Solving Module-Level Copper Wire Bonded ICs Failures

Posted by Greg Caswell on Sep 23, 2016 9:30:00 AM

Copper-Wire-Bonded-ICs.jpgOver the past two decades gold prices have increased nearly $1,000 per Troy ounce, necessitating a transition to replace gold with copper bond wires in integrated circuits (ICs). This seemingly small change is actually very significant. The differences between the metals’ properties require full optimization of each module in order to compensate for the nuances of the copper wire bonds (Cu-WBs).

Quality, Reliability and Durability

Of particular importance is copper’s hardness and tendency to oxidize. If either, or both, of these characteristics are ignored, several Quality, Reliability and Durability (QRD) issues arise:

Ball Bond Separation

Ball bonding requires Electronic Flame Off (EFO), meaning heating sufficient to melt the wire tip into a symmetrical, oxide-free ball of precise dimensions. A misshapen sphere can damage the pad or produce weak, partial bonds that may fail after molding or in the field under usage stress.

Copper is a harder metal than gold, which can make the bonding process more challenging. However, as copper purity increases (ideally >95 copper or above), the softer the wire is and it can create a more controlled, consistent bond and help prevent damage to the pad during bonding. Pure copper is more plentiful and cheaper than gold, which makes it a reasonable substitute for that precious metal; and, although its effectiveness is still being proven for bonding in harsh environments, pure copper presents fewer issues than its lower purity counterparts.

Oxidation & Corrosion

Gold is impervious to oxidization and corrosion, so exposure to air does not affect gold ball formation. Copper normally oxidizes, especially under high temperatures. This jeopardizes ball bonding because copper oxidization on sphere surfaces result in weak partial bonds that are susceptible to failure.

To combat oxidation during copper ball formation, it was initially thought that a nitrogen inert atmosphere was required. However, it has been subsequently discovered that a forming gas of 95% nitrogen and 5% hydrogen is ideal for preventing copper oxidation.

Pad or Die Damage

A great deal of thermosonic bonding energy is required to bond copper with aluminum. This excessive energy increases the risk of aluminum splash, typically resulting in one of two detrimental outcomes: the molten pad aluminum flows from under the ball bond, or the ball punches through the pad. Further, die material may be damaged or the electrical/electronic (E/E) features under the pads and/or the remaining pad aluminum may be too thin.

The Cu-WB bonding process must be controlled so that ≥0.2µm of the original aluminum pad thickness remains to maintain the pad strength required for preventing pad fractures or tears. Alternatively, thicker pads may be substituted for the typical 1µm aluminum thickness (used with gold wire bonds).

Stitch Bond Separation

Humidity, temperature or electrical bias of chlorine ion attraction to positively charged pads causes corrosion that will eventually separate a stitch bond. 

Since copper is susceptible to corrosion, using a molding compound with a pH of 4-6 and a low chlorine content at ideally <10 ppm is essential for alleviating corrosion failure risks. It is also imperative to minimize voids or irregularities in copper-aluminum IMC bonds that would otherwise allow moisture ingress, increasing corrosion degradation and separation stresses within the bond.

Module-level Thermal Cycling Tests

Considerable failure risks exist when using Cu-WBs and not all of them are identified at component-level testing. Module-level thermal cycling tests are accelerated life tests calibrated to focus on the structural integrity of the fully assembled circuit board, specifically as it relates to potential attachment solder fatigue caused by mismatched coefficient of thermal expansion (CTE) between the components, housings, and the board.

For Cu-WBs, module-level thermal cycling tests help identify and address known issues related to bond defects or under-optimized designs. Copper bonding quality is dictated by energy. Bonding energy that is too low could result in weak bonds that separate from pads; too high and bond pad wear-through or cracking could occur. Optimizing the Cu-WB bonding process to consistently hit the “sweet spot” is essential to producing more reliable integrated circuits that perform well during modular level testing and ultimately over the lifespan of a product.

Advanced CAE tools can make optimization easier. Sherlock Automated Design Analysis™ software simplifies and improves reliability prediction by virtually running thermal cycling, power-temperature cycling, vibration, shock, bending, thermal derating, accelerated life, natural frequency, CAF and more – assisting not only in copper related problem-solving, but problem prevention.

Contact us today to learn how Sherlock can work for you!

Are GPUs Reliable Enough for Autonomous Vehicles? webinar

Topics: Sherlock

Sign me up for updates and offers from ANSYS, including DfR Solutions, and our partners. I can unsubscribe at any time.