database structure is far away from the optimized structure in ATK

[jennylau1996]

Hi, I input the Sb2Se3 structure from database:

https://github.com/cryos/avogadro/blob/master/crystals/selenides/Sb2Se3-Antimonselite.cif

and asked ATK-DFT to relax the structure. The original force was huge. The structure changed a lot and looked very different from the literature figures after relaxation. I have chosen enough k points and I am quite confused of what I have got.

Could you please help me with that?

Thanks!

Jenny

[Jess Wellendorff]

Hi Jenny. We will need to see the atkpython script you used in order to comment on this.

[jennylau1996]

Hi, Jess -

Could you please have a look?

Thanks for your help!

Jenny

[Ulrik G. Vej-Hansen]

I suggest that you instead start by optimizing the cell - I think a lot of the distortion you see is because the atoms try to adjust to a "wrong" lattice constant. Additionally, the distortion is not as big as it seems. VNL shows bonds when the distance between atoms are smaller than some element-dependent number, and by increasing the so-called "fuzz factor", that number can be changed. If you increase the fuzz factor in both cases, they look more alike in the Viewer.

Also, I got an initial force of 1.5 eV/Å, which I would definitely not consider huge. Considering that the original structure was not optimized with your chosen DFT method, it is perfectly reasonable to see initial forces of that size, or even larger.

[jennylau1996]

I think this is a fair comment. Thanks, I will try that on!

Jenny