distance between some ions is very small for liquid

[jie_yao2]

Dear VASP group,

When I simulate the iron silicate melt at 8000 K using VASP 6. I always meet the warning "distance between some ions is very small" in the log file. I attached the POSCAR and POTCAR.

I might ask that is it possible to set for example RWIGS in the INCAR to reduce distance ?

How critical the warning is ?

And is there any possible solutions to solve ? (The POSCAR and POTCAR files are attached please)

thanks,

Jie

[ferenc_karsai]

Please also attach your INCAR, KPOINTS and OUTCAR file for that calculation.

[jie_yao2]

Sure. They are attached please.
The OUTCAR is too large to attached, therefore I attach log INCAR and KPOINTS.

Thanks.
Jie

[jie_yao2]

Hi Ferenc,

I just asked the same question during VASP workshop, I think it should be not very critical.

Thank you for your help,

Jie

[ferenc_karsai]

I have talke to Georg Kresse.
Although RWIGS is checked for the nearest neighbors, in the end it does not matter if ions get too close.

The important radii that should not overlap are the core radii for the projectors (RMAX in the POTCAR file - the units are in atomic units).
If these overlap then you will get most likely ghost bands and the calculation will not be correct.

For very high temperature and pressure you need to use very hard potentials.

Look for the potentials with most electrons in the valence and with the smallest RMAX. These are usually the ones with ending "_sv" or "_h".
So I would possibly try to swap the following:
Mg_pv -> Mg_sv
Fe_pv -> Fe_sv
O_h -> it's already the hardest
Si -> Si_sv_GW

I hope these will work for you.