Le Cao,
   There are a few tricks to use for long TMAH etches. First, make sure
you have a good seal around the top of the reflux condesnsor.  We often
put a strip of electrical tape around the top of the beaker to improve
the seal with the reflux condensor. Second, add about 17% v/v isopropyl
alcohol to the solution to stabilize the etch rate and selectivity as a
function of time.

Here's a link to one of our lab write ups on bulk silicon etching:
http://www.mems.louisville.edu/index3.html

FYI for everyone: We have a number of resources for the MEMS comunity at
www.mems.louisville.edu ,follow links to the resource manual for a
number of SOPs but also follow the links to the Micro Course list and
the experiments there in.

-Mike Martin
  U. of Louisville


Le Cao Hoai Nam wrote:
> Dear MEMS researchers,
>
> I am using TMAH (20wt%, 90 degree C, magnetic stirrer 250rpm) to etch
a
> through-wafer cavity (525um) of a Si (100) chip. The mask layer is a
> thermally grown SiO2 (600nm). To protect back side, I use an etching
jig
> made of stainless steel with O-ring around the chip. I found that the
> etching process is extremely unstable and unreproducible.