Dear Archana Sharma

You can either try by increasing the supercell or by increasing k-mesh. Since the number of atoms is large, so first try with increasing k-mesh. It seems like in z-direction, k-mesh is a little less, so try with 9*9*5, whether it works or not.

As I have worked in a similar system, I believe 2*2*2 supercell should work, only k-mesh might be the issue. Otherwise, a small supercell is the main reason behind such issues.

Dear Archana Sharma

Try to use phonopy with the finite-displacement method with the same supercell (https://phonopy.github.io/phonopy/vasp.html), it usually solves the problem of negative frequencies obtained by DFPT(IBRION=8). Also in most cases, this method of calculating the phonon spectra is faster.

Your calculated phonon dispersions contains large negative frequency at the gamma point, its not small negative frequency that you can remove by acoustic sum. I think either you haven't relaxed the structure properly, i.e., relaxed with very strict convergence criteria, or your structure is dynamically unstable. You can relax it again using very strict convergence criteria, use gamma center k-point, chose the k-point based on reciprocal lattice length, use correct smearing parameter, and cutoff energy, and finally symmetrize the force constant.

Hope these will significantly improve your phonon dispersions.

All above comments are valid and may work for your case. But please remind that the q-->0 is the long wave length which means there may be a "lower energy configuration" with long length distortion in respect to your cell size. In this case you can not fix it as it is nature of your material

Use the linear response method for better Phonon Dispersion calculation (quality). And the minor negative frequency coming from optical bands may be due to an artifact of approximation that you can ignore.

Should we allow ionic steps while performing calculations in displaced super cells created by phonopy

Dear Karthikeyan . M

No, you should set NSW=0 or 1

Thanks a lot for your immediate response,

1) Will it be the case always?? or are any exceptions there.. Actually, I am trying to calculate thermophysical properties using phonopy..

2) What is the reason behind restricting ionic movement

Dear Karthikeyan . M

1) Yes, when using the finite displacement method, you will need to calculate forces on atoms in displaced structures without relaxation (NSW=0/1).

2) The basic idea is that the PHONOPY generates displacements based on a symmetry of the structure, thereby reducing the number of calculations. You need to calculate the forces acting on atoms in each displaced structures. "After the atomic force calculations, the set of atomic displacements are expanded using the symmetry and then all the elements of force constans between atoms in a primitive cell and the supercell are fit to the symmetry expanded forces of atoms in supercells" (from https://phonopy.github.io/phonopy/reference.html). If you allow the atoms to move during force calculation, then you break the symmetry of the displacements (generated by PHONOPY) and get incorrect forces for this displacements.