carrier concentrations

[zhengjian4210]

Dear Nakib,

I have a question on carrier concentrations.

I saw your discussion with other users about chemical potential. Since I am not going to dope the system, I have the same number for both enref and chempt. But I finally got the result that the system is n-doped, with n = 2.47e19 cm^-3. So I want to know how to set up the parameters for the result without doping. The files are as follows:
elphbolt.out.log
epw.in.txt
epw.out.txt
input.nml.txt
nscf.in.txt
nscf.out.txt

Looking forward to your reply. Thanks.

Best regards,
Jian Zheng

[nakib]

Dear Jian Zheng,

I took a look at your input.nml file for elphbolt. Here's the relevant section:

&electrons
	spindeg = 2
	indlowband = 5 !Lowest transport band
	indhighband = 6 !Highest transport band
	indlowconduction = 5 !Lowest conduction band
	numbands = 8 !Total wannier bands
	enref = 6.7029 !eV, Conduction band minimum
	chempot = 6.7029 !eV
/

Above, you have set the chemical potential to the conduction band minimum. In other words, you have (strongly) n-doped the system. To get to the undoped (or, intrinsic) limit, you need to set the chemical potential in the middle of the band gap. (But there is a catch. This is silicon, and the DFT band gap is too small. You can mitigate this by using a scissor shift in elphbolt.)

A point to note: if you are going to put the chemical potential far away from the conduction and valence band edges, you have to make sure to use a large enough transport active Fermi window (fsthick).

Best,
Nakib

[zhengjian4210]

Dear Nakib,

Thank you for your reply. Is my setup reasonable like this? Among them, 6.68 is the Fermi energy output by epw.

enref = 6.68 chempot = 6.68 scissor=0.45 fsthick = 1

Best,
Jian Zheng

[nakib]

You have to check for yourself what scissor value you need to match the experimentally measured band gap.

As for the choice of where enref should be, it depends on what you want to study. The enref is defined to be the middle of the Fermi window that elphbolt builds. If you need transport in both the conduction and the valence bands, it will make sense to put enref in the middle of the experimental band gap and then use a large enough fsthick to capture enough states in both the valence and conduction bands. Your input file seems to only conduction bands as the transport active ones. So if all you care about is transport of the conduction bands, you can choose the enref to be at the conduction band edge and choose fsthick to only include enough states in the conduction band.

(@dpaulzc and @mrm24 might be able to comment further on this.)

Best,
Nakib

[zhengjian4210]

Dear Nakib,

Your explanation is quite clear. I will incorporate these adjustments in my following attempts.

Thanks.

Best,
Jian Zheng