UT theoretical chemistry code forum

Dear all
I have done the bader anlysis for the oxygen in 32 supercell Ni. my question is the meaning od MIN DIST in the ACF.dat and DISTANCE in BCF.dat
ACF.dat

# X Y Z CHARGE MIN DIST
----------------------------------------------------------------
1 0.0254 0.0254 0.0254 10.0014 1.1968
2 0.0254 3.5753 0.0254 10.0014 1.1968
3 3.5753 0.0254 0.0254 10.0014 1.1968
4 1.7749 5.3248 0.0308 9.9932 1.0938
5 1.7749 1.7749 0.1401 9.7957 0.3248
6 3.5753 3.5753 0.0254 10.0014 1.1968
7 5.3248 1.7749 0.0308 10.0057 1.0938
8 5.3248 5.3248 0.0076 9.8984 1.2639
9 0.0254 0.0254 3.5753 10.0014 1.1968
10 0.0254 3.5753 3.5753 10.0014 1.1968
11 1.7749 1.7749 3.6900 9.7959 0.7226
12 1.7749 5.3248 3.5807 9.9932 1.1867
13 3.5753 0.0254 3.5753 10.0014 1.1968
14 3.5753 3.5753 3.5753 10.0014 1.2779
15 5.3248 1.7749 3.5807 10.0058 1.1867
16 5.3248 5.3248 3.5575 9.8984 1.2639
17 1.7749 0.1401 1.7749 9.7957 0.3248
18 0.1401 1.7749 1.7749 9.7957 0.3248
19 0.0308 5.3248 1.7749 10.0640 1.0938
20 1.7749 3.6900 1.7749 9.7959 0.7226
21 3.6900 1.7749 1.7749 9.7959 0.7226
22 3.5807 5.3248 1.7749 10.0640 1.1867
23 5.3248 3.5807 1.7749 9.9780 1.1867
24 5.3248 0.0308 1.7749 10.0334 1.0938
25 1.7749 0.0308 5.3248 9.9549 1.0938
26 0.0308 1.7749 5.3248 10.0640 1.0938
27 0.0076 5.3248 5.3248 9.9904 1.2639
28 1.7749 3.5807 5.3248 10.0316 1.1867
29 3.5807 1.7749 5.3248 10.0640 1.1867
30 3.5575 5.3248 5.3248 9.9906 1.2760
31 5.3248 3.5575 5.3248 9.9301 1.2639
32 5.3248 0.0076 5.3248 9.8668 1.2639
33 1.7749 1.7749 1.7749 7.3878 0.9530

BCF.dat # X Y Z CHARGE ATOM DISTANCE
------------------------------------------------------------------------
1 0.0000 0.0000 0.0000 10.0014 1 0.0439
2 0.0000 5.3248 1.7749 10.0640 19 0.0308
3 0.0000 0.0000 3.5499 10.0014 9 0.0439
4 0.0000 5.3248 5.3248 9.9904 27 0.0076
5 6.9223 1.7749 1.7749 9.7957 18 0.3176
6 0.0000 1.7749 5.3248 10.0640 26 0.0308
7 0.0000 3.5499 0.0000 10.0014 2 0.0439
8 0.0000 3.5499 3.5499 10.0014 10 0.0439
9 1.7749 6.9223 1.7749 9.7957 17 0.3176
10 1.7749 0.0000 5.3248 9.9549 25 0.0308
11 1.7749 1.7749 1.7749 7.3878 33 0.0000
12 1.7749 1.7749 6.9223 9.7957 5 0.3176
13 1.7749 1.7749 3.7274 9.7959 11 0.0374
14 1.7749 3.7274 1.7749 9.7959 20 0.0374
15 1.7749 3.5499 5.3248 10.0316 28 0.0308
16 1.7749 5.3248 0.0000 9.9932 4 0.0308
17 1.7749 5.3248 3.5499 9.9932 12 0.0308
18 3.5499 0.0000 0.0000 10.0014 3 0.0439
19 3.5499 0.0000 3.5499 10.0014 13 0.0439
20 3.5499 1.7749 5.3248 10.0640 29 0.0308
21 3.5499 3.5499 0.0000 10.0014 6 0.0439
22 3.5499 3.5499 3.5499 10.0014 14 0.0439
23 3.5499 5.3248 1.7749 10.0640 22 0.0308
24 3.5499 5.3248 5.3248 9.9906 30 0.0076
25 3.7274 1.7749 1.7749 9.7959 21 0.0374
26 5.3248 0.0000 1.7749 10.0334 24 0.0308
27 5.3248 0.0000 5.3248 9.8668 32 0.0076
28 5.3248 1.7749 0.0000 10.0057 7 0.0308
29 5.3248 1.7749 3.5499 10.0058 15 0.0308
30 5.3248 3.5499 1.7749 9.9780 23 0.0308
31 5.3248 3.5499 5.3248 9.9301 31 0.0076
32 5.3248 5.3248 0.0000 9.8984 8 0.0076
33 5.3248 5.3248 3.5499 9.8984 16 0.0076
~

Thanks for help

In the ACF.dat file (atomic charge file), the MIN DIST value is the minimum distance from the suface of the Bader volume to the associated atomic nucleus.

In the BCF.dat file (bader charge file), the DISTANCE value is the distance from the Bader charge density maximum to the nearest atomic nucleus.

These values are defined on the web page:
http://theory.cm.utexas.edu/bader/

thanks graeme for the calrification

dear all,
I have done the bader analysis for oxygen (six valence electron) in octahedral site in 32 supercell of Ni (10 valence electron), with RWIGS = 1.286 (A) for Ni and 0.820 (A) for oxygen. the pseudopotentiel used is that of PAW.
If I look at the MIN DIST the Ni firsts neighbour of oxygen (atoms number 5, 11, 17, 18, 20 and 21) have a MIN DIST lfar from the value of 1.286 or 1.2779 (A)
in the other side, its suggested by graeme that if the smallest distance is found to be less than this cut-off radius then the results MAY NOT BE VAILD !!!!!!!!!!!!!

Can you give me more clarification. there is the ACF.dat output
Thanks for all
# X Y Z CHARGE MIN DIST
----------------------------------------------------------------
1 7.0744 7.0744 7.0744 10.0003 1.2779
2 7.0744 3.5753 7.0744 10.0003 1.2779
3 3.5753 7.0744 7.0744 10.0003 1.2779
4 1.7749 5.3248 7.0690 9.9932 1.1867
5 1.7749 1.7749 6.9597 9.7972 0.7226
6 3.5753 3.5753 7.0744 10.0003 1.2779
7 5.3248 1.7749 7.0690 10.0063 1.1867
8 5.3248 5.3248 7.0922 9.8984 1.2639
9 7.0744 7.0744 3.5753 10.0003 1.2779
10 7.0744 3.5753 3.5753 10.0003 1.2779
11 1.7749 1.7749 3.6900 9.7974 0.7226
12 1.7749 5.3248 3.5807 9.9932 1.1867
13 3.5753 7.0744 3.5753 10.0003 1.2779
14 3.5753 3.5753 3.5753 10.0003 1.2779
15 5.3248 1.7749 3.5807 10.0064 1.1867
16 5.3248 5.3248 3.5575 9.8984 1.2639
17 1.7749 6.9597 1.7749 9.7972 0.7226
18 6.9597 1.7749 1.7749 9.7972 0.7226
19 7.0690 5.3248 1.7749 10.0634 1.1867
20 1.7749 3.6900 1.7749 9.7974 0.7226
21 3.6900 1.7749 1.7749 9.7974 0.7226
22 3.5807 5.3248 1.7749 10.0634 1.1867
23 5.3248 3.5807 1.7749 9.9786 1.1867
24 5.3248 7.0690 1.7749 10.0340 1.1867
25 1.7749 7.0690 5.3248 9.9549 1.1867
26 7.0690 1.7749 5.3248 10.0634 1.1867
27 7.0922 5.3248 5.3248 9.9905 1.2760
28 1.7749 3.5807 5.3248 10.0316 1.1867
29 3.5807 1.7749 5.3248 10.0634 1.1867
30 3.5575 5.3248 5.3248 9.9906 1.2760
31 5.3248 3.5575 5.3248 9.9301 1.2639
32 5.3248 7.0922 5.3248 9.8668 1.2639
33 1.7749 1.7749 1.7749 7.3877 0.9530

It is good to be suspicious when the bader surfaces go inside the rcore distance. The pseudopotential tries to ensure an accurate charge density outside of the core, but as you get inside rcore, the density can not be trusted.

In your case, things don't look so bad. You should compare the MIN DIST to RCORE (not RWIGS). For O this is 0.8 Ang and for Ni it is 1.2 Ang.

It is also important to check that your results are converged. Make sure the charges do not change as you go to a finer fft grid. You can also use the O_h pseudopotential, which has a smaller rcore value of 0.58 Ang. As long as your charges do not change with the two different O potentials, and they are converged with the grid size, you can trust them.

thanks graeme for your help, but it still not clear to me in the case of firsts neighbours of oxygen which have MIN DIS of 0.72 (A) much smaler than the rcor value (1.2).
thanks

The minimum distance from the bader volume to the atom should be compared to the rcor value for that type of atom. You said that that atoms with a small min-dist (5, 11, 17, 18, 20 and 21) are oxygen atoms, so these values should be compared to the rcor of oxygen. When you use the hard oxygen, the rcor value is quite small (0.58 Ang).

thanks graeme
the atoms (5, 11, 17, 18, 20 and 21) are not oxygen but nickel atoms (i.e, the first nieghbours of oxygen atom located at octahedral interstitiel site).
thanks

Ah, I see what you mean. Yes, then the bader surface is well inside rcore for Ni. This doesn't necessarily mean that the analysis is wrong, but it is worrisome.

To make sure your results are not dependent upon the choice of pseudopotential, you could use the Ni_pv potential. This will significantly increase the number of core electrons (to 16 for Ni). If you get a similar (net) charge for the Ni atoms, I would trust the results much more. I would also be very surprised if the 16 valance electrons in the Ni_pv potential did not sufficiently reproduce the core charge density.

If you try this out, please let us know what you find. The pseudopotentials are the biggest source of worry for the analysis, so we would like to understand how sensitive the bader charges are, and what kinds of errors we can expect.