Author Topic: A problem of finite bias in spin calculation  (Read 100 times)

0 Members and 1 Guest are viewing this topic.

Offline qiuminz

  • Regular ATK user
  • **
  • Posts: 17
  • Country: cn
  • Reputation: 0
    • View Profile
A problem of finite bias in spin calculation
« on: May 8, 2019, 08:56 »
Dear QuantumwiseStaffs:
       Recently, I am tring to reappear this work :doi.org/10.1088/1361-648X/ab1732.  few days ago, I faced the converge problem:https://forum.quantumatk.com/index.php?topic=6045.0
      Thanks to Mr Anders Blom, his advice is useful , I get the converged result in 0V bias. But  when bias applied, the convergence is geting worse. Even small bias voltage such as  0.1V or -0.1V applied, the calculation still get converged.
     I have tested many LCAO Basis SET, and found that HGH basis set is suitable for the bulk and device configuration at 0V voltage.   And I found that  if i use GGA -PBE for nonspin calculation, all bias voltages will well converged. Is it means that my device configuration is  reasonable´╝č
    My  ATK version is 2019.03  and the author's  is Atk2015, does the different atk version cause these problem?
    In  nonspin calculation, only spin up electron(default) is consided, and in spin calculation with SGGA, spin up and down electron are consided. can we  only consider  spin up(down)  electron?  does these two types of Hamilton in density matrix cause the numberical error  in the NEGF  step?
   I will upload my py files , could you tell me the reason of the poor convergence?  If you want to reappear the work, what bias set you will use?
   PS; all the converge skills in the guidelines i have tried,  only Mr Anders Blom's advice is work in this calculation.

Offline qiuminz

  • Regular ATK user
  • **
  • Posts: 17
  • Country: cn
  • Reputation: 0
    • View Profile
Re: A problem of finite bias in spin calculation
« Reply #1 on: May 8, 2019, 09:01 »
In this calculation , 0.1 V is OK, 0.2,0.3,0.4V cant get converged.  In the artcile, they calculate from -1V to 1V