QuantumATK Forum

QuantumATK => General Questions and Answers => Topic started by: Dipankar Saha on July 30, 2015, 20:47

Title: ATK-SE/DFTB-CP2K/Extended Huckel
Post by: Dipankar Saha on July 30, 2015, 20:47

Hello,

The predefined  S-K basis set for MoS₂ ....I find CP2K, non self consistent ...

What does this  'non self consistent' mean_
a frozen core approx. ? ... confined atomic orbitals in Slater type representation??

Regards_
Dipankar

Title: Re: ATK-SE/DFTB-CP2K/Extended Huckel
Post by: Anders Blom on July 31, 2015, 13:07

It means there is no charge transfer and no electrostatics. It also means, don't use it for MoS2 :) It's not accurate in any way. The only reason we ship these DFTB sets is so one can test some calculations with DFTB, but for accurate simulations one needs to use parameters from DFTB.org (to the extent they are accurate) and/or do some verification testing on known systems, comparing to DFT.

Title: Re: ATK-SE/DFTB-CP2K/Extended Huckel
Post by: Dipankar Saha on July 31, 2015, 13:20

1)
What if... despite of using such a basis set ....one goes for the SCF iteration...??!!    /  Does not it bring some approximated atom-based electron density...in to the picture....??!!
2)
Isn't it still possible... to find a Hartree potential....based on the redistribution of Mulliken charges (Just like...as it happens with the consideration of the 2nd order term.... in any standard TB) ??

Thanks & Regards_
Dipankar

Title: Re: ATK-SE/DFTB-CP2K/Extended Huckel
Post by: Anders Blom on July 31, 2015, 14:03

1. It works technically, like you say. Only the problem may be that the parameters are fitted for non-selfconsistency (DFTB1 model, vs. the SCC DFTB2).
2. Sure, like I said in 1, it works, but are the results good? Depends on the quality and type of parameters you have.

Title: Re: ATK-SE/DFTB-CP2K/Extended Huckel
Post by: Dipankar Saha on July 31, 2015, 14:28

Ohhh....that's a relief (somewhat)...!!! :P  / Thanks a lot Anders for sharing your point of view...!!  :)

Best_
Dipankar

Title: Re: ATK-SE/DFTB-CP2K/Extended Huckel
Post by: Dipankar Saha on July 31, 2015, 17:22

Quote from: Anders Blom on July 31, 2015, 14:03

Sure, like I said in 1, it works, but are the results good? Depends on the quality and type of parameters you have.

While comparing  with the DFT calculation, will it suffice the case...if we merely look into the band dia. ??

Regards_
Dipankar

Title: Re: ATK-SE/DFTB-CP2K/Extended Huckel
Post by: Anders Blom on July 31, 2015, 21:16

It's a good start, and to be honest pretty much all you can expect to get from a DFTB calculation, i.e. there is not much more to compare. Well, DOS, but that's closely related to the band structure. But of course, just because the band structure is ok doesn't mean everything else will match too, esp. it doesn't say much about how the DFTB parameters will transfer to another system (diamond C to graphene, for instance, etc). However, at least DFTB seems much more transferable than orthogonal tight-binding.

Title: Re: ATK-SE/DFTB-CP2K/Extended Huckel
Post by: Dipankar Saha on August 2, 2015, 08:12

That's quite true......./ Thanks again Anders ...!!!  :)

Besides, I never worked with Hückel.... How about that?

For self consistent calculations....how to include onsite shifts?

For various elements of a composite system.....the basis types... which we use may not  be the same..... In that case,  whether the default vacuum levels for various atoms  do ensure the same energy zero for all atoms ....or, we need to modify those??