I've made many, many batches of Al cantilevers from a number of
different sources of Al, on different types of substrates, usually Si,
but typically with a thin oxide between the Si and cantilever to avoid
silicides.
Al makes a lousy cantilever - it is almost impossible to get a
repeatable set of stresses in the material - one day it will bend up,
another it'll bend down. ANY heat or plasma treatment (removing resist
using plasma strippers), including release by XeF2 (it DOES generate
heat) or wet etching, will radically change the stress gradient in the
material. The melting point of Al is just too low.
As has been mentioned, the stiction from a purely wet release will make
the problem worse.
Also, Al cantilevers will suffer from long term creep - the material is
just too soft.
It may be fine for quick and dirty experiments, but don't count on it
for any kind of production devices.
I suggest you look at different materials for cantilevers.
Tom Rust
Nanochip Inc