ABINIT Discussion Forums

I have not seen a nice phonon band structure for SrWO4 because there are strange dips for the highest phonon band, at the Gamma point.
I know that, This is due to the lack of LO-TO splitting for the ANADDB treatment of the first list of vector.
I edited the file trf2_5_B2EPS.freq, and Replaced these values ( line 29 and 125) by the correct value, including the LO-TO splitting,
that i found in the my file trf2_5.out, at the end.
unfortunately corrected phonon band structure become Worse than first step.
I have enclosed my output file for your observation.
line 29 :
-0.376331E-06 -0.376287E-06 0.470384E-03 0.470384E-03 -0.506413E-07 0.363048E-03 0.542676E-03 0.542676E-03 0.554947E-03 0.554947E-03 0.536647E-03 0.572050E-03 0.572050E-03 0.598734E-03 0.113532E-02 0.873685E-03 0.113532E-02 0.111480E-02 0.128509E-02 0.128509E-02 0.982144E-03 0.142328E-02 0.143733E-02 0.173006E-02 0.173528E-02 0.173528E-02 0.187384E-02 0.184687E-02 0.343300E-02 0.343300E-02 0.345099E-02 0.345099E-02 0.361504E-02 0.340873E-02 0.397949E-02 0.401029E-02
line 125:
0.363048E-03 -0.506413E-07 -0.376331E-06 0.470384E-03 -0.376287E-06 0.470384E-03 0.554947E-03 0.572050E-03 0.536647E-03 0.542676E-03 0.542676E-03 0.554947E-03 0.873685E-03 0.572050E-03 0.113532E-02 0.111480E-02 0.113532E-02 0.982144E-03 0.128509E-02 0.173528E-02 0.598734E-03 0.142328E-02 0.128509E-02 0.187384E-02 0.143733E-02 0.173006E-02 0.184687E-02 0.173528E-02 0.343300E-02 0.343300E-02 0.361504E-02 0.345099E-02 0.345099E-02 0.340873E-02 0.397949E-02 0.401029E-02

Dear Narges,

Your diagnostic is correct : the LO-TO splitting is the origin of the strange behaviour that you see.
However, note that the LO-TO splitting depend on the direction of the q wave vector.
Apparently, your material is anisotropic in such a way that the LO-TO splitting along the Z-Gamma and along
the Gamma-X directions differs. You have to take into account carefully this issue in doing the replacement that you mention.

However, alternatively, the easiest way to obtain a nice-looking phonon band structure is to increase the sampling of the
q wavevectors along the different lines. You sent me by mail, a few days ago, the trf2_5.in file, and I join it to my post, for the other members of the forum to see it.
You should edit the trf2_5.in file, and use a much finer first set of q wave vectors.
Instead of a total of 157 points, use between 1000 and 2000 points. As ANADDB calculations are quite fast, this should not be a problem for your computer.

Best wishes,
Xavier Gonze

As a complement, note that you have to give a number of sampling points that is proportional to the reciprocal length
of your segments. In your band structure trf2_6.out1.eps (that you also sent me a few days ago) it is clear that your X and P points are not located correctly,
because you have not provided enough points to sample Gamma-X and P-Gamma, while you have provided (relatively)
too few points to sample X-P segment.
Xavier Gonze

Dear X. Gonze,
thank you for your clear and helpful explanation.

Best Regards
Narges

Thank For Info.
Gclub