> After the structure is released etched from the
> silicon using XeF2, the aluminum bridge bows and twists due to stresses.
> What causes this and how can I prevent it?

I have some experience with low-stress sputtered tungsten, titanium,
and aluminum films.

Unless special care is taken, most sputtered metals have residual stress
that varies with the deposition conditions
(e.g., power/voltage, pressure, sputter gas, temperature).

Generally, the stress becomes more compressive at lower sputter pressure
and higher voltage/power.

You are probably more interested, however, in the stress gradient,
which causes the rolling up or down:
if the stress is more tensile at the top of the film than at the bottom,
it will roll up and vice versa.

If the metal has been through a high-temperature step,
annealing will change the stress, typically leaving it tensile
due to the slower contraction rate of the silicon substrate.

Furthermore, if a chemical reaction has occured
(e.g., oxide or silicide formation),
the reaction product at the surface (top, bottom, or edges)
will take up more or less space than the reactants,
leaving the product in compressive or tensile stress
with respect to the metal.

One approach to reducing the stress gradient is to keep the aluminum
at the same temperature during the entire sputter deposition.
Do this by slowing the deposition rate or heating the
wafer before deposition, if possible .

Another approach is to anneal the aluminum in a non-oxidizing environment.

For further reading stress in sputtered metal films,
see the series of papers by Thorton and Hoffman.

        --Kirt Williams

******************************************
Kirt R. Williams, Ph.D.
Research Staff Scientist
Lucas NovaSensor
1055 Mission Ct., Fremont, CA  94539-8203
Kirt_Williams@compuserve.com
(510) 661-6147  FAX (510) 770-0645
******************************************