Nate:

Your success with Ar in place of the O2 is most likely a result of
sputtered material from the etched surface acting as a passivant on the
sidewall.  I suspect that you will get equal or better sidewall results
with the O2 provided that you increase the O2/SF6 ratio.  The sputter
yield for oxygen is lower than that for argon.  You will need to
increase the O2/SF6 ratio to the extent that the passivation rate is
sufficient to produce an angled, non-undercut profile at the
corresponding silicon or silicon nitride etch rate.  This is what you
hope to observe. If your passivation rate is too low for the
corresponding silicon or silicon nitride etch rate, you'll continue to
see the undercut profile.  When you find the correct balance of sidewall
passivation and vertical silicon or silicon nitride etch rate, you
should find that the undercut is eliminated.

20-30mT is a reasonable range for this process.  (This seems low for a
parallel plate RIE system.)

With regard to the temperature, your wafer piece could be heating quite
a bit, to 100C or more, during the process.  The higher temperatures
could greatly confound your results, particularly if the temperature
varies with each set of conditions that you choose to investigate.  If
the chuck is flat, and you are not using backside helium, you might
consider a thin layer of vacuum grease or fomblin oil between the back
of the sample and the chuck to improve the thermal contact.  This may
not sound pleasant but it does work.  Either can be cleaned up with IPA.
In terms of control and hardware reliability, you might consider using
the fomblin or grease to attach the wafer piece to a dummy wafer.
Ideally, the dummy wafer will have a film that will not etch in
fluorine.  A silicon wafer with an aluminum film, for example, is a good
candidate.  Without the aluminum, the silicon will etch and consume much
of your fluorine.  While the peak temperature of the wafer piece will be
higher with the use of the dummy wafer relative to direct placement on
the chuck, you can limit the direct exposure of your chuck to the
plasma.  You can also isolate the grease or oil to wafers that will be
removed from the chamber.  You should also get better clamping and be
able to run backside helium cooling with the use of dummy wafers.

With regard to your question about whether to explore this chemistry
further or to look at other chemistries, you should be able to determine
if the O2/SF6 approach will be fruitful by running a few tests to see if
you can obtain a non-undercut, positively angled profile with the
oxygen-rich chemistry.  If you can produce a profile that has a straight
sidewall profile, then you should be able to then reduce the O2/SF6
ratio to find a mix of O2 and SF6 that produces a near-vertical profile.
If all of your results at the higher O2/SF6 ratios are undercut, then
you might consider alternatives.  You have at least a reasonable chance
with this chemistry so long as you don't mind the lower etch rates that
you'll get with the increased O2.  You may need to get the temperature
under control but this is likely to be the case for any chemistry that
you choose.  If you cannot get the positively angled, non-undercut
profile with O2 and SF6, you might consider adding in argon as a next
step.  The argon should provide an additional source of passivants and
help to slow the etch rate to produce the needed balance described
above.  Small amounts of argon (~10%) should have an influence on the
process.

You might be asking yourself at this point why the Ar/SF6 is not the
better choice over the O2/SF6, and the answer to this question is that
it may well be.  The ultimate choice between the O2 and the Ar is most
likely to be determined by the sputtering rate of the mask and the
appearance of "grass" in your etched layer.  You should expect a lower
selectivity to the mask with the use of Ar over oxygen and if the
sputter rate of the metal mask is too high, you might see micromasking
in the silicon or silicon nitride layers that leads to a grassy surface
texture.

Regards,
Robert Ditizio
Director of R&D
Tegal Corporation

-----Original Message-----
From: mems-talk-bounces+rditizio=tegal.com@memsnet.org
[mailto:mems-talk-bounces+rditizio=tegal.com@memsnet.org] On Behalf Of
Nate Lawrence
Sent: Tuesday, August 24, 2010 10:59 AM
To: General MEMS discussion
Subject: Re: [mems-talk] Anisotropic RIE Etching of Si and SiN

Thanks for the response Robert, its very thorough. The current problem
with
my sidewalls is that they are generally bowed in in the middle and I
would
like them to be straighter. I also get undercutting when I'm etching
with an
anisotropy of about .9, which isn't terrible but I would like to
improve.
The most successful chemistry so far actually being SF6/Ar.

I have tried a variety of pressures and powers, the most successful of
which
have been low pressures and medium high powers so I generally stick to
20-30mTorr and 200W. I usually cannot get the pressure below 20mT when
the
gases are flowing. All the etching I have done has been at 25C, I have
the
ability to change the temperature some but I don't think I can cool it
significantly below that.

I am processing small pieces of wafer, I has tried fixing them to the
chuck,
which I believe is not electrostatic, with tape to prevent them from
moving
in the chamber but have abandoned this as they seem to stay put as long
as
the chuck isn't wiped with IPA right before the chamber is pumped. I
have
thought about using paste fix them and increase the thermal contact
between
the chuck and sample but since the chuck isn't cooled that much I didn't
think this would make that big a difference.

All the recipes I have tried with SF6/O2 have been SF6 rich rather than
O2
rich, I will have to give a more O2 rich recipe a try, thanks!

When reading about using SF6 as an etchant I have seen mixed reports,
some
claiming that it is only suited for isotropic etching and others
claiming
that with the addition of O2 it can etch anisotropically, though from
the
responses I've gotten here it seems to be a common chemistry for
anisotropic
etching. Do you think exploring this combo further would be better than
looking other chemistries to improve the etching?

Thanks Again,

Nate