mcvittie at snf.stanford.edu
Mon Dec 5 10:45:11 PST 2005
See my replies to your replies below.
> Is there an alternative material that the top rign can be made from?
> Al2O3? Quartz? Any other suggestions? What are the STS fingers
> made from?
Quartz would be great but hard to fabrication and easily broken. Al2O3 would be
better than Al since it has a much lower sputter rate (7 x less), but Al2O3 may
still be a problem. Al2O3 is used for the clamp in the Lam and is why the SF6/O2
process can;t be used for deep Si etching in that tool. The Al2O3 clamp caused
grass problems in the Lam with F based etch chemistries. It is not a problem with
Cl and Br based chemistries. An engineer at Lam told me, we would have to go to a
quartz clamp to get around the grass problem. I believe the STS clamp fingers are
> > 1. Some years ago when we did contamination tests for all the etchers, we
> > ran wafers with and with the holder in STS1. We found a large Al
> > contamination level on wafers etched with the holder.
> How bad was the contamination? Was it possible to remove the Al?
TXRF with using a W source has a threshold/noise level for Al, which is around
Process Al(no clean) Al(with clean)
(10E10 at/cm2) (10E10 at/cm2)
STS(no holder) 13000+/-4000 < 7000
STS(w holder) 16000+/-4000 10000 +/- 4000
The Lam also showed lots of Al but it went below the noise level after cleaning.
> > 4. Sputter rates increase linearly with ion flux. Therefore, one expects 3X
> > the Al sputter rate for the present holder design in STS2. Thus, grass is
> > likely be a much worth problem with the holder in STS2.
> Given that the silicon etch rate should also scale with ion flux
> to first order, it is unclear that the grass problem would be
> worse for a similar etch in the new etcher than in the old one.
> It would seem to scale with etch depth in both cases.
Yes, the grass problem may scale with rate, so grass with the holder may not be
> The use of the magnet in the new etcher to turn-off the ion
> flux for part of the cycle may reduce any grass caused by ion bombardment
> of the aluminum ring.
Yes, after the magnet turns on the Al sputter rate is likely reduced. The resist
etch rate goes down by 2 X when the magnet is used.
> Granted there could well be non-linear effects and it may be
> very sensitive to the chamber environment, so it is unclear to
> me in which etcher the grass problem would be worse in.
I agree with you.
> 5. The holder increases the thermal resistance between the wafer and the
> > chuck. With 3X the power heating is more of a problem in STS2.
> > 6. I believe Elmer did try running the STS1 holder in STS2 while the STS2
> > startup engineer was here last year. My understanding in that the wafer
> > overheated.
> I agree that thermal issues will be more important in the STS2.
> A redesign of the holder with lower thermal impedance as
> a goal might solve this.
> Bottom line some sort of holder is very important to the full
> utilization of both STS etchers and the current one is far from ideal.
> Can we come up with a better one?
I agree there is a big need and it is worth looking into finding a solution. An
alternative approach is to use a carrier wafer with a better thermal bonding
material than resist. A number of groups are using a silicone free heat sink
material from Techsray (1978). One problem with this material is that it is
loaded with ZnO. If it is exposed to the plasma, it will very likely contaminate
the chamber with Zn. My understanding is that AMAT uses it for pieces in tools
where contamination is not an issue. It would be useful to do an experiment with
this material to see if it solves the problem. If it works bonding wafers, we
could focus on finding a cleaner alternative.
More information about the specmat