[Pw_forum] In the presence of electric field: which parameter help with convergence

[Christoph Wolf]

Dear Fedor,

thank you for your message. I have found a (fairly OK) solution to applying
higher fields that seems to work OK with graphene and is actually described
in example 10 of PW but I think the sample does not apply the technique. I
quote:

"To perform a calculations with an electric field, an estimate of
the optimized wavefunctions is needed to build the electric field
operator (See: I. Souza, J.Iniguez and D. Vanderbilt, PRB 69, 085106,
2004). Therefore when lelfield ==.true. a copy of the wavefunctions
is read from disk (i.e. startingwfc should be 'file')."


therefore one first runs the scf input file with everything present but the
efield turned off (efield_cart(1,2,3)=0.0). After converging this one
copies edits the efield (efield_cart(3)=0.02) and adds
"startingwfc='file'); this has allowed me to use much higher fields than a
"cold start" (startingwfc='random') and the behavior seems consistent with
literature! It seems that in the zero-field case all efield related inputs
have to be present, otherwise I ran into some I/O error ( Error in routine
gk_sort (1): array gk out-of-bounds)

Hope this helps people encountering convergence problems in the future!

With best regards,
Chris

On Fri, Jan 5, 2018 at 12:40 AM, Fedor Goumans <goumans at scm.com> wrote:
>
> Dear Christoph,
>
> While I don’t have an answer to your specific QE question, the
application of an electric field to a 2D material in a plane wave code may
be problematic to begin with.
> We had someone try out band gap closure in the ‘MoWSeS’ materials under
electric fields (which would amount to sawtooth potentials in PW codes if I
understood correctly).
> In that case PW codes would just give wrong behavior, while localized AO
codes with prober 2D pbc (and a homogenous electric field) would allow for
polarization across the surface.
> See e.g.:
https://www.scm.com/highlights/closing-band-gap-2d-semiconductors/
>
> Hope this helps,
> Best wishes,
> Fedor
>
> On Jan 1, 2018, at 13:13, Christoph Wolf <wolf.christoph at qns.science>
wrote:
>
> Dear all,
>
> a happy new year!
>
> I have recently played with 2-D materials (yes, graphene) and electric
fields. I encountered several situations where convergence is really hard
to achieve and I was wondering which of the following parameters is usually
considered to be helping with convergence
>
> &control
> ....
>     lelfield=.true.
>     gdir=3
>     nberrycyc=6
>    nppstr=1
> /
>
> &ELECTRONS
>  ....
>  mixing_beta=0.4
>  mixing_mode='TF'
>  efield_cart(3)=0.00275
> /
>
>
> the system is separated in z-direction by ~20A of vacuum. the cutoffs for
(NC) PPs were converged in the field-free case.
>
> What is a suggested "systematic" approach to get convergence in fields of
this magnitude?
>
> Your help is greatly appreciated!
>
> With best wishes for the new year,
>
> Chris
> --
> Postdoctoral Researcher
> Center for Quantum Nanoscience, Institute for Basic Science
> Ewha Womans University, Seoul, South Korea
>
> _______________________________________________
> Pw_forum mailing list
> Pw_forum at pwscf.org
> http://pwscf.org/mailman/listinfo/pw_forum
>
>
> Dr. T. P. M. (Fedor) Goumans
> Business Developer
> Software for Chemistry & Materials
> Vrije Universiteit, FEW, Theoretical Chemistry
> De Boelelaan 1083
> 1081 HV Amsterdam, The Netherlands
> T +31 20 598 7625
> https://www.scm.com
> https://twitter.com/SCM_Amsterdam
> https://www.linkedin.com/company/software-for-chemistry-&-materials
>
>
>



--
Postdoctoral Researcher
Center for Quantum Nanoscience, Institute for Basic Science
Ewha Womans University, Seoul, South Korea
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.quantum-espresso.org/pipermail/users/attachments/20180105/6616f229/attachment.html>