Help with Temperature Fluctuations in MD

[Rishikanta Mayengbam]

Dear users, 

I’m running an NVT MD simulation for a molecule on a slab (vacuum in y direction) in CP2K at 300 K using a Nose-Hoover thermostat, but I’m seeing extreme temperature fluctuations (up to 50,000 K within a few steps). 

Do you have any suggestions on how to stabilize the temperature? Any insights would be greatly appreciated.

Best,
Rishikanta

[Lucas Lodeiro]

Hi,

From your ener file it is clear that there is some high reorganization or reaction for your system... the potential energy falls from -6631 to .6669 Hartrees, 38 Hartrees of fall is not due to the normal AIMD oscillations or thermal effects. Did you optimize your slab structure before to run the MD?

Taking into account the atoms listed on the input you are simulating a perovskite Cs/FA/MA-PbBr_3, which has high reorganization at the surface, w.r.t. the bulk structure, when the cleavage to construct the slab is done. For Cs/FA/MA-Br surface, could be organic cation rotations, and for PbBr surface there is a high reordering due to surface states/dangling bonds, and also rotations of organic cations. Here you can read a paper dedicated to this features of perovskites: https://pubs.acs.org/doi/full/10.1021/acsomega.0c04420

Also, in your input, at fixed atom section you can use (if the complete list of numbers from 1 to 483 is needed): LIST 1..483

Regards - Lucas

--
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+uns...@googlegroups.com.
To view this discussion visit https://groups.google.com/d/msgid/cp2k/8119598f-54f0-4af3-b203-2bcc07165f3an%40googlegroups.com.

[Rishikanta Mayengbam]

Hi Lucas,

Thanks for your detailed analysis and suggestions. My system consists of an organic molecule above a CsPbBr₃ slab. The initial structure for MD was taken from a VASP-optimized configuration. However, I may need to reoptimize it again in CP2K before running the MD simulation.

I'll try your suggestions and see if they improve the results.

Thanks again for your help!

[Hasan Tunçer]

Hello,

I asked for help for the similar problem, you can see in the group. I was told to use REGION MASSIVE can help for a faster equilibration. Also, my comment could be about TIMECON, 1000 seems so high to me.

Regards,

Hasan

6 Şubat 2025 Perşembe tarihinde saat 16:30:13 UTC+3 itibarıyla Rishikanta Mayengbam şunları yazdı: