Dear Smritijit,

You can certainly use VCA, but how accurate it is depends on what you'd like to study. VCA was developed for intermetallic alloys, where the properties tend to be linearly dependent on the doping concentration ("Vegard's law"). It is generally found that small doping concentrations and similar (ideally, isoelectronic) substitutions give reasonable electronic properties (band-structures, DOS etc) and geometries, but poor formation energies; in particular, VCA tends to lead to anti-mixing such that alloys are not energetically favourable with respect to their two pure parent phases.

On a more practical note, I suggest you test the VCA for your system by performing calculations for higher dopant concentrations, where it is also possible to perform explicit substitutions (e.g. 10%). This will give you a reasonable idea of how VCA performs compared to actual doping models, though do bear in mind that the specific doping models you construct will be periodic and so also not "the true answer". Once you are satisfied that the VCA is giving reasonable agreement, you can use it for the lower dopant concentrations you actually want to model; and conversely, if you are not satisfied then at least you know VCA isn't good enough for your work.

Hope that helps,

Phil Hasnip

(CASTEP developer)

Dear Phil,

I would like to profit your discussion to ask you about how to build a VCA crystal using Materials Studio 8.0? I am totally new with MS and CASTEP. I have a perovskite KTa0.7Nb0.3O3 structure and I would like to simulate its phonon dispersion.

I have tried using the Crystal builder, but when I add the atoms, saying Ta and Nb, who are at the same position with different occupancy, the software doesn't allow me to add two atoms at the same position.

Thank you for your help.

Best regards,

Tra (postdoc @ ESRF)

Dear Tra,

I don't usually use Materials Studio, but I think you do this by editing an atom and specifying it as a "mixture" of multiple elements, rather than adding several atoms at the same site. Specifying the mixture directly for a given atom means that there is only one atom at each location in the crystal, but it is a "mixture atom".

In the CASTEP cell file this will appear as separate entries in the POSITIONS_ABS or POSITIONS_FRAC block, but the atoms at the same coordinates will have a MIXTURE tag with the same identifying integer like this:

Si 0.25 0.25 0.25 MIXTURE=(1 0.8)

Ge 0.25 0.25 0.25 MIXTURE=(1 0.2)

where the integer ("1" in the example above) identifies the mixture atom, and the real number ("0.8" and "0.2" in the example above) determines how much of that element is in the mixture. So the example above has a single mixture atom at 0.25, 0.25, 0.25 which is 0.8 Si and 0.2 Ge.

Hope that helps,

Phil

yes you can, With the following steps

Philip James Hasnip I have some trouble in the virtual pseudo-potential generation in the CASTEP. Must the two elements creating the virtual pseudo-potential be in the same valence state???

Not strictly, but it does tend to work best when they are "similar" -- meaning isoelectronic, or +/- 1 valence electron. The VCA is very crude, and generally only works when you expect the property you're interested in to vary linearly with composition.

What problems are you having?

Philip James Hasnip I am trying to mix two elements which can not be sure with the same valence state.

Dear shan wan Fei

To my knowledge VCA is expected to work reasonably well in the isovalent case

Fadla Abdelilah Oh, thanks a lot! Then what about the material with the mixed valence state for one element of which one will generate a virtual potential with other one element?