[lammps-users] thermal/conductivity of Cu crystal


I'm trying to calculate thermal conductivity of copper using fix thermal/conductivity. Somehow after convertion to SI units, I obtain values of the order of 10^-1 W/m*K. The same results for silver. Does anybody tryied this?

My script follows. I obtain linear temperature profile, calculate gradient (in K/m), and divide the swapped energy by time (in sec), cross-section (in m^2) and temperature gradient. Finally, convert units from ev to J.
Any ideas? Many thanks in advance.

# Cu thermal conductivity test ---------------------

timestep 0.02
units metal
boundary p p p
atom_style atomic

lattice fcc 3.615

region full1 block 0 50 0 50 0 50 units lattice
create_box 1 full1

create_atoms 1 region full1
pair_style eam
pair_coeff * * Cu_u3.eam

velocity all create 300 6549325

fix 1 all nvt 300 300 100
fix 2 all langevin 300 300 0.2 654678
thermo 100
run 1000
unfix 2
unfix 1
reset_timestep 0

compute ke all ke/atom
variable temp atom c_ke[]/(1.5*8.617e-5) # to get temperature in K
fix 3 all ave/spatial 10 100 1000 z lower 0.05 v_temp &
            file tmp.profile units reduced

fix tc all thermal/conductivity 1 z 20 swap 100

thermo_style custom step f_tc # swapped energy in ev
run 100000

So long as you're careful about units, it looks good to me.
Maybe Reese can comment.


The thermal transport carrier in metals are not dominated by phonon but electrons so classical MD simulation underestimate the conductivity a lot. So you’d better figure out that first except worrying about your lammps script.