Bowing of the membrane into the cavity can have a couple of causes. First, if
the cavities are sealed, during the bonding anneal the oxygen in the air sealed
in the cavity will be eliminated through oxidation, reducing pressure in the
cavity, resulting in the diaphragm deflection. Inplane stress in the diaphragm
can be calculated for a circular plate, given the deflection, and for the
dimensions mentioned, this would result in stresses on the order of 10-20MPa,
assuming a 10um center deflection.

Second, thinning of the wafers by means of lapping/polishing will exert
pressure on the diaphragms, which can also result in deflection into the
cavity. Also, because the diaphragm is fairly large, some non-uniform polishing
may be occurring, so that the diaphragm is of a different thickness than the
field areas. This would make it seem as if it is deflected. Do you know the
thickness variation across the diaphragm?

Lalitha Parameswaran
Massachusetts Institute of Technology