I am encountering problems with adhesion of Ti/Au layers to GaAs
substrates after electron-beam lithography with PMMA.  My process runs as
follows:

* Spin, expose and develop etch pattern with Shipley S1813 optical resist
system.

* Etch substrate in buffered HF/peroxide etchant followed by soak in DI
water.

* Spin double layer 495k/950k PMMA (180nm, 50nm) with 150degC pre-,
intermediate and post-bakes.

* Expose with 100kV e-beam writer.  Broad area clearing dose has proven
to be ~500uC/cm^2 (this seems high perhaps?).

* Dip develop in 5:15:1 MIBK:IPA:MEK for 10s and rinse in IPA for 1
minute.

* HCl dip for 90s.  I've considered brief O2 ashing at this stage to
remove resist residue, but am concerned about the effects on fine features
(~40nm).

* Evaporate 20nm Ti, 100nm Au and lift-off in hot acetone.


Adhesion problems become apparent with subsequent ultrasonic ball bonding.
Unprocessed GaAs chips in the same evaporation have no adhesion problems.

Can anyone suggest a likely cause?  Could a small residue of cross-linked
PMMA be forming in the exposed areas?

Regards,

Jonathan Griffiths