NPT for a subset of a bigger system

LAMMPS LAMMPS Mailing List Mirror
1

Hi,

I have a simulation box containing 500K atoms. The final configuration has some atomic defects in it and well equilibrated.

Now, I have selected an inner box from this entire simulation box such that the inner box contains the defect region. The inner box has about 20,000 atoms in it.

I performed a NPT run with this small simulation box so as to make the final pressure to be equilibrated about zero.

Here, I am facing the problem. The structure/arrangement of the atoms inside this inner box is seen to be changed significantly. In some region, they are disordered, in other region the group of atoms move as if they are driven by a torque. It seems to be strange.

In the simulation part, I have given the input configuration file in the LAMMPS specified format.
For doing the NPT, I have used:

fix NPT all npt temp 300.0 300.0 0.1 iso 0.0 0.0 0.4

Is there something to be done extra when equilibrating a structure which is a subset of a bigger system ?

Thanks for your time.

Regards,

2

There is not enough information in your post to tell why you are
seeing strange behaviour. However I have a question:

  When you run the new NPT simulation, did you
1) create a new DATA file containing only the mobile atoms? ("defect region").
  Or did you
2) use the original DATA file containing all of the atoms in your system?

You can run the simulation either way. However to keep it simple, I
suggest you use option 1 (create a new small DATA file with only the
movable atoms). This is much easier to do than trying to run an NPT
simulation using a subset of the atoms in your system.
Cheers

Andrew
P.S.
If you are using method 2, you should not be using:

fix NPT all npt temp 300.0 300.0 0.1 iso 0.0 0.0 0.4

Instead, you should define a group (eg. "mobile") containing only the
atoms you wish to move, and use:

fix NPT mobile npt temp 300.0 300.0 0.1 iso 0.0 0.0 0.4

There may be other necessary tricks to make this work well
(neigh_modify exclude), but I will let others comment on those.

3

Andrew,

Thank you very much for your reply.

I am using 'Method-1" : (create a new DATA file containing only the mobile atoms (“defect region”).)

But, a NPT run on such a system makes a distorted atomic arrangement.

I could not find the way out !

Thanks,

4

Hi,

I have a simulation box containing 500K atoms. The final configuration has
some atomic defects in it and well equilibrated.

Now, I have selected an inner box from this entire simulation box such that
the inner box contains the defect region. The inner box has about 20,000
atoms in it.

I performed a NPT run with this small simulation box so as to make the final
pressure to be equilibrated about zero.

Here, I am facing the problem. The structure/arrangement of the atoms inside
this inner box is seen to be changed significantly. In some region, they are
disordered, in other region the group of atoms move as if they are driven by
a torque. It seems to be strange.

i don't think it to be that strange. i think that your overall
strategy is flawed.
how should it be possible to do a volume relaxation of a system that
is embedded into a larger, system that you keep constant? how do you
guarantee, that you have the perfect continuation of your subsystem,
so that you can use periodic boundaries without creating massive
stress along the surfaces where you cut out the subsystem?

axel.

5

Another thing to consider is the decay length of the stress field
created by your atomic defects. You don't really comment on the type
of defects that you have (vacancy clusters, dislocations or just some
misplaced atoms here and there) so its hard to judge if the stress
field of the defects could also be affecting the stability of your
system. Yet, as a general rule the decay length of stress field can be
on the order of the defect size (longest wavelength roughness if one
is talking about a rough surface). If you chop your initial system too
much those defects could start interacting through the periodic
boundaries...
Carlos

6

Thank you for all of the comments.

Dr. Axel mentioned about the continuation of the periodicity of the sub-system to avoid stress. I agree and care was taken to avoid such situations as far as possible.

Dr. Carlos also mentioned about the nature of the defects. The sub-systems consists of many types of defects arising out of the radiation knock. We did a detail defect analysis and found that this region contains defects like vacancy and interstitial clusters, stacking faults and Shockley partial dislocations.

It is understandable that these defects will interact with their images through periodic boundary conditions. Here, I have a query. How to understand that what should be the size of the sub-system so that such interactions can be avoided ? Is it a hit and trial method ?

Thanks for your time.

Regards,

7

Thank you for all of the comments.

Dr. Axel mentioned about the continuation of the periodicity of the
sub-system to avoid stress. I agree and care was taken to avoid such
situations as far as possible.

sorry, but in these cases there is no "as far as possible". either you
have a proper perfect continuation or not. if not, you have a mess at
your hands. it is pointless to argue about secondary effects, when
there is a fundamental problem in your approach.

axel.

8

Prithwish,
Let’s say you indeed managed to satisfy perfectly the periodicity of your system when you do the cut. Still, notice that you went from 500K atoms to only 20K. At zero average pressure your modified box should be displaying volume oscillations from the NPT integration substantially larger than the original box. This will result in additional stresses for the small system. You could try damping the dynamics of the box oscillations and see if that helps.
Carlos