Can we consider van der Waals interations in device simulation？

[yeshizhuo]

Dear QuantumATK staffs and users:
  Happy new year! I have read some papers using QuantumATK software. Some of the authors stated that they have considered van der Waals interations in their device simulations.    And one of the authors said he have included the van der Waals interation in his calculations even using the semi-emprical Extended Huckel calculator. I think it is really amazing!  As known, QuantumATK give a online tutorial about the DFT-D method and the basis-set superposition error .(https://docs.quantumatk.com/tutorials/dispersion_corrections_and_bsse/dispersion_corrections_and_bsse.html) This is very helpful to understand the interaction between atomic layers. What's more, it tell us the existence of the so-called basis set superposition error (BSSE). I have tested the DFT-D method combined with the counterpoise correction in my device simulations. I have used the GGA and Pesudo Dojo parameters (with version 2018.06-SP1). The log file give a warning that  DFT-D is not included in the device simulation. I have to give up a good idea. That's why I am so amazing about the above papers. Maybe I have written the scripts in a wrong way. And I want to know whether we can consider van der Waals interations in device simulation. If that's possible, can someone give a help document or a script file. I will appreciate you sincerely. If that's impossible, can the QuantumATK engineer complete this function in the next version? (I know nothing about the  implementation of NEGF. If that's too difficult or impossible in principle, please fell free to tell me)

[yeshizhuo]

And if that's impossible in principle. How should we treat these papers that considering the van der Waals interations in device simulations using QuantumATK. (One of my friends is reviewing one of them  . It really troubles him)

[Petr Khomyakov]

Judging on general grounds, I would say including correction terms for van der Waals interaction makes sense when doing geometry optimization of a van der Waals system. The obtained structure can then be used to build a device configuration with no van der Waals terms included, since they would have a minor or perhaps no effect on the electron transmission, as they do not cause any hybridization of states.

So, it is very likely that the authors mean structural relaxation done with van der Waals corrections, but not the actual device calculations to get conductance or current for a fixed device structure.

[yeshizhuo]

Got it! Thanks for your patience and wisdom.