[Pw_forum] converge problem in calculating the vacancy on TiO2 surfaces

[Ari Paavo Seitsonen]

Dear Vega,

   Just my guesses and suggestions: One could try mixing of the density, 
like

  &electrons
     mixing_beta   = 0.7
  /

(or a value even smaller than 0.7, or some other parametres for the 
mixing) since with a defect you might now have defect levels in the band 
gap, leading to a more difficult a convergence. Eventually one could also 
switch on the broadening of the occupation numbers, eventually switching 
it off for the physical/production jobs, but just to check if the problem 
in convergence comes due to this.

   And in the original paper the authors discuss spin-polarisation, one 
might test also this (nspin = 2), with different initial magnetisations 
[starting_magnetization(xx) = ...] for the atoms surrounding the defect.

     Greetings,

        apsi


On Sun, 1 Mar 2009, vega lew wrote:

> Dear all,
>
> I am planning to calculate the Oxygen vacancy properties of TiO2 surfaces.
> The calculation for the ideal surfaces are quite normal and the results are
> in line with the one that literature reported.  But, when I tried to
> calculate the anatase 101 surface with an Oxygen vacancy (O2c was removed
> directly from the stoichiometric Anatase 101as described in J. Phys. Chem. C
> 2007, 111, 16397-16404, I encounter the problem of convergence for SCF
> cycle. I tried to add 20% more  empty band to the system the problem still
> laid in. How could I overcome the problem?
>
> my input file was attached at the bottom of the mail
>
> thank you for reading. Any hints will be appreciated.
>
> vega
>

...

>
> &SYSTEM
>                       ibrav = 8,
>                   celldm(1) =19.3269,
>                   celldm(2) = 1.1094,
>                   celldm(3) = 1.7450,
>                         nat = 71,
>                        ntyp = 2,
>                       nosym = .true.,
>                   ecutwfc   =   30,
>                   ecutrho   =  300,
> /
>
> &ELECTRONS
>
> /
>
> &IONS
>                ion_dynamics = 'bfgs',
> /
>

-- 
-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-=*=-
   Ari P Seitsonen / Ari.P.Seitsonen at iki.fi / http://www.iki.fi/~apsi/
   IMPMC, CNRS & Université Pierre et Marie Curie
   Tel: +33-1-4427 7542, Fax: +33-1-4427 3785, GSM: +33-6-6736 3820