QuantumATK Forum

QuantumATK => General Questions and Answers => Topic started by: isaac117 on July 10, 2012, 03:49

Title: Charge-missing problem, plz help(structure pre-optimized with VASP)
Post by: isaac117 on July 10, 2012, 03:49

 I have a capacitor test calculation using the ATK 2008, and the charge-missing problem keeps happening in the first few steps of the TwoProbe Calculation, no matter how I tune the parameters. So I wonder maybe the structure itself is not well tuned, although it's well-converged in the VASP code. So I come here to look for help...and I have attached my files below.

Title: Re: Charge-missing problem, plz help(structure pre-optimized with VASP)
Post by: zh on July 10, 2012, 04:45

Now we don't provide the support for the ATK2008.  You can try the latest version of ATK. 

For the charge-missing problem of a two-probe system, it is a typical case that may occur.  You can follow the discussion in the following thread to tune the parameters in your script file:
http://quantumwise.com/forum/index.php?topic=73.msg349#msg349
http://quantumwise.com/forum/index.php?topic=308.msg1784#msg1784

Title: Re: Charge-missing problem, plz help(structure pre-optimized with VASP)
Post by: isaac117 on July 10, 2012, 05:15

    I'll try the latest version of ATK, and thanks for sharing these useful posts. I've read them before, they show that the cause of the problem seems to be very complicated...and maybe no effective solution. However, in postnikov's post he said he solved this problem by shift the atomic coordinates. I guess I should have a try, but I don't quite understand why it works and how to shift those coordinates. I knew ATK2008 is very old version, but I believe there is a way to solve this problem within it, and can help us to work with ATK better.

Title: Re: Charge-missing problem, plz help(structure pre-optimized with VASP)
Post by: nori on July 10, 2012, 17:32

First of all, you should not use 'NeutralAtom' but 'EquivalentBulk' if the electrodes are homogeneous.
In addition, I feel mixing parameters could be improved as mixing = 0.01 and history step = 20.

I've not looked into your structure carefully but it may be better that a few unrelaxed electrode layers are inserted in the central region.

Title: Re: Charge-missing problem, plz help(structure pre-optimized with VASP)
Post by: isaac117 on July 11, 2012, 03:28

   Thanks for the help. I run the calculation with parameters changed as you said. EquivalenBulk run leads to unconvergence in EquivalentBulk Calulation step(in about 200 steps ), and increasing diagonal mixing parameter and history steps doesn't improve the calculation. In the central region I have 3 surface layers of the electrodes, but they are relaxed. I'll try insert some "bulk" Platinum layers between the surface layer and electrode layer.

Title: Re: Charge-missing problem, plz help(structure pre-optimized with VASP)
Post by: kstokbro on July 11, 2012, 07:15

Are you running with the latest version now?

Title: Re: Charge-missing problem, plz help(structure pre-optimized with VASP)
Post by: isaac117 on July 11, 2012, 15:51

Er, not yet. We have a newer version of ATK on the server, but someone else is using it... So I have to wait, or solve the problem within ATK 2008.