 0.0000 2.4975 6.0426 10.0248 1.1731 38.0717
12 0.0000 4.9950 6.0426 10.0367 1.1794 38.2643
13 2.1629 -1.2487 6.0426 10.0253 1.1794 38.0826
14 2.1629 1.2487 6.0426 10.0368 1.1767 38.2611
15 2.1629 3.7463 6.0426 10.0183 1.1773 37.9892
16 4.3258 -2.4975 6.0426 10.0362 1.1731 38.2525
17 4.3258 0.0000 6.0426 10.0170 1.1796 37.9757
18 4.3258 2.4975 6.0426 10.0249 1.1767 38.0707
19 1.4419 0.0000 4.0374 9.9688 1.1607 10.8194
20 1.4419 2.4975 4.0374 9.9780 1.1633 10.8567
21 1.4419 4.9950 4.0374 9.9722 1.1567 10.8337
22 3.6048 -1.2488 4.0374 9.9785 1.1570 10.8585
23 3.6048 1.2487 4.0374 9.9722 1.1609 10.8337
24 3.6048 3.7463 4.0374 9.9682 1.1633 10.8174
25 5.7677 -2.4975 4.0374 9.9729 1.1633 10.8358
26 5.7677 0.0000 4.0374 9.9687 1.1570 10.8191
27 5.7677 2.4975 4.0374 9.9787 1.1607 10.8589
28 0.7210 1.2487 2.0078 9.9686 1.1619 10.8257
29 0.7210 3.7463 2.0078 9.9792 1.1613 10.8683
30 0.7210 6.2438 2.0078 9.9726 1.1594 10.8422
31 2.8839 0.0000 2.0078 9.9785 1.1595 10.8649
32 2.8839 2.4975 2.0078 9.9737 1.1658 10.8461
33 2.8839 4.9950 2.0078 9.9694 1.1613 10.8289
34 5.0468 -1.2488 2.0078 9.9724 1.1614 10.8406
35 5.0468 1.2488 2.0078 9.9687 1.1595 10.8255
36 5.0468 3.7463 2.0078 9.9784 1.1619 10.8651
--------------------------------------------------------------------------------
NUMBER OF ELECTRONS: 360.00000


The number of the valence electrons sounds reasonable. However, I found that the atomic volume of the surface atom (in the first and fourth layer) is nearly four times larger than that of the bulk atom (in the second and third layer).
It is known that the Wigner-Seitz radius of the surface atom should be larger than that of the bulk atom. But in my case, the volume seems to be too large. So, I wonder if there is something wrong with my Bader result.

Any suggestions will be appreciated.

[andyfan]

My INCAR file for the Bader calculation is also attached.

SYSTEM = default

Start parameter for this Run:
ISTART = 0
ICHARG = 2
INIWAV = 1

Electronic Relaxation:
ENCUT = 400
ENAUG = 650
PREC = Accurate
IALGO = 48
NELM = 600
NELMIN = 8 # surface = 8, bulk = 4
NELMDL = -12 # When IALGO = 38, 0
EDIFF = 1E-04
GGA = PE
LREAL = Auto
WEIMIN = 0
NGX = 100
NGY = 100
NGZ = 240

Ionic Relaxation:
NSW = 1
IBRION = -1

DOS related values:
SIGMA = 0.2
ISMEAR = 0

Spin polarized:
ISPIN = 2

LAECHG=.TRUE.

[graeme]

The calculation is fine. In a Bader calculation, all space is assigned to atoms (or more precisely, charge density maxima). For a surface, all of the vacuum space is assigned to the surface atoms. The charges on these atoms look very reasonable.

[andyfan]

Hi, graeme,

Thanks very much for your reply. Now I know that the atomic volume of the surface atoms from Bader calculation contains the vacuum space. Then how can I get the actual atomic volume of the surface atoms? Can I simply subtract the vacuum volume from the Bader calculated atomic volume of the surface atoms?

I have an additional question. I want to calculate the d-band center of the metal surface (e.g., Ni(111)). It is known that the d-band center is sensitive to the Wigner Seitz radius of the surface atom, and the Wigner Seitz radius of the surface atom is suggested to be generally larger than that of the bulk atom(correct me if I am wrong). Therefore, I want to use the Bader calculated atomic voulme to estimate the Wigner Seitz radius of the surface atom, and then put this value into the INCAR file (RWIGS tag) of VASP to calculate the DOS.

Is this idea reasonable? And what about your suggestions? Thanks.

[graeme]

The Bader definition of atomic volumes is very large for surface atoms. Unfortunately, there is no alternative definition within this theory.

I think it would be better to integrate the density of states within the Bader volumes. A student in my group, Wenjie Tang, has the code for this and will post it very soon. This does, however, require a small modification to the vasp code. An advantage of this is that one does not have to specify the integration radius around atoms. Also, the sum of the partial DOS equals the total DOS.

[andyfan]

Thanks very much.

If possible, would you please send a copy of the code to me? Or please let me know as soon as you post the code.
My email adress is: fanchen061@mail.ecust.edu.cn

Thanks.