Hi Jeffrey,

Aside from the numerous surface treatments/modifications you can
employ to improve adhesion at the SU-8/Glass interface - many of which
will have a detrimental effect on adhesion, if the not performed
carefully, and (arguably) little beneficial effect even when followed
to the letter - you could try to mechanically fasten your SU-8 layer
to the glass surface. I would recommend this particularly in your
case, as you need to dice your wafer into individual dies at a later
stage. Mechanical interlocking can be achieved using standard MEMS
processing techniques - so no need for a DRIE to etch glass. I have
explained an approach that works in the following paper:
M P Larsson et al 2006 J. Micromech. Microeng. 16 S161-S168

A natural limitation would be the minimum thickness of your SU-8
layer, but the process is certainly scalable, subject to a few tweaks,
to cater for the needs of most.

Good luck and let us know how you get on.

Regards,

Michael