In terms of problem solving I would review the following factors.

1. The accuracy of your flows and whether you have accidentally gone
into the polymerization regime. The SiO2 etch works by being close to
the polymerization mode, but not in it, so there is carbon on the oxide
surface, and RIE bombardment. If you flows are inaccurate this could be
a problem, you might be in a polymerization mode if not enough O2, and
slow etch if too much O2 which is removing the carbon. Also, if your gas
flows don't come on at the same time you might initially polymerize.
Once a wafer is polymerized form the plasma, it is hopeless to save it.

You should routinely take a rate of rise of the pressure for a given gas
flow to get an independent check of the MFCs.

2. The fact that you can wet etch it quickly rules out a sub-oxide, but
it might be that your film is SiOxHy and doesn't need as reducing plasma
and it might be the cause, but this is far fetched.

3. If you have another etcher, give it a try. This would pin-point the
first etcher you tried as a problem. It is a partitioning step.

4. Do the usual equipment verifications of power, gap, flow, etc.

5. Can the etcher etch any other films? This will focus on the specific
items of the SiO2 etch that are different from the etches that work.

6. What does the monitoring of your wafer etch establish. Are the
emission signals normal, does the plasma strick, etc.

7. What can be see on the wafers after etch.

8. Loading: Is the recipe set up for a certain percentage of resist
coverage and you are procesing a wafer with another percentage of resist
coverage? For example is the recipe is normally used for a wafer with
only a small percentage of resist coverage and you are using it for
contact or vias or some layer with only a small percentage open area,
you might be shifting the plasma chemistry. The resist as it is eroded
and aslo as it desolvates in the hot plasma chamber contributes to the
chemistry. Perhaps you are using a new resist or underbaked one.

Hope this helps.

-----Original Message-----
From: mems-talk-bounces@memsnet.org
[mailto:mems-talk-bounces@memsnet.org] On Behalf Of Chord Chen
Sent: Tuesday, September 18, 2007 2:25 PM
To: mems-talk@memsnet.org
Subject: Re: [mems-talk] Silicon oxide depositon and etch problems

We routinely use CHF3/O2 RIE for oxide etch, with a gas flow of 50 sccm
CHF3 and 2 sccm O2. The etch rate is about 30 nm /min but is decreasing
if you do a long etch and use resist mask due to polymerization. The
rate is also dependent on the deposition condition of TEOS.

Long