Current Issue The International Reference for Chip-Scale Electronics, Flip-Chip Technology, Optoelectronic Interconnection and Wafer-Level Packaging January - February 2002 Email the editor

Lee Levine Contributing Editor

In high-performance semiconductors, we see the trend toward copper metallization with low-K dielectric layers progressing.

As density (finer-pitch lines) and on-chip clock speeds increase, Al metallization can't meet the required performance standards achieved by Cu metallization. This, in turn, drives the development of processes necessary for interconnecting, assembling and packaging the copper metallized chips.

Technical Requirements

The International Technology Roadmap for Semiconductors [public.itrs.net] gives an assessment of the technical requirements and timetable for future semiconductor development.

Development, however, is progressing at a faster pace than previously forecast. Last year, users beat the 1999 roadmap by employing 130 nm lines (45 µm ball-bond pad pitch).

IBM's Power 4 chip, boasting copper metallization for faster speed and lower power consumption, marked this arrival.

Cu wafer metallization can be bonded with either copper or gold wire using either ball- or wedge-bonding techniques. Au wire bonding is a more conservative starting point, but the monometallic system-Cu wire bonded to copper wafer metallization-offers a reliability advantage.

There are three major issues to be addressed with Cu wafer metallization and Cu wire bonding: oxidation, hardness and corrosion.

• Oxidation: Cu oxidizes readily at wire-bonding temperatures and must be protected from oxidation to preserve bondability. Even slight oxidation, detectable by its small color variation, is enough to destroy yields. There are two methods you can use to successfully prevent die oxidation.

  1. The first alternative is the use of thin (< 100 A°) inorganic, friable glass/ ceramic coatings. These coatings will crack during bonding, exposing the oxide-free Cu bond pad to the Au or Cu wirebond. During bonding, the coating is swept aside and the bond forms between the clean, oxide-free bond pad and the bonding wire (ball or wedge). The coating is thin enough to remain in debris fields within the bond area without lowering bond quality. You can apply these coatings in assembly (an advantage), either before or after probe.

  2. The second alternative involves capping the bond pads with additional metallic layers (Al or Au). Capping requires at least three additional front-end mask steps and is more expensive. Au caps have been shown to achieve reliable results with both Au and Cu wire. The reliability of Al caps has only been documented with Au wire.

• Hardness: On the second point, Cu is harder than gold and requires more energy for deformation.

  The additional hardness, combined with bond pads that contain low-K dielectric internal layers, causes another difficulty.

  Most low-K dielectrics are soft and spongy. During wire bonding, both the wire (ball or wedge) and the bond pad deform together, forming the bond. The soft, spongy layers within the structure allow the top metallization layer to cup and deflect under the ball, preventing the co-deformation necessary for bond formation.

  Development of stiffer dielectrics, designing bond pad structures for improved stiffness (thicker metallization layers, tungsten plugs) and optimization of the wire-bonding equipment will all be required to solve this problem and achieve stable manufacturing processes.

• Corrosion: Cu wire bonding has demonstrated long-term reliability. It's been shown in earlier studies, however, that you must select packaging materials carefully. The studies demonstrated that the use of antimony-based fire retardants in molding compounds may cause catastrophic corrosion failures. Molding compounds containing Cl and other halides are also detrimental. Preventing corrosion of the Cu wire or metallization inside the package is critical and encapsulant choices must be tested carefully to achieve reliability.

Conclusion

All of the reliability and manufacturing issues regarding Cu wafer metallization and Cu wire bonding can be resolved by good engineering. I've been a proponent of copper wire bonding since 1986, and I'm pleased to see the growing use of this technology as an emerging trend.