If you want such an accuracy over a large area, angle alignment will be very
critical. I used to prefer low magnification for angle alignment since depth
of field (hence seperation) is not a problem, but for this you need good
contrast (e.g metal, not a shallow recess etch). I used to place large L
patterns (20-50um wide, 300-500um long legs) on corners of my dies, then
align angle at low magnification and check the alignment at high
magnification.
Unfortunately I do not have any alignment marks handy from those days..
If you have a good angle alignment verniers or crosses will give you pretty
decent alignment, 1um was not a great deal, 0.5 um was attainable for
samples w/ edge beads removed.

Good luck
Erhan Ata

-----Original Message-----
From: Glen Landry [mailto:landry@anvil.nrl.navy.mil]
Sent: Wednesday, May 28, 2003 6:10 AM
To: mems-talk@memsnet.org
Subject: [mems-talk] Photolithography alignment marks


I am looking for other people's ideas for alignment marks.  I am using a
Karl Suss MJB3 aligner and I want at least 0.5 micron accuracy.  The marks
I am using consist of a ~100micron feature for rough alignment, and then a
vernier.  The vernier consists of five rectangles, 2 microns wide, and
separated by 5 microns.  The rectangles on the second mask are separated by
5.5 microns in such a way that when alignment is correct, the center
rectangles align.  When alignment is 0.5micron off, the rectangle one from
center is aligned.  It looks like this (excuse the crude drawing):

   O     O     O     O     O       base vernier
O      O      O      O       O    top vernier (aligned)

These marks can get me to better than 0.5micron, but I have to almost be in
vacuum contact before I can see my alignment clearly.  I have thought of
better alignment marks, but at $600/mask I want to get it right the first
time.  I am wondering if it would work better if the top marks were
larger/smaller than the bottom marks (I am using a dark field mask) or if I
would still be able to get as good of an alignment if my marks were say
5microns wide so I could see through them without being in very close
contact.  If anyone has a CAD/PDF design of marks that work for them,
please e-mail me.  Thanks.

Glen Landry