unequal bader charge of H in H2

[zengfeng]

Hi,

I calculated the bader charge of H in H2 but got unequal bader charge value. The electron density was obtained from Abinit. And the electron density file was transformed into a cube file with CUT3D. The cube file was used as the input for bader charge calculation. The command ./bader -vac off was used. Could you please help to find out what the problem is? Many thanks in advance.

Attached is the input cube file.

The resuls are as following,

ACF:
# X Y Z CHARGE MIN DIST ATOMIC VOL
--------------------------------------------------------------------------------
1 18.8972613 20.0836232 5.9722905 1.0981100 0.70331689425.5111023
2 18.8972613 21.4903517 5.9722905 0.9017874 0.47244579342.6334817
--------------------------------------------------------------------------------
VACUUM CHARGE: 0.0000
VACUUM VOLUME: 0.0000
NUMBER OF ELECTRONS: 1.9999

AVF:
Atom Volume(s)
------------------------------------------------------------------------
1 ... 1
2 ... 2
------------------------------------------------------------------------

BCF:
# X Y Z CHARGE ATOM DISTANCE
-------------------------------------------------------------------------
1 18.8973 20.3250 5.9723 1.0981 1 0.2414
2 18.8973 21.2489 5.9723 0.9018 2 0.2415
-------------------------------------------------------------------------

[graeme]

There should be two solutions to this problem. One is to increase the grid density and use the standard algorithm. The errors in atomic charges should drop linearly with grid density.

Second, you can use the "-b weight" flag to switch over to a more accurate method by Hu and Trinkle. This should converge quadratically with grid density, and should always be more accurate than the standard "-b neargrid" method.

[zengfeng]

[quote="graeme"]There should be two solutions to this problem. One is to increase the grid density and use the standard algorithm. The errors in atomic charges should drop linearly with grid density.

Second, you can use the "-b weight" flag to switch over to a more accurate method by Hu and Trinkle. This should converge quadratically with grid density, and should always be more accurate than the standard "-b neargrid" method.[/quote]

Hi, many thanks for your reply. The second solution worked. But when I try to use "-vac off" and "-b weight" at the same time. An error occured as below. Could you please help with this problem. Thanks!

zengfeng@ubuntu:/tmp$ ./bader -b weight -vac off H2

GRID BASED BADER ANALYSIS (Version 0.95a 02/26/16)

OPEN ... H2
GAUSSIAN-STYLE INPUT FILE
FFT-grid: 180 x 180 x 54
CLOSE ... H2
RUN TIME: 0.54 SECONDS

SORTING CHARGE VALUES ... DONE
CALCULATING FLUX ... DONE

INTEGRATING CHARGES
0 10 25 50 75 100
PERCENT DONE: **********************
RUN TIME: 114.16 SECONDS
62200 87 90 1 0.150145156671534
some volume not assigned
zengfeng@ubuntu:/tmp$

[graeme]

There is clearly some problem related to the weight method in the vacuum region. I'll need to look into that. For now, I think you can do pretty well using a small vacuum value.

With the default (1e-3) I get:

# X Y Z CHARGE MIN DIST ATOMIC VOL
--------------------------------------------------------------------------------
1 18.8972613 20.0836232 5.9722905 0.9618664 0.7033168 61.2245996
2 18.8972613 21.4903517 5.9722905 0.9618659 0.4724457 61.2253149
--------------------------------------------------------------------------------
VACUUM CHARGE: 0.0762
VACUUM VOLUME: 18645.6947
NUMBER OF ELECTRONS: 1.9999

with 1e-4 I get:

# X Y Z CHARGE MIN DIST ATOMIC VOL
--------------------------------------------------------------------------------
1 18.8972613 20.0836232 5.9722905 0.9931706 0.7033168 151.1176006
2 18.8972613 21.4903517 5.9722905 0.9931708 0.4724457 151.1186098
--------------------------------------------------------------------------------
VACUUM CHARGE: 0.0136
VACUUM VOLUME: 18465.9084
NUMBER OF ELECTRONS: 1.9999

and with 1e-5 it is looking pretty good:

# X Y Z CHARGE MIN DIST ATOMIC VOL
--------------------------------------------------------------------------------
1 18.8972613 20.0836232 5.9722905 0.9988621 0.7033168 307.9906236
2 18.8972613 21.4903517 5.9722905 0.9988624 0.4724457 307.9919818
--------------------------------------------------------------------------------
VACUUM CHARGE: 0.0022
VACUUM VOLUME: 18152.1620
NUMBER OF ELECTRONS: 1.9999

[zengfeng]

[quote="graeme"]There is clearly some problem related to the weight method in the vacuum region. I'll need to look into that. For now, I think you can do pretty well using a small vacuum value.

With the default (1e-3) I get:

# X Y Z CHARGE MIN DIST ATOMIC VOL
--------------------------------------------------------------------------------
1 18.8972613 20.0836232 5.9722905 0.9618664 0.7033168 61.2245996
2 18.8972613 21.4903517 5.9722905 0.9618659 0.4724457 61.2253149
--------------------------------------------------------------------------------
VACUUM CHARGE: 0.0762
VACUUM VOLUME: 18645.6947
NUMBER OF ELECTRONS: 1.9999

with 1e-4 I get:

# X Y Z CHARGE MIN DIST ATOMIC VOL
--------------------------------------------------------------------------------
1 18.8972613 20.0836232 5.9722905 0.9931706 0.7033168 151.1176006
2 18.8972613 21.4903517 5.9722905 0.9931708 0.4724457 151.1186098
--------------------------------------------------------------------------------
VACUUM CHARGE: 0.0136
VACUUM VOLUME: 18465.9084
NUMBER OF ELECTRONS: 1.9999

and with 1e-5 it is looking pretty good:

# X Y Z CHARGE MIN DIST ATOMIC VOL
--------------------------------------------------------------------------------
1 18.8972613 20.0836232 5.9722905 0.9988621 0.7033168 307.9906236
2 18.8972613 21.4903517 5.9722905 0.9988624 0.4724457 307.9919818
--------------------------------------------------------------------------------
VACUUM CHARGE: 0.0022
VACUUM VOLUME: 18152.1620
NUMBER OF ELECTRONS: 1.9999[/quote]


It works. The minimun vacuum charge density could be set as 1e-5. Many thanks.