In answer to your question, the maximum thickness of SiO2 you can grow
thermally is best approximated using the Deal-Grove model
(first-order) since it is a thick layer.  Practically, if you are
using wet (steam) oxidation at atmosphere at 1100C, you can grow 1um
in a couple of hours, more than that takes longer.  2um is achievable
in about 15 hours, 3um in about 48.  It's a function of how long
you're willing to grow thermal oxide.  As far as thermal stress - I
have in the past used 3-3.5um thick thermal oxide layers for optical
applications (e.g. waveguides) without severe loss.  It does not have
pinholes or such if you use it for an etch mask either (wet or dry).
To my knowledge, as long as you start with a bare silicon wafer with
no patterns, you will not get thermal stress gradients.  Now, however,
if there are patterns on it - that's a different issue.  If the
patterns are smaller than the desired thickness of the oxide layer, or
the spacing between them is in that range, thermal stress can develop.
  You'd have to provide more details on your process and feature set
before I could really provide advice.
   -Eric
> ---------- Forwarded message ----------
> From: "guda reddy" 
>
>  I have question for SiO2 as a mask while etching Si in KOH. I have been
suggested that the maximum thickness of the thermally grown SiO2 that i
could use is only 1 micrometer and if i go beyond that there would some
thermal stress and thermally grown SiO2 is not stable. I would really like
to known whether there are any papers or any book that mentions about this.

If there are, could any of you suggest me. I would really appreciate for your
help.