Thanks for answering me!
In fact, The recipes where I took notice that the tube is fluched by NH3 before
and after the step of the silicon nitride deposit are set up to make
stoechiométric nitride. Or, my objective is to deposit low stress (silicon-
rich) nitride. So, I think that instead of sweeping the tube with NH3, doing it
with DCS will be more convenient, seeing the nature of the deposit I want. I am
waiting for your suggestions. Thank you very much in advance.

best regards;
Nader

Pierre Huet  wrote:
I think the rationale may be that for the first few seconds or so of
deposition, the deposition parameters in the tube are not stabilized;
the gas flows, the total pressure, the precursors' partial pressures,
etc. have not reached steady-state. During this unstable period, if you
have a temporary excess of DCS vs NH3, the deposited film may be
silicon-rich, which may not be what you want. Same thing when the gas
flows are cut, at the end of the deposition. Making sure that the tube
is flushed with NH3 before and after deposition ensures that there is
never an excess of DCS vs NH3, so the film composition will be
homogeneous throughout its thickness.