Optical Properties using GW approximation

[hpaudel]

Hi,

I am trying to calculate the optical properties of a TiO2 slab with 240 atoms and a Folic Acid molecule attached on its surface. The structure is optimized using VASP 4.6. There is no GW approxmiation in VASP 4.6 and the gap is well underestimated. So, I ran the job in VASP 5.2 to calculate the optical properties. I followed the steps given in GW approximation recipe. However, I could not get the WAVECAR and CHGCAR file from the first run to use for the non-self consistent calculation. The electronic iteration does not get converged. The job stops after few iteration cycle. I tried the same calculation in bulk of Bi2Se3. But it worked there. So what is the problem here? Is it due to bad geometry or something else.

Following is my INCAR file:

SYSTEM = APTMS-TiO2
ISTART = 0
ICHARG = 2
EDIFF = 0.0001
EDIFFG = -0.01
ENCUT = 450
LREAL = .TRUE.
ISPIN = 1

AMIX = 0.2
BMIX = 1.0
LOPTICS = .TRUE.

#LORBIT = 11
#LVTOT = .TRUE.

LWAVE = .TRUE.
LCHARG = .TRUE.
NELM = 1
NSW = 0
IBRION = -1
LMAXMIX = 6
ISMEAR = -5
SIGMA = 0.2

# IMPROVE ELECTRONIC CONVERGENCY
NSIM = 4
ALGO = N
POTIM = 0.10 # time step for ionic-motion
TEIN = 0.0 # initial temperature
TEBEG = 0.0
TEEND = 0.0 # temperature during run

Thank you.

hpaudel

[giacomo giorgi]

Hi hpaudel,
These settings usually work in my calcs


SCF:

SYSTEM = XXX
ENCUT = XXX eV !plane-waves cutoff - always set in eV
EDIFF = 1E-XX
ISMEAR = 0 !gaussian; default is MP order 1,w sigma0.2
SIGMA = 0.05 !this is small smearing of 0.1eV


NSCF:

SYSTEM = XXX
ICHARG = 11
ALGO = Exact
LOPTICS = .TRUE.
NELM = 1
ENCUT = XXX eV !plane-waves cutoff - always set in eV
NBANDS = XXX
ISMEAR = 0 !gaussian; default is MP order 1,w sigma0.2
SIGMA = 0.05 !this is small smearing of 0.1eV


GW:

SYSTEM = XXX
ALGO = GW0
NOMEGA = XX
LPEAD=.TRUE.
ENCUTGW= XXX
LWANNIER90=.TRUE. ! needed to write Wannier90 files?
LWAVE=.TRUE.
LOPTICS = .TRUE.
ENCUT = XXX eV !plane-waves cutoff - always set in eV
NBANDS = XXX ! (same as NSCF run)
ISMEAR = 0 !gaussian; default is MP order 1,w sigma0.2
SIGMA = 0.05 !this is small smearing of 0.1eV


cheers,
Giacomo