Hi James

You may want to try mounting your device using the method described in
the following paper.

3-D Integration of 10-GHz Filter and CMOS Receiver Front-End
Choi, T. Sharifi, H. Sigmarsson, H.H. Chappell, W.J. Mohammadi, S.
Katehi, L.P.B.
IEEE TRANSACTIONS ON MICROWAVE THEORY AND TECHNIQUES
2007, VOL 55; NUMB 11, pages 2298-2305

You will then be able to use low spin speeds and pattern thicker
resist. The AOE may also be an option if cost is not an issue.

Best
Leo

On Wed, Feb 4, 2009 at 5:33 AM, James Paul Grant  wrote:
> Thanks to all who have contriubuted to this topic.
>
> Yes I'd love to use a thicker resist such as AZ9260 however I do not have
> this available to me. The thickest resist we have is AZ4562. Normally if I
> spin AZ4562 on a 3 inch wafer at low spin speeds (e.g. 1500 rpm) I can
> obtain a resist thickness of ~ 10 microns. I know I create other problems
> for myself with such a low spin speed (non-uniformity, more edge bead etc.)
> however these are not show stopping problems. Unfortunately my substrate is
> a CMOS chip of 5 mm by 5 mm and if I try to spin AZ4562 at 1500 rpm the
> resist essentially does not spin and I got a very non-uniform coating. The
> resist does spin nicely at spin speeds greater than 4000 rpm. So essentially
> I have 6.2 microns of photoresist to play with and have to etch 10 microns
> SiO2. Up until now I cannot get the selectivity required for my process
> (~2:1). I can achieve 1.5:1 however the etch rate is incredibly slow (~20
> nm/min). Note I only have RIE processes available to me for etching SiO2
> (there is also one ICP process for deep silicon etching).
>
> I contacted Surface Technology Systems (STS) for details about their
> Advanced Oxide Etch (AOE). This is an ICP based process and from the paper
> released by STS using this process selectivities of up to 12:1 can be
> achieved using photoresist. The gas chemistry is C4F8/H2 by the way. I was
> also told if a metal mask is used selectivities of up to 200:1 are possible.
> I will try and convince my line manager we should try this AOE however it is
> all dependent on cost!
>
> Thanks to Morten as well for sharing his process.
>
> James

Xiaoguang "Leo" Liu
Birck Nanotechnology Center,
Purdue University,
1205 W.State Street, West Lafayette, IN, 47906 USA
liu79@purdue.edu