Hi,
We have a very odd and interesting adhesion problem with a (250A TiW, 10
wt.% Ti) / Ag 6500A) metal contact pad sputter deposited / patterned onto an
SiO2 coated substrate (1um oxide on Si). We've run this process for over a
decade with very few problems.
However, recently we've seen a problem. After the metallization steps
(sputter/resist pattern/etch) the metal appears ok (thickness and
resistivity are all in control). The defect only appears on the TiW/Ag pads
after a down stream surface modifying NF3 plasma (we are making the a
polyimide layer ultra hydrophobic). The defect presents itself as a Blister
under the metal (blister is ~10um in diameter and 1um high). SEM & FIBs
pictures show that the Ag is still attached to the TiW...but the TiW is
lifting. Auger shows that there is no contamination from the Ag down to the
Si02.
Some other odd observations to note:
*       We have 3 contact pads per chip. Sometimes the defect on affects one
pad, whil eth eothers are perfectly clean. Somtimes I'll see 10 chips in a
row where this is the case...always the same pad affected - then of
course... all pads or no pads will be affected for a while.
*       Not all lots in the line are affected. Not all wafers are affected
in a lot.
*       The affected product has a sister product that is identical except
for the fact that it has a TiW Au layer patterned elsewhere on the wafer
first. The product with the TiWAu does not exhibit the defect at all...mmmmm
*       Sometimes the lblistering area actually gets a small bubble on top
of it...no auger or xps results back from the lab yet to id "tiny bubble".
Any ideas on what could be causing it?
One theory is that the Fluorine is penetrating the silver (very porous
looking layer) and attacking the underlying TiW.
Another, there is a residue that affects the adhesion of TiW to the SiO2.
Once the wafer goes into the fluoring plasma, the vaccum causes the TiW to
lift of the SiO2.
Other ideas are VERY welcome.

Gerry Overton
Wafer Fab Eng Manager,
i-STAT Canada