How do we calculate bandstructure using hybrid functionals in VASP?
I am trying to calculate band structure in VASP using the HSE06 hybrid functional but I get this error. In my calculation, I have done scf and then nscf(bands) calculations but the nscf calculation stops and gives the error message below. I have attached the script here(band.sh) and the POTCAR file I am using. Kindly let me know how I can solve this.
==============ERROR===================
| internal error in: mkpoints_full.F at line: 673
| Two k-points are equivalent 10 11
| If you are not a developer, you should not encounter this problem
| Please submit a bug report.
Dear Sylvester Makumi ,
For band structure with PBE or LDA functionals, we use nscf calculation. You need ICHARGE=11 and for KPOINTS, you need to give high symmetry points based on your crystal structure which can be determined by :
Article Setyawan, W. & Curtarolo, S. High-throughput electronic band...
or,
https://www.materialscloud.org/work/tools/seekpath .
Whereas, for HSE, you need to have an SCF calculation in order to get WAVECAR and IBZKPT then use them in the next step for HSE calculation. The IBZKPT file should be renamed as KPOINTS file for your HSE calculation.
Bests,
Nima
Dear Nima,
Thank you for your response. I have done as you have said however, I get a bandstructure that looks ugly.
Hi Sylvester Makumi
I am wondering whether you just copied IBZKPT into KPOINTS, or also added high symmetry points with zero weight from OUTCAR of PBE band structure to it or not? Can you share your KPOINTS file for HSE bands?
In HSE band structure, bands at initial KPOINTS are random which are taken from the IBZKPT file. Real band structure is from the point where you provided the high symmetry path. like in this example at
https://www.vasp.at/wiki/index.php/Si_bandstructure
Hybrid calculation using a suitably modified KPOINTS file
Explicit k-points list 18
Dear Sylvester Makumi
The reason that you get such zigzag band is because you doesnt provide the WAVECAR for your calculation.
As mentioned in VASP:
Please note that step two requires a WAVECAR obtained from a standard DFT run (not an HSE calculation), otherwise the resulting conduction bands often have a zig-zag structure.
one again with reading the WAVECAR in your calculation with ISTART tag and provide the WAVECAR file in the directory.
Bests,
Nima
Thank you Nima Ghafari Cherati and Payal Wadhwa for your contribution. However, I have not succeeded yet. I get a bandstructure with two sections, it looks like it calculated the bands over the k-points from IBZKPT and the ones I supplied with zero weight. I attach here the KPOINT file and the bandstructure I get.
Hi Sylvester Makumi
you can see in your KPOINTS file that first you have random KPOINTS from IBZKPT and then you have actual high symmetry path for band structure with weight equal to zero.
These random KPOINTS from IBZKPT is giving you the zigzag band structure, which you need to skip. Your actual band structure starts from the point where high symmetry path starts, which is around 2.25 k-distance in your attached image of band structure.
So just skip the initial kpath, then you will get the HSE band structure, which will also have same curves like PBE.
Hope this helps!