How to set thermostat properly in CVD simulation?

Dear LAMMPS users:
I’m trying to simulate CVD reaction and molecules are deposited into the simulation box continuously. I intended to keep the tempreture of the system around 1300K by using NVT ensemble, while the fix NVT all nvt temp 1300 1300 10 command seems failing to work on the deposited molecules. Here is part of my input file:

reset_timestep 		0
fix			        NVT all nvt temp 1300 1300 10
molecule 		    benzene benzene.txt
region		        depo block -20 20 -20 20 -20 20 side in
fix			        d all deposit 5000 0 10000 43 region depo near 2 mol benzene vz 1 2 target 0 0 0 units box
timestep		    0.0002    	
run			        50000000
unfix		        NVT

What should I do to make the tempreture of the whole system under control?

That would be quite unphysical.

In an experiment, you only have the heat dissipation from the substrate to the bulk.
So at best you should thermalize only a lower part of your substrate. To do that efficiently, you probably should switch away from a Nose-Hoover thermostat, which is intended for bulk systems in equilibrium to correctly sample the NVT ensemble with a weak coupling, but not for dissipating large amounts of kinetic energy effectively. Instead you should be using fix nve and then a dissipative thermostat for the bulk region of the substrate.

Please keep in mind that depositing atoms or molecules introduces energy into the system. If you add them too fast, the temperature of the entire system will go up substantially. If you have exothermal chemical reactions, even more so.


Thank you Axel, I learned a lot from your professional reply!