17.5um thin Cu wire bonding for fragile low-K wafer technology

This paper describes the challenges of a 17.5um thin bare Cu wire bonding on aluminum bond pads for a fragile low-k wafer technology, on a BGA package. Previous evaluations have so far focused on 20um and 25um bare Cu wires as a suitable low cost replacement for Au wires. To improve performance, more fragile low-k wafer technology is being developed. In the past, some key technical challenges experienced are the substantial aluminum splash and potential cracks underneath the bond pads. This paper will focus on using a thin Cu wire of 17.5um diameter to address the smaller BPO of 44um, with very minimal aluminum splash, and prevent any cracks on a fragile low-k construction. As copper is a harder material than aluminum, to make a good bonding interconnect between the two metals and yet addressing these concerns, is critical. Also, the bonded ball bond should have a sufficient amount of remnant aluminum to survive reliability test. As Cu oxidation is rampant, an inert environment of a forming gas of 95N₂5H₂ is needed to contain this aspect in order to achieve a good wire bond. To address the small BPO and with the aluminum splash which will pose the likelihood of shorting between adjacent ball bonds, a small bonded ball bond size is important. Process parameter optimization coupled with wire bonder hardware and software capabilities are crucial. Acknowledging these concerns, the selection of the capillary was given much thought. Such a capillary must achieve the small bonded ball bond desired, and in association with process optimization, minimize the aluminum splash, ensure remnant aluminum, and finally to address any potential damage to the fragile lowk construction underneath the bond pads. The prescribed approach would be to use a lower set of bond process parameters. Hence, the selected capillary design must be able to achieve such criteria. Evaluations on a wire bonder equipped with a Cu kit, using a 17.5um thin bare Cu wire together with - special low-k capillary design, and with a bonding temperature of 150 deg C, showed promising initial data. At time zero, all traditional wire bond process buy-off criteria are met. However, some lifted ball bonds were observed during the wire pull, with the readings greater than 2gf criteria. Further optimization and refinement is needed. Of course, the necessary reliability test data must follow suit. Also, cross-sectional analysis did not reveal any damage or cracks to the fragile low-k regions below the bond pads. Cu wire to aluminum bond pad interface creates an intermetallics phase that is prone to corrosion after certain stress test conditions. Some studies have shown that the addition of Pd, or Palladium, can overcome this corrosion. With this in mind, the evaluations also include using a Pd coated Cu wire of similar size. Like the bare Cu wire, time zero data showed viability for the Pd coated wire as well. In fact, initial data showed an improvement over the bare Cu wire for standard wire bond process buy-off criteria. The above evaluations have demonstrated that a 17.5um thin bare Cu as well as Pd coated thin Cu wires showed somewhat similar positive initial results on a low-k wafer. Of course there are still many technical challenges as well as reliability considerations to overcome, before the proliferation of 17.5um Cu wire on such a fragile wafer technology.