How to differentiate Raman and Infrared in VASP phonon calculation?

Message

lisakite99 · #1 Post by **lisakite99** » Thu Jul 08, 2010 5:45 pm

I did a \gamma point phonon calculation using VASP5. 2. I have got all the frequencies in the OUTCAR, but how to differentiate Raman and Infrared, and how to assign symmetry (A_u, A_g, B_u, B_g etc) for each frequency. Does anyone know about that? Your help is greatly appreciated.

1 f = 95.480183 THz 599.919686 2PiTHz 3184.876003 cm-1 394.874440 meV
2 f = 95.476662 THz 599.897561 2PiTHz 3184.758545 cm-1 394.859877 meV
3 f = 95.439165 THz 599.661959 2PiTHz 3183.507775 cm-1 394.704801 meV
4 f = 95.435781 THz 599.640695 2PiTHz 3183.394884 cm-1 394.690805 meV
5 f = 93.122700 THz 585.107178 2PiTHz 3106.238807 cm-1 385.124667 meV
6 f = 93.117966 THz 585.077434 2PiTHz 3106.080903 cm-1 385.105089 meV
7 f = 93.110185 THz 585.028544 2PiTHz 3105.821356 cm-1 385.072909 meV
8 f = 93.078693 THz 584.830677 2PiTHz 3104.770908 cm-1 384.942670 meV
...

alex · #2 Post by **alex** » Fri Jul 09, 2010 7:33 am

I fear there is no automatic way in VASP. You would have to look at every mode and assign the symmetry of the vibration manually.
If you have access to the CRYSTAL code, try that. Assignment is automatically done.
There is one workaround, if you do not need the symmetry itself. Just calculate the IR intensity. That helps a bit.

Hth

alex

panda · #3 Post by **panda** » Mon Jul 12, 2010 7:20 pm

can you post your INCAR to show how you set this up please?

lisakite99 · #4 Post by **lisakite99** » Mon Aug 02, 2010 6:33 pm

Alex, thanks for your kind reply. How to calculate IR intensity in VASP?

To do \Gamma point phonon calculate

Firstly, fully relax your structure and save WAVECAR

Secondly, you can start a phonon calculation by set INCAR as

PREC= Accurate
ADDGRID = .TRUE.
ENCUT= 500.00 eV
ISTART=1
ICHARG=0
IBRION= 6
NFREE=4
POTIM=0.015 !displacement of atoms in Ang.
IALGO= 48
LREAL=Auto !needed for good convergence
NSIM = 4
LCHARG = .False.
LPLANE=.TRUE.
~

forsdan · #5 Post by **forsdan** » Mon Aug 02, 2010 10:00 pm

I do not have any input on the IR-Raman issue. However, I want to add some comments on your INCAR file. Please note that the frequency calculations can be very sensitive to your choice of parameters. It is often essential to use accurate forces (EDIFF=1e-8) and small finite displacements (0.005-0.01 at maximum). In your case NFREE=4 together with POTIM=0.015 lead to a maximum displacement of 0.03 Ã… which should be questioned if it is still within the harmonic regime. It is really recommended to perform tests with respect to POTIM and check how the frequencies change. However, keep in mind that for small displacements the EDIFF must be tighter to allow for accurate force evaluation.

An alternative to finite displacements is also to use DPFT (IBRION =

, which I personally find more convenient. However, given that both methods are converged w.r.t. the settings the results should be next to equal.

Cheers,
/Dan

<span class='smallblacktext'>[ Edited Tue Aug 03 2010, 12:06AM ]</span>

alex · #6 Post by **alex** » Wed Aug 11, 2010 9:24 am

Hi, sorry for the late answer. My frequency related part of INCAR looks like:

Frequency calculation
NSW = 1 ! due to special program gimmicks
IBRION = 5
NFREE = 2 ! two displacements per coordinate
ISIF = 0 ! no stress tensor calcn.
POTIM = 0.02 ! actual displacement
IDIPOL = 3
# LDIPOL = .TRUE.
DIPOL = 0.5 0.5 0.5

This works only for slabs. Here the slab is located around the middle of the unit cell, hence DIPOL ist set to the centre. the surface is perpendicular to c. This will compute the dipolmomentchange at every position of frequency calculation. There is a script (dyndip) by, I believe, Andreas Eichler to do the math afterwards and to give nice IR spectra. But: By default no intensities within the surface plane. So this might not be enough in your case.
Dipol correction is not applied here.

Cheers,

alex