CCL:G: [Summary] IR intensities from VASP



Dear CCL-ers!

As there are more people interested on the most recent approach of calculating IR intensities in VASP code, I have decided to put a tutorial on my webpage:
http://homepage.univie.ac.at/david.karhanek/downloads.html
Ralf has inspired me to share in here and I hope it will be useful.

The vibrational intensities using linear response are available in simply one VASP 5.* calculation and subsequent evaluation of the OUTCAR with my BASH script. Feel free to copy-paste and use it.

Regards,
- David -
__
Dr. David Karhanek
Post-Doctoral Researcher – Group Dra. Lopez
Institute of Chemical Research of Catalonia (ICIQ)

Av. Paisos Catalans 16 – 43007 Tarragona (Spain)
Phone +34 977920229 – Fax +34 977920231
dkarhanek**iciq.es - www.iciq.es



On 05/04/11 16:09, Ralf Tonner tonner++chemie.uni-marburg.de wrote:

Sent to CCL by: Ralf Tonner [tonner ~~ chemie.uni-marburg.de]
Dear CClers,

after several very helpful answers to my above-mentioned post and initial investigations, I want to summarize the results.

In VASP, the intensities of IR-active vibrational frequencies are not calculated directly.
CASTEP can do this - Raman activities can be calculated as well (thanks to Phil Hasnip).

If calculations with VASP are undertaken, two approaches can be followed to obtain the respective IR-intensitites:
a) change of dipole moment w.r.t. normal modes
("old" approach, also used in most molecular QC programs, e.g. Gaussian. See e.g. Preuss, Bechstedt Phys. Rev. B 73 2006, 155413 for an application to VASP)
b) density functional pertubation theory (a.k.a. linear response theory).

For a), a finite difference calculation has to be carried out with displacement of all atoms in all directions (might be reduced by symmetry considerations). The dipole moment has to be calculated for all displacements.
Afterwards, the dipole moment change w.r.t the normal modes has to be calculated which is proportional to the IR intensity of the respective mode. Several programs are available for this task (thanks to Alfred Gil Arranz and Kacper Druzbicki).
This method is restricted to isolated molecules and slabs in VASP.

For b), a linear response calculation has to be carried out and the born effective charges have to be determined for all atoms. Afterwards, the method published by Gianozzi&Baroni (see JCP, 100, 1994, 8537 and Rev. Mod. Phys. 73, 2001, 515) can be applied to derive IR intensities.
Many thanks to David Karhanek (ICIQ, Spain) for introducing me to this method and providing a script implementing this scheme.

This latter approach is also followed e.g. by CASTEP and seems to be more reliable and more general.

Kind regards,
Ralf.