Hello Martin,

  Here's a little trick, but it's probably not accurate enough. I used it
with little samples, not whole wafers. (1) The sample is stuck frontside
down with dental wax on a glass slide and photoresist is spun on (make sure
the resist not only covers the sample backside but also a good part of the
glass slide as well). (2) Turn sample over.
 (3) Now the mask is aligned to the sample frontside and the resist exposed
through the glass slide (... the resist on the sample backside is not
exposed). (4) After development turn sample over. (5) Align the mask to the
patterns in the sourrounding resist on the glass slide. (6) Expose and
develop the previously unexposed resist on the sample backside. Regards,

-- Werner