I realise this thread is getting a bit old, but it is a recurring theme.

The Bosch process can achieve vertical sidewalls, or slightly undercut.  It
is not able to achieve positive profile (i.e. 85 degrees) because of the
mechanism of sidewall protection.  If the hole gets wider as the etch
progresses, then etchant will penetrate behind the passivation in an
irregular pattern, creating ledges in the profile.  The exception is at the
bottom of narrow holes, which will tend to be slightly positive profile as
they close up (due to lack of ion bombardment at the etch face).

As the gentleman from Alberta implied, the way to achieve positive profile
in silicon is to use a cryogenic process.  This is not switched (unlike the
Bosch process) and relies on silicon oxy-fluorides to protect the sidewalls.
It requires a substrate stage that can be cooled to around -100 degrees
Centigrade, together with good thermal contact to the back of the wafer.  It
does not work well on small samples on carrier wafers!

Unfortunately, this is only an option if you have an Oxford Instruments or
Alcatel DRIE system.  STS systems do not have cryo stages and cannot be
retrofitted (as the o ring seals are too close to the cold area).

Martin