[Pw_forum] negative phonon frequencies again :: graphene

Martin Gmitra martin.gmitra at gmail.com
Tue Dec 18 21:02:56 CET 2012


Dear Shouting,

Thank you for your answer. I have thought that sum rules can be applied only
for q=0. The output of the dynmat.x says:

 %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
     Error in routine readmat (1):
     Acoustic Sum Rule for q != 0 ?
 %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

The q=(0 0.01 0) in the middle figure. Perhaps you have meant some other way?
I changed asr='crystal' for matdyn.x input and recalculate phonon dispersions.
It improves, but the negative frequencies are not healed completely.

Does it mean that there is no unique way to avoid the problem?

Best,
Martin Gmitra

Universität Regensburg
Institut I - Theoretische Physik
Universitätsstraße 31
D-93040 Regensburg
Deutschland/Germany

On Tue, Dec 18, 2012 at 7:10 PM, Shouting Huang <huangshouting at gmail.com> wrote:
> Dear Martin,
>
> I suggest you apply sum rule after ph.x calculations, which will eliminate
> negative frequencies in some cases.
>
> Best,
> Shouting Huang
>
> On Tue, Dec 18, 2012 at 11:38 AM, Martin Gmitra <martin.gmitra at gmail.com>
> wrote:
>>
>> Dear all,
>>
>> Let me open question of negative (imaginary) phonon frequencies one
>> more time to the forum.
>> I would like to calculate phonon dispersions for graphene. I can not
>> get rid of negative frequencies
>> for ZA mode in vicinity of Gamma point. Please have a look at attached
>> figure, where different USPP
>> with LDA and GGA have been used for different energy cut-offs. One
>> gets impression that increasing
>> q-mesh for interpolation removes the problem but it is not a case. I
>> have found a post (no further replies)
>> http://qe-forge.org/pipermail/q-e-developers/2010-June/000030.html
>> where the phonon dispersion does not suffer from the negative
>> frequencies at all. I would appreciate
>> if Huang Shouting provide more details of his inputs. I have to say
>> that I have tried many variations
>> of input parameters like (using C.pbe-n-rrkjus_psl.0.1.UPF if not
>> specified else):
>> -- experimental lattice constant and relaxed within used PP
>> -- variation of c/a, increasing vacuum up to c/a=13
>> -- with and without inversion symmetry (shifting atomic z positions out of
>> zero)
>> -- k-mesh for SCF calculation containing K_point=(0.3333333 0.57735026
>> 0) (yield very wrong phonons)
>> and without the K_point
>> -- k-mesh with and without shift (turning on/off Gamma point)
>> -- varying ecutwfc and ecutrho parameters for single mode calculation
>> q=(0.0  0.01  0.0) to inspect convergence,
>> see middle figure, for different PPs and cut-offs/I
>>
>> -- decreasing tr2_ph threshold for single mode calculation q=(0.0
>> 0.01  0.0), bottom figure, does not make
>> the frequencies positive for selected negative frequency (ecutwfc=57,
>> ecutrho=340), see arrow at middle figure.
>>
>> None of this variation brings me to success. Could you please comment
>> or give some advices?
>>
>> Best regards,
>> Martin Gmitra
>>
>> Universität Regensburg
>> Institut I - Theoretische Physik
>> Universitätsstraße 31
>> D-93040 Regensburg
>> Deutschland/Germany
>>
>>
>> Below are examples of input files used.
>>
>>
>> ----------------------------------------------------------------------------------------------
>>
>>  &CONTROL
>>                  calculation = 'scf' ,
>>                 restart_mode = 'from_scratch' ,
>>                       outdir = './tmp/' ,
>>                   pseudo_dir = './espresso/pseudo/' ,
>>                       prefix = 'grp' ,
>>  /
>>  &SYSTEM
>>                        ibrav = 4,
>>                    celldm(1) = 4.65857947376452 ,
>>                    celldm(3) = 6.09876841964 ,
>>                          nat = 2,
>>                         ntyp = 1,
>>                      ecutwfc = 37 ,
>>                      ecutrho = 320 ,
>>                         nbnd = 8 ,
>>  /
>>  &ELECTRONS
>>                     conv_thr = 1.0d-10 ,
>>                  mixing_mode = 'plain' ,
>>                  mixing_beta = 0.7 ,
>>              diagonalization = 'david' ,
>>  /
>> ATOMIC_SPECIES
>>     C   12.01078  C.pbe-n-rrkjus_psl.0.1.UPF
>> ATOMIC_POSITIONS crystal
>>     C      0.333333333    0.666666666    0.500000000   1  1  0
>>     C      0.666666666    0.333333333    0.500000000   1  1  0
>> K_POINTS automatic
>>  13 13 1   0 0 0
>>
>>
>> ----------------------------------------------------------------------------------------------
>>
>>  &INPUTPH
>>                       outdir = './tmp/' ,
>>                       prefix = 'grp' ,
>>                       fildyn = 'grp_nearG.dyn' ,
>>                        trans = .true.,
>>                 alpha_mix(1) = 0.7,
>>                     amass(1) = 12.01078,
>>                   iverbosity = 1 ,
>>                       tr2_ph = 1.0d-12 ,
>>                      nmix_ph = 4,
>>  /
>> 0.0  0.01  0.0
>
>
>
>
> --
> -------------------------------------------------------------------------------
> Shouting Huang
>
> PHD student in Department of Physics
> Washington University in St. Louis
> Room 364, Compton Hall
> 1 Brookings Drive, St. Louis, MO 63130
> -------------------------------------------------------------------------------
>




More information about the users mailing list