Greetings!
In energy and relaxation calculations, I do a convergence test to determine the best number of K points. I do this by varying the k points number N x N x 1 and getting the energy. When the energy difference between the N x N x 1 and (N-1) x (N-1) x 1 is no longer greater than 0.001 eV, I take (N-1) x (N-1) x 1 as the K points settings.
My question is this: Is there a systematic method similar above in order to determine the best number of k points for DOS calculations? I read in the VASP manual that it should require more k points, but how should I know how many is enough?
Thank you very much.
http://cms.mpi.univie.ac.at/vasp/vasp/Accurate_DOS_Band_structure_calculations.html
Hi Matthieu,
Thank you for the reply. Actually, for the energy calculations, I did not look much into ISMEAR but simply used ISMEAR=0, SIGMA=0.1 (Gaussian smearing). The reason is that my system is hypothetical (I do not know yet if is it insulator or metal) and studies with similar systems used the same ISMEAR settings. I only did the convergence test for the k points sampling size.
Regarding comparing DOS obtained with the different mesh size, may I ask if there is a "mark" of what we can say as correct DOS? Like some characteristic of the graph which I can use as criteria to say that I do not need to increase the sampling any further?
Lastly, thank you very much for the tip regarding ISMEAR=-5. I was planning to use that settings once I started doing the DOS calculations.
If ever I find anything helpful regarding this, I'll let you know. Thanks again!
Hi Shaira,
I also followed the methods explained here when I first started computation modeling using Materials Studio.
Just recently I started using vaspkit with my new research . In vaspkit there is an option to create a KPOINT mesh using your POSCAR file and the program will then give you an option of which mesh quality to use but it will also tell you which quality is optimal to be used. Using vaspkit decreases the amount of time spend om finding the KPOINT for your research.
The best way to decide whether your KPOINTS are sufficient to determine the correct DOS is to see the bandgap from your DOSCAR. If the bandgap is same as that obtained from band structure calculations, it means your KPOINTS are dense enough to determine correct DOS. If the band gap is not the same in your DOS and band structure calculations, it means you have not taken a dense enough KPOINTS grid to determine the correct DOS. Thus, go on increasing the KPOINTS grid (i.e., make it dense enough) till the bandgap obtained from BS and DOS calculations are the same.
You can also use vaspkit program for getting KPOINTS file easily for your DOS and BS calculations.
- Badges
- Science topic
- Similar topics
- Filing