I've never had to do it, but I've read in several papers that people often
coat the final mold with a layer of silane groups in order to minimize
adhesion between the PDMS and the mold. They mention is in section 2.2.2 in
this very useful paper:

Components for integrated poly(dimethylsiloxane) microfluidic systems
Jessamine M. K. Ng, Irina Gitlin, Abraham D. Stroock, George M. Whitesides *
Department of Chemistry and Chemical Biology, Harvard University, Cambridge,
MA, USA

If the suggestion given previously about spinning the second layer before
developing the first layer doesn't work, then you could try coating the
surface as they do in the paper (I suspect though that the previous suggestion
should solve your problem very well).


good luck,
Joe Grogan

Quoting Vishwanath Somashekar :

> Hi,
> I have been trying to put two layers of SU-8 on to a silicon wafer.
>
> Cleaning of wafer includes rinsing in ethanol and then with DI water
> followed by heating oven at 120 for about an hour.
>
> I then plasma treated(o2) the wafer to improve the adherence of SU-8
> to the wafer.
>
> I then exposed it and finally developed it after baking it.
>
> Now to lay the second layer of SU-8, I plasma treated the developed
> wafer and then spun coated this wafer with SU-8 and then followed it
> by Soft bake followed by exposure to UV and then baking it and finally
> developing it.
>
> When I cast PDMS to it, a part of the second layer came with the PDMS
> after baking. Could anyone of you let me how to over come this
> problem.