>I don't have bosch process fitted in my
>ICP system. Someone could tell me if it's possibile to simulate the
>process with alternate SF6, CF4 plasmas (adding H2 or Ar if needed)?

During the passivation cycle, you need to deposite a film of
plasma-polymerized fluorinated monomers (PPFM). This should be achieved
using a mixture of CF4 and H2 process gases. Take a CF4 plasma as starting
point, and add H2 until you deposit instead of etching silicon. Not too much
dc bias. Deposition rate should be about 50 nm/min, or so. (Literature:
Riccardo d'Agostino, H. Yasuda, for instance). Deposit for 5 minutes, and
you should see a film on a flat wafer (colour). If you carefully scratch,
you should see the film peel of, like thin resist layer. This will be your
passivation step.

Etching cycle: SF6 plasma etches blank Si not covered by polymer film. You
need to remove the polymer film at the floor of your feature using ion
bombardment. Place a polymer covered wafer inside your system, and record
the polymer removal rate as function of the dc bias (DekTak, alpha stepper).
Maybe additon of Ar may help here. The polymer removal rate at low/zero dc
bias is about the polymer etching rate at the sidewalls of your features.
Find a compromise between high polymer removal rate at the trench floor (dc
bias), low polymer etching rate at the trench sidewall (~zero dc bias) and
selectivity to mask. Your etching cycle must be substantially longer than
the time required to clean off the polymer deposited at the trench floor
during your passivation cycle.

I hope this gets you started. Further parameter optimization is required to
achieve well defined profiles.

B. Volland