Hi,

I am trying to calculate the thermal conductivity of a spherical copper particle in motion. I am using the Green-Kubo relation to do so. For some reason, I get a value which is much much higher (roughly 10 times higher). I used the same script to calculate the thermal conductivity of argon and that worked perfectly. I made sure that the volumes I am using are correct and that I am calculating the kinetic energy, potential energy and stresses of the correct group (as indicated in the manual page for the heat/flux compute).

Does anyone have a rough idea of what might be going on?

Thanks in advance for any help I get

Adding to Carlos comments,

You should actually get about 5%-10% of the experimental value if everything is right. Just because mostly electrons contribute to the thermal conductivity in Copper. Saying that, did you think about deleting center of mass velocity (as you particle is in motion) when you calculate virial and convective components of the instantaneous heat flux. I am not sure but that might be causing the issue (given you are using right force-field and not making some fundamental mistake as carlos suggested).

Regards,

Vikas

Dear Carlos and Vikas,

Thank you very much for your responses. The value I am aiming at is based on both, experimental data and other MD simulations.

Regarding deleting the center of mass velocity, do you mean from the per-atom kinetic energies? If so, is there a way to define a compute that does that? As for the stresses required for the heat/flux compute, I am using the virial parameter which ignores the kinetic energy terms.

Thanks again

Dear Carlos and Vikas,

Thank you very much for your responses. The value I am aiming at is based

on both, experimental data and other MD simulations.

Regarding deleting the center of mass velocity, do you mean from the

per-atom kinetic energies? If so, is there a way to define a compute that

does that? As for the stresses required for the heat/flux compute, I am

using the virial parameter which ignores the kinetic energy terms.

For starters, you can just make sure that the particle is not drifting or

rotating. The velocity command has options to zeroed translational and

rotational components of the system velocity as a whole.

NOTE: I never calculated \kappa for finite size systems thus there might be

other things to consider.

Carlos