My Community

Dear Sir,

After doing the static calculation for DOS of a metal system, i find the integrated DOS and the projected DOS provided in DOSCAR are inconsistent. For example, up to Fermi the integrated DOS is 4, but if i calculate the DOS manually by summing up the projected DOS and integrating, the result is only about 3.2 up to the Fermi level. Why this inconsistency happens?Does this mean the projected partial DOS data are not correct?

regards
sincerely,
tom
08/30/09

Linglin,

you have to learn about how the projection is carried out: the density is projected into spheres. If you go back to your solid state course you have learned that maximum packing of equally sized spheres will result in a density of about 70% ... In your case you have diffently sized spheres. You have about 80% of the density. That's where the difference comes from ...

cheers

alex

alex, thanks very much for your reply. i still wondering if i wanna use these spherically projected DOS to integrate for the occupation number of certain orbit, for example the d-orbital occupation number, how should i do the integration? i mean the direct integration from the projected DOS of d in DOSCAR cannot produce the right number of electrons on the orbit, right?

Right. But this is the only thing you'll get (as far as I know). So you have to live with it. If you have spin polarized system, there is a nice trick, if you are interested in e.g. unpaired electrons. You use the (alpha - beta) block in CHGCAR (the 2nd set), look at it, if the charge is localised. If yes, you go back to PDOS and compare it with your projected value (probably something around 0.8 e).

If you do not spin polarized systems, just stick to your numbers, explain clearly, what you have done to get them and then do your argumentation.

Hth

Alex