Study of Au Ball Bond Mechanism and Reliability on Pd/Ni/Cu Substrate

Abstract

Microelectronic wire bonding is a manufacturing process used to electrically connect integrated circuits with circuit boards or other substrates. Conventionally, balls are molten at the end of a Au bonding wire and subsequently bonded on Al metallization of a integrated circuit. However, Pd/Ni metallization has recently been used for its improved mechanical properties. The bondability, bonding mechanism, and reliability of Au ball bonds on Pd are studied in this thesis. The substrates were produced in this project using three different materials. The base material is polished Cu in the shape of a coupon (1.0 cm × 1.0 cm × 0.5 mm). Cu coupons are plated with Ni (1.0 μm) using an electroless process, followed by electrolytic plating of a layer of Pd (0.3 μm), resulting in an arithmetic mean roughness of the surface of 0.08 μm (baseline sample, sample 0). Higher roughness values of 0.2, 0.4, and 0.5 μm are artificially produced by rolling (sample 1), sanding (sample 2), and sandblasting (sample 3), respectively, on the Cu surface before plating Ni and Pd. A 25 μm diameter Au wire is used for bonding on the polished and roughened substrates with a process temperature of T = 220 °C, and it was found that ≈ 4 % to ≈ 18 % less ultrasonic amplitude was required for successful bonding on the roughened substrates compared to the polished substrate. Bondability is measured by shear testing the ball bonds. An average ball bond strength achieved on the polished substrate is 130 MPa. This value is lower on the roughened substrate with the exception of the sandblasted substrate. Long-term thermal aging at 250 °C was performed with ball bonds on samples 0-3 for durations of ≈ 300 h. The reliability of the bonds is characterized by non-destructive contact resistance analysis during aging and destructive cross section analysis after aging. Contact resistance values for the ball bonds range from 1.6 to 3.5 mΩ at 20 °C before aging, and does not correlate with roughness. For the baseline sample, contact resistance of the ball bonds decreases during aging by -6 % (median value), which indicates electrical integrity of the interconnections at high temperature. This decrease possibly is due to interfacial gap filling by Au or Pd diffusion. In contrast, the contact resistance increases for the roughened samples 1-3 and changes are 0.4, 5, and 14 %, respectively (median values). A conclusive explanation for this increase has not yet been found. After 250 h of aging, a TEM analysis showed Au to Pd diffusion in the baseline sample with a diffusion depth of ≈ 0.1 μm Au. No intermetallics, voids, or contamination is found on the interfaces after aging according to nanohardness, SEM, and TEM analyses. No bond lift-offs or electrical opens were found for the aging temperature and durations chosen. No conclusive evidence for the presence of Au-Pd intermetallics or voids is found.