Why 3d xz and yz DOS are not equal?

[blazing]

Hello, everybody!
i perform static calculation of DOS (no relaxation). calculated partial Fe 3d xz and yz DOSs fairly different. But this planes of supercell is equivalent. so i would expect 3d xz and yz DOSs to be equal. why they so different? I tried both ISYM=0 and ISYM=2 (i use PAW) but the result is the same.

i use G-centered mesh of 12x12x4 k-points and following INCAR and POSCAR

SYSTEM = Si-Fe

ISTART = 0

PREC = High

IALGO = 38

LORBIT = 11

VOSKOWN = 1

ISMEAR = -5

SIGMA = 0.2

ICHARG = 2

ISPIN = 2

MAGMOM = 2*2.2 22*0

ENCUT = 500.00 eV

RWIGS = 1.45 1.32

EMIN = -12.0

EMAX = 12.0

NEDOS = 4500

NELMIN = 7

ISYM = 0

EDIFF = 1E-04

NELM = 100

LMAXMIX = 4

_________________________________________________________

Si-Fe
5.46629641000000
1.0000000000000000 0.0000000000000000 0.0000000000000000
0.0000000000000000 1.0000000000000000 0.0000000000000000
0.0000000000000000 0.0000000000000000 3.0000000000000004
2 22
Direct
-0.0000001665126154 0.9999994757712332 0.9999991992102141
0.4999999242219825 0.4999994044806253 0.0000008427821678
0.2502413656913599 0.2502414702442717 0.0755009127318464
0.7497579098538607 0.7497580882263163 0.0755009320541033
0.4999999705984418 0.0000001275445618 0.1612810753170716
0.0000000216216688 0.5000000754537444 0.1612810494549297
0.7499987234778644 0.2500014356763398 0.2463117951690840
0.2500013019109163 0.7499987895118092 0.2463117847534330
0.0000000311016059 0.0000000945531129 0.3308557753582138
0.5000000142102849 0.5000001140466511 0.3308574548101630
0.2500016165761108 0.2500016947636573 0.4153651401647637
0.7499984291188296 0.7499985260657783 0.4153651332107411
0.5000000305006924 0.0000001093129547 0.4999999875402176
0.0000000163364906 0.5000001018218755 0.4999999848917632
0.7499984252517670 0.2500017248790424 0.5846348729943670
0.2500016491229251 0.7499984894648687 0.5846348812144275
0.0000000366874459 0.0000001149173411 0.6691441679722935
0.5000000242707757 0.5000000879383629 0.6691425623870116
0.2500012275193796 0.2500013031597314 0.7536881966810683
0.7499988361922804 0.7499988878872008 0.7536881993653227
0.5000000096212143 0.0000000487501474 0.8387189415755849
0.0000000663320584 0.5000000988598432 0.8387189406447495
0.7497563792510331 0.2502437502701064 0.9244990589111671
0.2502441570436415 0.7497559864004311 0.9244991108052957



With much appreciation


Mikhail
<span class='smallblacktext'>[ Edited ]</span>

[blazing]

I also calculated DOS for bulk Fe (bcc) with 2 atoms per cell. The unexpected result was that 3d xz, yz and xy partial DOS were unidentical. maybe i make some mistake in my calculation set up? or what is it otherwise?

The setup was following:
KPOINTS

Automatic mesh
0
Monkhorst Pack
20 20 20
0. 0. 0.

POSCAR

Fe
2.836639314
1.0 0.0 0.0
0.0 1.0 0.0
0.0 0.0 1.0
2
Direct
0.00 0.00 0.00
0.50 0.50 0.50

INCAR :

SYSTEM = Fe

ISTART = 0

LORBIT = 11

PREC = High

IALGO = 38

NELMIN = 7

NEDOS = 4500

VOSKOWN = 1

ISMEAR = -5

SIGMA = 0.2

ICHARG = 2

ISPIN = 2

MAGMOM = 2*2.5

ISIF = 7

#NSW = 40
NSW = 0

#IBRION = 2
IBRION = -1

EDIFF = 1E-06

ENCUT = 600.00 eV

#EDIFFG = .1E-02


With much appreciation


Mikhail

[forsdan]

In the DOS calculations I've performed for bulk bcc Fe, the 3d xy, yz, xz partial DOS for each atom (both spin components seperately) are in very good agreement with each other (I just doublechecked my previous data). There is only minor differences due to the k-point mesh I used in that case. So I have to come back to the comment I gave in the first post you made. Are you sure that you are comparing the correct columns? See:

http://cms.mpi.univie.ac.at/vasp-forum/ ... php?4.5940

Your input-files look reasonable, except although NEDOS = 4500 is very a high number. I think that will result in a very spikey/peaky-looking DOS. I think it's betting to keep it at least under 1000 or even at the default 300.

Cheers,
/Dan Fors

<span class='smallblacktext'>[ Edited Wed Aug 19 2009, 09:08AM ]</span>

[panda]

Try this and see if it works:
Fe
2.836639314
-0.5 0.5 0.5
0.5 -0.5 0.5
0.5 0.5 -0.5
2
Direct
0.00 0.00 0.00
0.50 0.50 0.50

This is how I would orient an (fcc) cell. I think the way you have your orientation set up the planes are in fact not equal. I don't know, try it and see. The worst that can happen is the same thing, the best that can happen is that you get the result you want!

[admin]

please note that there is no symmetrization of the local partial DOS done by vasp after the scf cycle. Therefore, minor differences eg in the local partial DOS of symmetry equivalent atoms may occur

[panda]

Sorry, Is it then necessary to switch symmetry off when doing the calculation? ~Thanks