Adsorption Molecular Dynamics

[Seunghwan_Kwon]

Hello,

I am doing molecular dynamics of oxygen adsorption on Zirconium Carbide (100) surface,

already discussed in this paper: 'https://doi.org/10.1021/acs.jpcc.6b01460'

According to the paper, the oxygen adsorbs on specific site,

but in the attached files, oxygen does not adsorbs.

Rather, it forms carbon monoxide and go away in my attached XDATCAR.

What would be the mistake that I have made in INCAR?

I think POSCAR, KPOINTS, POTCAR do not have any problems.

Thank you!

[manuel_engel1]

Hi,

It can be quite challenging to reproduce such results with a purely stochastic MD (molecular dynamics) approach. There are no obvious mistakes in the INCAR file that I can see. Adjusting parameters such as GGA, ALGO, LREAL and the specific MD tags can change the result, but it is difficult to judge without knowing the system. A more sophisticated approach would be to use biased MD to constrain certain degrees of freedom via the ICONST file. Please consult the VASP wiki for more information.

[Seunghwan_Kwon]

Thanks for the reply

I have one more question.

I am planning to constrain the atomic distance so that carbons don't go away.

However, I doubt that it is reliable simulation to constrain artificially the movement of atoms.

Also, it seems unphysical that ZrC crystal structure collapses at 1000K, well below the melting temperature (about 3800K).

So, is constraining the distance appropriate strategy?

Thank you for reading such ambiguous problem.

[manuel_engel1]

No worries, I'm trying to assist you as much as I can. Biased/Constrained MD is a standard technique that gives accurate results if applied correctly. Here is a good reference that explains the method in more detail.

However, before attempting to do this, I would recommend increasing the accuracy of your MD simulation. The low melting temperature of 1000K indicates that the simulation is somehow not correct. I would first suggest changing "ALGO=Very Fast" to "ALGO=Fast" to see if the crystal still collapses. Playing around with the PREC parameter and the MD related tags might also improve the situation.