Wirebonding at higher ultrasonic frequencies: reliability and process implications

Abstract Higher-frequency ultrasonics have been utilized to improve the bondability of difficult substrates, i.e., substrates that would not bond or that bonded poorly using conventional ultrasonics (nominally at 60 kHz). A systematic study of the influence of higher-frequency ultrasonics on bond strength and the bondability of various substrates is reported. The studies were carried out using two essentially identical thermosonic ball bonding machines, one bonding at nominally 60 kHz and the other at 100 kHz. The only differences between the bonding machines were the ultrasonic generators’ operating frequency and the transducer horns. Key to the study was the ability to make the bonding experiments as controlled, repeatable, and independent of all variables (except frequency) as possible. Control techniques included setting the electronic flame-off to produce consistently sized free-air balls; monitoring the ultrasonic voltage and current waveforms; and picking force, dwell, energy, and substrate heat settings that would allow strong bonds to be formed at both frequencies. Wirebonds (ball bonds) in this study were evaluated primarily by the ball bond shear test. Statistical methods were used to determine whether the differences in the means and variances between comparable samples sets (one bonded at 60 kHz and the other bonded at 100 kHz) were significant. Results of our studies indicate that significant differences exist between bonding at nominally 60 kHz and bonding at 100 kHz. In particular, we describe effects associated with (1) the ball shear strength before and after thermal aging (temperatures up to 200 °C) for both 60- and 100-kHz bonds, (2) the influence of substrate-metallizations combinations on the geometry and strength of the bonds at the different frequencies, and (3) the sensitivity and control of the overall bonding processes.