Structure factors

[rugfri]

Hello,

I noticed that the structure factors calculated by xrayutilities are different than that of Vesta.

Example
abs(xu.materials.Si.StructureFactor([1,1,1]))=38.06110892913849
while
Export the structure to a cif (xu.materials.Si.toCIF(...))
import it to Vesta, Utilities, Structure Factors gives |F|=61.0337

Maybe I do not compare things correctly?
Could you please comment?
Thank you!

[dkriegner]

No idea what equations Vesta is using, so its difficult to comment on this. Likely one should compare it for a series of hkl to understand if at least its both internally consistent.

If you think there is some bug here please let me know.

[rugfri]

I built a small comparison in file attached
XU_Si_F_Vesta_XU.txt

Vesta documentation is here, at p. 110-111 there is the Structure Factors part.
https://jp-minerals.org/vesta/archives/VESTA_Manual.pdf

I would think there is bug unless I am not using abs(xu.materials.Si.StructureFactor()) in the right way(?)
xu Si222 values are suspicious.

I will now try to benchmark Vesta against another program.
There may be variations due to different tabulated form factors (and sometimes neglected f', f'' ...).

Thanks.

[dkriegner]

Ok, I only now see that you did not read the manual of StructureFactor.

So there is no bug, but the function does not work as you use it:

Signature: xu.materials.Si.StructureFactor(q, en='config', temp=0)
Docstring:
calculates the structure factor of a material
for a certain momentum transfer and energy
at a certain temperature of the material

Parameters
----------
q :     list, tuple or array-like
    vectorial momentum transfer
en :    float or str, optional
    x-ray energy eV, if omitted the value from the xrayutilities
    configuration is used
temp :  float
    temperature used for Debye-Waller-factor calculation

Returns
-------
complex
    the complex structure factor

If used correctly you will get absolute values close to ones from Vesta and of course close to 0 for the 222 reflection.

[dkriegner]

Good that its clear. While I think the current interface of StructureFactor is needed (e.g. for the simulations). I will consider adding a dedicated function which works with (h,k,l) input.

[dkriegner]

@rugfri would you have any wishes for #235 ?

[rugfri]

nothing specific, @dkriegner.
Indeed it is only a nice to have.
The current interface is more general, the (h,k,l) input is only a shorthand valid in some special cases.
Thanks.
Sorry for the late reply

[rugfri]

@rugfri would you have any wishes for #235 ?

@dkriegner I may have a one which I guess can be related to the StructureFactor calculation and its use in simulation, as you already mentioned.
How would it be possible to simulate 2D (or even 3D) sections of reciprocal space for a given stack?
This is somehow an expansion of the typical case of the truncation rod simulation covered in the examples.
With the current version I guess one would have to generate the stack and then pass to StructureFactor the array of Q-points (or via a loop?).
Thanks!

[dkriegner]

With the current models this is not possible. There are no parameters which describe any mosaicity or block sizes. so the physical input parameters are missing for such models. The structure factors would not be the problem. but generalizing all the rest is a very big adventure.