Lost symmetry ?

Message

forsdan · #1 Post by **forsdan** » Sat Jun 17, 2006 6:23 pm

Dear VASP-users,

I'm performing a vibration calculation for a bcc-iron system with the input

ENCUT = 420 eV
ENAUG = 650 eV
EDIFF = 1E-5
ISMEAR = 1
SIGMA = 0.1
VOSKOWN = 1
IBRION = 5
LREAL = .FALSE.
PREC = HIGH
NSW = 1
POTIM = 0.02
NFREE = 2
MAGMOM = 4 4
ISPIN = 2

and

bcc:
2.831000000000000
1.0000000000000000 0.0000000000000000 0.0000000000000000
0.0000000000000000 1.0000000000000000 0.0000000000000000
0.0000000000000000 0.0000000000000000 1.0000000000000000
2
Selective dynamics
Direct
0.0 0.0 0.0 T T T
0.5 0.5 0.5 T T T

The cutoff, k-points and smearing have all been checked for convergence. The question mark arises when I look at the force constant matrix:

-5.725020 0.000000 0.000000 5.728042 0.000000 0.000000
0.000000 -4.614101 0.000000 0.000000 4.614101 0.000000
0.000000 0.000000 -17.044915 0.000000 0.000000 17.044908
5.728042 0.000000 0.000000 -5.731064 0.000000 0.000000
0.000000 4.614101 0.000000 0.000000 -4.614101 0.000000
0.000000 0.000000 17.044908 0.000000 0.000000 -17.044902

which produce the frequencies 12.21, 7.08, 6.35, 0, 0, 0 THz approximately. I expect the three zeros but what I can't figure out is why the three other eigenvalues ain't equal. Shoudn't displacements in x-,y- and z-direction produce the same force constants due to the symmetry in a bcc-lattice ?

I've also tried to decrease the POTIM value to 0.01 which yields 12.22, 7.09, 6.37, 0, 0, 0 THz instead, but the problem still remains. But even if I ain't in the harmonic regime shouldn't the symmetry still be there ? Can the smearing have something to do with the lost of symmetry or what may be the cause of it ?

Best regards

Dan Fors

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

forsdan · #2 Post by **forsdan** » Sat Jun 17, 2006 7:25 pm

I've also tried ISMEAR = -5, yielding 12.24, 6.73, 6.42, 0, 0, 0 THz

best regards

Dan Fors
<span class='smallblacktext'>[ Edited Sat Jun 17 2006, 09:26PM ]</span>

forsdan · #3 Post by **forsdan** » Mon Jun 19, 2006 7:57 am

I tried to increase the supercell to contain 54 and 128 Fe atoms respectively with POTIM = 0.02 and PREC = HIGH. If I displace one iron atom in the bcc-lattice I now obtain force constants which are almost identical as expected. So I suppose my problem was due to the size of the supercell so my next question is rather: Why does the size effect the force calculation ? Or is it something else I havn't considered ?

Best regards

Dan Fors

tjf · #4 Post by **tjf** » Mon Jun 19, 2006 12:03 pm

If you do an sdiff on the OUTCARs what differences do you see other than the obvious ones from making the supercell larger? Changes to the identified symmetry elements or anything?

forsdan · #5 Post by **forsdan** » Wed Jun 21, 2006 9:10 am

The only difference as I can see is a difference between the identified point symmetry group in the 2-atom system and the 54-atom system. However, the symmtry groups are the same for the x-, y- and z-direction within the system itself, so it doesn'y explain why I get a difference. The scf-loops simply end at different total energy for the same displacement in the different directions with the tendency that the zz-components in the force matrix are larger than the xx and yy-components.

I've now also tried to increase cutoff, k-points; tested different smearing methods, smearing widths and POTIM-values, but I can't work myself around the problem (if there now is a problem). I'll will look through the OUTCAR more in detail but any other suggestions what to do ?

Best regards

Dan Fors