If you want a permanent bond, use lower power and shorter time. 30W
for 15 seconds if you can. Also, the age of the PDMS matters. If it
has been off the mold for a couple days, then you should clean it in
dilute HCl:water (1:5) for 5-10 min before doing the plasma treatment.
I got into the habit of always doing the HCl step to give me peace of
mind that the bond would work. Also, I've found sometimes it helps to
let the bond rest for 24 hrs before using the device. I had a recent
experience where I bonded several samples to glass cover slips, pulled
on one of them 5 min later and it came off. The next day I came back
to find that all of the others that were done identically were very
well bonded. This might just be superstition, I haven't investigated.

Obviously also make sure your Si is clean (solvent clean with acetone
& IPA at the very least).

That's what has worked for me.

Hope it works for you
-Joe Grogan

Quoting willfeng@bu.edu:

> I am looking for advice to bond a PDMS flow cell to a silicon chip.
> When I attempt to send fluid through the 20um wide and 120um high
> channels, the PDMS lifts off the silicon wafer. I usually clean the
> PDMS and the silicon using 20sscm flow of oxygen with an RIE, at
> 100W 50mTorr for 1 minute. The PDMS flow cell is approxiately 10mm x
> 5mm. Has anyone had success with a PDMS sample of that size? Thank
> you for your time.
>
> Willie Feng