have problem optimizing nanotubes(convergence....)

Message

srikanth.pedireddy · #1 Post by **srikanth.pedireddy** » Thu Jul 16, 2009 10:36 am

hi all,
I am a new user of VASP . I have a problem, in the optimization of nanotube containing B, N and C elements (all together). Unit cell consists of about 64 atoms. It runs for many days(approx. 9 days) and complete without converging. I tried with different pseudo potentials but it didn't help me solving problem . Please some one help in correcting the input file. I am pasting here the INCAR file i have used.
System =ortdigBNC6
PREC = HIGH
ISTART = 0
ICHARGE = 2
INIWAV = 1
IALGO = 48
IWAVPR = 11
POTIM = 0.5
ISYM = 2
NSIM = 4
LWAVE = .TRUE.
LCHARG = .TRUE.
EDIFF = 1E-05
EDIFFG = -0.005

NSW = 100
IBRION = 2
ISIF = 3
LCORR = T
ENCUT = 300
# DOS related values:
EMIN = 10.0
EMAX = -10.0
ISMEAR = 2
<span class='smallblacktext'>[ Edited ]</span>

forsdan · #2 Post by **forsdan** » Thu Jul 16, 2009 2:35 pm

Well, I assume it is a convergence issue with respect to the ionic relaxation. One suggestion is to run with ISIF = 2 instead of ISIF = 3. The ISIF = 3 option will allow for both volume and shape relaxation of the supercell, so if your vacuum regions (in the directions where the nanotube has a finite extension) are too small you'll have an interaction between nanotubes in different supercells due to the periodic boundary conditions. This will cause the nanotubes to approach each other. Such a relaxation behavior can of course take a lot of time. (Please check the coordinates and the superell volume/shape after the calculation to confirm that this actually happens.)

So unless you want this to happen then ISIF = 3 is not an appropriate choice for a nanotube. Instead fix the volume and shape by using ISIF = 2, which will allow for atomic relaxation within a fixed supercell. This will definitely give a faster ionic convergence.

EDIT: Please also make sure that the self-consistency cycle for each ionic iteration converges within NELM (default 60) steps, so that the forces in each step are reliable.

Hope this helps
/Dan Fors

<span class='smallblacktext'>[ Edited Thu Jul 16 2009, 09:27PM ]</span>