[Wien] Fwd: Phonon dispersion for wurtzite ZnO

[Luis Ogando]

Dear Wien2k community,

   Some days ago, I sent the the message below, but I got no answer.
   I would really appreciate if someone could take a look and point if some
parameter looks like inadequate. Obviously, I would thank any hint.
   Thank you again !
   All the best,
                 Luis



---------- Forwarded message ----------
From: Luis Ogando <lcodacal at gmail.com>
Date: 2018-06-05 12:19 GMT-03:00
Subject: Phonon dispersion for wurtzite ZnO
To: A Mailing list for WIEN2k users <wien at zeus.theochem.tuwien.ac.at>


Dear Wien2k Community,

   I have no experience in calculating phonon dispersion, but I am trying
to calculate it using Wien2k 17.1 and Phonopy.
   I am getting "strange" results  for wurtzite ZnO and I would highly
appreciate your comments. Basically, I got a nice plot for the 2x2x2
supercell with negligible negative frequencies for the acoustic modes at
gamma. Trying to check convergence, I increased the supercell to 3x3x2, but
the negative frequencies became not negligible. The next step was to
increase the cell to 4x4x3 with more k-points, but the negative frequencies
are still there. Moreover, some of the optical modes look like not
converging.
   LO-TO splitting are not taken into account.
   Well, I will describe my steps and, please, give me some suggestions or
indicate the errors.

1) Structural optimization:
   1.1) I optimized the wurtzite ZnO cell using the "[5]  VARY A and C
(2D-case)" option.
   1.2) 3% of RMT reduction (1.65 for Zn and 1.42 for O)
   1.3) non-magnetic configuration
   1.4) LSDA
   1.5) Core/valence separation energy : -7.0 Ry (valence states for Zn: 3P
(local orbital), 3D and 4S // valence states for O: 2S (local orbital) and
2P)
   1.6) RKmax: 8.0 and 8.5 (convergence tests)
   1.7) K-grid :  (8 8 4) and (9 9 4) (convergence tests)
   1.8) Emax: 1.5 Ry (no SOC)
   1.9) Gmax: 14
   1.10) MSR1a
   1.11) tolf : 0.05 (case.inM)
   1.12) run_lapw -ec 0.0001 -cc 0.0001 -fc 0.01 -i 200

Results:
   Sigma ~ 1.7x10^{-3}
   Differences in optimized "a" and "c" for the two RKmax and K-grid sets ~
0.004%
   When optimizing "u" with the optimized "a" and "c", the residual forces
where lower than 3.2x10^{-3} mRy/Bohr

2) Supercell:
   2.1)0% of RMT reduction (1.97 for Zn and 1.69 for O)
   2.2) non-magnetic configuration
   2.3) LSDA
   2.4) Core/valence separation energy : -7.0 Ry (valence states for Zn: 3P
(local orbital), 3D and 4S // valence states for O: 2S (local orbital) and
2P)
   2.5) RKmax: 8.0
   2.6) K-grid : (4 4 2) for the 2x2x2 supercell
                     (3 3 2) for the 3x3x2 supercell
                     (4 4 2) for the 4x4x3 supercell (trying to check if
the k-grid plays a role)
   2.7) Emax: 1.5 Ry (no SOC)
   2.8) Gmax: 14
   2.9) run_lapw -p -it -ec 0.0001 -cc 0.0001 -fc 0.01 -i 200

=> All the K-grids had "no shift" !

   Finally, two questions:

1) When I built the P1 1x1x1 supercell, I simply separated the equivalent
atoms, preserving positions and the gamma angle equal to 120. Phonopy finds
the correct original space group. Is this the correct procedure ?

2) There is a paper where the authors (Prof. Blaha among them) calculate
the phonon dispersion for wurtzite ZnO using Wien93 (PRB 53, 11425 (1996)).
In this calculation, the 3P states for Zn are core states. Do you believe
that this is relevant ?

   I am sorry for the huge email !
   Thank you !
   All the best,
                 Luis
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://zeus.theochem.tuwien.ac.at/pipermail/wien/attachments/20180611/6dfdaaaf/attachment.html>