The ABINIT Users Mailing List ( CLOSED )

["Pietro Delugas" <delugas@dsf.unica.it>]

Hi Erik
  As you have 2 modes developing a dipole alog z direction,
  it  may happen that the non analicity correction term mixes them
yielding two different eigenmodes at different frequencies. The
generalization of the
 Lyddane Sach Teller ( formula 64 in the Gonze-Lee paper ) relation tells
that the product of the 2 LO modes energies increases compared with the
product of  the TO  frequencies.  If I remember well anaddb prints out an
estimate of the dielectric constant based on this relation check if it is
consistent with the calculated value of the dielectric tensor zz component.
  regards
   pietro


> I am doing a calculation of the Gamma point phonons in LiNbO2, which seems
> to complete ok.  After including electric field calculations and using
> anaddb to examine non-analyticities in c direction to find LO modes I see
> some strange behavior.. namely that one frequency actually decreases (85.5
> meV -> 70.0 meV), and particular eigenmode vanishes and is replaced by
> another eigenmode.  All three of these modes have A2u symmetry.
> Polarization in x-y direction behaves as one would expect.
>
> Following PRB55, 10355 (1997), I see that it is reasonable that the
> eigenvectors might change, but the change in eigenfrequency should be
> positive, I think.
>
> Attached are input files for abinis and anaddb.  The mode which has
> appeared is in the last one in the last list of eigenvectors: #24, with an
> energy of 96.3 meV.
>
> ~Erik Ylvisaker
>