Crystal Structure InGaZnO4

[chitrapandy]

Hi
I have downloaded the crystal structure of InGaZnO4( CIF file) from Springer Material. When I import into the builder following error occurred.

Traceback (most recent call last): File "/opt/QuantumATK/QuantumATK-S-2021.06/lib/python3.8/site-packages/AddOns/ConfigurationLoaders/ConfigurationLoaders.py", line 300, in load configurations = importCIFFile(filename) File "zipdir/NL/IO/CIFFile.py", line 841, in importCIFFile File "zipdir/NL/IO/CIFFile.py", line 816, in configurationsFromCIF File "zipdir/NL/IO/CIFFile.py", line 732, in bulkConfigurationFromBlock File "zipdir/NL/IO/CIFFile.py", line 600, in bravaisLatticeFromBlock File "zipdir/NL/IO/CifHandler/CifHandler.py", line 375, in __getitem__ File "zipdir/NL/IO/CifHandler/CifHandler.py", line 457, in GetCifItemKeyError: 'Item _cell_length_a not in Cif block'

Currently, I am using the S-2021.06 version. Please clarify the above error.

Thank you.

[NW]

Hi,

You can use the Databases option to find the structure from Material Project or COD.

[chitrapandy]

I found the material structure when I import it into ATK builder I am getting errors.

[Anders Blom]

Pls attach the file you are trying to import. Seems it's missing some info but maybe it's due to some formatting issue in the CIF file

[chitrapandy]

Please find the attached cif file for the InGaZnO4 crystal structure.

[Anders Blom]

This is not a CIF file... And whatever file format it is supposed to be, it does not have any real information about the system - no unitcell, no element information, and apparently not even positions of atoms.

[chitrapandy]

Thank you for your response.
I am unable to attach the *.cif file.

I have downloaded the InGaZnO4 crystal structure from the below link.

https://materials.springer.com/isp/crystallographic/docs/sd_1402632

and try to open in the builder and an error occurred.

Thank you

[Anders Blom]

Thanks for letting us know. This file has several problems, and in some respects we should work on improving our parser to provide better error messages.

Fundamentally this structure is not directly usable for simulations because it contains partial occupancies, as seen from the lines
M1 '0.500Ga + 0.500Zn' .6c .3m 0 0 0.2141 1 ? '?'

You can work around that by importing the structure as Ga2InO4, then make a supercell and replace random Ga atoms by Zn (50% concentration).

But even with that, the file doesn't import, seemingly due to some minor mistake in the file or our parser.

I fixed it by removing everything unnecessary in the file, and making the change as mentioned to just import it as Ga2InO4. The file which works to import is attached.

[chitrapandy]

Thank you for your time and the file.
Meantime I got the  InGaZnO4 Crystal structure from materialsproject which is working without any errors.
I would like to generate a amorphous structure. will the amorphous pre-builder is good choice or MD simualtion?
Can you give your suggestion?
Thank you

[Anders Blom]

MD is more systematic, and you have more control, esp. since it starts from the actual crystal you have. The prebuilder is more brute-force, kind of just throwing the right ratio of atoms in a pot and stir it. The output still has to be equilibriated of course, but it does save time by skipping the melt part, and it does generate a very random structure initially, whereas melting the crystal might retain some of the long-range order. You do need a decent idea of the density to begin with for the prebuilder though. I think it's worth trying the prebuilder in your case, but you still need to do MD on it.

Btw, if this is a system you want to study in great detail, it could be really worthwhile to fit an MTP forcefield for it. That way you can do all kinds of thermal and mechanical studies with MD in hours rather than weeks, in the end! The fun part is that you can basically automatically generate the MTP while amorphizing it with DFT, which you have to do anyway.

[Anders Blom]

Actually, nobody has done a forcefield for IGZO before as far as I know, so that would be extremely valuable in general. We could provide support and guidance for that effort, it could actually be a great test-case for some new features we will introduce in December. You can reach out to me separately on email to discuss this further.

[chitrapandy]

Actually, nobody has done a forcefield for IGZO before as far as I know, so that would be extremely valuable in general. We could provide support and guidance for that effort, it could actually be a great test-case for some new features we will introduce in December. You can reach out to me separately on email to discuss this further.

I tried to reach you but it says blocked personal messages.

[chitrapandy]

I am trying to simulate IGZO amorphous structure following the below tutorial.
 
https://docs.quantumatk.com/tutorials/amorphous_structures/amorphous_structures.html#amorphous-structures

I lower the density of the unit cell and do melting and I get the error to do optimization because of atoms leaving the boundary.

My question is  do I need to optimize this structure ? Because  i want to lower the density and make a cubic structure and do optimize after cooling process.

Here I attached the  log file for you reference.

[Anders Blom]

I didn't know I had blocked PMs, not on purpose. But I read the replies here - just a bit slowly :-)

I suspect the forcefield you are using is not suitable for IGZO and gives very wrong forces. In fact, I don't know of any forcefield specifically designed for this material.

Did you run the script in parallel (MPI)? This is generally not recommended unless you have a very large system with 1000+ atoms, because the domain decomposition is not efficient otherwise. You can see that in the output, but the error message is a bit confusing because how can an atom leave the domain if there is only one... We could look into that (if you don't mind sending the script) but in a sense it doesn't matter because the fundamental issue is that the forcefield itself does not work for this material.

[Anders Blom]

Hard to identify the forcefield when it is written out explicitly, is this a build-in forcefield in QuantumATK?