Can semicondutors act as an electrode?

[wanghao1889]

Dear ATK experts:
    Nowadays, I am quite confused about whether the semicondutors can act as an electrode. In the guide "Transport calculations with ATK", it says that "In systems with semiconductor electrodes, the screening length can be quite long, and a very long central region may be required before the potential approaches the bulk value". However, in many published articles (https://pubs.acs.org/doi/abs/10.1021/acs.jpcc.7b07921), I found that they used the intrinsic semicondutors acting as the electrode as shown in my attached figures. So, as far as i am concerned, if the system we studied does not have a interface, just an intrinsic semicondutor nanoribben, can we just using the extended semicondutors nanoribben layers as the electrode,
without doping it? Wow,  I am quite confused now. I will appreciate your suggestions.
    Best wishes!

[Ulrik G. Vej-Hansen]

There is nothing fundamentally wrong with using intrinsic semiconductors as electrodes in a ATK-NEGF device calculation. However, it requires careful convergence of the length of the central region, to ensure that the potential from the scattering region is screened properly.

[Petr Khomyakov]

Strictly-speaking, intrinsic semiconductors can only be used as a material for perfect , ballistic leads attached to a resistor in the central region where the actual scattering takes place, and attached to bulk reservoirs of electrons in the left and right infinity.  I note that the reservoirs are then not explicitly included in the simulations in spirit of the Landauer-Büttiker formalism of the electron transport that is equivalent to nonequilibrium Green's-function formalism (used in QuantumATK) within the framework of the independent particle approximation adopted for ground-state density functional or semiempirical calculations of the electron transport.

In this way, only the geometrical part of contact resistance (=conductance of bulk, perfect semiconductor leads) is taken into account, and metal contact resistance that arises from the details of the actual metal-semiconductor interface is neglected, i.e., the contacts are assumed to be transparent.

If you want to model more realistic electrodes, you may attach the  leads made of the intrinsic semiconductor material  to some metal electrodes. In some cases, having metallic-like electrodes can be achieved by heavy doping the semiconductor leads.

It would be instructive to read some textbooks on the electron transport to understand these matters, e.g., a commonly-used book by S. Datta   https://www.cambridge.org/core/books/electronic-transport-in-mesoscopic-systems/1E55DEF5978AA7B843FF70337C220D8B.

[wanghao1889]

Thanks for your reply! However, I am still confused with that. As they use the intrinsic semicondutor electrode shown in the figures above, the Electrostatic Differential Potential should never be flat near the electrode when applying a bias, as there is actually no scattering point in the perfect nanoribben.
So my question is that can I use the extended semicondutor layers as an electrode to study the IV of the intrinsic semicondutor nanoribben without doping it,  for example MoS2 nanoribben? Is the explaination  of the author above correct?
   
 

[Petr Khomyakov]

Doing non-zero bias calculations for a perfect system with no scatters in the central region does not make much sense. Please read the book by Datta, see also the Forum post at https://quantumwise.com/forum/index.php?topic=4970.msg21537#msg21537.

In general, you may have a screening problem with the leads made of an intrinsic semiconductor materials for zero and non-zero bias calculations of devices (with scatters) if the induced potential in the central, scattering region is long-ranged, making the flat band assumption invalid.  Then you have to use (i)  metal electrodes or doped semiconductor, or (ii) impose the Neumann boundary conditions to have flat bands in the leads.  Unfortunately, the latter is not implemented in QuantumATK.