If you could process a free-standing silicon membrane (possible??) with a known
(large ~cm^2)) area and density, then you could weigh it using an Cahn
electrobalance model G-2 and divide by area*density to obtain the thickness ..
the Cahn has a mass resolution of 50 ng! I've done this with sputtered aluminum
films of 100-nm thickness on glass slides to verify a profilometer measurement.
I measured the glass slide before and after a simple NaOH etch removal of the
aluminum, and got excellent agreement. Interestingly, the Cahn manual describes
measuring the masses of 'membrane filters' in Section 5 under special
techniques.

Good luck!

Thomas E. Wilson
Professor of Physics
Marshall University
Science 154
One John Marshall Drive
Huntington, WV  25755-2570
Tel: 304.696.2752
FAX: 304.696.2494