[lammps-users] Number of atoms in a layer during thermal conductivity computations

LAMMPS LAMMPS Mailing List Mirror
1

Steve, LAMMPS users,
While doing thermal conductivity computations in silicon (with Stillinger Weber potential) using fix/thermal conductivity, I notice while keeping the thickness of the each layer in the domain at 0.14 nm, the number of atoms in the various layers in the domain fall between 32 and 34, i.e. a fairly steady value.
However, for computations with copper (with EAM potential), I notice that the number of atoms in the layers (again 0.14 nm thickness) ranges between 11 and 32, i.e. a fairly wide spread.
I am wondering why this behavior is different in copper as compared to silicon. Is it because of the difference in crystal structure (diamond versus fcc)?
All the temperature profile plots in copper appear reasonable.

Also, based on some results I am seeing for thermal conductivity versus domain length, the phonon physics in copper is also not clear - but I guess that is a separate question/issue.

Thanks,
Sreekant

2

I don;t know. If you think something is wrong (which is unlikely),
then I'd visualize the system or dump out the coords and do a couple
snapshots manually to verify LAMMPS is binning things correctly.

Steve

3

Dear Sreekant,
Crystal direction (along with you have these thin slabs) might be an issue. Try increasing this .14 nm to about .5 nm and see you get similar density across. 1.4 Angs seems to low thickness value.

Regards,
Vikas