Now, LORBIT and RWIGS

Message

dcurulla · #1 Post by **dcurulla** » Thu Aug 30, 2007 3:28 pm

Hi again!

I have tried to do LDOS using LORBIT = 11 and the resulting LDOS is weird in the sense that it fully misses important features at 5 eV (5sigma) and close to the Fermi level (2pi). I have checked the OUTCAR file, and to my suprise the "total charge" was only of ~180 electrons out of a total of 210.

So, I have decided to use LORBIT = 5 and I am playing with the CO molecule in order to optimize the values for RWIGS. I hope that they will be transferable, won't they? Has somebody else done this already? Could you please share with me what RWIGS have you used for CO?

Thanks

Dani

aron · #2 Post by **aron** » Thu Aug 30, 2007 6:46 pm

I find it useful to look at the valence charge density. That can give you a ratio to work with between different ions. Then scale it with respect to the bond lengths, and adjust until you get the best compromise between overlap/electron counting.

#3 Post by **admin** » Fri Aug 31, 2007 1:49 pm

the 'total charge' in OUTCAR is the total LOCAL charge (ie that part of integrated charge density which lies in the volumes of the spheres around the atoms), NOT the total number of electrons.
If the difference seems very lare to you:
Check the number of electrons in OUTCAR (NELECT), if the total number of valence electrons is wrong in OUTCAR, there must be an error in POSCAR (not all atoms defined in the line giving the number of atoms per type)

concerning the DOS please check the total DOS if that peak is a narrow peak. (the integrated DOS has a step without a visible peak at this energy). If so, increase NEDOS such that the energy interval given by dividing the E-range by NEDOS is smaller than the peak width. RWIGS only influences the weight of the states, if the peak is not visible in your total-DOS-plot as well, increasing RWIGS will have no effect.
[ Edited Fri Aug 31 2007, 04:08PM ]