Dear Valerio you can try with this reverse technique, first make SU-8 channel on CMOS chip with the help of a channel photo mask having aligning mark   which exactly match or align with the align mark on CMOS chip to achieve microchannel at exact location with micrometric accuracy. Once the channel is made, then cover/bond with a cured PDMS slab by O2 plasma treatment. Hope this will help you..... good luck.

Dear Valerio, if you were dealing with a very small volume, like only several pieces, and if the main obstacle in your work was the high stickiness of the PDMS especially when the plasma treatment done, you may try to do the alignment before the plasma treating. Do the alignment under microscope, adding some water or alcohol may help with lubricating, fix one edge of the PDMS part then lift it off from the Si chip, so that the PDMS part can be folded back to the Si chip after the subsequent plasma treating. A flexible thick PE film will be needed to support the soft PDMS material during the operation until the bonding all done. the alignment accuracy will mainly depend on how you do the edge fixing. Just for your reference...

Dear Pradip,

Thank you for your encouraging suggestion. Your idea is great but I will keep it as a B-plan: the SU-8 fabrication of uchannels directly on the chip seems to be a little bit too risky for me since the chip is very small and I need high aspect ratio. So, it would work but there is a high probability of failure during the entire process. The PDMS bounding is more reversible than SU-8 fabrication on the chip surface. Any other idea?

Dear Xiagu,

Thank you for your great suggestion. I need micrometric precision, I think your method would not provide me the needed accuracy: too much operations between alignment and actual bonding. Do you have any other idea?

Thank you very much for your time!

Dear Bernhard Wolfrum, 

Thank you very much for your answer which I found really promising! Indeed, I tried with something similar we have in the cleaning room (flip chip bonder) and it seems I could reach very high precision in alignment. However, the problem is that after O2 plasma exposure I have very short time to bond the PDMS (1 minute - as you for sure know). So the time is too short to use the flip chip bonder.  

For this reason, I am thinking about using a different binder for the PDMS still using the flip chip bonder since I need high alignment. Do you use any type of adhesive when you use your bonding method? I am thinking about using SU-8 as binder so when cured the PDMS will be well sealed on the chip. Do you think it could work?

Thank you very much for your time and for your expertise.

Valerio