Temperature in NVT AIMD rised to extermely high

[Jessie wang]

Dear all,

I'm trying to do NVT AIMD for a system (a cluster surrounded by water molecules). The water molecules was generated by packmol with tolerance 2.0.

I was set temperature to 298.15K, but after several steps, it rised to hundreds even thousands K.  I was used NOSE-HOOVER thermosat, and tried CVSR also. it does not work.

 Here I attached my input file, and any help would really appreciated. 

Thanks

Best,

Jessie

[Lucas Lodeiro]

As a first glance, the problem is probably located on your initial geometry... Packmol generates *good* solvation spheres, but they are far away from the minimal energy geometry, and probably the forces on atoms at the first step are very high. This explains why your kinetic energy rised a lot, a huge amount of potential energy is changing to kinetic energy. When I run this type of AIMD, usually relax the structure first with classic MD with OPLS-AA forcefield, to get a better geometrical guess to start the AIMD. Otherwise, you can do a geometrical relaxation (a non-tight relaxation) in ORCA at the same level to get a better geometrical guess. 

Also, I would add D3(BJ) dispersion correction to your calculation.

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 on the web visit https://groups.google.com/d/msgid/cp2k/7856dc7f-d236-477b-8c20-98fdd1d3037bn%40googlegroups.com.

[Jessie wang]

Thank you Lucas. 

I've applied  geo-opt before AIMD, temperature still rised a lot. But now I know the temperature rising is because the initial geometry from your advice. I'll try to improve initial geo. And it is also a good advice to add D3 dispersion correction.

I have another question, when we run AIMD, the outer sphere of water molecules sometimes run too far away from our system, and actually, the box is in a right size. What should I do to limit the motion of water to make it not run too far.

Best,

Jessie

[Lucas Lodeiro]

Hi Jessie,

If your optimized geometry still get high temperature at the beginning of AIMD, maybe it reached a flat profile of the PES or a metastable structure with high energy. In this case I would run an AIMD with NH thermostat in the MASSIVE setting with a short timescale, to couple thightly each DoF with the thermostat, maybe this helps to thermalize the system, obviously after that you will need to equilibrate with a global thermostat and reasonable timescale.

It is reasonable that some solvent molecules fly out of the system when the temperature rises a lot.... With an equilibrated system this would not happen. By the way, there are restraints in CP2K to add confinement potentials.

Regards - Lucas

To view this discussion on the web visit https://groups.google.com/d/msgid/cp2k/5dd3975a-13ef-451e-9b73-ff023bb05605n%40googlegroups.com.

[Jessie wang]

Thank you so much Lucas!!

Very helpful advices !

Best,

Jessie

[Niharendu Choudhury]

Hi Jessie,

I think the best option is to run a short simulation using classical MD and then start ab-initio simulation from the final configuration obtained from the classical MD.

Regards

Niharendu

[Jessie wang]

Dear All,

I changed thermostat to CSVR/MASSIVE and timecon to 50, after 1 ps, the temperature goes down gradually. Then I changed the thermostat to NH/GLOBAL with timeco 1000, but for now the temperature is  too low for the system. I set the thermostat as 298.15K, and now it 250K. 

I wonder what's going on there, and what should I do to correct this.

I'd really appreaciate if anyone provide any comments.

Best,

Jessie