[Wien] Re : Supercell problem of Al2O3

[Michael Fischer]

Hello,

sorry, I was not aware that you cannot select the space group in 
hexagonal setting in structgen - as I said, I do not have much 
experience with rhombohedral and hexagonal cells in WIEN2k.

I think you should do as follows: Choose H lattice type and enter the 
atoms separately. You have to enter 4 aluminium and 6 oxygen positions. 
The other ones should be generated by the lattice type. In order to 
find out which positions you need to enter, I suggest you use "WYCKPOS" 
of www.cryst.ehu.es. There, you enter space group 167 and look at the 
coordinates given for Wyckoff positions 12c (Al) and 18e (O). For x or 
z, you substitute the coordinate you have calculated.
E.g. for Al at the position (0 0 0.35), i.e. z=0.35, you need to enter 
the four positions:
0 0 0.35
0 0 0.15
0 0 0.85
0 0 0.65
Alternatively, select P lattice type and enter all 12 and 18 atoms 
respectively (more tedious). In addition to the above, you need to 
apply the transformations for all four Al and six O-positions.
  (2/3,1/3,1/3) +
(1/3,2/3,2/3) +

The cause why your second try did not work is easily explained: If you 
enter only two atoms and do not choose the correct space group, 
structgen does not know how to generate the symmetry-equivalent atoms. 
So you end up with a cell that has way too few atoms.

Greetings
Michael