[CP2K-user] [CP2K:16635] Re: Spin polarization and multiplicity in H-abstraction on a SiC surface where total magnetic moment changes

[Marcella Iannuzzi]

Dear Senja, 

With smearing activated, by default the multiplicity is not conserved.
Have you already tried the keyword

   - FIXED_MAGNETIC_MOMENT 
   <https://manual.cp2k.org/trunk/CP2K_INPUT/FORCE_EVAL/DFT/SCF/SMEAR.html#FIXED_MAGNETIC_MOMENT>

in the section 
CP2K_INPUT <https://manual.cp2k.org/trunk/CP2K_INPUT.html> / FORCE_EVAL 
<https://manual.cp2k.org/trunk/CP2K_INPUT/FORCE_EVAL.html> / DFT 
<https://manual.cp2k.org/trunk/CP2K_INPUT/FORCE_EVAL/DFT.html> / SCF 
<https://manual.cp2k.org/trunk/CP2K_INPUT/FORCE_EVAL/DFT/SCF.html> / SMEAR 
<https://manual.cp2k.org/trunk/CP2K_INPUT/FORCE_EVAL/DFT/SCF/SMEAR.html>?

Regards
Marcella

On Thursday, February 24, 2022 at 9:26:37 AM UTC+1 senja.r... at gmail.com 
wrote:

> Dear CP2K developers and users,
>
> I have a question regarding spin polarized calculations and the use of the 
> multiplicity and relax_multilplicity tags. 
>
> I would like to calculate the H-abstraction energy on my Si-terminated 
> 4H-SiC (0001) surface. 
>
> Initial structure: each Si surface atom has one dangling bond which is 
> passivated by hydrogen atom (non magnetic groundstate, magnet moment = 0).
>
> Final structure: one Si-H bond is broken, the H atom is moved away from 
> the surface. The dangling bond and H atom now have a spin polarized ground 
> state in which both unbound electrons occupy one spin channel (spin 
> polarized groundstate, magnetic moment = 2).
>
> I have plotted my results for VASP and CP2K in the attached figure. The 
> nonmagnetic calculations show good overlap (blue and purple curve).
>
> For vasp, one can fix the total magnetic moment with NUPDOWN = 0/2. If one 
> sets this to 2 and starts with a ferro magnetic initial guess for all atoms 
> (MAGMOM = N*1), the orange curve is retrieved. One can see that the 
> groundstate is nonmagnetic until 3.4A and then becomes ferromagnetic. 
> Additionally, one can not set NUPDOWN and let it relax (red and green curve 
> for two different initial guesses). In this case, VASP is close to the 
> correct transition but not perfect. 
>
> In CP2K, once I turn on the magnetization with UKS and MAGNETIZATION = 1 
> in the &KIND section, the curves are all over the place and hop in and out 
> of different magnetic states. I tried to tune this with the MULTIPLICITY 
> and RELAX_MULTIPLICITY tag without success. 
>
> My questions are:
>
> 1. How can I fix the total magnetic moment so that I could reproduce the 
> orange VASP curve with CP2K?
> 2. What's the influence of the multiplicity on this calculation (I tried 
> the default 1 and the triplet/3 settings). I suppose my system is similar 
> to the O2 example where the Si-dangling bond and H electron are similar to 
> the triplet state.
> 3.What does relax_multiplicity do? It seems to improve some data points 
> (grey vs. brown curve) but not all.
> 4. Which print settings should I use to get an overview of the magnetic 
> moment total (like printed in OSZICAR VASP) and on each atom (comparable to 
> LORBIT = 11 in VASP) 
>
> I would like to start using CP2K more but need to understand the spin 
> polarization. Can I improve my results to come closer to VASP (or maybe 
> even better)?
>
> Thank you in advance!
>
> Best wishes,
> Senja Ramakers (PhD student at Ruhr-Universität Bochum)
>
>

-- 
You received this message because you are subscribed to the Google Groups "cp2k" group.
To unsubscribe from this group and stop receiving emails from it, send an email to cp2k+unsubscribe at googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/cp2k/d4f93141-cf1f-423e-8538-5d4e3d9168a1n%40googlegroups.com.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20220224/410af254/attachment.htm>