Negative frequencies on your plot are suspicious. Otherwise, the lowest frequency peak corresponds to acoustic phonons. Their dispersion should be linear; therefore, the density of states should be proportional to the frequency squared. Looks like it is.

Then, you have a number of optical phonon bands. Evidently, you have a lot of atoms in a unit cell.

And, there is no scale along x axis.

Dear Igor Goliney , Thank you for your reply

On X-axis frequency in Tetrahertz (THz) . I calculated phonon DOS from DFT using VASP code and post process by phonopy code. People usually face this negative frequency problem. After possible improvement if negative frequencies do not go then people correlate structure at 0K is unstable.

I am new to such type of calculations. I do not know how much negative frequencies are acceptable. Is there any criteria ? Do people report negative frequencies ?

Shilendra

Dear Benedikt Klobes, thank you for your reply and suggestions !

Is there any way for removing negative frequencies by changing calculation parameter ? I am using VASP code in combination with PHONOPY code  for calculating phonon DOS. My calculation detail is following

Sructure Optimization INCAR tags

ISTART=0

ICHARG=2

ENCUT=600

ISMEAR=0

LREAL=Auto

ISIF=3

EDIFF=1.0E-8

EDIFFG=-0.001

IBRION=1

POTIM=0.5

ADDGRID=.True.

NSW=100

NPAR=8

KPOINTS

Automatic mesh

0

Monk

5  5   6

0. 0. 0.

After making supercell 2*2*1 by phonopy, I did phonon calculation with INCAR tags

ISTART = 0

ICHARG = 2

PREC = Accurate

ENCUT = 600.000

IBRION = 8

EDIFF = 1.0e-08

IALGO = 38

ADDGRID = True

ISMEAR = 0

SIGMA = 0.1

LREAL = False

LWAVE = False

LCHARG = False

KPOINTS

Automatic mesh

0

Monk

2   2  3

0. 0. 0

Thank You

Shilendra