[CP2K-user] [CP2K:21655] Re: Help Needed: CP2K SCF Convergence Issues on RuO₂ (110) Surface with OER Intermediates

[Marcella Iannuzzi]

Hello,

This last  suggestion is wrong, the outer_scf is only useful with the 
preconditioned OT method.

Other comments:
ALPHA should be in general smaller, < 0.01, if the  convergence is 
difficult.
There are other parameters that seem not optimal, like a rather low cutoff.
The posted output shows a series of not converge SCF cycles. The calculated 
forces are then wrong and modifying the coordinates with wrong forces is 
not a good idea.
Have you tried to use 3D PBC with standard periodic Poisson solver, rather 
than the 2D with MT?
It is advisable to use more recent and consistent basis set and pseudo 
potentials, e.g. BASIS_MOLOPT_UZH and POTENTIAL_UZH

regards
Marcella


On Friday, July 11, 2025 at 1:20:02 AM UTC+2 ann... at gmail.com wrote:

> The system is a RuO₂ (110) surface modeled with three layers — the bottom 
> layer is fixed during the calculation, while the top two are relaxed. I'm 
> using diagonalization for the SCF method. For the clean surface (*) and the 
> *O intermediate, SCF converges without any issues. However, once I 
> introduce H-containing intermediates like *OH and *OOH, the SCF becomes 
> extremely slow and often oscillates without converging at all.
>
> I've tried increasing MAX_SCF to 500, set Alpha to 0.1, and switched the 
> optimizer to CG, but the results are still not ideal. Since the system is 
> metallic, I didn't use the OT method.
>
>  Hello,
>
> First of all, no matter whether you are using OT or not, you certainly 
> need OUTER_SCF.  For example:
>
>       &SCF
>          MAX_SCF 20
>          EPS_SCF 1E-6
>          &OUTER_SCF T
>             EPS_SCF 1E-6
>             MAX_SCF 128
>          &END OUTER_SCF
>       &END
>
> Also, have you actually tried using OT with your system? Tweaking its 
> options like ENERGY_GAP might be helpful. And if it converges, you can 
> restart without OT from resulting wfn file.
>

-- 
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 visit https://groups.google.com/d/msgid/cp2k/d5a67577-3c09-4fb3-955f-f8ba927838aan%40googlegroups.com.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20250710/b9f5f68c/attachment.htm>