I am studying the phonons of the hexagonal GST (Ge2Sb2Te5) system. I used vasp to optimize the structure and then phonopy to generate the displacements. Using vasp to calculate the forces and then phonopy to post process the results I got the following phonon dispersion. There appears to be small imaginary frequencies that could indicate dynamical instabilities. I have found literature where stable phonon dispersion curves were generated and have tried to improve my results. In my experience, including born effective charges can help improve the phonon dispersion by adding a correction for long range interaction of polar compounds. I tried to do this but the results for my DFPT calculation are strange to me. The born effective charge symmetry is largely broken. I have included the born effective charges from the vasp OUTCAR. Does anyone have insight if I am doing something wrong for this calculation?
I'm not sure what your INCAR looks like but if you have LREAL set to anything but .FALSE. you will usually get unreal imaginary frequencies during phonon calculations