I took my "Physical Vapor Deposition" book, published by Airco Temescal
in 1976. It is the classic work on the subject and useful.
Unfortunately, I don't know of any contemporary replacement book. If
your university library doesn't have it see if www.abebooks.com has it
for sale used and get a copy. It really should be republished.

However, as applied to this problem I did some reading. I list some
various points.

1. The deposition of Au is different than Al for a variety of reasons
which I won't go into here. They are related to complex issues of the
deposited atom on the substrate at the moment of deposition. So success
with Au doesn't guarentte success with Al.

2. First you may have a problem with a vicous flow region above your
target that is excessive. Solid angle of deposition of materials is
determined by the radius of the vicous flow region and not the spot size
of the e-beam on the wafer substrate. A too large vicous region would be
caused by either a beam current/voltage that is too high. Or failure in
the cooling of the Al target in the crucible. The vicous flow region is
where the vapor pressure above the target is high enough that there
isn't molecular flow.

3. Ionization: The e-beam going through the vapor cloud of Al atoms
above the spot size can ionize the vapor. These can sputter at the wafer
and also they have scattering occuring on their path to the substrate
and don't come in at the designed solid angle of deposition. Again it is
caused by a high current/power e-beam interacting with the e

4. Substrate temperature: I am presuming that your substrate is not
heated, but the heat of condensation of the deposited material can warm
up your substrate. Heating up the substrate besides improving step
coverage could also cause outgassing of the resist and provide atoms in
the path of deposition causing scattering. A lower deposition rate would
reduce substrate heating.

5. Outgassing: Outgassing could cause scattering. You might want to have
some additional time for degassing of the wafers in the vacuum before
initiating deposition. Just because you have reached base pressure
doesn't mean that there isn't degassing of materials going on. Also, a
bake prior to deposition could be helpful. Not so hot a bake that you
can't dissolve the resist later. The rotating planetaries and the
surfaces if they have very thick coatings can provide spaces in blisters
and such for outgassing also. The chamber walls need to be clean as well
as the planetaries.

6.  Your e-beam evaporation spot if made smaller would would afford
poorer step coverage. Again this is related to having a smaller solid
angle of incoming deposited atoms to the substate. This can be effected
by focus of the beam and the energy of the beam which makes the puddle
larger.

7. I would check the pressure during deposition. Your pressure given
sounds like the pressure the system reached before deposition started,
but seems very low for a pressure when deposition starts.

8. Make sure that the e-vap puddle isn't boiling. This would be caused
by too high an energy beam. However, I think you would see spitting of
metal.

9. Finally, for your two bi-layer features make sure there is a negative
re-entrant angle for the resist for your small features as well as your
large features.

10. I don't think you are overbaking your resist since it is lifting in
the large areas. This is probably due to that for the large feature a
defect in the step coverage is found and the solvent can penetrate.

Edward H. Sebesta
Independent MEMS and Semiconductor Engineer.



-----Original Message-----
From: [email protected]
[mailto:[email protected]] On Behalf Of James Paul Grant
Sent: Friday, January 16, 2009 4:04 AM
To: General MEMS discussion
Subject: Re: [mems-talk] Lifting-Off 1um metal


Thank you to all who have replied so far to my questions

I shall state a few other relevant points:

1. The Al is e-beam evaporated at a base pressure of ~ 1x10-7 mbar.
Sputtering deposition tools are also available.
2. Deposition rate is 0.3 nm/s
3. The bi-layer resist thickness is 1.3 um while the metal thickness is
1um. Now I know one should always have a resist thickness at least two
time greater than the metal thickness however I can add the caveat that
one of the groups here at Glasgow Uni have successfully lifted off 1.2
microns Au using the same resist thickness (1.3 microns) - and they say
their process is very repeatable! The problem my colleague has is he is
already using the thickest e-beam resist he has available. He could
probably sneak another 200 nm from spinning the resist at lower rpm
which may of course aid lift-off.
4. I told him to try sonication which he has done for around 10 minutes.

No joy.
5. All processing was completed within 2 days (i.e. substrate clean,
resist spinning, bake, exposure, development, ashing, HF dip, Al
evaporation)
6. He has not attempted to deposit thin Al

I had a look at his sample for him and the metal in large areas has
lifted off however in the fine feature areas (~1 um feature size) the
metal has not even begun to lift-off.

His next step will be to deposit 500 nm instead of 1um. I hope lift-off
is more successful this time.

Thanks for all your helpful comments,

James