HMDS does not work for SU8. In fact I have noticed the adhesion to get
worse. I've had better success using the same vacuum bake without the
HMDS treatment, and quickly spin the wafer while it is still warm.
This will improve adhesion. Also, a slow ramp is critical for the soft
bake and the post bake. About 5C/2min is slower.

On Thu, Jan 27, 2011 at 8:08 PM, Mikael Evander  wrote:
> Hi all.
>
> I just wanted to let you know that I did buy the SU-8 3005 and while the
adhesion definitely is better on glass than the 2000 formulation, it is
unfortunately not good enough. All patterns survive the development and look
nice but they are not very well anchored to the surface. As I need them to stay
there as permanent structures I would need to improve the adhesion a bit
further.
>
> All wafers are still piranha cleaned and either singed or put through HMDS in
a YES oven (since Microchem claims an improved adhesion with HMDS for the 3000
formulation). I've tried an HF-dip to clean and roughen up the surface but it
didn't seem to improve anything. I've overexposed a wafer on purpose to make
sure that it's not a matter of not cross-linking the layer closest to the glass
surface, unfortunately no effect there either.
>
> What I have left to try, as far as I can think as least, is:
> - trying longer PEB times and doing very slow ramping (although I can't see
any tendencies to cracking and my layer is really thin)
> - testing a couple of hours treatment in NaOH as I've read that it might help
> - try a plasma treatment of the wafer before putting it in the singe oven
> - try a silanization (APTMS/AEAPS) of the wafer before putting it in the singe
oven
>
> If anyone has any other ideas/suggestions they are still very welcome :)
>
> best regards
>
> Mikael