Dear all,
I am trying to compute the Kapitza resistance between an Epoxy network reinforced by a SWCNT with the length of 5 nm. As you can see in the attached Lammps input file, to equilibrate the system at 300 K, I used Energy minimization followed by a NPT thermostat (for 100 up to 500 ps) and at the end, I used a NVE thermostat for 50 ps.
In the next step, I fixed the CNT and the Epoxy’s temperature to 350 and 300 K, respectively, using NVT thermostat for 200 ps. To simulate the heat transfer process, the NVT is switched off and the system is allowed to relax by performing. NVE simulation. The plan is to compute Kapitza through the recording of the CNT and epoxy temperature during the relaxation process.
My problem is whenever I switched to NVE the temperature in my system goes up and especially in the case of final NVE I faced the error " Bond between atoms is missing". As I have already read in the Lammps mailing list, the reason could be that my system contains highly overlapped atoms. I have also found that the fix NVE/limit can be useful in this situation.
I am wondering since by using fix NVE/limit, we are eliminating atoms movement and as a result, we are also limiting the temperature gradient in the system, could still using this fix instead of fix NVE ( in the equilibration process and in the final step) be a proper choice for me? (because in my case the temperature profile in the system is very important)
If no how can properly reduce overlapping in my system?
Any comment would be highly appreciated.
Dear all,
I am trying to compute the Kapitza resistance between an Epoxy network
reinforced by a SWCNT with the length of 5 nm. As you can see in the
attached Lammps input file, to equilibrate the system at 300 K, I used
Energy minimization followed by a NPT thermostat (for 100 up to 500 ps) and
at the end, I used a NVE thermostat for 50 ps.
In the next step, I fixed the CNT and the Epoxy's temperature to 350 and 300
K, respectively, using NVT thermostat for 200 ps. To simulate the heat
transfer process, the NVT is switched off and the system is allowed to relax
by performing. NVE simulation. The plan is to compute Kapitza through the
recording of the CNT and epoxy temperature during the relaxation process.
My problem is whenever I switched to NVE the temperature in my system goes
up and especially in the case of final NVE I faced the error " Bond between
atoms is missing". As I have already read in the Lammps mailing list, the
reason could be that my system contains highly overlapped atoms. I have also
found that the fix NVE/limit can be useful in this situation.
I am wondering since by using fix NVE/limit, we are eliminating atoms
movement and as a result, we are also limiting the temperature gradient in
the system, could still using this fix instead of fix NVE ( in the
equilibration process and in the final step) be a proper choice for me?
(because in my case the temperature profile in the system is very important)
If no how can properly reduce overlapping in my system?
i don't think (initial) overlapping atoms is the issue in your simulation.
it is more likely that you have chosen too large a time step or that
there is a problem with your force field parameters (e.g. values
incorrectly set to zero).
using fix nve/limit will treat the symptom and not the cause and will
render your simulation useless, as the reason for using fix nve is,
that you do not have any manipulation of the kinetic energy or the
volume of the system, but fix nve/limit *will* affect the kinetic
energy.
as for bad energy conservation, you may also consider picking a
tighter convergence for the ewald summation (and for better
performance, you may want to switch to PPPM instead).
axel.