Number of electrons from Projected DOS

[Masoud]

I have simulated a bulk system that includes transition metals and oxygen atoms. For the projected DOS calculations, the k-point meshing was 16 x 16 x 16 Monkhorst pack. The energy range consists of 301 data points.

According to my knowledge, if one integrates the orbital sum of projected DOS for each atom up to the Fermi level, it must equal the number of electrons attributed to that specific atom. And then, summation of all these atomic occupations should equal the total number of electrons in the cell.

When I integrate the total DOS up to the
Fermi level, I do get correctly the total number of electrons (162 electrons). However, the sum of occupation numbers for all atoms is much less (144 electrons).

I would greatly appreciate that anyone let me know

1) if one can expect the sum of partial occupations be equal to the total number of electrons or not.

2) If yes, how can one calculate the occupation of orbitals on individual ions in a more accurate way to see the above equality.

Thanks in advance

[forsdan]

This question has been discussed before in the threads. The projected DOS only accounts for the charge inside the RWIGS-radius so you'll never get the total number of electrons since there are interstitial regions that are not included.

You can optimize the projections by chosing the RWIGS-values as large as possible but with the condition that they don't overlap each other, but remember that there is however no unique way to choose the values in a multicomponent system. You'll however never get 100% of the electrons.

For further discussions I recommend you to search the threads in the forum regarding this issue.

Best regards
/Dan Fors
<span class='smallblacktext'>[ Edited Tue Mar 25 2008, 06:17PM ]</span>