2008/2/20, Kevin Paul Nichols :
> Regarding Michael's point below, that cracking in SU-8 is due to
> over-exposure:

> While intuitively this makes sense, in practice it's not what I've seen. In
> my experience, a wide exposure matrix, starting well below the optimal dose,
> always shows obvious, macroscopic cracking when *under*exposed. These
> disappear entirely when overexposed, even when highly overexposed (though of
> course, eventually your features eventually disappear as well...)
>
> Anyone else care to weigh in on this?

There is another way of getting rid cracks.
I mean shortening PEB time, and also lowering
its temperature (e.g. 75 celsius deg instead of 90).
Of course, there is a need to use longer exposition
time (i.e. overexposition).
There are some papers about this problem,
e.g. this one:

Authors: Bo Li, Miao Liu, Quanfang Chen,
Title: Low-stress ultra-thick SU-8 UV photolithography process for MEMS,
Publisher: SPIE, year: 2005
Journal of Microlithography, Microfabrication, and Microsystems,
volume: 4, number: 4, page: 043008

In abstract authors write:
[...]
   However, the fabrication of thick SU-8 MEMS has experienced
   severe problems such as cracks, distortions, or delaminations during the
   fabrication process and/or postservices, due to the large internal stress
   generated during the photolithography process.
[...]
   Analytical results show that the post-exposure bake (PEB) temperature
   is the main factor in developing the resulted internal stress. Under the
   guidance of analytical results, an optimized UV photolithography process
   for the fabrication of ultra-thick low-stress SU-8 patterns is developed
   with conventional (simple) equipment. A low PEB temperature of 55°C
   reduces the internal stresses by more than 70% compared to those
   fabricated with the recommended procedure. Experimental results
   indicate that cracks, distortions, and delaminations are eliminated from
   the fabricated SU-8 structures using the newly developed procedure.
   In addition, the patterned SU-8 has a Young's modulus of 2.5 GPa
   and an ultimate strength of 50 MPa, which is about 50% higher than
   previous reported values.
[...]

It looks like revolution, but it is not as simple IMHO.
Anyway, cracks vanished in my case (at expense of
slightly lower quality of prepared channels).

There is one more paper considering the importance
of slow cooling after PEB, but it is obvious for
peopel here, I think...

Regards -
Olgierd