> I would like to suggest using XeF2 as the last release step
> for the AFM cantilevers.  XeF2 is a gentle vapor-phase etchant
> that attacks Si at a rate in the order of 1um/min, depending
> on the exposed Si area.  Using this vapor-phase etchant, we
> have released Al structures as thin as 1500A and it could
> probably be even thinner.  The only question is if XeF2 would
> attack the SiN.  From our experience, XeF2 may attack certain
> types of SiN but not others.  The easiest way to find out is
> to try it out.  We have a XeF2 etching system at UCLA if you'd
> like to etch.
XeF2 is indeed a gentle (and simple) silicon etch.

I measured the etch rate of several materials in XeF2 in a setup in the U.C.
Berkeley Microfabrication Laboratory.

I found that when etching a whole 4-inch wafer, the ER of silicon varied with
location from about 0.1 to 0.3 um/min. Small samples were etched at over 1 um/
min.

I found that our approximately stoichiometric silicon nitride was etched at
about 12 nm/min, but, curiously, our low-stress, silicon-rich silicon nitride
was barely touched, being etched at about 0.2 nm/min. Results will likely vary
with deposition conditions and location, so it is advisable to run your own
experiment.

XeF2 is also readily masked by photoresist. I saw zero etching of two brands of
photoresist.

Kirt Williams
Berkeley Sensor & Actuator Center