[Wien] Site occupancy in multi-atom cell

[Stefaan Cottenier]

> I'm using Wien2K to work on my sample Li(Ni,Co,Mn)1/3O2. The atomic ratio
> Ni:Co:Mn=1:1:1, and these 3 particles randomly occupy same
> crystallographically equivalent site (for simplicity, let's assume it's
> the A-site in perovskite). In some other softwares to simulate the
> scattering process, the unit cell can be generated by specifying a
> occupany number(1/3 for current case) for each particle when putting it
> onto the site. However, I didn't find such a parameter when using
> StructGen. Does anyone have experience to bulid such structure with
> Wien2K? Is there similar function as the "occupancy number" in Wien2K?

No. There are 2 ways to approximate such a situation, but they have serious
drawbacks:

1) use the virtual crystal approximation (VCA), in which you put an
'average' atom on a specific site. This can be done by changing the Z in
case.struct. For your case, it would be Z=26.67 . You can imagine that this
is a serious approximation, but because your elements have very similar Z,
there might be something useful in it.

2) make your cell larger (e.g. 2x2x2), and put your 3 kinds of atoms
randomly at their specific positions (and probably you'll have to relax the
structure after that). Then, you have to repeat this for many different
positions, and average the property you are interested in over all those
calculations. You can imagine this is very time-consuming, especially
because a 2x2x2 might be too small to accomodate all useful configurations,
and you would need 3x3x3 or more.

Is there a good reason why you want to use wien2k for your problem? If not,
go to a code that has CPA implemented, which allows to do something like VCA
but with a better physical foundation (still an approximation, however).

Stefaan