[Wannier] Polarization of isolated H20 (dipole moment)

Giovanni Pizzi giovanni.pizzi at epfl.ch
Thu Jul 5 10:03:34 CEST 2012


Dear Pedro,

one possible reason for discrepancy is due to the finite k-sampling of 
the Brillouin zone for the evaluation of the WFs. The derivatives in k 
space are approximated with finite differences, and the WFs become 
periodic; this can induce some error on the calculated WFs centers and 
spreads. In particular, for Gamma-only sampling, the periodicity is the 
same of the lattice. You can try to use a finer k mesh (2x2x2 or larger 
instead of gamma-only) and see if things improve (note that we expect 
that the bands are flat in k-space; this is only a way of reducing the 
error on the calculated WFs due to the finite k-sampling).

There is also another detail to which you should pay attention: if the 
plane-wave calculation is converged with respect to the box size. Try to 
increase the cell size from 10.6 angstrom to larger values until the 
value of the dipole (calculated by QE) doesn't change significantly 
anymore. For small box sizes, in fact, you may have spurious effects due 
to the interaction of one water molecule with its periodic images.

Hope it helps,
Giovanni Pizzi


On 07/02/2012 08:42 PM, Pedro Augusto F. P. Moreira wrote:
>  Dear all,
>
>  I have not calculated correctly yet the spontaneous polarization of a 
> water molecule as I mentioned in my last e-mail. Here, I gonna detail 
> my calculation steps and put my inputs and main output and hope that 
> someone could give any advice.
>  As I said, I did two simulation: (1) as Cantele suggested: my h2o.scf 
> input  had tefield=.true. (it can be seen below); (2) And using th 
> Wannier funciton centres which I calculated with following sequence of 
> commands:
>
> 1) pw.x < h2o.scf >scf.out
> 2) pw.x < h2o.nscf >nscf.out
> 3) wannier90.x -pp h2o
> 4) pw2wannier90.x < h2o.pw2wan > pw2wan.out
> 5) wannier90.x h2o
>
> My inputs and main outputs are bellow. The polarization, using WFs, 
> were calculated by the vectorial sum of ion positions times their 
> atomic number less twice the WF centre sum.
>
>  I repeated all calculations, using a "relax" command instead of "scf" 
> command in step 1. The positions used on the following inputs 
> (h2o.nscf, h2o.win and WF polarization calculations) were the relaxed 
> atomic positions.
>
>  The polarizations, which I found, are:
>
>   Methods  |   Polarizations (Debye)
>  (1) tefield |    1.8179
>  (2) WFs    |    1.8966
>
>  using 'relax' command:
>  (3) tefield |    1.7828
>  (4) WFs    | 1.8544
>
>  Anyone knows why the polarization values are so discrepant. I note 
> that the discrepancies are greater if the atoms are dislocated to the 
> center of the cubic cell.
>
>   With best regards,
>
>   Pedro Moreira
>
> ##h2o.scf########################################
> &control
>     calculation='scf',              ! it was substituted for 'nscf' 
> and 'relax' in calculations with these names
>     restart_mode='from_scratch',
> pseudo_dir='/home/pedro/Documentos/espresso-5.0/pseudo',
>     outdir='/home/pedro/Documentos/espresso-5.0/exec1',
>     prefix='h2o',
>     tprnfor = .true.,
>     nstep = 500,
>     forc_conv_thr = 1.9e-4,
>     tefield=.true.
>     dipfield=.true.
> /
> &system
>     ibrav = 0,
>     nat = 3, ntyp = 2,
>     ecutwfc = 70.0, ecutrho = 700.0,
>     edir=3
>     eamp=0.D0
>     eopreg=0.1
>     emaxpos=0.5
> /
> &electrons
>     electron_maxstep = 500,
> /
> &ions
> /
> ATOMIC_SPECIES
>  H 1.00790 H.blyp-van_ak.UPF
>  O 15.9994 O.blyp-van_ak.UPF
> ATOMIC_POSITIONS angstrom
> O    0.0  0.0  0.0
> H    0.77 0.0  0.62
> H   -0.77 0.0  0.62
> CELL_PARAMETERS angstrom
>    10.6  0.00  0.00
>    0.00  10.6  0.00
>    0.00  0.00  10.6
> K_POINTS {gamma}
> ##############################################
> ##scf.out#######################################
>      iteration #  6     ecut=    70.00 Ry     beta=0.70
>      Davidson diagonalization with overlap
>      ethr =  4.01E-07,  avg # of iterations =  2.0
>
>      negative rho (up, down):  0.293E-02 0.000E+00
>
>      Adding external electric field
>
>      Computed dipole along edir(3) :
>         Dipole                0.7152 Ry au,          1.8179 Debye
>         Dipole field          0.0011 Ry au
>
>         Potential amp.       -0.0403 Ry
>         Total length         18.0280 bohr
>
>
>      total cpu time spent up to now is      134.0 secs
>
>      End of self-consistent calculation
>
>           k = 0.0000 0.0000 0.0000 ( 39751 PWs) bands (ev):
>
>    -24.8436 -12.5614  -9.1444  -7.1431
>
> !    total energy              =     -34.42311963 Ry
>      Harris-Foulkes estimate   =     -34.42315705 Ry
>      estimated scf accuracy    <       0.00000072 Ry
> ##############################################
> ##h2o.win######################################
> num_wann        =  4
> num_iter        = 100
>
> begin atoms_cart
> O    0.0  0.0  0.0
> H    0.77 0.0  0.62
> H   -0.77 0.0  0.62
> end atoms_cart
>
> begin projections
> random
> end projections
>
> begin unit_cell_cart
>    10.6  0.00  0.00
>    0.00  10.6  0.00
>    0.00  0.00  10.6
> end unit_cell_cart
>
> mp_grid    : 1 1 1
> gamma_only : true
>
> begin kpoints
> 0.0 0.0 0.0
> end kpoints
> ##############################################
> ##h2o.pw2wan##################################
> &inputpp
>    outdir = './'
>    prefix = 'h2o'
>    seedname = 'h2o'
>    spin_component = 'none'
>    write_mmn = .true.
>    write_amn = .true.
>    write_unk = .false.
> /
> ##############################################
> ##h2o.wout_scf##################################
>  Writing checkpoint file h2o.chk... done
>
>  Final State
>   WF centre and spread    1  ( -0.000000, -0.257416, -0.127263 )     
> 0.53681556
>   WF centre and spread    2  ( -0.399319,  0.000000,  0.338531 )     
> 0.49244402
>   WF centre and spread    3  (  0.399320, -0.000000,  0.338532 )     
> 0.49244441
>   WF centre and spread    4  (  0.000000,  0.257415, -0.127263 )     
> 0.53681748
>   Sum of centres and spreads (  0.000001, -0.000001,  0.422537 )     
> 2.05852147
>
>          Spreads (Ang^2)       Omega I      =     1.816167784
>         ================       Omega D      =     0.000000000
>                                Omega OD     =     0.242353686
>     Final Spread (Ang^2)       Omega Total  =     2.058521470
>  ------------------------------------------------------------------------------
>  Time for wannierise            0.012 (sec)
>
>  Writing checkpoint file h2o.chk... done
> ##############################################
> ##relax.out####################################
>      Computed dipole along edir(3) :
>         Dipole                0.7014 Ry au,          1.7828 Debye
>         Dipole field          0.0011 Ry au
>
>         Potential amp.       -0.0395 Ry
>         Total length         18.0280 bohr
>
>
>      total cpu time spent up to now is      688.6 secs
>
>      End of self-consistent calculation
>
>           k = 0.0000 0.0000 0.0000 ( 39751 PWs)   bands (ev):
>
>    -24.9139 -12.7725  -9.0391  -7.1487
>
> !    total energy              =     -34.42371576 Ry
>      Harris-Foulkes estimate   =     -34.42371607 Ry
>      estimated scf accuracy    <          4.4E-10 Ry
>
>      The total energy is the sum of the following terms:
>
>      one-electron contribution =     -64.27367874 Ry
>      hartree contribution      =      33.58470996 Ry
>      xc contribution           =      -8.36936347 Ry
>      ewald contribution        =       4.63384732 Ry
>      electric field correction =       0.00076917 Ry
>
>      convergence has been achieved in   4 iterations
>
>      Forces acting on atoms (Ry/au):
>
>      atom    1 type  2   force =    -0.00001778    0.00000017 0.00004093
>      atom    2 type  1   force =    -0.00002047   -0.00000010 -0.00001330
>      atom    3 type  1   force =     0.00003825   -0.00000007 -0.00002763
>
>      Total force =     0.000069     Total SCF correction = 0.000038
>      SCF correction compared to forces is large: reduce conv_thr to 
> get better values
>
>      bfgs converged in   6 scf cycles and   5 bfgs steps
>      (criteria: energy < 0.10E-03, force < 0.19E-03)
>
>      End of BFGS Geometry Optimization
>
>      Final energy   =     -34.4237157619 Ry
> Begin final coordinates
>
> ATOMIC_POSITIONS (angstrom)
> O        0.000004575  -0.000000364   0.015670447
> H        0.774640336   0.000000070   0.612163521
> H       -0.774644910   0.000000294   0.612166031
> End final coordinates
> #################################################
> ##h2o.wout_relax###################################
> Writing checkpoint file h2o.chk... done
>
>  Final State
>   WF centre and spread    1  (  0.000004, -0.255729, -0.113705 )     
> 0.53766112
>   WF centre and spread    2  ( -0.400232,  0.000000,  0.346762 )     
> 0.48541250
>   WF centre and spread    3  (  0.400236, -0.000000,  0.346762 )     
> 0.48540610
>   WF centre and spread    4  (  0.000005,  0.255727, -0.113705 )     
> 0.53766277
>   Sum of centres and spreads (  0.000013, -0.000002,  0.466114 )     
> 2.04614249
>
>          Spreads (Ang^2)       Omega I      =     1.801532201
>         ================       Omega D      =     0.000000000
>                                Omega OD     =     0.244610290
>     Final Spread (Ang^2)       Omega Total  =     2.046142491
>  ------------------------------------------------------------------------------
> ##################################################
>
> Em 29-06-2012 18:06, Pedro Augusto F. P. Moreira escreveu:
>>    Dear all.
>>
>>    I am trying to calculate the spontaneous electric polarization of a
>> single water molecule as a test.
>>    I am using Quantum Espresso (pw.x and wannier90). I did by two
>> methods:  (1) using  tefield=.true. as suggested by Cantele
>> (http://www.democritos.it/pipermail/pw_forum/2009-March/011936.html) and
>> (2) by Wannier function centres (WF).
>>    I run a set of simulations with different initial atomic positions.
>> The first simulation, I used the positions mentioned by Cantele and
>> found P = 1.82 D and P =1.84 D by method (1). By WF, P= 1.90 D and 1.92
>> D. The first values are calculated with BLYP functional and the second
>> ones with PBE functional.
>>    After that, I moved initially all atoms by a vector (1.0  0.0  0.0)
>> angstrom, relaxed the molecule and did again the calculations above. I
>> found for Cantele way: P = 1.78 D and 1.80 D. By WF, P = 6.33 D and 6.37 D.
>>    My questions are:
>>
>>    (a) I expected to find values closer to 1.87 D in all calculations,
>> but all they are diverging from this value by, at least, 2 %. Should I
>> really expect it ?
>>    (b) Anyone knows what can be happening in WF calculations when I moved
>> the atoms? I know that if the atoms were moved beyond the unit cell, a
>> constant factor would be summed to polarization. But, I think that is
>> not my case because the dislocation vector was 1.0 A , while my cubic
>> cell has 10.6 A edges.
>>
>>    With best regards,
>>
>>    Pedro
>>
>
> -- 
> Pedro Moreira
>
> IFGW - Unicamp - Brazil
>
>
>
>
> _______________________________________________
> Wannier mailing list
> Wannier at quantum-espresso.org
> http://www.democritos.it/mailman/listinfo/wannier


-- 
Giovanni Pizzi
Post-doctoral Research Scientist
EPFL STI IMX THEOS
MXC 319 (Bâtiment MXC)
Station 12
CH-1015 Lausanne (Switzerland)
Phone: +41 21 69 31159

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.quantum-espresso.org/pipermail/wannier/attachments/20120705/0fead94a/attachment.html>


More information about the Wannier mailing list