[Pw_forum] phonon question :how to get improved frequency ? lo-to can change the relative strength of IR peak ?

[xu yuehua]

hello all :
i have some problem about phonon at  G point .
firstly ,i use the same major parameter as the literuture listed when
compute the phonon at G
BUT my results are a bit smaller than literature :
the relax section:
&CONTROL
calculation  = "relax",
  prefix       = "relax_specify_ibra_408_6_20+force_nogauss",
  pseudo_dir   = "/home/xx/intel_espresso-3.2/pseudo",
  outdir       = "/raid/xx/pwscf/tmp/",
  etot_conv_thr=1.D-6
  forc_conv_thr=1.D-6
/
&SYSTEM
  ibrav     = 4,
  a=20,b=20,c=2.9,cosab=-0.5,cosac=0,cosbc=0,
  nat       = 15,
  ntyp      = 2,
  ecutwfc   = 29.98744D0,
  ecutrho   =200
/
&ELECTRONS
  conv_thr    = 1.D-9,
  mixing_beta = 0.2D0,
/
&IONS
  pot_extrapolation = "second_order",
  wfc_extrapolation = "second_order",
/
 ATOMIC_SPECIES
H  1.0  H.pbe-van_bm.UPF
O 15.999 O.pbe-van_bm.UPF
ATOMIC_POSITIONS { crystal }......
.....
...--------------------------------------------------------------------
scf section:
&CONTROL
calculation  = "scf",
  prefix       = "fiveringwater_specify_ibra_408_6_20_force_noguass",
  pseudo_dir   = "/home/xx/intel_espresso-3.2/pseudo",
  outdir       = "/raid/xx/pwscf/tmp/",
  /
&SYSTEM
  ibrav     = 4,
  a=20,b=20,c=2.9,cosab=-0.5,cosac=0,cosbc=0,
  nat       = 15,
  ntyp      = 2,
  ecutwfc   =29.98744D0,
  ecutrho   =200
/
&ELECTRONS
  conv_thr    = 1.D-9,
  mixing_beta = 0.2D0,
/
 ATOMIC_SPECIES
H  1.0  H.pbe-van_bm.UPF
O 15.999 O.pbe-van_bm.UPF
ATOMIC_POSITIONS { crystal }..........
------------------------------------------------------------------------------------.......
phonons of h2o at Gamma
 &inputph
  tr2_ph=1.0d-14,
  prefix='fiveringwater_specify_ibra_408_6_20_force_noguass',
  epsil=.true.,
  amass(1)=1.0,
  amass(2)=15.999,
  outdir='/raid/xx/pwscf/tmp/',
  fildyn='fiveringwater.dynG',
 /
0.0 0.0 0.0
------------------------------------------------------------------------------

my results:(cm -1)(via lo-to split)



3539.38

 3540.60

 3541.95

 3544.93
 3553.70

the not via lo-to results are :

3538.754218

3539.743464

3541.102871

3541.989623
3543.425339

but the literature results are :
3547
3553
3555
3558
3559
 you may my results are acceptable .but i think if i use the same  parameter
to compute phonon as literature  and
Energy cutoff:408ev
Monkhorst-Pack
k-point grid:1*1*6
and the same PBE-USPP,the cell shape and size,
(the literature only lists these parameters)

 the above difference are rising from the relax and scf section, in these
work ,i use the same k-points and the energy cut as phonon .
 i have changed a lot parameter in relax and scf.but the result are lower
than the literature:
i  enlarge the encut to 500ev,k points to 1,1,12
ecutrho to 300RY,
ALSO I have change tr2_ph to  tr2_ph=1.0d-18,but the results are the same
general .
my question ,is there any solution to lift my frequency?because i am a
newer.,i have no idea.

another question is if i use the lo-to split ,the big strength of  IR peak
shifts to higher frequency,for example : 3538 peak is the tallest in not via
lo-to ,but after lo-to spilt the
biggest strength is 3553
what the cause ?
please help me about above two question .thanks a lot






Xu Yuehua
physics Department of Nanjing university
China
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.quantum-espresso.org/pipermail/users/attachments/20070620/c8a18afd/attachment.html>