Charge Doping

[Fats]

Greetings ATK'ers, I'd like to utilize realistic doping levels in a two-probe structure (n-Si/insulator/n-Si), so I've opted to employ charge doping using the "charge" keyword in my electrode calculators (see below).  As a test, I'm taking an unrealistically high doping level to compare with substitutional n-type doping.

Code

left_electrode_calculator = LCAOCalculator(
    exchange_correlation=exchange_correlation,
    numerical_accuracy_parameters=left_electrode_numerical_accuracy_parameters,
    poisson_solver=left_electrode_poisson_solver,
    charge=-1
    )

When I run the calculation, however, the placement of the Fermi energy falls directly in the middle of the Si band gap (as if there was no charge doping).  This disagrees with the bulk calculation I did using the same electrode structure, k-point density, and doping level.  In that case, the Fermi energy lies extremely close to the Si CB (as expected).  Am I missing something in my input file?  Thanks for the help.

[zh]

I guess that the use of charge doping by changing the total number of electrons  may be only suitable for the bulk and molecule configurations, not for the electrodes in a device configurations.

[Anders Blom]

I also played with the charge to tune the Fermi level in semiconductors. I didn't use such extreme levels, since you turn Si metallic basically, I was more in the 0.01 range, and it was possible then to move the Fermi level from the middle of the gap all the way up to the CB.

So, I'm a bit surprised about the device behavior... I'll have a look and see if there is something missing - not in your script, but perhaps in the algorithm, to handle this.

[Anders Blom]

It turns out that a bug (or rather, a feature not implemented yet) prevents this from working properly for devices. This is now fixed and will be part of ATK 12.2.2 (coming out in 1-2 weeks), along with an updated implementation of our complex band structure routines.

[Anders Blom]

We're getting close to having 12.2.2 finalized. It was delayed because we found a really important solution to a performance issue in the transmission spectrum that had been troubling us for a while. So hopefully now 12.2.2 can be out relatively quickly.

[Anders Blom]

12.2.2 is now available - sorry for the delay!