Optimizing Geometry of 2D material sheet

[Subhban15]

Hello!

1. Could anyone please tell me whether I must use SGGA or LSDA or MSGGA while using the 'new calculator' for optimizing a 2d material defective sheet with 'Optimize Geometry'?

2. While trying to optimize the geometry of the same, I tried keeping k = 9x1x9 (C being the direction of transmission, my sheet consists of 7x1x7 = 49 atoms) but it took huge time to simulate so I reduced the k to 3x1x3.
I notice that I got better result with low k.
With high k the sheet seemed more deformed on the right than on the left but with 3x1x3 it seems more symmetric. Why is that? (The defect is very near to the centre but, a little to the left)

Thanks for the help!!! 

[Umberto Martinez]

1. that is entirely up to you. You can start by researching your system through the literature.

2. from your description your set up is not clear. Please, add your full scripts if you want more help.

[Subhban15]

Thank you for your reply.

Let me rephrase the question, should the k-point sampling in the new calculator (DFT) affect accuracy of the simulation?

[Anders Blom]

Yes

[Subhban15]

Dears Anders,

With k being 9,1,9 the optimized sheet looks as if it is crumpled on one side of the surface of the sheet (asymmetrical crumpling) whereas with 5,1,5 as k-point sampling the optimized sheet looks more symetric. What could be the reason behind it? And which one could be considered more accurate?

Also, would please tell me, out of the two which would be a better way to optimize:
1. Optimizing the defective nanosheet first and then making a device
or
2. Making a device out of the defective sheet and then optimizing the whole thing

Thank you very much for your time!

[Anders Blom]

Very hard to say from just a few simple parameters.  More k-points is always more accurate in the asymptotic limit, but special systems like graphene can have a non-monotonous convergence depending on which precise k-points are included. You can try 18x1x18 next. It may also depend on strain and other parameters (if you e.g. use a too small basis set).

[Anders Blom]

As for device first or not, it if you make the device first you have more atoms and have less freedom (the distance between the two electrodes is fixed) but on the other hand it can be hard to make the device from a defective sheet, unless you are careful with constraining the edges or you do the joining of perfect sheet (electrodes) with the defective central part manually.

[Subhban15]

unless you are careful with constraining the edges

What are the ways in which I can constrain the edges? I mean how do I control it to stay unaffected/less affected?

Also, 18,1,18 is not a great option for me because it will need a lot more time to simulate, and the system that I work on cannot handle it with easy.

[Jess Wellendorff]

You can constrain the edges of the central region (more precisely: the electrode extensions into the central region) in two ways, both using functionality in the GeometryOptimization widget:
1) either by completely fixing the atoms, or
2) by assigning the electrode extensions to a "ridgid body", which will relax rigidly without changing the internal coordinates of the electrode extension. See this tutorial: http://docs.quantumwise.com/tutorials/device_relaxation.html.

[Subhban15]

Thank you Jess Wellendorff and Anders Blom for your help!
I was going through the tutorial the other day and found it very helpful.


By the way, what does "rigidly relax" practically mean? I mean, to what level does the system relax? What happens internally? Is there a paper on what principle quantumwise uses to relax the atoms?

[Jess Wellendorff]

It simply means that all atoms in the "rigid body" are allowed to move only as a single unit during geometry relaxation: The forces on each of the atoms inside the body are collected into a single force acting on the center-of-mass of the body, and this is the force that is minimized for those atoms during geometry optimization. The bond distances between the atoms inside the rigid body does not change, so the crystal symmetry of the lattice inside the rigid body is conserved during geometry optimization. A well known trick from molecular dynamics simulations.

[Umberto Martinez]

Also the reference manual is a god place to have a look at:
http://www.quantumwise.com/documents/manuals/latest/ReferenceManual/index.html/ref.rigidbody.html
scroll down to the notes.

here, you can also find the description of more options for the RigidBody constrain such as fix_cartesian_direction and fix_cartesian_direction