# Making substrate atoms in a simulation immovable (simulation of a polymer melt on a substrate)

I have a silicon substrate, with a polymer melt placed on top of it.
Objective is to apply gravity fix on the melt (thus letting it fall) onto
the silicon substrate below. The silicon substrate I want to model as an
immovable thing. (number of atoms is too large for my system).

How does one make a group of atoms immovable in LAMMPS? I tried defining my
dynamics only on the melt group of atoms, but lammps is telling me 'All Pair
Coeffs Not Defined' and 'All Angles Coeffs Not defined, etc until I define
all interactions between Si-Si etc also. And once I define all interactions,
it tells me atoms not in image)

How does one go about simulating such systems?

Regards
Brahm

Define the system as if all atoms would move (assign
pair styles, coeffs, etc). But then do not include
the substrate atoms in the integration fix, e.g. fix nve.
If they are not integrated, they wonâ€™t move.

Steve

I have a silicon substrate, with a polymer melt placed on top of it.
Objective is to apply gravity fix on the melt (thus letting it fall) onto
the silicon substrate below. The silicon substrate I want to model as an
immovable thing. (number of atoms is too large for my system).

How does one make a group of atoms immovable in LAMMPS? I tried defining my

not time integrating them is one method. setting their force to zero
(after also initializing the velocities to zero) is another.

dynamics only on the melt group of atoms, but lammps is telling me 'All Pair
Coeffs Not Defined' and 'All Angles Coeffs Not defined, etc until I define

this is a different problem. you need to have parameters for all
interactions. however, you can exclude (non-bonded) interactions that
you don't want or need to compute from the neighbor list, similarly,
you can just remove the definition of bonds and angles, etc.

all interactions between Si-Si etc also. And once I define all interactions,
it tells me atoms not in image)

that is yet another kind of error and hints at a bad/inconsistent
initial geometry.

you have to be more systematic in building your system properly
tracking down error messages to the real error.

How does one go about simulating such systems?

if you look through the mailing list archives, you'll see plenty of
discussions explaining exactly that thing. if your simulation doesn't
work, it is due to *other* problems.

axel.