On Fermi Level

[rose]

Hi
it is well-known that mono-layer MoS2 and WS2 can affected by the substrate and may show p or n type behavior, but, in pristine suspended structure, MoS2 has n type characteristic while WS2 shows p type. I try to find the bandstructure of both MoS2 and WS2 but in most cases i got intrinsic ( Fermi at the center of band gap) or sometimes p type one. What should I do to simulate this natural n-type characteristic of MoS2?
p.s: i knew about the doping option in atk but this is natural n-type characteristic not adatom of functionalization effect.

thanks
Rose

[Daniele Stradi]

Dear Rose,

I would rather say that if you try to simulate MoS2 using the MoS2 unit cell and without introducing defects, then the system must be intrinsic, if the Brillouin zone is sampled accurately enough.

I think a more correct question to ask would be which is the physical origin for the "natural" n-type characteristics that you mention.

If you want to simply consider an average doping, then the two doping schemes implemented in ATK and described here should suit your purpose: http://docs.quantumwise.com/technicalnotes/doping_methods/doping_methods.html

Regards,
Daniele.

[rose]

Thanks.
I check your idea and improve the k point sampling and I got various type of Fermi level in the band gap. Some show n-type (small values of k point) and some show p-type (large number of k) characteristics. Interestingly the sample did in the quantumwise as the example of transmission of mos2 present slightly p-type characteristic while it is not mentioned explicitly in the text but it can be concluded from the figure.

I would wonder which technique of doping in atk (Explicit charge/Atomic compensation charges) is more effective in considering n-type mos2 as the channel of a transistor? how can I apply it?

thanks
Rose

[Daniele Stradi]

Hi Rose,

for semiconductors such as MoS2, the exact position of the chemical potential inside the gap is heavily dependent on the k-points sampling. If the Brillouin zone is undersampled, there will be an error in the position of the chemical potential, and  the system will appear as doped, even if the system is still charge neutral.

Regarding the doping method, the main difference between the two methods is that in the "Explicit charge" method the extra charge redistributes over the entire unit cell, whereas the "Atomic compensation charges" method allows to localize the extra charge in some specific region of your simulation cell. 

If you plan to study a transistor-like geometry of the type metal-semiconductor-metal, then the "Atomic compensation charges" is the method of choice, because it will allow you to dope the semiconductor without doping the metal, see for example: https://journals.aps.org/prb/abstract/10.1103/PhysRevB.93.155302

If you just want to add an intrinsic doping to a bulk material, then I would use the "Explicit charge" method.

Regards,
Daniele.