Nice to see you around. Below you will find my script to compute the lattice parameter for silicon using at 300K. This script worked on lammps-12Feb07.
boundary p p p
lattice diamond 5.44
region box block 0 4 0 4 0 4
create_box 1 box
pair_coeff * * si.sw Si
mass 1 4.664280d-23
velocity all create 300.0 8487 mom yes
neigh_modify every 3000000 delay 0 check no
neighbor 0.75 bin
fix 1 all npt 300.0 300.0 0.001 xyz 1.01325 1.01325 0.001 drag 0.5
run 2500000 every 40000000 NULL
Let me know if you succeed to compute the thermal conductivity for silicon, I have some values already computed from 300-1000K (also from Carlos Gomes) using Green-Kubo and different number of unit cells. I have also the associated lattice parameters. I am really interested to compare my values with other publications and also with your results. Among other things, it would be interesting to compare how many time steps are needed using NEMD to achieve convergence in the conductivity and the associated statistical errors.