[lammps-users] volume change under NPT


I am trying to relax a small polymer network with about thousands of atoms in LAMMPS to get a correct density. I minimized the energy first and run MD for 20000 steps under NPT ensemble. The problem is that the volume of the box keeps increasing to several times of the original one while temp and pressure fluctuate in a normal range. I wonder if this is normal or i got something wrong. Here is the input file. Thanks!

units real
atom_style molecular
boundary p p p
pair_style lj/cut 10.0

read_data custom.dat

pair_modify mix geometric
pair_coeff 1 1 0.0038 3.195
pair_coeff 2 2 0.02378 3.8983
pair_coeff 3 3 0.02378 3.8983
pair_coeff 4 4 0.01935 3.6621
pair_coeff 5 5 0.02393 3.4046
pair_coeff 6 6 0.02393 3.4046
pair_coeff 7 7 0.0860 4.03

thermo 100
minimize 1.0e-4 1.0e-4 1000 10000

dump 1 all atom 100 custom.dump

fix 2 all npt 300.0 300.0 100 xyz 1.0 1.0 1000.0 drag 2.0
run 20000


That sounds bad --- the volume should stabilize to a fairly small finite value. There are a number of things that could possibly be the problem here that you'll need to investigate. Here are some possible problem areas:

1) Your initial configuration might be a poor one. Minimization is no cure-all. You might try cycles of minimization and dynamics, gradually increasing the timestep size.
2) Your NPT control constants might be too aggressive, although they appear to be reasonable.
3) Your force field might be incorrectly specified. If you're simulating a polymer network, why are there no bonds defined? If atoms are supposed to be bonded, but no bonds are given, there will be a strong repulsion and the system will expand.
4) You might try NVE for a long while until you get equilibrated before going to NVT or NPT.
5) Are the dimensions on your box and particle spacing all reasonable? Are units consistent?