Hello Mark,

First, make sure that the imaginary freq is not a mistake of phonon calculation at Gamma point. If you are sure about then you may use the command:

dymmodes2xyz.pl POSCAR DISPLACECAR modes.dat (moviefolder) (freq.dat) (numimages) (dist)

To create displacements along the eigen vector of each mode. The output will be XYZ files in (moviefolder) directory. You can visualize each file using Jmol. In simple cases, you will be able to determine characteristics of the modes.

More details of the command can be seen: http://theory.cm.utexas.edu/vasp/scripts.html

Kien,

Thanks for your response.

My issue is in the construction of the DISPLACECAR, following an IBRION =5 calculation, which as opposed to a dynamical matrix calculation does not require a DISPLACECAR to run. I know my eigenvalues, eigenfrequencies, and modes from the IBRION =5 calculation, and can build the required *.dat files manually. However, I cannot seem to build a working DISPLACECAR. I have tried using dymseldsp.pl with the initial and resultant POSCARs of my structural optimization, and while this does return "movies", they display non-physical behavior; when I know that my system is very close to a local minima, as it has only one imaginary frequency, which is at ~45 cm^-1.

If there is only one frequency you can build your file by hand. I usually convert CONTCAR to CONTCAR.xsf, then I copy the eigenvalues from OUTCAR and paste them in the CONTCAR.xsf aside the coordinates. Then you can open CONTCAR.xsf with xcrysden, use the option "f" ("shift f" to modify the options) and you will see the representation with vectors.

Regards