Hi Christoph,

> I exposed these layers with 120sec, 60sec and 30sec (with 9mW/cm2) and I'm
> observing some kind of overexposure (extrem with 120sec and nearly invisible
> at 30sec).
I am wondering about the dose - regarding the documentation from Microchem
and my experience even 30sec (270 mJ/cm2) is at the higher end of the useful
range for this thickness. What is the reason to over-expose the SU-8 in such
a way? Do you just want to learn something about failure mechanism or do you
want to use this for final processes?

> [Pictures at http://www.imtek.de/micro-optics/SU8/SU8_2010_Exp.jpg]
As far as I understood the pictures, the unwanted cross-linked areas are
only at the top of your layer and felt down during (after) the development.
If you look "live" at the structure, could you confirm this?

> All three wafers are treated the same way, except the exposure dose.

> Is this kind of a diffusion of the activated acid or just internal
> reflections
> between the substrate (Pyrex) and the mask?

I don't think it's diffusion. Diffusion takes place during the PEB (which
was the same for each structure). The only difference between the 3
structures is the amount of acid generated during exposure (which forms the
concentration gradient for the possible diffusion) and this could not differ
in such a significant way.

So two things could be an explanation:
1. Reflection at the boundary Pyrex/air (chuck) - do the same structures and
over-exposure on a silicon wafer show the same effect?
2. The use of broad-band UV (especially small wavelength) - assuming your
using no I-line (365 nm) filter, there could be the excitation of the photo
initiator in such a way that it acts as an undirected "light source". This
could only happen in the top layer (due to the strong absorption of small
wavelength UV in the SU-8). Consequently in this top layer the refraction
index will change due to partial cross-linking and "guide" the UV light
further (so something like the "T-topping effect"). If your using broad-band
UV, could you limit it to 365 nm and see if the effect still appears with a
corresponding over-exposure?

I am curious to know about the reason of your problem, because I saw similar
effects in smaller dimensions even with "normal" exposure dose, but never
investigated it systematically, the use of an I-line filter solved (reduced)
the problem.

Best regards
Jochen