You wrote:

Hello folks,

I am looking for a way to make fairly thick microlenses on a patterned
silicon wafer.  The end result I'm hoping to achieve is an array of
hemispherical lenses with a diameter on the order of 20um (though it would
be nice to go even thicker as well if needed).  The lens needs to be
transparent in the visible spectrum (~ 400-800 nm).  Ideally, I'd like to
use glass/PMMA/PDMS, but am open to suggestions.

I've read of groups using photoresist reflow / grayscale lithography to
create the lens shape.  The profile is then transferred to an underlying
layer by RIE.  However, I don't know of any suitable material I could
deposit with the desired thickness (for example, I am limited to ~7um of
SiO2).  Is there anything suitable that I could put down and etch
anisotropically?  I have an ECR etcher (plumbed with SF6 and O2) and an RIE
(CHF3/O2) at my disposal.

Would it be easier to follow the molding approach that people seem to be
using.  (Creating a master and replicating in PDMS).  In this case, I was
hoping my mask aligner could be used for aligning/shaping the lens material..
Once in contact, I could give it a UV flood exposure.  Does this sound
feasible or do I need to apply heat at the same time for any decent molds?

Anyone have any experience making something similar?

Thanks,
Alasdair

Here's a crazy idea:

Cover the surface with a layer of 20µm glass beads, and reflow just enough to
get them to slump together into a single mass, but not enough to self level.

One source might be:
http://www.thermo.com/com/cda/landingpage/0,10255,1865,00.html or you might try
Scotchlite beads, IIRC they are a very high index glass.

I'd reflow under vacuum to prevent trapping air at the glass/silicon interface.
You'd need to adjust the time and temperature for best results.

You could probably even control the microlens placement with etched pits or
walls.

Hope this helps,

Dean Hopkins
MEMS process engineer
Silicon Valley, USA
(408) 426-0501