problem with High temperature and viscosity

Dear Lammps Users,

I am working towards calculation of viscosity of colloidal particles (ranging from 50-200nm in size) with implicit solvent using Green-Kubo method in REAL units from the attached input script(LJ/CUT and NVE/LANGEVIN combination).
The script has been adapted from the example input provided at lammps website for liquid argon viscosity calculation.

The simulation gives me very high temperatures when simulating with the following lines included in script then I get very high temperatures and high viscosity(e^2 to e^10 Pa. S.)

but when i remove the unfix 2 command from the script I get almost no diffusion from the colloidal particles.

I have began working with lammps just couple of months ago and still following through the learning curve.
if there is someone who has also faced similar problem and solved the issue, would it be possible to give my input a quick look and help a newbie in better understanding the stragy and implementation??

gk_visc.in (3.9 KB)

I suggest before you measure viscosities you verify you can
just run reasonable NVE, NVT, and NPT dynamics with your
system. E.g. at different densities can you measure good
T and P values, can you see diffusive motion, etc.
You don’t say what your pair style is. Nve/sphere is not
needed for most pair styles. For pair colloid it is optional,
see the doc page.

Steve