IC Wafer Thinning

1
IC Wafer Thinning

• Desired thickness goal for front-end ICs on
  modules is 150m
• Prototype results obtained so far have used
  mechanical thinning(grinding) of bumped wafers.
  The typical process is
• Bumps are protected by a layer of
  photoresist(also useful during dicing)
• Additional protection is provided by grinding
  tape attached to front of wafer, then
  released(UV) after grinding is complete
• Absolute thickness can be controlled to about 10
  microns, uniformity across wafer is better than
  this.
• Dicing of thinned wafers has not been problem.
• AMS
• Dummy(4) wafers(4) successfully thinned(165m)
  and dummy modules(5) made.
• Active(6) IC wafers(2) thinned(156m and 171m)
  and active 16-chip modules made and tested
  successfully in beam.
• Thickness of 150-175m appears achievable.
• Yield analysis in progress(bumps damaged during
  thinning and thin die lost during handling after
  dicing).

2
IC Wafer Thinning

• IZM
• Dummy wafers(6) have been thinned successfully
  to 200-210m. Dummy module built, X-rayed and
  looks OK.
• No active wafers thinned yet.
• Greater thickness and non-uniformity(edge bead)
  of photoresist coating limited thickness to about
  200m. Breakage, edge chipping below this gt use
  thinner photoresist, and this program has been
  launched.
• Also greater interaction with tape and
  photoresist - see photos. No bump damage but a
  concern. Seen also in AMS, but not as much.
  Reduced by tuning UV exposure, using different
  tapes but need more experience.
• Thickness of 200m demonstrated, plan in place to
  move towards 150m in next few months.

Before photoresist removal
After photoresist removal
3
IC Wafer Thinning

• Sofradir
• Dummy wafers(4) thinned to 150m and 200m.
• Dummy modules made successfully.
• No active wafers thinned yet.
• Vendor experience
• Thinning of wafers bumped by AMS and IZM done, so
  far, by Okamoto Corporation in the U.S. This used
  robotic equipment(except for removal of grinding
  tape and packaging) and is inexpensive(about 20
  chf/wafer(6) in large quantities). Very large
  capacity
• Thinning of wafers bumped by Sofradir done by
  MICROPOLISH in France. Cost so far much
  higher(order of magnitude or more) than Okamoto.
• Investigating other vendors.
• Have done one wafer using Atmospheric Discharge
  Plasma technique that is a non-contact method
  aimed specifically at thin bumped
  wafers(SmartCard market). Will keep eye on this
  technique.