Since my first XeF2 post last month, I have had many inquiries about this
silicon etchant. Many of the same questions have been asked several times, so
it seems appropriate that I should summarize here.

ksjp

** How did you find XeF2 in the first place?
While discussing TMAH+silicic acid as a possible CMOS post-process etchant, a
colleague of mine, Eli Yablonovitch, suggested that XeF2 might be just the
thing we were looking for. We spent several days calling chemical supply and
excimer laser companies, all of whom denied that XeF2 could exist, and we
ultimately gave up. Several weeks later, in a discussion on polysilicon
stringer etching, Mike Hecht of JPL mentioned that he had used XeF2 to etch 300
microns of silicon and stop on 50 Angstroms of SiO2, and that XeF2 could be
purchased from PCR (see below). Mike reassembled his old reactor at JPL, and we
etched the first CMOS chip together at UCLA.

** it sublimates [...] is the gas fed into an evacuated chamber?
Yes, we have a small glass bell jar on an aluminum plate with ports for N2, a
vacuum pump, manometer, and the XeF2 line. For reasons that are as yet unclear,
we get the best performance when we pulse etch (pump/etch/pump) rather than
allow a constant flow of gas into the chamber.

** What does it etch?  Not etch?

Etches silicon and polysilicon at up to 10 microns per minute. Does not etch
    (at all, as far as we can tell):
    photoresist
    thermal oxide, PSG, BPSG
    aluminum
    gold
    NiTi (supplied courtesy TiNi Alloy Co.)
    Si3N4 (haven't tested low stress yet)
    acrylic

** It sounds like this is a 'dry' etch
Yes, and an extremely gentle one. We have undercut some structures that are so
fragile they can be destroyed by blowing on the chip, but which survive the
etch. No liquid or adhesion forces.

** Are there any tricks?
Dehydrate your sample (e.g. 140C for 10 minutes) before etching, otherwise you
may have problems with the formation of a polymer-like film on the silicon
surface. Neil Bartlett (discoverer of the first noble gas compound in 1962)
thinks that this film is probably the silicon equivalent of PTFE.

** How much anisotropy, if any, can you achieve?
Close to none. The etch is nearly isotropic.

** How smooth are the sidewalls and bottom surfaces
Very rough. All surfaces show something like 1-2 microns roughness.

** Is it safe?
On contact with moisture, it turns to xenon and HF, but at the room temperature
sublimation rate, this produces a hazard comparable to an open beaker of HF -
caution is necessary.

** Where can I buy it and how much does it cost?
PCR Inc
P.O. Box 1466
Gainsville, FL 32602
(904) 376-8246
371-6246 (fax)

The cost is $45/gm, which may seem expensive, but the etch is extremely
efficient, and one gram of XeF2 can last for many etches. Also, the cost goes
down rapidly with larger orders to a few dollars per gram.

** What are some of the references?
M.H. Hecht, et al., "A novel X-ray photoelectron spectroscopy study of the Al/
SiO2 interface", J. Appl. Phys, 57(12), 1985, pp5256-5261.
E. Hoffman, et al., "3D structures with piezoresistive sensors in standard
CMOS", MEMS 95, pp.288-293.
F.A. Houle, "Dynamics of SiF4 desorption during etching of silicon by XeF2",
J.Chem Phys. 87(3) pp1866-1872.
D.E. Ibbotson, et al., "Comparison of XeF2 and F-atom reactions with Si and
SiO2", Appl. Phys. Lett. 44(12), 1984, pp.1129-1131.
H.F. Winters, J.W. Coburn, "The etching of silicon with XeF2 vapor", Appl.Phys
Lett. 34(1), 1979, pp.70-73.

** Can you send me some papers on this?
We can send you our MEMS 95 paper, which doesn't give much detail about XeF2,
but does describe our etch setup and procedure. Send me email if you want a
copy.