Perhaps a good methodology would build on what Joe describes--the
deposition of an oxide layer onto an Au substrate, and after bonding,
removal of the unbonded oxide in the microfluid channels, perhaps
through use of an acid to ionize the oxide. There is a good paper by
the Whitesides lab (discussed recently on this forum) on solubility of
PDMS with acids/bases; I can't remember what the paper said about
acid/base solubility. It's probably very low due to PDMS's
hydrophobicity.

Best,
Philip H.