Author Topic: DFT calculation result different from EH method  (Read 339 times)

0 Members and 1 Guest are viewing this topic.

Offline weixiang

  • Heavy ATK user
  • ***
  • Posts: 52
  • Country: us
  • Reputation: 0
    • View Profile
DFT calculation result different from EH method
« on: September 13, 2018, 18:30 »
Dear experts,
I am studying the the e-ph effect in GNR transistor devices using the STD method. I calculated the PLDOS for the pristine as well as STD configuration of a GNR device using both EH and DFT method.  In the following images, the left one is the PLDOS result for pristine configuration using EH method, (test3_0.0V_pristine_start_point.py) the middle one is the PLDOS of STD configuration using EH method, (test3_0.0V_start_point_EH.py) and finally the right one is the PLDOS of STD configuration using the DFT method. (test3_0.0V_start_point_DFT.py)


It is noticed that the left and the middle PLDOS are similar, only except that the middle PLDOS is messier.
Realized that the EH method is not suitable for calculating non-perfect configurations like STD, I tried to use the DFT method to do PLDOS calculation as shown in the right image. However the right image looks different from the other two in that the middle band gap region shifted down.

 I believe  there must be one of them being wrong.  Then I calculated PLDOS for a pristine configuration using both EH and DFT method as shown in following images: the left one is for EH method (test_EH.py) and the right one is for DFT method (test_DFT.py)



The two results are different,  the DFT one has small band gap while the EH one has larger band gap. What would be possible reason for this difference?
I have attached all my scripts for calculating these PLDOS and bold their names.
I don't if there is any setup that I did improperly in my DFT scripts, Please advice if I am wrong.


Offline Petr Khomyakov

  • QuantumWise Staff
  • Supreme ATK Wizard
  • *****
  • Posts: 1025
  • Country: dk
  • Reputation: 17
    • View Profile
Re: DFT calculation result different from EH method
« Reply #1 on: September 13, 2018, 21:10 »
The two results are different,  the DFT one has small band gap while the EH one has larger band gap. What would be possible reason for this difference?
Why would you expect the two methods to give the same band gap?

Offline weixiang

  • Heavy ATK user
  • ***
  • Posts: 52
  • Country: us
  • Reputation: 0
    • View Profile
Re: DFT calculation result different from EH method
« Reply #2 on: September 13, 2018, 21:33 »
Because I assume both methods can calculate the band structure feature with reasonable accuracy. So I think the results of the two method should be same or very similar to each other.

Offline Petr Khomyakov

  • QuantumWise Staff
  • Supreme ATK Wizard
  • *****
  • Posts: 1025
  • Country: dk
  • Reputation: 17
    • View Profile
Re: DFT calculation result different from EH method
« Reply #3 on: September 13, 2018, 22:32 »
Because I assume both methods can calculate the band structure feature with reasonable accuracy.
Why do you assume that both methods can calculate the band structure with reasonable accuracy for your particular system of study?  Did you do any benchmarks for the corresponding bulk system, for instance?

Also, your device seems to be shorter than the screening length, see https://docs.quantumwise.com/tutorials/atk_transport_calculations/atk_transport_calculations.html.

Offline Anders Blom

  • QuantumWise Staff
  • Supreme ATK Wizard
  • *****
  • Posts: 4974
  • Country: dk
  • Reputation: 78
    • View Profile
    • QuantumWise
Re: DFT calculation result different from EH method
« Reply #4 on: October 2, 2018, 22:02 »
The 3 words "DFT", "bandgap", and "accurate" are dangerous to put in the same sentence. It is a topic that requires great care. ATK contains many methods for improving the band gap in DFT (which typically is moderately to severely underestimated), but one should carefully study this on the bulk material before attempting a full device calculation, in case the gap is important (which isn't always the case).