The ABINIT Users Mailing List ( CLOSED )

["D. R. Hamann" <drh@bell-labs.com>]

Dear Yong,

From this email, I have come upon an even more basic problem than
convergence (for your PbTiO3 supercell, at least). The file ptn_ela.in
you sent me specified values for acell, rprim and xred which were
inconsistent with the final values of these variables in your GS
relaxation run output, ptn_geo.out. If you used the WFK file from this
GS run (as you stated) for your RF calculation, these GS wave functions
were inconsistent with the RF geometry, which is taken from the RF .in
file, not the WFK file. There probably should have been a warning about
this in the log file, but I'm not sure. These RF results are thus
meaningless.

This isn't the issue with your basic 1-unit cell, however, since you
calculated the GS wave functions as part of the RF run. This is where
you should solve your convergence problems before returning to the
supercell.

As to the symmetry issue you discuss below, I'm not sure if this
represents a breakdown of the symmetry in the relaxation run or simply a
rotation in the xy plane away from the conventional Cartesian axes while
still preserving the initial space group symmetry.

Best regards,
Don Hamann


PS: This was re-sent because it was distributed to the Forum list with
an uninteligible subject.

Yong Liu wrote:

> Dear Hamann:
> Thank you for your answers.
> Since I have checked the calculation of structural relaxtion for PTiO3
> superstructure again, I found that the the final rprim is not consist
> with the initial symmetry as following,
> -6.9104342826E-01 6.9104344962E-01 2.1193850757E-01
> 6.9104342826E-01 -6.9104344962E-01 2.1193850757E-01
> 6.9104342826E-01 6.9104344962E-01 -2.1193850757E-01
> while the spacegroup 87 need the rprim(0,0) = -rprim(0,1). Why did
> this happen?
> Now I repeat the GS calculation with a bigger ecutsm to check the
> conv! ergence.
> Thank you again.
>
> */"D. R. Hamann" <drh@bell-labs.com>/* 写道：
>
>     Dear Yong,
>
>     I'm replying to both this and your later Forum inquiry about the
>     negative elastic constants of your 4-formula-unit PbTiO3 supercell.
>
>     Using the DDB file you sent me privately for this tetragonal
>     structure,
>     I calculated the phonons with anaddb both with and without the
>     acoustic
>     sum rule (a summary is attached). Without asr, there are 3 negative
>     (unstable) frequencies at qpt=0 0 0. These could be small
>     violations of
>     the asr causing the 3 acoustic modes to have small negative values
>     instead of zero, although your values are large. (I note that the
>     asr =
>     0 results from anaddb duplicate the phonon frequencies in your abinit
>     output file.) When asr is imposed, these 3 modes ! should be
>     forced very
>     close to zero. However, as you can see in the attached, imposing asr
>     gives 7 unstable modes with large negative eigenvalues. Since the
>     elastic constants are calculated with the asr, the negative
>     relaxed-atom
>     Cij are a correct representation of an unstable structure. The warning
>     in your original anaddb output about large asr violations was in fact
>     the problem.
>
>     For your 4-unit PtTiO3 superstructure, your basic abinit output has 11
>     unstable modes, so I am not surprised that even your rigid-ion
>     diagonal
>     elastic constants all come out negative for this highly unstable
>     structure. I'm surprised that you seem to have been able to relax the
>     structure in the first place.
>
>     Now, why the problem? Abinit has been used successfully for phonon and
>     elastic constant calculations for a number of ferroelectric and
>     paraelectric perovskites. While your ecut=45 looks quite large, the
>     psp's you are using have very localized states. In particular,
>     Psps_for_tests/22ti.psp.mod treats the 3s and 3p semicore states as
>     valence. All I can suggest is further convergence studies to get the
>     phonons of the basic cell converged with at most small asr errors
>     and no
>     unstable optic modes.
>
>     Incidentally, with ecut=45, ecutsm=0.5 is much too small (see the
>     technical discussion of ecutsm in the input variable list to
>     understand
>     why) and could produce spurious errors in the elastic constants even
>     with good phonon convergence. I'd suggest ecutsm = 0.05 * ecut,
>     approximately.
>
>     Good luck,
>
>     Don Hamann
>
>     Yong Liu wrote:
>
>     > Dear Abinitors:
>     > I've done the RF calculation about the elastic constants of
>     tetragonal
>     > PbTiO3, Cij consists with the previous results of experiments and
>     > simulations except for C12, C44, and C66. Consindering the error
>     > induced from the k-point grid (involving 115 k-points) and numerical
>     > error (ecut 45hartree), the p! ositive C12 and C66 are
>     acceptable. But
>     > the negative C44 totally confused me. Or this negative elastic
>     > constant has some physical means?
>     > P.S.
>     > Cij (Relaxed ion, GPa)
>     > 2.4239904 1.5071663 0.7684601 0.0000000 -0.0000001 0.0000000
>     > 1.5071663 2.4239877 0.7684603 0.0000000 0.0000001 0.0000000
>     > 0.7684603 0.7684601 0.5754473 0.0000000 -0.0000001 0.0000000
>     > 0.0000000 0.0000000 0.0000000 -0.5976210 0.0000003 0.0000000
>     > -0.0000001 0.0000001 -0.0000001 0.0000003 -0.5976204 0.0000000
>     > 0.0000000 0.0000000 0.0000000 0.0000000 0.0000000 1.0505907
>     > Results from Journal of materials research, 12:2623-2627(1997)
>     > c11 = 237Gpa, c33 = 60Gpa, c44 = 69Gpa
>     > c66 = 144Gpa, c12 = 90Gpa, c13 = 70Gpa
>     >
>     >
>     > Sincerely.
>     > Yong Liu
>     >
>     >
>     
> -------------------------------------------------------------------------------------
>     > Yong Liu,
>     > Institute of Inorganic Materials,
>     > Department of Materials Science and Engineering,
>     > Zhejiang University,PR China
>     > Doctoral student
>     >
>     >
>     ------------------------------------------------------------------------
>     > 雅虎1G免费邮箱百分百防垃圾信
>     > 雅虎助手¨D搜索、杀毒、防骚扰
>
>
>     -- 
>     D. R. Hamann Phone: 908-582-4454
>     Director, Theoretical Materials Fax: 908-582-4868
>     Physics Research (retired) email: drh@bell-labs.com
>     Bell Laboratories
>     Lucent Technologies
>     700 Mountain Ave, Room 1D-371
>     Murray Hill, NJ 07974-0636 USA
>
>     -outvars9: echo values of input variables ----------------------
>
>     Flags :
>     elaflag 3
>     instrflag 1
>     Miscellaneous information :
>     chneut 1
>     First list of wavevector (reduced coord.) :
>     nph1l 1
>     qph1l
>     0.00000000E+00 0.00000000E+00 0.00000000E+00 1.000E+00
>
>     Phonon wavevector (reduced coordinates) : 0.00000 0.00000 0.00000
>     Phonon energies in Hartree :
>     -7.023109E-04 -7.023109E-04 -6.750167E-04 1.545292E-04 1.545292E-04
>     3.871682E-04 5.365783E-04 8.008861E-04 8.008861E-04 1.097496E-03
>     1.097496E-03 1.331143E-03 2.188541E-03 2.188541E-03 2.820736E-03
>
>
>
>     -outvars9: echo values of input variables ----------------------
>
>     Flags :
>     elaflag 3
>     instrflag 1
>     Miscellaneous information :
>     asr 2
>     chneut 1
>     First list of wavevector (reduced coord.) :
>     nph1l 1
>     qph1l
>     0.00000000E+00 0.00000000E+00 0.00000000E+00 1.000E+00
>
>     Phonon energies in Hartree :
>     -5.573373E-03 -5.265437E-03 -4.493790E-03 -3.079270E-03 -1.679307E-03
>     -1.277654E-03 -1.014457E-03 -5.438577E-04 7.192503E-04 1.024782E-03
>     2.225236E-03 2.586592E-03 3.904726E-03 4.464320E-03 4.605258E-03
>     Phonon frequencies in cm-1 :
>
>
>
>     -outvars9: echo values of input variables ----------------------
>
>     Flags :
>     elaflag 3
>     instrflag 1
>     Miscellaneous information :
>     asr 1
>     chneut 1
>     First list of wavevec! tor (reduced coord.) :
>     nph1l 1
>     qph1l
>     0.00000000E+00 0.00000000E+00 0.00000000E+00 1.000E+00
>
>
>     Phonon wavevector (reduced coordinates) : 0.00000 0.00000 0.00000
>     Phonon energies in Hartree :
>     -5.573373E-03 -5.265437E-03 -4.493790E-03 -3.079270E-03 -1.679307E-03
>     -1.277654E-03 -1.014457E-03 -5.438577E-04 7.192503E-04 1.024782E-03
>     2.225236E-03 2.586592E-03 3.904726E-03 4.464320E-03 4.605258E-03
>
>
>
>
> Sincerely.
> Yong Liu
>
> -------------------------------------------------------------------------------------
> Yong Liu,
> Institute of Inorganic Materials,
> Department of Materials Science and Engineering,
> Zhejiang University,PR China
> Doctoral student
>
> ------------------------------------------------------------------------
> 雅虎1G免费邮箱百分百防垃圾信 <http://cn.mail.yahoo.com>
> 雅虎助手-搜索、杀毒、防骚扰 <http://cn.zs.yahoo.com> 


-- 
D. R. Hamann                            Phone: 908-582-4454
Director, Theoretical Materials         Fax:   908-582-4868
  Physics Research (retired)            email: drh@bell-labs.com
Bell Laboratories 
Lucent Technologies
700 Mountain Ave, Room 1D-371
Murray Hill, NJ 07974-0636 USA