My question is both specific and general. While looking through the materials project database I found this entry. The band structure reported and density of states are dramatically different it seems and I am unsure what to make of this.

Does materials project determine band gap via band structure or density of states and how is such a large disagreement in energy alignment possible (or is it not in general).

enter image description here

  • 5
    $\begingroup$ It looks like there is a scale alignment problem. If you "move" the DOS graph down, you can "fix" the problem. From dataserver, they reported a gap of 2.283 eV, maybe the Fermi was shifted to zero in only one graph. $\endgroup$
    – Camps
    Jan 3 at 14:22
  • $\begingroup$ @Camps I somewhat agree but when you look at the band structure, there is nowhere to get a 2.283 eV band gap. $\endgroup$ Jan 3 at 18:36
  • $\begingroup$ This system is AFM. This is not reflected in your bandstructure but partially considered in your dos plot. I mean the bandstructure is mixing the up-spin and down-spin but the dos consider only the up spin. $\endgroup$
    – Jack
    Jan 4 at 0:56
  • $\begingroup$ @Jack It considers both, spin up and spin down match in this structure so they overlap. The magnetic moments should exactly cancel out on the same elements and the AFM structure is degenerate. I think spin orbit effects will break this, but they are definitely not considered here. I am starting to think that the BS calculation and DOS may be in different magnetic configurations though based on your answer. Might be hard to confirm this without reproducing the calculation though. $\endgroup$ Jan 4 at 3:42
  • 1
    $\begingroup$ What kind if plotting program have you used to draw this ? $\endgroup$
    – Chi Kou
    Jan 5 at 6:32
  • In short, you can't completely believe the data supported by the Materials Project.

  • Specific to the Co2W2O8 entry, there are many self-contradictory points shown on their webpage. This can be seen easily when you take a look at the INCAR and POSCAR files downloaded from their website:


    ALGO = Fast
    EDIFF = 0.0006000000000000001
    ENCUT = 520
    IBRION = 2
    ISIF = 3
    ISMEAR = -5
    ISPIN = 2
    LASPH = True
    LDAU = True
    LDAUJ = 0 0 0
    LDAUL = 2 2 0
    LDAUTYPE = 2
    LDAUU = 3.32 6.2 0
    LMAXMIX = 6
    LORBIT = 11
    LREAL = Auto
    LWAVE = False
    MAGMOM = 2*0.6 2*5.0 8*0.6
    NELM = 100
    NSW = 99
    PREC = Accurate
    SIGMA = 0.05 

    Co2 W2 O8
    4.700949 0.000000 -0.062527
    0.000000 5.766850 0.000000
    0.000000 0.000000 5.061702
    Co W O
    2 2 8
    0.500000 0.342041 0.250000 Co
    0.500000 0.657959 0.750000 Co
    0.000000 0.827445 0.250000 W
    0.000000 0.172555 0.750000 W
    0.259471 0.374817 0.899680 O
    0.740529 0.374817 0.600320 O
    0.259471 0.625183 0.399680 O
    0.740529 0.625183 0.100320 O
    0.217971 0.108689 0.431476 O
    0.782029 0.891311 0.568524 O
    0.217971 0.891311 0.931476 O
    0.782029 0.108689 0.068524 O
    • It seems this INCAR will be used to relax the structure in FM spin-polarized mode (ISPIN=2, MAGMOM = 20.6 25.0 8*0.6). However, the website shows the magnetic ordering is AFM.
    • What's more, the main initial magnetic moment should be placed on Co atoms rather than W atoms.
    • If you use ISMEAR=-5, it is not necessary to set SIGMA.
  • Return to the disagreement between the band structure and density of states (DOS) disagreement. I think the bandgap of 2.283 eV shown on their webpage is read from the DOS. However, the band structure and DOS are not completely exhibited no matter what the data are calculated with AFM or FM spin-polarized mode. I mean there should be two spin channels in their band structure and DOS.

  • Therefore, I strongly suggest you do some own calculations with the structure supported by the Materials Project to obtain reasonable results.

  • $\begingroup$ Thanks for giving an answer to this question, I think everything you have mentioned is correct. $\endgroup$ Feb 21 at 21:29
  • $\begingroup$ @TristanMaxson You are always welcome. Interestingly, I obtain a downvote for this question. $\endgroup$
    – Jack
    Feb 21 at 22:38

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.