spin-orbit coupling of Bi2Te3

Message

bioggg · #1 Post by **bioggg** » Fri Dec 27, 2013 4:30 pm

Dear all,

I am trying to reproduce the band structure of Bi2Te3. However, my output magnetizations are zero after relaxation. Would you please help me out?

This is my INCAR file.
PREC = Accurate
ISTART = 0
ICHARG = 2
IBRION = 2
NSW = 30
NELMIN = 5
ISIF = 3
ENCUT = 228
ISMEAR = 0
SIGMA = 0.2
ISPIN = 2
EDIFF = 1.0E-08
EDIFFG = -1.0e-07
IALGO = 38
LREAL = .FALSE.
ADDGRID = .TRUE.
LWAVE = .TRUE.
LCHARG = .TRUE.
LORBIT = 11
NEDOS = 5001
MAGMOM = 0.0 0.0 0.0 2.0 -2.0 (Te1,Te1,Te2,Bi,Bi)
ISYM = -1

THANK YOU So MUCH!

WolverBean · #2 Post by **WolverBean** » Sun Jan 05, 2014 5:32 am

I have had this problem before too, in a different system. It seems that in antiferromagnetic systems, VASP somehow loses the magnetization information during geometry relaxation. The effect is unpredictable: sometimes it will take 7 or 8 steps, and lose the magnetization on the 9th step. I was able to avoid this problem by using DFT+U, but my system had transition metals to which applying a +U correction made sense. I'm not sure if a +U correction is as meaningful for Bi.
If anyone else has experience with this problem, please chime in!

bioggg · #3 Post by **bioggg** » Mon Jan 06, 2014 10:55 pm

Thank you so much for sharing your experience.

mickeyshaughnessy · #4 Post by **mickeyshaughnessy** » Tue Jan 07, 2014 12:27 am

You could try relaxing in one step and then running a second calculation with fixed (relaxed) atomic positions. This may not be the lowest energy atomic positions for the antiferromagnetic state, but at least your magnetic moments will be correct.