Pquest and CF4

Michael Wiemer mwiemer at stanford.edu
Thu Sep 29 13:53:58 PDT 2005

Pquest users,

Over the last couple of weeks GaAs users - mostly Harris group people who
are by far the group with the most people using the tool - have discussed
the question of opening the tool for CF4 etching. The question was
basically, "Should CF4 etching of SiO2 and other materials widely used in
the lab be allowed in the Pquest and if so, under what circumstances".
Nickolai Belov proposed a test to investigate the effects of CF4 on GaAs
etching to help answer this question. In the end, little support was voiced
for the tests proposed by Nickolai Belov (test CF4 effect on GaAs etching),
or for Flourine etching in the Pquest in general. I will try to summarize
the main points which came out in the discussion in this email.

Currently, O2 with small amounts of CF4 is used to clean the chamber as well
as etch some polymers (BCB for example). No one really had a problem with
this as it is a necessary cleaning procedure, nor with the BCB etch, being
so similar to the clean. 

The issues, as well as I can summarize them here, follow below. While
reading, keep in mind that this is not a "etch anything" tool. It is the
only III-V etch tool in the lab and its integrity as a III-V etch tool must
be maintained. This can only be done with good policy (I fully recognize
that policy on this tool has not always been stellar). And any good policy
is simple, so here we go....

1.) CF4 polymerization

Instead of using an oxygen rich plasma like the CF4/O2 clean, the proposed
etching would involve little to *no* O2 and 100% CF4. With this kind of
etch, there is the concern that this etch will deposit/build up a
teflon-like polymer film in the chamber. 

2.) F affects the GaAs (and, more dramatically, the AlGaAs) etch rate.

This this is well documented. There is no desire among the users to re-prove
this. Conditioning the chamber after a F etch can return the chamber to good
Cl2/BCl3 GaAs etching condition. The question is what kind of conditioning
and how long should the conditioning be. However, there was little desire to
put the effort into researching this. Just like tests which may address
problem 1 above, the test to address this problem is time consuming and may
ultimately be inconclusive. And any test only tests 1 particular condition -
other CF4 users in the future would surely want to alter the process a
little and then we can be faced with the same problems all over again - this
leads to the next problem....

3.) What about other people and the future? 

One of the biggest problems which came out during the discussion is that
etching of SiO2 opens the door to more SiO2 etching. 90% of the lab users
are Silicon people and if the word gets out that the Pquest can be used for
SiO2 etching, there would probably be significantly more interest in the
future. Considering there are at least 6 SiO2 etch tools in the fab, why do
we have to make the Pquest another one? People do not want to open this tool
to SiO2 (Fluorine) etching and then deal with these issues above continually
as the next and then the next person decides that Pquest is their best-bet
tool for their Fluorine etch. If we make an exception here, or there, we
will CONSTANTLY have to deal with issues of process compatibility. The tool
must be maintained as a III-V etch tool. Why do these questions never come
up with the P5000? Answer: because there is a strong policy on that tool
about what gases/materials can be run in each chamber. 

We propose that CF4 etching is eliminated from the "list" of possible Pquest
etches. High O2-low CF4 flow etches will still be OK. And the week will stay
split: GaAs etching Wed-Sat, non-GaAs etching Sun-Tues. If you want to use
CF4 in the Pquest and are unhappy by this, perhaps the best way to
communicate is to get together (email is so difficult and time consuming).
Contact people: Rafael Aldaz (aldaz at stanford.edu) and Mike Wiemer
(mwiemer at stanford.edu). We are in contact with other GaAs users.

More information about the pquest mailing list