Calculate thermal conductivity for single layer graphene

Hi all,

I want to calculate the thermal conductivity of graphene by using NPT to reach equilibrium volume at 300K then change to NVE to collect heat flux for Green-Kubo method.

Problem is the boundary condition should be periodic in in-plane direction but if i set boundary to be p p f, i can’t control the pressure in z direction for NPT. Not sure what can i do to control the pressure.

So i’m considering to set the boundary to be p p p first in NPT and change it to p p f in NVE to collect heat flux. Will there be any problem doing this?

I will be grateful for any guidance.


I don’t think classical MD is a good method for this kind of system, since the in-plane thermal conductivity of graphene is not entirely ballistic but also provided through the electrons.

That said, for a single sheet it makes not sense to run variable cell in the non-periodic dimension. Using a fixed boundary is the correct approach as that box size is effectively infinity. Thus you should not “control” pressure in that direction (it should be zero anyway, it only is not for the simple technical reason that you define a finite box).
If you study the documentation for fix npt more carefully, you will learn that it is possible to couple each direction to the Nose-Hoover barostat independently and can also choose to either enforce isotropic change of the x- and y- dimensions (by coupling them) or treat them independently (which is probably the more physical approach).

But then again, you will have to adjust the computation of the heat flux to be a 2d process and also you will only able to measure the ballistic contributions. You may also have to be very careful about getting well converged results.