Annealing for obtaining the structure of material by AIMD simulation？

[zhijian fu]

Dear all,

I want to obtain the structure of (ZrO2)92(Y2O3)8 by annealing from 1200 K(the melting state) and the AIMD simulations,

but I don't know how to set the input file.

I build an input file (in attachment)by referring to the cp2k/test examples, but the AIMD simulation can not be done.

Could you give me some advice?

[Marcella Iannuzzi]

Dear ?

In your input there are so many problems that it is difficult to know where to start advising. 

But in the first place, are you really convinced that you want to obtain the structure of  (ZrO2)92(Y2O3)8 from the melt with AIMD and a simulation cell of ~20 Å per side?

I would consider this a hopeless project. In general nucleating a structure from a melt is a so complex problem that there is a whole research branch on this and for sure straightforward MD is not the solution, not even using much larger boxes and classical force fields. 

Anyway, have you used DFT to check the electronic properties of the melt and compared to any known property?

Or do you have any experience with DFT?

Have you tested the DFT settings, like functional, basis set, pseudo potential, vdw corrections.

A PW cutoff of 400 eV is extremely small and it is not going to give any reasonable result. 

Also LDA is in general not sufficient to obtain reasonable results. 

Looking at your MD settings, you would like to run a constant pressure/constant temperature simulation, setting a thermostat, but with annealing. It is not clear how a thermostat should behave trying to keep a constant temperature while an annealing factor is applied. 

Do you have any experience with molecular dynamics simulations?

Moreover the annealing factor is 0.75, which means that the temperature initially set at 1200 K will be decreased to less than 300 K in 5 MD steps, which translates in about 3 fs. 

Do you think that this is realistic?

Regards

Marcella

[zhijian fu]

Dear Marcella,

Thanks for your reply!

I have used VASP for doing AIMD simulations, the CP2K input parameter comes from the VASP input file.

Because VASP NPT simulation is not stable, I want to use CP2K NPT ensemble for doing AIMD simulations.

As you suggested, I change a small system (ZrO2)26(Y2O3)3, and the system has 93 atoms and about 10 angstroms per side.

In fact, the PAW energy cutoff for VASP is 400 eV, and I change the cutoff for CP2K calculation (in attachment).

In the input set of CP2K, the basis set and potential is DZVP-GTH-PADE, does the DZVP-GTH-PADE belong to LDA ?

I don't know how to set input for annealing temperature, annealing velocity, and thermostat.

In fact, I want to anneal from melting state (1200 K) to 900 K with about the annealing velocity 300K/1.5ps (2*10^14 K/s),

and then do NVT ensemble simulation at 900 K from the structure of the NPT (1200 ->900 K NPT simulations), until reach 300 K, and then calculate the

phonon density of states at 300 K.

That is NPT( 1200->900 K), and then NVT(900 K), and NPT(900->600 K), and then NVT(600 K), and NPT(600-300 K), and then NVT(300 K) simulations.

And how to set input file for annealing temperature, annealing velocity, and thermostat?

Thanks,

Zhijian

[Thomas Kühne]

Dear Zhijian, 

yes PADE refers to a Paderborn fit of the well known PW92 LDA XC functional. 

Since using the regular ANNEALING statement the velocities are rescaled, this 

inherently is not compatible with any thermostat! After some time you will recognize 

that the thermostat will desperately induce kin. energy to compensate for the 

rescaled velocities to maintain the initial set external temperature. Hence, please 

use TEMPERTURE_ANNEALING, which, however, rescales the external temperature 

within the thermostat instead of the velocities (T ~ v^2). 

Best, 

Thomas

--
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/11a50782-6e7c-47d9-b93c-5cf6adffd3f6n%40googlegroups.com.
<ysz.inp>