Daniel:

Your starting process looks complicated for an application that might be
better served with a simpler approach.  You don't mention the mask that
you are using but you might consider removing the BCl3 after a brief
breakthrough step and removing the CHF3 altogether for the case in which
no selectivity is required between the GaAs and the AlGaAs.  A
breakthrough step using Ar/Cl2 is commonly used to initiate the etch in
GaAs etch processes to remove surface contaminants, mask residues, etc,
but since you do not have argon, BCl3 might be a reasonable alternative.
I don't see how the CHF3 would help in this process other than to
potentially improve the mask selectivity but you probably won't have
much difficulty getting reasonable selectivity to either a hard mask or
PR.  You can expect very high etch rates for GaAs with chlorine.  You
might also consider increasing the Cl2 flow from 2sccm to >10sccm to
ensure that you are not starving the reaction on the wafer.  If you were
to try the Cl2-only process with a BCl3/Cl2 breakthrough step, and find
that additional issues remain with selectivity, roughness, profile
angle, etc, then you might consider additives to address these issues at
that point.  With your current process, the roughness that you see could
be due to low etch reactant supply, surface contaminants, polymer
deposition from the CHF3, or a combination of these effects.  Starting
with a simpler chemistry, Cl2, should help to simplify the process
development.

For the case in which you desire high selectivity between the GaAs and
AlGaAs, you might consider adding CF4 to Cl2, starting at about 10% to
determine how much of an increase in selectivity you can get.  CF4
contains the fluorine that you need without the complexity of the
polymeric deposition that you get with CHF3.  The achievable selectivity
will depend largely on bias power so you may need to keep the bias power
low.  If the bias power is too high, then you could sputter the
underlying AlGaAs even with large amounts of fluorine.  Lowering the
bias power should allow you to produce a more chemically reactive, and
less physical process.  Splitting the process into a low selectivity
main etch step followed by a reduced bias power, CF4-containing overetch
to stop on the AlGaAs might also help.

Regards,
Robert Ditizio

-----Original Message-----
From: mems-talk-bounces+rditizio=tegal.com@memsnet.org
[mailto:mems-talk-bounces+rditizio=tegal.com@memsnet.org] On Behalf Of
D.Grimm@ifw-dresden.de
Sent: Sunday, August 15, 2010 8:22 AM
To: General MEMS discussion
Subject: [mems-talk] RIE on III-V semiconductor superlattice

Hi guys,

I am starting to do RIE on III-V semiconductor superlattices.

However, most recent literature deals with high-selective etching
processes (GaAs <-> AlGaAs).

My aims are the following:

- etching depth around 1um
- high as well as no selectivity against AlAs
- high smoothness because I need to pos-process the sample.

We have in a parallel plate reactor the following gases available:
Cl2, BCl3, SF6, CHF3, CF4, He, N2 and O2.

Do you have any hints for recipes? I tried on a GaAs/InGaAs/AlAs/GaAs
structure 2sccm Cl2/8BCl3/12CHF3 (or He) but the results are not so
encouraging: Very rough etching borders as well as etching bottom
(Ra~30nm).

Greetings
Daniel