Frequency underestimation problem

[dattr18]

Dear Yuzie007,

I just get started to use upho for phonon unfolding. I tested using AlN having WZ structure.
First, I generated a 2x2x2 supercell (32 atoms) from the primitive cell. I used this 32 atoms supercell to calculate phonon dispersion using phononpy, by generating 1x1x1 supercell to get atomic displacements (basicaly keep the same size). The calculation was interfaced with QE. This calculation produced force constants of the supercell.

To run the upho, POSCAR and POSCAR_ideal were created. The POSCAR_ideal contains
X
5.823848962318
1.00000000000 0.00000000000 0.00000000000
-0.50000000000 0.86602540378 0.00000000000
0.00000000000 0.00000000000 1.60812126409
X
4
Direct
0.666666667 0.333333333 0.000000000
0.333333333 0.666666667 0.500000000
0.666666667 0.333333333 0.381069307
0.333333333 0.666666667 0.881069307

The unfolded phonon dispersion has pretty similar pattern with what is assumed for WZ-AlN. The only thing not good is that the phonon frequency is underestimated. Maximum frequency reaches close to 4 THz (figure attached), while it should be 27 THz in the normal phonon calculation.

To check the phononpy calculation, I plotted the phonon dispersion of the supercell, it looks OK in terms of phonon frequency range.

Could you please give me some comments on this problem, I very appreciate.
Thank you!

Best regards,

Dat

[dattr18]

Here also comes what the band.conf file contains
DIM = 2 2 2
PRIMITIVE_AXIS = 1 0 0 0 1 0 0 0 1
BAND = 0.0 0.0 0.0 2/3 -1/3 0.0 1/2 -1/2 0.0 0.0 0.0 0.0 0.0 0.0 -1/2
BAND_POINTS = 51
BAND_LABELS = \Gamma K M \Gamma A
FORCE_CONSTANTS = READ