Temperature change during NVE for double-walled carbon nanotubes

LAMMPS LAMMPS Mailing List Mirror
1

Dear Lammps users,

I am simulating the oscillation behavior of double-walled carbon nanotubes (DWCNTs), in which the outer tube is fixed and the inner tube is free. Airebo potential is used. First we run NVT ( 800 K) for 20 ps. After that, we run NVE for 2 ns. We plot the temperture histories of the DWCNTs (the whole system), the inner tube and the outer tube. We know that, the outer tube is fixed, so the temperature is zero K. Before 1.3 ns, the temperature of the system is about 800 K during NVE, but the temperature of the inner tube is about 1600 K ( I think that the kinetic energy of the inner tube oscillation introduces a higher temperature of the inner tube). The temperature of the inner tube suddenly rises sharply to 4.4861e+06 K at a certain moment (for example, 1.3 ns in this case), causing the inner tube to collapse. Could anyone tell me what happened during NVE. How can we avoid this happening?

The corresponding codes are listed as following,

2

Dear Lammps users,

I am simulating the oscillation behavior of double-walled carbon nanotubes (DWCNTs), in which the outer tube is fixed and the inner tube is free. Airebo potential is used. First we run NVT ( 800 K) for 20 ps. After that, we run NVE for 2 ns. We plot the temperture histories of the DWCNTs (the whole system), the inner tube and the outer tube. We know that, the outer tube is fixed, so the temperature is zero K. Before 1.3 ns, the temperature of the system is about 800 K during NVE, but the temperature of the inner tube is about 1600 K ( I think that the kinetic energy of the inner tube oscillation introduces a higher temperature of the inner tube). The temperature of the inner tube suddenly rises sharply to 4.4861e+06 K at a certain moment (for example, 1.3 ns in this case), causing the inner tube to collapse. Could anyone tell me what happened during NVE. How can we avoid this happening?

Did you also include your frozen part in the NVT? Did you include the frozen part in the NVE? I assume that in both cases you should not.
Also, divergence with NVE can be due to a wrong (too high) timestep.

Julien

3

Dear Julien,

Thanks a lot for your advice.

In my simulations, I assume that the frozen part is “boundary”, and the others is “mobile".

Maybe I should modify the script to:

4

Just don’t time integrate (fix nve) the boundary atoms. Then
they should not move. And only measure the temperature
of the mobile atoms (compute 10 mobile temp). And monitor
that temp (c_10) in your thermo output.

Steve