My Community

Dear Forum Members,

I hope you're well.

I've encountered unexpected negative frequencies while analyzing phonon results, I've attached a zip file. Could someone please review them and provide feedback on any potential mistakes or other factors causing this discrepancy?

Thank you for your help.

Hi,

In general, there are several reasons that may explain the occurrence of negative frequencies:

1) The supercell is not large enough. However, this problem would not concern the phonon dispersion at the GAMMA point, which in your case is already negative.

2) The geometry that you are using does not correspond to a minimum. In particular, the positions of the atoms in the unit cell should first be optimized before doing a phonon calculation. Did you do it?

3) The parameters of the calculations (e.g., basis-set size, number of k-points, etc.) are not good enough. However, in the present case you chose parameters that seem to be quite good, so this should not be the problem.

Besides, you have used the DFPT method:
https://www.vasp.at/wiki/index.php/Phon ... ion_theory
https://phonopy.github.io/phonopy/vasp-dfpt.html#

Have you also tried the method based on finite differences:
https://www.vasp.at/wiki/index.php/Phon ... ifferences
https://phonopy.github.io/phonopy/vasp.html

should i need to optimize(relax) supercell?which ISIF will be better?

No, you can do the optimization with the small original cell. To avoid negative frequencies it should be enough to optimize only the atoms positions. However, ideally it would be better to optimize everything (cell shape, volume and atoms positions), which is probably feasible since your system is very small. The description of ISIF is here: ISIF.

thank you for your response.will you kindly check my INCAR file and suggest me tags to add.