Mallik,

If you are looking to bond the glass surfaces directly, then your best
bet is to use a plasma based activation set-up (although it is not
absolutely necessary).  However, the requirements for this type of
bonding are pretty severe, especially since, for biomedical
applications, you will probably be looking for waterproof seals.

To accomplish this, there are a few requirements.  Most important is the surface
roughness, which needs to be <1 nm (0.5 nm is better - this is a normal
roughness for prime silicon). Additionally, there are some
limitations on the amount of bow, although this requirement is material
specific (for instance, on standard thickness silicon, wafer bow <25 um
is acceptable, since the silicon can flex a bit; on 1mm glass, the bow
has to be less, since the substrate is not as flexible). Finally, the
TTV is also a requirement, although, again, it is material specific, and
for the same reason as bow.  A typical TTV for standard thickness
silicon is 5 um.

If these conditions are all met, then your pair would be a candidate for
direct bonding.  The general minimum requirements for direct bonding of
the glasses is that the wafer surfaces are extremely clean.  Then,
simply bringing them in contact with each other (taking care to initiate
the bond from a single point) will cause some adhesion to occur.  Wet
chemical methods (such as a reversed RCA - or SC - clean) and plasma
activation methods can enhance this initial bond force, although with
wet chemical methods, the initial bond strength is, at best, still weak.

For the wet chemical activation, a follow up anneal at ~1000 - 1100 C
would be required to make the bond permanent and water tight.

This is where the primary benefit of the plasma system comes in.  A
plasma activated bond requires, at the most, a 300 C anneal, so no
furnace is required.

In our experience, however, the source of plasma is pretty critical.
Inductively coupled plasma systems have been seen to rarely work.  RIE
or capacitively coupled systems have been seen to work in some
instances, but tend to run a risk in a few areas.  The process window of
these systems is relatively small, as the risk exists of roughening the
surface with the plasma.  Additionally, since the intent of RIE systems
is to etch, often, the RIE system will have contamination issues
(particles, etc.).

The systems EVG manufactures have been demonstrated to result in high
strength direct bonded substrates with anneals at temperatures less than
(sometimes much less than) 300C.  We have also demonstrated that we can
plasma activate a substrate more than 20 times with no measurable
increase in the surface roughness of the substrate, allowing a much
wider process envelope.

Best Regards,
Chad Brubaker