Band-gap calculation

Hello, everyone!

My question is devoted to incorrect band-gaps (see I am looking for band-gap in MnPS3 and I know experimental value. However, I am a new VASP user, thus, it’s problem for me to understand good or bad INCAR file. I have found but the band-gap value is so far from exp value (more than 50%). I would like to ask how I can change INCAR file from this mp-8613 in GGA & LDA for correct (may be with 50% error) band-gap value?

Hi @A-cubed!

As you’re aware, DFT calculations typically systematically under-estimate the electronic band gap. The accuracy is related to the choice of exchange-correlation functional used – such as LDA or PBE (a GGA functional). All band structures currently reported on Materials Project use PBE even though we know there is a systematic error. These calculations can still be useful, because they can accurately describe other features of the band structure and densities of states, even if the magnitude of the gap is too small.

However, this is a fundamental limitation as a result of the physics behind DFT theory and the PBE functional. Therefore it is not a question of a ‘good’ or ‘bad’ INCAR. Even a ‘good’ INCAR will give an under-estimated band gap if you use PBE.

The important question to ask is what you actually want to know. If you already know the experimental band gap, what new information are you hoping to get from a DFT calculation?

There are ways to improve calculation of band gaps from DFT by using a more computationally-expensive exchange-correlation functional. I’ve had good experiences with MBJ, a meta-GGA, which is not too expensive (they take 2-3x longer than PBE) but the calculations can be tricky to converge. For inorganic materials, HSE06 is currently one of the best functionals we can use and often gives very reliable band gaps, but these calculations are very computationally expensive, which is why we don’t run them on all materials on the Materials Project. The appropriate INCAR tags here are GGA, METAGGA and, for HSE06, LHFCALC and HFSCREEN.

Just a warning though, these aren’t easy calculations to start with if you’re new to VASP, but good luck, and I hope this helps!




Dear Matt,

Thank You for Your response! It is very important for me!
I had only one question about the quality of mp-8613. And, I understand, everything is OK with this calculation and the small band-gap magnitude comes from the used functional. Thank You very much!

1 Like

Dear mkhorton,

I am new to vasp and hence searching for a solution I came across this question. I am trying to find band structure using METAGGA= MBJ tag with LDA POTCAR. The DOSCAR file is giving me a fine band gap 1.15eV but I am not able to plot the band structure. i.e. I also want to find the band gap from the band structure. Also, the energy is coming out to be positive. I think there is a convergence problem. But I am finding it very difficult to figure out. I have used the same procedure for plotting the band structure for LDA and GGA and they are coming fine. How can I solve this?
Thank you very much.