initial velocity of inserted atoms using fix_deposit

Hi,dear Ray.
Why do you say that the whole surface and the incident atoms should not be
thermostated? I do not understand it. Do you mean that I just thermostat
the substrate in the initial equilibrium piriod and unfix the langevin
after that and then insert the new atoms?

You should first equilibrate the whole substrate with thermostat, then in
your deposition simulation only thermostat the bottom portion of the
substrate so the surface (and the O atoms) are not thermostated.

It troubles me that I relax the substrate by fix-nve and langevin method
to 900K. The temperature keeps 900K well. However, after I fix_deposit the
oxygen atoms into the system with vz=-0.05. The temperature droped to 450K.
I tried to change the vz but nothing helps. Could you please tell whether
my thermostat is right?

There are multiple mistakes in your input script but I don't have the time
to go over them one by one. You are over-complicating deposition
simulations. Just take an equilibrated surface, thermostat it reasonably,
then deposit O atoms/molecules with reasonable intervals. Everything else
is redundant. Talk to your advisor about your setup.