update on etching rate

Luigi Scaccabarozzi scaccag at stanford.edu
Wed Aug 21 11:03:39 PDT 2002


Hi all,
Len asked me to report about my recent experiments with the pquest.
I found out, as first suggested by Jim (McVittie), that, at same etching
conditions (=same recipe) the etch rate depends essentially on 2
parameters:
1. impedance of plasma-sample-carrier wafer, which we
can assume constant in time, but in general not uniform along the sample.
2. Temperature of the substrate, which in general is not constant in
time

I always used the recipe GIGI2 (Ar:BCl3:Cl2=15:10:1.5, ECR=200, RF=75,
p=2mTorr, heater temp=20C, cooling water temp=0C), which gives nice,
straight walls with baked 1813 phptoresist.

about the 1st point:
I noticed that huge nonouniformities in etch rate are present when the
sample (even small) has not uniform contact with the carrier, for example
when using copper tape only on part of the sample. That's because,
according to Jim, ions go mainly toward the low impedance regions (the
ones with copper).
Also, if Copper is everywhere, but the contact between carrier and sample
is uneven, we have the same kind of non-uniformities.

So, I decided to go back to the old rail carrier: if the sample back
surface is clean and we don't put anything between sample and carrier the
etch depth is uniform (~ +-50A) along the whole sample (I checked with 2"
GaAs wafer).
Of course this method has some disadvantages (point 2):
The contact sample-carrier is uniform, but thermal conductivity is poor,
so the temperature of the sample increases during the etch.
This leads in the first minutes (until stabilisation of the temperature)
to INCREASING etch rate.
But, when calibrated the etch time, the results are totally repeatable
(since the contact is always the same), as I have verified.

Also, I could estimate that the "averaged" etch rate without copper, at
~20C, for bulk GaAs and my recipe is around 1600-1900 A/min (2-3 min
etching) corresponds more or less to etch with copper at 40-60C.
So I guess that the temperature of the sample roughly increases to 40-60C
during etching WITHOUT copper.

Finally, if we set the temperature of the heater and cooler, so that the
temperature of the chuck keeps constant around 40C, the temperature
does not increase further during etching (at least for my recipe), and
if also using copper, and leaving inside the chamber to stabilize
temperature, I guess the etch rate should be constant during etching, but
I haven't tried this out.

I hope this info can be useful, and please give me feedback if you find
same or different behaviors.
Thanks
Gigi





More information about the pquest mailing list