Report from Florence

1
Report from Florence

• Bonding of Modules L12p

Mirko Brianzi 20 April 2004
2
Bonding Setup in Florence
There is a new Pull Tester in our system, now we
are finally able to make good pull test.

• Dage 3000

Delvotec 6400
3
Status

• Since March bonded 22 new modules (6 chips).
• For this L12p modules the loop height is very
  important, so it took a lot of time to set all
  the loop parameters and solve some problem.
• Now a new program for Delvotec is ready and it
  automatically bonds all the front side of the
  module, making all the loop shapes more
  reproducible (as much as possible).
• The failure rate for these modules has been
  3fail/module because they are the first ones and
  we had to fix some problem, actually the last
  ones hadnt any fail.

4
PA moving

• The main reason of failure is the PA not glued at
  the edge and this make often fail someone of the
  last bonds.

Solution
Making some pull test on bad bonds we can say
that anyway too weak bonds are well visible.
Insert of teflon
5
Gold pad moving

• Similar problem bonding Bias to Filter Pad
  because the bond position is outside the Jig.

bonds
Lateral View
silicon
carrier
support
end of jig
6
Dirt

• Sometimes bonding pads on PA are dirty and this
  has caused some fail.

7
PA gap
Very difficult to reproduce exactly the same loop
height because the mechanical assembly of the
module influence the bond loop.

• For example
• PA-Si gap can vary between 100µm and 220µm.

100
302 000 200 35 277
220
302 000 200 35 278
8
Vacuum

• Because the mechanical difference between each
  module, sometimes the vacuum flex the silicon
  during bonding

so the central bonds can result 50µm higher (or
more).
9
Readout bonds
340
240

• Loop heights are as much as possible (within the
  specifications).

Specifications min. 200µm max. 350µm?
10
Bias bonds

• Bonds in this module are too high but they have
  been corrected to 250µm.

370
Specifications min. 200µm max. 350µm?
11
Filter bonds

• Its very difficult to stay inside the
  specifications.

525
420
In case that some bond is too low, we could
accept it if its above a safe limit (350µm?).
3kV/mm (dry air) 500Vgt170µm
Specifications min. 500µm max. 600µm
12
Pull Test on PA

• Results of Pull Test on test bonds on PA
• for 8 modules

Mean force 9.6g Standard deviations 0.22 -
0.37g Abs. minimum value 8.4g
13
Pull Test on Readout bonds

• Results of Pull Test on Readout bonds
• for 5 modules

Mean force 11.1g Standard deviations 0.69
0.99g Abs. minimum value 9.0g
( Correction Factor 1.4 )
14
Pull test correction

• Measured geometry of bonds used for pull test to
  define the correction factor.

Angles can vary depending on the exact point
where the wire is pulled 25-27 for PA test
bonds gt factor 1 16-20 for readout bonds gt
factor 1.4
Factors 161.81 181.62 201.46 251.18 271
.10 301.00
These factors seems very high
15
Pinholes after bonding

• Found after bonding
• 2 pinholes not present
• in the Database.
• 1 of them is probably caused by a fail during
  bonding.

The correspondent bond have been removed in
this case do we have to update the Bonding
Database?
16
Conclusion

• New bonding program for Delvotec include also the
  automatic pattern recognition and it is fast and
  efficient (using some trick to avoid failure on
  PA).
• No problems with Data Base, it has been updated
  to the last version.
• Good and uniform pull test results.