Author Topic: Use CNA or DXA to identify twin boundaries in a luster?  (Read 515 times)

tuy15

  • Newbie
  • *
  • Posts: 3
Use CNA or DXA to identify twin boundaries in a luster?
« on: January 28, 2019, 09:04:16 PM »
Dear OVITO team and users,

I am using OVITO to identify the twin boundaries in a cluster(e.g. an icosahedron cluster I attached)? I tried common neighbor analysis(CNA) and Dislocation analysis (DXA) tools, but neither of them works. The output is just the structure type labeled in different colors.

Should I expect result show me clear boundaries, like the one from your manual? 
(The attached picture found from https://ovito.org/manual/particles.modifiers.dislocation_analysis.html)

I am not sure if I missed any steps. I would appreciate if anyone could share the experience.


Best regards and thanks,
Tina
« Last Edit: January 28, 2019, 09:09:24 PM by tuy15 »

Constanze Kalcher

  • Administrator
  • Sr. Member
  • *****
  • Posts: 301
Re: Use CNA or DXA to identify twin boundaries in a luster?
« Reply #1 on: January 29, 2019, 11:52:44 AM »
Dear Tina,

could you please explain what crystal structure your cluster has and also how you define twin boundaries? From the screenshot I cannot get a good idea of the structure yet.
Note that for using the Dislocation analysis tool you need to specify the lattice type of the input crystal and currently, only the following crystal structures  are supported: FCC, HCP, BCC, and diamond.
So if your cluster does not have one of these crystal structures the DXA is probably not the right tool for your setup.

But I can of course try to look into the problem further if you upload your structure here.

-Constanze

tuy15

  • Newbie
  • *
  • Posts: 3
Re: Use CNA or DXA to identify twin boundaries in a luster?
« Reply #2 on: January 29, 2019, 05:04:34 PM »
Hello Constanze,

Thanks for your reply! Attached is the crystal structure, which is nearly a perfect icosahedron(ovito CNA analysis suggests that it contains FCC and HCP structure). I think the edges shared by two neighboring triangle facets should be the "boundaries" I mentioned. Thanks again for your help!


Best,
Tina

Constanze Kalcher

  • Administrator
  • Sr. Member
  • *****
  • Posts: 301
Re: Use CNA or DXA to identify twin boundaries in a luster?
« Reply #3 on: January 29, 2019, 07:53:46 PM »
Hello Tina,

thank you for the picture, but could you upload the coordinates so I can load them into OVITO myself?

Also note that what is shown in the picture from the manual are dislocations in a FCC crystal. All (green) FCC atoms have been deleted so that only (red)
atoms in a HCP environment are visible -  indicators for planar defects like stacking faults and coherent twin boundaries in FCC.

-Constanze

tuy15

  • Newbie
  • *
  • Posts: 3
Re: Use CNA or DXA to identify twin boundaries in a luster?
« Reply #4 on: January 29, 2019, 09:11:55 PM »
Hello Constanze,

Attached is the atom coordinates in lammps trajectory format. When I use CNA to identify the structure, the surface atoms are always identified as "unknown" since they don't have as many neighbors. This might make it difficult to identify the boundaries, since they are on the surface of crystals.  Thanks again for your help!

Best,
Tina

Constanze Kalcher

  • Administrator
  • Sr. Member
  • *****
  • Posts: 301
Re: Use CNA or DXA to identify twin boundaries in a luster?
« Reply #5 on: January 30, 2019, 01:25:45 PM »
Hello Tina,

thanks for uploading the structure. Yes, that the surface atoms will be identified as "Others" is completely expected. Interestingly, inside the cluster there is a mixture of atoms that are identified both as fcc and hcp and the center atom is sitting in a perfect icosahedral environment. Note, that a twin boundary in the fcc structure would be a {111}-hcp plane and I can't really identify that in your cluster, but I think you maybe mean something else by twin boundaries?

Regardless, you can see that there is some sort of symmetry in the arrangement of the of the atoms colored by different structure types. I attached a little video for you that shows your cluster, one time viewed from a "corner atom" and one time the viewing direction is perpendicular to a facet. You can use the Slice modifier
http://ovito.org/manual_testing/particles.modifiers.slice.html
 to slice through your structures like in the animation. I used the functions "Pick three atoms" and "Align plane normal to view direction" to find the correct slicing planes and animated the slice using the auto-key mode.
http://ovito.org/manual_testing/usage.animation.html

-Constanze
« Last Edit: January 30, 2019, 01:36:41 PM by Constanze Kalcher »