Procedure: |
Preparation: Start
by making sure you have the machine reservations in CORAL,
and then “engage”. Make sure the removable bowl
is inserted and that all labware you intend to use is of the
SU8 purple-dot classification, including developer bowls and
tweezers. All wafers processed in photo are re-classified as
gold, TRL’s red-dot classification after leaving the
photo area. Your wafers should be clean and dry before starting,
often by Using a piranha type clean followed by an 10 min+
oven or hotplate dehydration. Do not use HMDS prior to coating.
Coating: If Using hot plates, level two
exhaust vented hot plates, and make sure they have smooth
piece of disposable aluminum foil on top of the heating
element. Choose a flat metal hotplate top for uniform conduction
heating: consider the difference in heat transfer between
the desired uniform conduction heating and indirect conduction
and convection heating of a wafer only partially in contact
with the hot plate. Set them at 65 and 95C. If you are
Using the SU8oven, make sure it is at the correct temperature,
and read the Microchem website for the correct bake times.
Never use any box oven other than the SU8oven, and never
use a hotplate or oven unless it has been foil protected.
We will present a recipe for SU8-2050 targeting ~ 100u, Using hot plate
heating, but other viscosity 2000 series resists are available, and other
thicknesses for 2050 are realizable.
Set the SU8spinner on the two stage “speed 1&2” setting;
speed 1 at 500 rpm for 10 seconds and speed 2 for 30 seconds at your
recipe specific speed; 1500 rpm in our case. Put your wafer on the appropriately
sized chuck and make sure it is well centered.
Note: your wafer MUST be bigger than the chuck in all directions, or
resist will be sucked into the spinner by the wafer vacuum.
Static dispense resist onto clean, dehydrated, wafers. Our experience
shows better results for pouring from the bottle from a very low position,
preventing thin filaments from dropping down over the wafer caUsing bubble
defects. When the resist pour covers just over a quarter of the area
of the wafer, twist the bottle and lift to terminate the pour without
strings and without getting resist on the threads. Don’t put the
cap on over resist covered threads as it will dry and create thousands
of particle defects, if you can ever get the cap off again. After moving
the wafer to the hot plate, and clean the tweezers with acetone or PM
Acetate, rinsed with 2-propanol, always allowing waste solvent to drain
into the funnel into the solvent waste bottle. Replace the solvent waste
cap when done; the funnel may not be left in place! Let the SU8 remain
and cure in the removable coater bowl; staff will periodically replace
this item. Do not let solvents go into the coater bowl, as they will
prevent the SU8 from curing, and lower its viscosity enough to allow
it to flow down towards the coater shaft.
Use a swab with a squirt of acetone on it to “edge bead” clean
the wafer, or wet resist will be deposited upon the chucks of downstream
equipment such as the hot plates and aligner.
Put the wafer on the 65C hotplate pre-bake for about 4 minutes in our
SU8 2050 targeting 100u thickness example, then move it to the 95C hotplate
without cooling or stopping, for the 20 minute softbake. At the end of
the softbake, after cooling, the wafer may not still be wet! If so, the
heating was likely non-uniform. If the wafer is still wet, the SU8 would
ruin your mask and possibly damage the aligner, so it may not proceed
further until at least a surface film hardening occurs.
Exposure: reserve and engage “EV1” under
CORAL, do not choose the shorter wavelength “KS2”.
See the EV1 SOP for specific machine instructions. Select
the 4” gold (red-dot) chuck; and center the removable
SU8 separation disk on the chuck. Expose, in our example,
for between 320 and 530mJ/cm2, which at 10mW/cm2, means
between 30s and a min, through your negative tone mask.
If Using a transparency in flood expose mode, don’t
forget to put a clean blank mask on top to make it conformal
to the wafer. Make sure the separation disk is cleaned
and returned to its proper storage location when done.
*Consider Using a UV340 filter to eliminate the undesirable shorter wavelengths,
which scatter and absorb readily in SU8, caUsing a “T-topping” effect
of exposing, and rendering insoluable, the top layer of resist. This
filter works very well with chrome masks, but when Using transparency
masks, the bubbles naturally present in the Clearfield part of the film
act like minute masking points, and you end up with small perforations
in your SU8 noticeable after developing. To limit this effect, overexpose
by several minutes to blur these defects as much as possible. In all
cases with a filter, overexpose by about 40% to allow for transmission
losses.
Post Exposure Bake: (PEB) Put on the 65
and 95C hot plates for 1 and 10min respectively in our
example; if you get tensile cracking, consider going back
to the 65C plate briefly to cool more slowly.
Develop: Develop in PM Acetate, PGMEA,
or SU8 developer, in purple dot Pyrex bowl. If fully developed,
rinse developer (displacing the solvent) off of the wafer
with 2-propanol (IPA). If a milky white color is apparent,
it is from partially dissolved SU8; your wafer may need
more developing. Dry the IPA with the N2 gun and inspect.
If complete, pour developer into solvent waste and rinse
the Pyrex bowl well with IPA, and return it to the purple
dot shelf.
Failure Modes: If resist dissolves in
areas where it should remain, either the PEB was too cold
(non-uniform) or the time was too short, or the exposure
time was too short. If resist doesn’t dissolve where
it should, in areas not near exposed features, the softbake
or PEB was too long or hot. If resist doesn’t dissolve
only in areas adjacent to exposed features, it was either
over exposed or the mask was in poor contact, or you may
need some type short wavelength filter. If resist
separates or lifts off from wafer, cool the wafer down
more slowly after PEB by turning the hot plate off and
leaving the wafer in place to cool slowly for 10 minutes.
|