lm-decomposed DOS calculation of LiMnO2

[ornis]

Hello Everyone,

I am trying to calculate the lm-decomposed density of states of LiMnO2 using GGA+U with ISPIN=2 and LORBIT=11. In the layered LiMnO2 (C2/m, no. 12), the Mn ions are in the center of MO6 octahedra and the Mn ions are in +3 oxidation states with high spin. Thus the Mn ions have 4 electrons in 3d orbitals. In line with the crystal field theory, the 3d orbitals degenerate in t2g and eg states, where, t2g have 3 of the valence electrons and eg have 1 electron. Now, due to the Jahn-Teller effect, the spin-up channel of the eg state will further split into dz2 and dx2-dy2, as reported in J. Phys. Chem. C 2015, 119, 21904−21912.

However, my DOS calculation showed that in the spin-up channel of Mn 3d, the dxy and dyz orbitals have large contributions above the Fermi level, which should not be the case. The spin-down channel is found to be in agreement with the reported studies. But the spin-up channel does not match with the reference as well as theory. I have checked both the ferromagnetic and antiferromagnetic case, but the DOSs are similar. I have also checked with higher NBANDS. Any suggestion to resolve this issue would be appreciated. thank you very much

Following is my INCAR and POSCAR file. I have used Li_sv, Mn_pv, and O pseudopotentials of PBE scheme. The attached pictures are the reference and my DOS plots.

INCAR:

ISTART = 0
ALGO = Normal
EDIFF = 1E-7
EDIFFG = -0.01
ENCUT = 550
IBRION = 2
ICHARG = 2
ISYM = 2
ISIF = 3
ISMEAR = -5
SIGMA = 0.05
ISPIN = 2
LDAU = True
LDAUJ = 0 0 0
LDAUL = 0 2 0
LDAUPRINT = 2
LDAUTYPE = 2
LDAUU = 0 5.1 0
LMAXMIX = 4
LORBIT = 11
LREAL = .FALSE.
LWAVE = .FALSE.
LCHARG = .FALSE.
MAGMOM = 2*0 2*5.0 4*0
NELM = 100
NSW = 100
NBANDS = 36
PREC = Accurate
AMIX = 0.2
BMIX = 0.00001
AMIX_MAG = 0.8
BMIX_MAG = 0.00001


POSCAR:

Li2 Mn2 O4
5.488399
1.000000000000000 0.000000000000000 0.000000000000000
0.000000000000000 0.523240864492009 0.000000000000000
-0.421857691964654 0.000000000000000 0.887942587579657
Li Mn O
2 2 4
Direct
0.000000000000000 0.500000000000000 0.500000000000000
0.500000000000000 0.000000000000000 0.500000000000000
0.000000000000000 0.000000000000000 0.000000000000000
0.500000000000000 0.500000000000000 0.000000000000000
0.228170000000000 0.500000000000000 0.231487000000000
0.771830000000000 0.500000000000000 0.768513000000000
0.728170000000000 0.000000000000000 0.231487000000000
0.271830000000000 0.000000000000000 0.768513000000000

[admin]

Both spectra are very similar. You are using different orientation of the lattice vectors,
therefore your eg orbitals differ from those in the reference paper.
For nicer DOS (pDOS) use higher number of points (increse NEDOS from 301 to e.g. 2000).