I am using LAMMPS package and rNEMD method to calculate thermal conductivity in a composite polymeric system and one graphene sheet. I know that for a simple system, using this method, a linear temperature gradient should be imposed on the system after reaching steady-state and using Fourier’s law we can get thermal conductivity. However, I am not sure that the same approach could be applied to measuring the thermal conductivity of composite materials, as for such materials, a temperature jump will happen in the temperature profile.
I would appreciate any help or suggestions on this question.
This is not really a LAMMPS question, but rather a question about the method. That is something that is considered off-topic for this forum categories. You may want to survey the published literature to see if there are studies for similar systems and what people did in that case, or consult/collaborate with a researcher that has a publication record in this field and thus a suitable level of expertise. Depending on your own seniority in the field, the letter step may be something that you have to defer to a more senior person in your research group.
I wonder what the logic is for choosing the rNEMD method to solve your problem. There are other options for calculating the thermal conductivity (with well-explained examples in the LAMMPS examples directory) that you can choose.
Also, based on what I understood from your words, your geometry is rather complex and is likely to cause problems in calculating the thermal conductivity. My guess is that in addition to the problem you mentioned, you will have additional difficulties while increasing your system size to converge your thermal conductivity values.