Problem with converting real units to metal for water simulation


I am going to simulate a system consisting of bunch of water molecules with SPC/E model for water molecule interactions. When I use real unit for LJ potential parameters, the simulation results including the thermodynamics are reasonable. But, when I use metal unit (with conversion factor as 1 kcal/mol = 0.04337 eV), the temperature grows unreasonably under npt simulation. What is the reason for this observation? I have attached the parameter for LJ for both of unit systems. I will appreciate your help in advance.

your damping coefficients are too large metal units. Remember, the unit of time in metal is ps. I think that is the reason.
you have to reduce it by 3 orders of magnitude when you convert from real to metal units.