QuantumATK Forum

QuantumATK => General Questions and Answers => Topic started by: dwh on August 24, 2012, 09:58

Title: charge transfer
Post by: dwh on August 24, 2012, 09:58

Hello,may I calculate the charge transfer from graphene to SiO2 in the system graphene placed on SiO2 ? If maybe, how can I set up the system?

Title: Re: charge transfer
Post by: kstokbro on August 26, 2012, 09:29

Use the interface builder to find a common supercell for graphene and SiO2.
Relax the system, try initial configurations with different vertical displacements of the systems to find the optimal geometry.
Calculate the mulliken population, this will give you the charge transfer.

Title: Re: charge transfer
Post by: monir_apece on April 3, 2017, 07:28

I was trying to build a structure like graphene on the top of SiO2 using interface tool. I have used displacement tool so that graphene layer could be 1.43 Angstrom away fromn SiO2 layer.  I was getting following error consistently
 
 ** Back Engine Exception : Diagonalization error, overlap matrix not positive definite. This may be caused by atoms that are too close to each other or situated in equivalent positions, or a too low value of interaction_max_range. Check the configuration, basis set and/or try to increase the interaction_max_range.
** Location of Exception : eigensolutions.cpp:2091

 
 Please advise me how to move graphene so that its 1.43 Angstrom away from SiO2 layer.

Title: Re: charge transfer
Post by: Anders Blom on April 3, 2017, 09:51

You have 8 atoms that are exactly overlapping (they are on top of each other).
Use the Selection tools>Close neighbors plugin in the Builder to find and delete them.

Title: Re: charge transfer
Post by: monir_apece on April 3, 2017, 17:45

Thank you Dr. Blom! Its working fine now. To save computational time what are the steps I could take.

1. Relax graphene while constraining SiO2 as fixed. Once graphene relaxed then optimize the whole structure again.
2. Should I use DFT-D2 for van der waals correction?
3. FFT2D for Poisson solver with C direction dirichlet  instead of periodic?
4. Is there any influence by changing History steps? in Iteration Control.

Title: Re: charge transfer
Post by: Anders Blom on April 3, 2017, 19:11

1. Or the opposite...
2. After a general optimization you can include D2
3. No, just use FFT
4. Sure, but not entirely predictable. Sometimes a lower number of steps leads to better convergence, but if you don't have problems (and don't run out of memory) just leave it at default

Title: Re: charge transfer
Post by: monir_apece on April 3, 2017, 22:27

Dear Dr. Blom,

So, we dont have to use D2 for geometric optimization, right?  or use D2 for final optimization.

Title: Re: charge transfer
Post by: Anders Blom on April 3, 2017, 23:17

You can pre-optimize without it, but if you want to include it in the total energy, then you also need to optimize with D2.

Title: Re: charge transfer
Post by: monir_apece on April 5, 2017, 16:23

Dear Dr. Blom,

Should I check "No SCF Iteration" to save computational time?

Title: Re: charge transfer
Post by: Anders Blom on April 5, 2017, 16:37

No. Sure it saves time, but also gives wrong results :) Basically this option only works for tight-binding models.

Title: Re: charge transfer
Post by: monir_apece on April 14, 2017, 16:33

Dear Dr. Blom,

For band structure calculation does DFT-D2 has any impact?

Title: Re: charge transfer
Post by: Anders Blom on April 15, 2017, 02:48

Indirectly yes, since it changes the structure and the structure influences the band structure.