From mcvittie at snf.stanford.edu  Thu Mar 28 12:06:23 2002
From: mcvittie at snf.stanford.edu (Jim McVittie)
Date: Thu, 28 Mar 2002 12:06:23 -0800
Subject: Arcing Problem
Message-ID: <3CA377BF.7F257BC0@snf.stanford.edu>

P5000 Users,

This week we had to replace some expensive parts in the oxide etch
chamber because they
were damaged from arcing. After discussing the problem in AMAT, we
concluded that the
problem was most likely from  running the chamber at too high of a bias
voltage. It was
estimated that the bias voltage may have been as high as 1600V. As you
know, the one does
not directly set the bias voltage in a process recipe. Instead, the bias
is developed from the RF
current with a resulting bias value which depends on the pressure,
power, B field and gas mixture. With a typical power setting of 500 to
600W, the most important parameter determining bias is the pressure. To
avoid arcing in the future, Applied is recommending that we prohibit any
process with pressures below 60 mT. The exception is the breakthrough
step in the Al etch process which uses 30 mT at 300W for 15s to go
through the Al2O3 layer.

We will continue to let users modify the standard etch process on the
P5000, but I need approve all proposed process modifications before
hand.

For your interest the process that lead to the arcing problem was:
O2=20, He=20, Pressure=20mT, Power=600W, B=40G, Time=1000s

    Jim McVittie



From mcvittie at snf.stanford.edu  Thu Mar 28 14:29:36 2002
From: mcvittie at snf.stanford.edu (Jim McVittie)
Date: Thu, 28 Mar 2002 14:29:36 -0800
Subject: P5000 Logbook
References: <3CA377BF.7F257BC0@snf.stanford.edu>
Message-ID: <3CA39950.222DBE28@snf.stanford.edu>

P5000Users,

In trying to understand the recent arcing problem, I found that some users
are not fill out the log books. When problems occur, it is extremely
important that we have a good log of the recent
history of the chamber. In particular, I want to know # of wafers run, total
rf on-time, any
process changes for a standard process, bias voltages and anything unusual
that occurred during the run. If we continue to have a problem with getting
the logs fill out, some delinquent users may be removed from the P5000
qualified user list for a week or two. I expect, we can get compliance
without having to go that far.


By the way, there is still some debate going on about whether the 20 mT/600W
process was the cause for the arcing damage.

    Thanks,   Jim McVittie