Problem in understanding mobility calculation using HSE corrected electronic structure in Quantum Espressso

[souravrudra]

Dear developers,
I have a basic doubt regarding mobility calculation of carriers using Quantum Espresso and EPW. As DFT underestimate the band gap of silicon and some other materials we need to do HSE calculation in QE to correct evaluation of electronic structure. HSE calculation in QE is not support with nscf calculation, but epw needs nscf calculation. So after searching in internet i found that we need some prefix.eig file from HSE or other hydrid functional calculation that contains the same no of kpoints and nbnd as in nscf.in file. So I am not getting the sequence of the caculations. I thought first we have to perform hse calculation in a seperate folder, at the end of hse calculation with wannier90 it generates a pfefix.eig file in the, then creating a new directory, copy the prefix.eig file into the newly created directory and calculate normal scf, phonon, nscf and epw. In case of epw calculation in the input file we have to write eig_read=.true. and follow the same procedure as in tutorial.

Is the above described procedure correct for mobility calculation of material like Si with HSE corrected electronic structure?

I have attached one prefix.eig for 30 bands 2 k points file here generated by wannier90 interpolation. Can you check whether the prefix.eig file generated by wannier90 is correct in format or not?

Please correct my doubt.

Thanks with regards
Sourav Rudra
PhD student
JNCASR, Bengaluru, India

[hpaudya1]

Hi Sourav,

Please read this input description https://docs.epw-code.org/doc/Inputs.html#eig-read
You need to provide the same number of bands as in the nscf calculations and all k-points through 'prefix.eig' file.

To obtain 'prefix.eig' file with HSE calculation, use uniform k-points (provide a list of k-points you use for nscf input) in scf.in.

You may need to work a little bit on the formatting of 'prefix.eig' given by the wannier90 code to make it compatible to EPW, which should be as follows (for MgB2);

nbnd =, k-points =, 16 27
0.0000000 0.0000000 0.0000000 #coordinate of k-point
-4.86220 #eigenvalues
4.52220
7.91910
7.91910
9.17740
13.79120
13.79120
15.67190
17.47370
20.23630
24.42660
24.42660
26.25060
26.96580
26.96580
27.42350
0.0000000 0.0000000 0.2919000
-4.00870
0.95210
8.17820
8.17820
14.15360
14.15360
14.94580
15.51600
17.44260
20.74670
20.74670
20.88070
22.55800
22.55800
24.67120
27.84270
......
......

I hope it helps.

Please remove the duplicate post that you just copied.

Happy EPWing,
Hari

[souravrudra]

Thank you for clarifying me. I will try with this. Also, Is there ant tutorial example regarding this?
I want to know about the scissor term in epw input, it says that it is used for band gap correction. So if i am doing a calculation without any band gap correction in a semiconductor and running EPW to calculate hole mobility with in some energy window near fermi level, including scissor then will it shift the conduction band minima upward? Recently I am doing mobility calculation in which the scf output file write highest occupied, lowest unoccupied level (ev): 12.5905 12.5761 . It means a negative gap, fermi level croses both VB and CB . But it is a semiconductor with opening gap with hse or dft+U calculation, so at the time of hole mobility calculation boths states (VB and CB)are taken into account within the width of energy window near fermi level. That will affect the hole mobility value. In this case will applying a significant amount of scissor vale in epw input will help to solve the problem i.e. it will correct the gap and shifts the CBM to upwards such that fermi level do not cross the CB.

Thanks with regards
Sourav Rudra
PhD student
JNCASR, Bengaluru, India