Patrick,
                    I have just spent some considerable time trying to solve
a similar problem, albeit with a polymeric scratch resistant coating on
polycarbonate components of a device that we are developing. There are
several approaches that you might want to try, depending upon several
factors such as the thickness of the photoresist film you require. Basically
(as you probably know), the film characteristics are determined by (i) fluid
viscosity (ii) molecular weight (iii) spin speed and (iv) spin cycle time.
Of lesser influence are acceleration and deceleration during the spinning
cycle itself. The hardcoating that we were trying to coat was similar in
viscosity and molecular weight to most of the standard Shipley resists. We
need a film thickness of less than 2 microns - no problem, except for the
edge effect.! We found that the edge effect was of the order of 30 microns
in height, surrounding an otherwise homogeneous 2 micron film.
We tried several methods to eliminate / reduce the edge effect, which I will
describe in some detail:

[1]    We solved the problem by directing a high velocity jet of gas (He, or
N2) onto the substrate edge. The gas flow was directed onto the very edge of
thy substrate by using a hypodermic needle. If the substrate was spun in an
anticlockwise direction, then the gas-flow was directed onto the substrate
edge, about 2mm from, and normal to the actual edge, pointing away from the
center of the substrate. The gas-flow was furthermore angled at about 10
degrees to the plane of the substrate. You will have to find the velocity of
gas flow that works siutably for your application - naturally the gas-flow
needs to be greater than the velocity of the passing substrate edge!!
However - this method worked GREAT for us. Our edge effects, went from 30
microns on a 2 micron thick film (50mm diameter substrate) to a very
tolerable 3 microns. It is important however that the spinning process is
started, and the fluid coating given time to spead out evenly over the
substrate BEFORE applying the gas flow. Otherwise you end up with a mess!
Also, it is important that the gas flow is stopped, just before the spinning
cycle is finished, otherwise the gas jet may leave a mark on the substrate /
coating. We probably could have eliminated the edge effect entirely, by
experimenting a bit more with gas flow velocities, timing and spin speeds -
however we got to the stage where we could easily live with the modified
edge effect.

[2]    Reducing the viscosity of the coating fluid - use a suitable solvent.
This works well, but will obviuosly influence the final thickness / spin
speed relationship.

[3]    Increase the spin speed. Obviously this will also affect the spin
speed / thickness relationship. However you can also use a combination of
[2] and [3]. For example, we found with our particular coating, that
sp[inning at 5000 rpm resulted in a film thickness of 1.5 microns, and an
edge effect of 6 microns - unacceptable to us for our application. However.
spinning at 10,000 rpm, resulted in a film thickness of 1 micron, and an
edge effect of 1.1 microns in height (0.1 microns above the film thickness).
We made our edge effect measurements using a dektak stylus profiler.


I hope that this is of some value to you, please do not hesitate to contact
me if I can be of further assistance.

Regards,

Karl H. Cazzini (Ph.D)
Conifer group & Associates
Tel: (508) 620 8845
Fax: (508) 620 8839
-----Original Message-----
From: Patrick Leech 
To: MEMS@ISI.EDU 
Date: Monday, January 31, 2000 11:59 PM
Subject: Removal of edge bead


>
>Dear Colleagues,
>I would like to know of a simple and effective way of removing the edge
>bead after spinning of AZ4620 resist on a 3" silicon wafer. At present,
>after spinning at 4,000 rpm, a 2-3 microns lip of resist develops around
>the rim of the wafer.
>
>Thank You,
>
>(Dr.) Patrick Leech,
>CSIRO Division of Manufacturing Science and Technology,
>Private Bag 33, Clayton South MDC,
>Victoria, Australia, 3169.
>
>