[lammps-users] problem with sw potential

                I am trying to do some thermal interface problems with silicon and other materials. I tried this program for silicon (4X6X8) with one end fixed using sw potential. When the ensemble shifts to NVE, the temperature is increasing to almost 650K. Below is the code i used for the program. I tried different things like longer NVT (2nanosec's) with a shorter time step, but nothing worked. If i use tersoff potential, i don't have this temp rise problem. I am not sure what the problem is.

   units metal
   dimension 3
   boundary s s s

  atom_style atomic
lattice diamond 5.43
region box block 0 4 0 6 0 8
create_box 1 box
create_atoms 1 region box
mass 1 28.01
pair_style sw
pair_coeff * * Si.sw Si
   region 8 block INF INF INF INF INF 1.0 units box
   group left region 8
  velocity all create 10.0 873586443 dist gaussian
   fix 1 left setforce 0.0 0.0 0.0
   velocity left set 0.0 0.0 0.0 units box
   fix 12 all nvt temp 10.0 10.0 0.2 drag 5.0
     thermo_style custom temp pe ke press
   thermo 10
   timestep 0.0005
   run 500000
   unfix 12
   fix 5 all nve
   timestep 0.0005
    run 2000000



have you visualized what is happening in your simulation?
that might have a hint for what is going wrong.

also you are using fix nvt across the group all, but only
a part of those atoms are moving (due to fix setforce and
zeroed velocities). that should throw fix nvt off, trying to
adjust the temperature to have the whole block at 10K.
wouldn't it be better to use these settings?
or only nvt/nve time integrate the mobile atoms?

group mobile subtract all left
compute mtemp mobile temp

fix 12 all nvt temp 10.0 10.0 0.2 drag 1.0
fix_modify 12 temp mtemp

thermo_style custom step temp pe ke vol
thermo_modify temp mtemp