I just implemented an EBR process that gives a wafer edge resist removal
and is perfection.

The EBR ridge is a known issue with EBR processes, but it has its fixes.

I am assuming that you are doing EBR with a novolak resin based positive
resist.

1. Use an EGMEA type solvent instead of acetone. One product is AZ70/30.
Basically you don't use acetone, but you use the solvent system of the
resist itself to do the EBR. It doesn't have to be the exact solvent
system of your resist, but a solvent that is similar to those used to
make novolak resin resists.

Acetone has a known tendency to creat resist ridges inside the EBR ring
and that is why the industry switched away from acetone in the 1980s. I
think early 1980s.

Acetone also absorbs the N2 used for dispense systems and then seltzers
(bubbles like a soft drink, soda, Coca-Cola) when it drops in pressure
will in transit through the dispense system. DON'T USE ACETONE.

2. Do your EBR at 1500 to 2500 rpm. The resist just inside the radius
path of the EBR stream footprint has a tendency to absorb the solvent,
become partially liquid and pull itself into a ridge. If you are
spinning at a high speed such as 1500 to 2500 rpm you will cause this
semi-soft material to flow outward into the EBR stream footprint path.

A good recipe is to do an EBR at some radius from the center, then move
out a small distance from the center and do some EBR for about 4-5
seconds. What happens is that this semi-liquid resist will flow out and
you can remove it.

3. Ask your vendor for assistance. It is not impossible that they might
be of help. More importantly, the first question your boss will ask you
is, "Did you contact the vendor." You can say yes and be done with it.

All humor aside, though I am not entirely joking. The vendor should be
able to help you with this. The EBR ridge is a known issue that was
solved in the 1980s. They should be able to give you good advice on
this.

Ed


-----Original Message-----
From: [email protected]
[mailto:[email protected]] On Behalf Of Bibbotson
Sent: Monday, March 10, 2008 6:12 PM
To: [email protected]
Subject: [mems-talk] Edge Bead Removal

Dear all,

I was hoping someone out there has some experience of the edge bead
removal
(EBR) of conventional positive photoresists from silicon/silicon oxide
wafers.

I've recently commissioned the automated EBR on our Karl Suss RC8
spinner but I've done some profiling and discovered that although the
EBR reduces the width of the edge bead, it actually increases the peak
heights.

Without EBR the photoresist ends up with a 2mm wide edge bead of up to
1.6um in height (for a spun film of thickness of 660nm).

With EBR, a new bead is created at the new edge of the photoresist where
the solvent hits.  This is only 1.2mm wide but has a thinner peak of up
to 3.5um high!  These peaks are only ~100um wide.

Should edge bead removal normally create a highly vertical edge or is it
normal for new peaks to be introduced?  Will the new thinner peaks have
less of an effect on high resolution hard contact lithography or is edge
bead removal not going to help at all unless I get a perfect vertical
edge without additional peaks?

About my process:  I am spinning the wafer at 1250RPM whilst the solvent
(acetone) is dispensed at pressure by an automated arm through a
syringe.
1250RPM is the maximum speed the system will allow and high speeds are
recommended to prevent backside contamination.

Many thanks for any comments in advance,

Bob Ibbotson
Heriot-Watt University