Thanks a lot, Roger. I did try pure SF6. And it didn't work. Since I tried a
lot of different recipes and failed, I'm afraid that SIMOX tech would
generate some SiO2 even in the device Si layer and this would create an
issue for my RIE to the Si. Maybe it's a naive idea. Anybody had the
experience of etching through the device layer and stopping at the oxide
layer using the SIMOX SOI?

If it's not the issue of the SOI, I will focus on my recipe again.

Thanks.

Jay

On Mon, Jun 9, 2008 at 2:31 PM, Roger Shile  wrote:

> If you use pure SF6 (no O2) you should get an isotropic profile.
>
> Roger Shile
>
> -----Original Message-----
> From: [email protected]
> [mailto:[email protected]] On Behalf Of Jie Zou
> Sent: Sunday, June 08, 2008 9:45 PM
> To: General MEMS discussion
> Subject: [mems-talk] RIE on SOI
>
> Hi,
>
> To make a movable structure, I need to etch through the device layer. I
> tested the ICP-RIE recipe on a Si wafer and got an isotropic profile
> (1um/min). But when I applied this recipe to the SIMOX SOI wafer (200nm
> device layer and 350nm oxide layer), I only get an anisotropic profile
> and
> can't make any undercutting. Is it related to the SOI wafer? Would a
> bonded
> wafer be better? My supervisor prefer a smart-cut type SOI wafer, but I
> can't find any vendor in US selling that.
>
> I have tuned my recipe many times and didn't get improvement. I
> increased
> the source (ICP) power and lower the bias power.
>
> My recipe is 48 sccm SF6, 32 sccm O2, source power (ICP): 160W, bias
> power:
> 360W, Pressure: 80mTorr. It gives an anisotropic profile.