Phonon calculation with negative frequencies (imaginary models)

Dear Julian,

Thanks for your great software and your kind support in this community.

I am working on phonon calculation of UO2 with two U vacancies recently. I have created the 5 x 5 x 5 supercell fractional coordinates with lammps and converted it into GULP input format, as you can see in the attached gin file. I output the frequencies by using output command in the GULP input script. What I expect is the 3*N phonon frequencies at gamma point. However, it outputs several negative frequencies, which mean imaginary models, as explained in the gout file. It seems like supercell does matter to the phonon calculation, since I have tested single cell and got zero negative frequencies. Is there any suggestions from you?

Thanks for your kind support.


I am sorry I cannot upload the gin and gout files, since new users are not allowed to upload files.

Dear Qingyu,
This just means that the phonon dispersion is unstable away from gamma for the single cell & that your system wants to break symmetry and distort. If you find the point in the Brillouin zone where the imaginary modes are at their lowest, then this tells you the size of the supercell you need to make to in order to break symmetry and remove the imaginary modes. Alternative, it just means the force field you have input isn’t very good for UO2.

Dear Julian,

Thanks for your kind reply, I will check the structure and the force field.