About phonon and eigenvector

[fang]

Dear Abiniters,

Hi,I have two questions.
(1). I try to calculate phonon frequencies at gamma point of the hexagonal phase of GaSe.
And this is my input(a part):
rfphon 1
rfdir 1 1 1
nqpt 1
kptopt 2
tolvrs 1.0d-8
ngkpt 6 6 2
nshiftk 1
iscf 5
ixc 11
diemac 12
nband 30
ecut 30
nstep 300

But I got the negative phonon frequency:
Phonon wavevector (reduced coordinates) : 0.00000 0.00000 0.00000
Phonon energies in Hartree :
-9.336844E-05 -9.176884E-05 1.059004E-05 3.827890E-05 9.361631E-05
9.510691E-05 2.332906E-04 2.334972E-04 2.682767E-04 2.684224E-04
5.393799E-04 5.406935E-04 8.573927E-04 8.578686E-04 8.654459E-04
8.659214E-04 8.796197E-04 8.800295E-04 8.874010E-04 8.878130E-04
1.050585E-03 1.072045E-03 1.327610E-03 1.327859E-03
Phonon frequencies in cm-1 :
-2.049200E+01 -2.014093E+01 2.324246E+00 8.401248E+00 2.054641E+01
2.087355E+01 5.120136E+01 5.124672E+01 5.887994E+01 5.891190E+01
1.183802E+02 1.186685E+02 1.881759E+02 1.882804E+02 1.899434E+02
1.900478E+02 1.930542E+02 1.931442E+02 1.947620E+02 1.948524E+02
2.305767E+02 2.352868E+02 2.913767E+02 2.914313E+02

Then I increase ecut to 40 Ha, I got all positive frequency:

Phonon wavevector (reduced coordinates) : 0.00000 0.00000 0.00000
Phonon energies in Hartree :
5.416151E-05 5.552603E-05 1.431128E-04 1.436117E-04 2.497892E-04
2.501201E-04 2.829831E-04 2.832167E-04 3.348836E-04 3.380225E-04
6.304938E-04 6.320513E-04 8.714120E-04 8.718104E-04 8.794336E-04
8.798505E-04 8.925478E-04 8.928840E-04 9.002979E-04 9.006614E-04
1.086787E-03 1.105199E-03 1.356178E-03 1.356358E-03
Phonon frequencies in cm-1 :
1.188708E+01 1.218656E+01 3.140964E+01 3.151912E+01 5.482238E+01
5.489502E+01 6.210762E+01 6.215888E+01 7.349846E+01 7.418737E+01
1.383774E+02 1.387192E+02 1.912528E+02 1.913403E+02 1.930134E+02
1.931049E+02 1.958916E+02 1.959654E+02 1.975925E+02 1.976723E+02
2.385221E+02 2.425632E+02 2.976466E+02 2.976861E+02

But it's seems incorrect,there are no acoustics mode.
when I calculate Relax, the input parameter tolmxf 10 d-7,toldfe 10d-8,is that enough?
Or I should increase ecut to calculate phonon?
By the way, I use GGA-PBE-HGH: Ga-q3.HGH and Se-q6.HGH
Someone can give me better suggestions?

(2).When after a phonon calculation at Gamma point how could I obtain the
phonon eigendisplacements(that is how the atom move)?

Thank you to all

with regards
fang

[ilyes]

Dear Fang,

I think with a little bit of search in the abinit forum, you should find your answers (Read the Nettiquette before posting). However, concerning your two questions:
you should use anaddb code to get your results by imposing the acoustic sum rule in the input of the anaddb file (asr =1 or 2) to get vanishing acoustic modes at Gamma and use in the same input file eivec = 3 to get the eigenvectors.

Best regards

Ass. Prof. Ilyes Hamdi
Department of electronics and automatics
Institute of applied sciences and technology Gafsa
Tunisia

[fang]

Dear Prof. Ilyes Hamdi,
Thank you for the suggestions.
I will try again.
Thanks very much.

Fang.

[fang]

Dear Prof. Ilyes Hamdi,
I study the tutorespfn lesson 2,and abinit forum.
But I still have some questions.
The tutorespfn lesson 2 has many q points, but in my job only gamma,
Can I only ues the input eivec and ars to run? It's seems not ok.
What input variables I need to add?
I can't understand the tutorespfn lesson 2, it's too complicated,
and I have no one can discuss, so I come here,sorry.
Thanks very much.
Fang.

[ilyes]

Dear Fang,

Before starting any response function calculations to get the phonon spectra of a given material, you should read the paper behind the theory, namely:

1) Responses to atomic displacements and macroscopic electric fields : detailed description of an implementation based on plane waves and separable pseudopotentials. X. Gonze, Phys. Rev. B55, 10337 (1997).
2) Dynamical matrices, Born effective charges, dielectric permittivity tensors and interatomic force constants
X. Gonze and C. Lee, Phys. Rev. B55, 10355 (1997).
Then, DO the response function 1 and 2 tutorials.

For a brief summary, the abinit code is able to calculate the response to an atomic displacement (the interatomic force constants (IFCS) at a certain q vector) using the density functional pertubation theory (DFPT). From the IFCS, one can obtain the phonon mode caracterized by a q vector.
To get the phonon spectra of a material, you should calculate the IFCS (or the DDB if you want) with abinit (from DFPT) at a certain number of q vectors (corresponding to a non shifted Monkhorst-Pack (MP) grid), then use the anaddb to get the IFCS at any q vectors (q vectors that do not belong to the MP grid) by fourier interpolation.