Your structural relaxation is very well done (in fact, I typically use EDIFFG=-0.001 and it usually works just fine, although I then use "external" displacements and not DFPT). Some possible sources of your imaginary frequencies:

1. ENCUT (could be higher, hard to say without knowing what compound you have and what pseudo's you use).

2. k-point sampling (sometimes an issue you don't say what you use).

3. It could be that your result is correct and that the structure is in fact dynamically unstable. Do not forget this possibility!

Then there is also

4. Your structure could be stabilized by phonon-phonon interaction, i.e. effects going beyond the harmonic approximation. Then you need to go to more sophisticated methods that either compute self-consistent phonons or derive phonon spectra from molecular dynamics trajectories.

Hi all,

Many thanks for your reply I will try all possibility and will revert back........

negative phonons often indicate the structure is not stable at T=0. do you have "perfect" cell or does it contain defects (dopants, vacancies?) if the latter is the case, you will need huge cell so the defects dont see each other thru PBC.

Check the pressure in outcar file. Phonopy method is very sensitive to stresses in the structure. Use a large enough supecell. Another, indeed, the results are correct and your structure is thermodynamically unstable. Your accuracy of calculation is good enough.

Dear Ashish Bhatanagar,

Have you resolved negative frequency in phonon density of states plot (phonon DOS).   I am also getting negative frequencies in phonon DOS.

Thank You

Shilendra

Dear Torbjörn Björkman  first thank you for always help and suggestions !

I am also getting negative frequency in phonon DOS for my system having 24 atoms in a unit cell.

I took 2*2*1 supercell (96 atoms) and I optomized structure after that I did phonon calculation with IBRION=8 tag.

As you suggested in earlier post in your suggestion number 4

Your structure could be stabilized by phonon-phonon interaction, i.e. effects going beyond the harmonic approximation. Then you need to go to more sophisticated methods that either compute self-consistent phonons or derive phonon spectra from molecular dynamics trajectories.

My query is that phonon spectra from ab-initio MD trajectory how will be different from  DFT and as you told about phonon-phonon interaction. How does calculate phonon-phonon interaction which code is suitable ?

For ab-initio MD I know from VASP code and in VASP IBRION=0 for MD then how will calculate pphonon spectra from MD by VASP code ? Should we use different code for ab-initio MD ?

My final query is that how would I confirm my calculation is right and my system is thermodynamically unstable that is why I am getting negative frequency only by looking experimental results?

Do people report negative frequency in literature ?

Any basic literature for understanding the phonon calculations and basics of this ?

Thank You

Shilendra

Hi Artem Kuklin, how would a larger supercell help?

Hi Hitanshu Sachania , in my current experience it is not so strict. I guess a difference between 2x2 and 3x3 is small, though sometimes it can help to enhance your spectra. And of course, it depends on the original size of the unitcell.

Hi Artem Kuklin, thank you for your interest. I'm a bit confused about the DIM tag. What if the unit cell is not cubic? Do we try to make it cubic by taking such dimensions as 2x3x6 or do we still just make it uniformly bigger with 2x2x2 or 3x3x3 or further?

From what I read in the SourceForge forum for phonopy, the size of the supercell must be decided wrt converging thermal properties, but that would be very expensive. Do you have any suggestions regarding the size that must be taken?

Hitanshu Sachania , actually from my experience, the supercell size is not so critical, it just should be enough to reproduce interatomic forces to build a force constant matrix. Then, if one of the dimensions of your unit cell is (for example) twice less than the other ones, I would recommend to increase it more when building a supercell. Don't forget your supercell always refers to the unit cell in the end. So if you have a tetragonal unit cell with a=4, b=4, c=6, I intuitively would do 3x3x2. The only thing I strictly recommend is to try to preserve as high group of symmetry as possible. Also, it could be that at 0K the certain structure may be unstable while it could be stabilized by increasing temperature.

Hello everyone,

Phonon calculations (Phonon Band Structures, Phonon DOS, and thermal properties) in materials science using VASP and phonopy are explained on the Youtube Channel. Please find the link below:

https://bit.ly/2UtvbHE

https://www.youtube.com/watch?v=FX7WjL074g4

Best regards,

Rasoul Khaledialidusti