Erkin,
     Keeping the under-layer from dissolving in the developer requires some
sort of serious hardbake, preferably in a vacuum oven with a nitrogen purge.
  This won't be perfect, but it would help with your problem.  You could
also try using two very different resists with the upper layer having a
developer chemistry that does not attack the lower layer.  Finally, you
could look at SiO2 or Si as a sac layer that you deposit once and pattern
twice with the dimple pattern being a short timed etch (preferably RIE).

Good Luck!

-Lou

>From: "erkin seker" 
>Subject: [mems-talk] MEMS switch contact dimple
>Date: Wed, 31 Aug 2005 09:46:31 -0400
>
>Hi,
>
>I am trying to build MEMS DC switches using both cantilevers and bridges.
>I want to fabricate a contact dimple using a bilayer photoresist process.
>I could get the dimple topography I wanted on the sacrificial layer by
>consecutively spinning and patterning two layers of AZ4110 in the past.
>However, now when I spin the second layer, it seems to dissolve the first
>layer where the dimple indentation is.
>
>Does anybody have a relatively straight-forwards recipe for fabricating
>contact dimples?