All -

I would be glad to hear from anyone who has experience etching suspended
structures from LPCVD silicon nitride who might have seen the same
problem I'm about to describe.  We have Si3N4 membranes 1um thick and
3mm on a side, from which we etch a long, narrow island suspended by
legs ~15um wide and 1mm long.  This was difficult to do until we
accepted a suggestion that it's necessary to support the membrane
underneath in order to release the suspended structure.  So this is the
fabrication process:

Pattern suspended island on Si3N4/Si/Si3N4 wafer with Shipley PR
Release membrane with KOH etch (back side windows patterned previously)
Sputter 450nm Al on back side of wafer (underneath membrane)
Si3N4 RIE to release suspended structure
Al wet etch

The process works well and reliably, which is great, HOWEVER: there's
quite strong evidence that the suspended silicon nitride island gets
very hot during the Si3N4 RIE (much more than 200C). This is very
surprising, as there's a very large metal layer in direct contact with
the membrane/island. Essentially the heat deposited by RIE seems to
prefer to flow out through the thin narrow legs of the bridge (thermal
resistance = ~10^6 K/W) than to flow through the membrane (only 1um
thick) into a metal layer in contact with the platter.

Does anyone know (or has anyone measured) the thermal transport across
an LPCVD Si3N4/Al interface? The thermal interface resistance would have
to be anomalously ENORMOUS to explain the results I am seeing. But the
LPCVD nitride is a funny material in some respects, and I cannot rule
this out.

Appreciate any input.  Thanks!

--
Dr. Aaron Datesman
Post-Doctoral Research Associate
Materials Science Division
Argonne National Laboratory
9700 S. Cass Avenue
Bldg. 223, B-217
Argonne, IL 60439

630-252-9154 (office)
630-252-7777 (fax)
773-899-1095 (cell)
datesman@anl.gov