Dear all,
I would like to report a possible bug in the program bader as well as a work-around.
It seems that bader is having difficulty interpreting .cube files for periodic monoclinic
unit cells, and possibly periodic non-orthogonal unit cells in general. Tests were
performed using the CPMD software package RHO_TOT.cube files that were carefully
reformatted to be consistent with the VASP chgcar format, i.e., z-index slowest, no
periodic points along cell edges, grid origin at (0,0,0).
1) The program can read .cube file of monoclinic unit cells, if the charge density is
not periodic. For example, if a single atom is placed in a monoclinic unit cell,
with four electrons placed directly on top of it, then bader correctly predicts the
electronic charge of this atom to be 4.0000:
ACF.dat:
# X Y Z CHARGE MIN DIST ATOMIC VOL
--------------------------------------------------------------------------------
1 0.3532 0.0000 1.0619 4.0000 0.0000 143.4118
--------------------------------------------------------------------------------
2) However, using .cube files of monoclinic unit cells with periodic charge density,
spurious results are obtained. For example, more than one partial charge is predicted for
symmetrically equivalent atoms Zr (indices 1-4), O1 (indices 5-8), and O2 (indices 9-12)
in monoclinic zirconia:
ACF.dat:
# X Y Z CHARGE MIN DIST ATOMIC VOL
--------------------------------------------------------------------------------
1 2.3448 0.3890 2.0660 9.3881 1.5524 67.9484
2 6.5824 5.3133 2.8931 8.2448 0.6895 59.8227
3 1.5389 4.5352 7.0251 9.3881 0.6357 67.9484
4 5.7765 9.4595 7.8523 2.9462 0.0420 22.2457
5 8.8014 8.1910 1.5403 7.3307 1.0269 82.0238
6 0.1257 3.2667 3.4188 7.6903 0.5748 85.7468
7 7.9955 6.5818 6.4994 11.5677 0.7713 117.8390
8 -0.6801 1.6575 8.3779 7.6718 0.0370 81.2965
9 5.3235 2.5301 0.2063 9.8293 0.3260 114.0894
10 3.6036 7.4543 4.7528 7.3307 0.5857 82.0238
11 4.5177 2.3942 5.1654 7.3307 1.3184 82.0238
12 2.7977 7.3184 9.7119 7.2811 1.0574 87.7052
--------------------------------------------------------------------------------
NUMBER OF ELECTRONS: 95.99944
3) Now, if this .cube file is converted into a VASP chgcar file by rewriting the header
and multiplying every electron density value by the cell volume, then the partial charges
obtained are consistent with the symmetry of the system:
ACF.dat:
# X Y Z CHARGE MIN DIST ATOMIC VOL
--------------------------------------------------------------------------------
1 1.2408 0.2059 1.0933 9.3888 0.9898 10.0702
2 3.4833 2.8117 1.5310 9.3888 0.9898 10.0702
3 0.8144 2.3999 3.7175 9.3888 0.9898 10.0702
4 3.0568 5.0057 4.1552 9.3888 0.9898 10.0702
5 4.6575 4.3345 0.8151 7.2803 0.8869 12.9926
6 0.0665 1.7287 1.8092 7.2803 0.8869 12.9924
7 4.2311 3.4829 3.4393 7.2803 0.8869 12.9926
8 -0.3599 0.8771 4.4334 7.2803 0.8869 12.9924
9 2.8171 1.3389 0.1092 7.3307 0.9614 12.1576
10 1.9070 3.9447 2.5151 7.3307 0.9614 12.1576
11 2.3906 1.2669 2.7334 7.3307 0.9614 12.1576
12 1.4805 3.8727 5.1393 7.3307 0.9614 12.1576
--------------------------------------------------------------------------------
NUMBER OF ELECTRONS: 95.99944
It would be useful to know the exact formatting/specifications for .cube files of
non-orthogonal cells with periodic electron densities required by bader. Previous
posts at the UT theoretical chemistry code forum seem to imply that periodic .cube
files can be handled without difficulty.
Alternatively, users could be instructed to convert .cube files of non-orthogonal
cells with periodic electron densities into VASP chgcar format. After the .cube file
is read and "GAUSSIAN-STYLE INPUT FILE" is printed by bader, an additional
statement to this effect could be printed.
Sincerely,
Yves
Bader and .cube files of non-orthogonal cells
Moderator: moderators