Actually 6um of thermal oxide is impossible, so
that would be out of the question.  If you were
doing some sort of CVD, it would have to be LPCVD
because otherwise the stresses in the oxide would
crack itself trying to get that thick.

It seems the only option for you is optimizing
your photoresist or finding another photoresist
that works better for you.

Are you sure it is the hard bake that is causing
your PR to reflow?  It is possible with such a
thick PR you are actually having an angle
associated with your exposure.

Nik
Electrical Engineering
Penn State University

At 10:45 AM -0500 9/1/06, Hongjun-ECE wrote:
>If oxide is preferred I would recommend thermal oxide which is stronger as
>mask material than PECVD oxide. However, for Liu's case, using all kinds of
>oxide will meet challenges of selectivity. The best selectivity of the best
>oxide is ~1:100. So 550 um wafer needs >6 um oxide. It is hard to get so
>thick.
>
>Strong baked thick resist is still the solution for his case. If still not
>enough, combined mask of resist and oxide is an option. If the feature size
>is not critical, it SHOULD strong enough.