Michael,

The shrinkage of the film during crosslink is not the only reason for
the tensile stress in the film; in addition, the thermal processing of
the material has an impact.

The only way I can think of to create a residual compressive film stress
in the SU-8 is to somehow process it in such a way that the substrate it
is on will shrink more during the thermal curing steps.  One way would
be to use a substrate with a greater CTE (coefficient of thermal
expansion) than SU-8, but this is a bit tough, since, if I remember
correctly, SU-8 has a CTE on the order of 50 ppm/C.

Another possibility (although I'm not sure how you would pull this off)
would be to somehow heat the substrate while at the same time
maintaining a lower temperature in the SU-8.

Best Regards,
Chad Brubaker