Author Topic: monkhorst paramaters and periodicity for the scattering region  (Read 3281 times)

0 Members and 1 Guest are viewing this topic.

Offline cwolowiec

  • Regular QuantumATK user
  • **
  • Posts: 13
  • Reputation: 0
    • View Profile
Hi,

When I define the brillioun zone monkhorst parameters in the VNL NanoLanguage Tool, the resulting script returns these parameters for both the left and right electrodes. In the same resulting script, however, there are no brillioun zone parameters defined for the central (or scattering) region. I am curious how ATK performs the brillioun zone integration for the scattering region during the scf calculation and whether the scattering region is treated as periodic like the electrodes or as non-periodic during the scf calculation. Any insight is appreciated.

Thank You,

Chris



Offline Anders Blom

  • QuantumATK Staff
  • Supreme QuantumATK Wizard
  • *****
  • Posts: 5411
  • Country: dk
  • Reputation: 89
    • View Profile
    • QuantumATK at Synopsys
Re: monkhorst paramaters and periodicity for the scattering region
« Reply #1 on: January 29, 2009, 21:42 »
The scattering region takes the k-point specification from the electrodes in the X and Y directions (or, rather, A and B, as defined by the electrode unit cell vectors). In the Z (transport) direction, however, there is no periodicity. This is the whole point of the open boundary conditions that allows ATK to apply a finite bias across the structure; electrons flow through the system which is in non-equillibrium.

During the initial "equivalent bulk run" which is turned on by default, the system (Left electrode + Scattering region + Right electrode) are turned into a bulk system and run as such for the purpose of creating a better initial guess for the density matrix that that provided by neutral atoms. During this calculation (which can be turned off, and should be turned off if the equivalent bulk system does not fulfill periodic boundary conditions geometrically or parameter-wise, for instance in case of anti-parallel spin polarized electrodes), the k-points are (NA,NB,NC=1) where NA and NB are taken from the electrode k-point specification. There shouldn't be any need for k-points in C (Z, that is) since the system is quite long in this direction, and anyway the equivalent bulk calculation is just an approximate first step towards the real results.