Hi Stanley,
You do not list  the metal that was applied nor the following operations
of the prime wafers versus the test wafers, but I would be inclined to
think that considering your comments of lift off versus cratering ( the
glass) that there is either an oxide build up on the metal or excess
material from the dicing operation on the bonding pad  that will not
allow the bonding wire attachment.
If your time/power and weight are normal for the wire bonding operation
as compared to the test wafers, then surely there is surface interference.
Not sure what to offer to resolve this, but at least check your post
dicing cleaning process to insure full residual material removal and
also try to reduce any temp excursions of the wafers/ bonding pads that
would create a higher oxide level

Best
Ken

Stanley R. Shanfield wrote:

>We are trying to manually wire bond to thin (~.2 micron) Pt/Au on glass.
Although we get good bonds on test wafers at the time the metal is deposited,
the bonding is very unreliable after it has gone through more processing (i.e.
at wire pull test, the wire falls off, leaving the pad intact,& undamaged; or
the pull strength is low). The wafer sees anodic bonding conditions at one point
in our process sequence, and the sawing process can leave scratches on the pads
(the scratches do not penetrate the metal).
>
>Going up in bonding energy, time and force is a concern, since the glass
sometimes forms fissures under the bonding area at more extreme bonding
conditions, and a subsequent wire pull creates a pull-out from the glass. It
does not seem to help anyway. These bonds have to survive a ~350C temperature
excursion, so a pull-out is a concern.
>
>Has anyone ever dealt with a problem like this? Any advice or observations?