Reliability of assembly of chips and flex substrates using thermosonic flip-chip bonding process with a non-conductive paste

Abstract

This study investigates the reliability of the assembly of chips and flex substrates using the thermosonic flip-chip bonding process with non-conductive paste (NCP). The high-temperature storage (HTS) test, the temperature cycling test (TCT), the pressure cooker test (PCT) and the high-temperature/high-humidity (HT/HH) test were conducted to examine the reliability of chips that are bonded on flex substrates. The environmental parameters used in the various reliability tests were consistent with the JEDEC standards. After the reliability tests, a peeling test was performed and the microstructure of the tested specimen observed to evaluate further the reliability. The bonding strength increased with the storage period in the HTS test. After the peeling test, a layer of copper electrodes was observed to be stuck on gold bumps over the fractured morphology of the chips when the chips and flex substrates were assembled using an ultrasonic power of 14.46 W, indicating that the bonding strength between the gold bumps and the copper electrodes was even higher than the adhesive strength of the layers that were deposited on the flex substrates. The HTS test yielded sufficient thermal energy to promote atomic interdiffusion between gold bumps and copper electrodes. Metallurgical bonding between the gold bump and the copper electrode occurred, improving the bonding strength. In the assembly of chips and flex substrates without the application of ultrasonic power in bonding process, the adhesive strength of NCP was highly reliable after HTS test, because the bonding strength was maintained after HTS test for various storage periods. The typical failure mode of PCT was interfacial delamination between NCP and flex substrates. Approximately 80% of the specimens exhibited full separation after PCT at 336 h when chips and flex substrates were assembled without applied ultrasonic power to the bonding process, revealing that the NCP cannot withstand the PCT and lost its adhesive strength. Applying an adequate ultrasonic power of 14.46 W in the bonding process not only improved the bonding strength, but also enabled the bonding strength to be maintained at high level after PCT. The high bonding strength was attributable to the strong bonding of the gold bumps on the copper electrodes after PCT for various storage periods. This experimental result demonstrates that ultrasonic power can increase the reliability of PCT on chips and flex substrates that were assembled with the NCP. The bonding strength of the gold bumps on the flex substrates did not change significantly after the TCT, revealing the great reliability of TCT on chips and flex substrates that were assembled using the thermosonic flip-chip bonding process with the NCP. The bonding strength of chips bonded to flex substrates increased with the storage periods of the HT/HH test if ultrasonic power was applied to bonding process. Neither delamination nor any defect at the bonding interface was observed. The reliability of the HT/HH test for chips bonded on flex substrates using the thermosonic flip-chip process with the NCP fulfills the requirements stated in the JEDEC standards. According to the experimental findings of various reliability tests, the chips that were bonded to flex substrates using the thermosonic bonding process with NCP met the JEDEC specifications; with the exception of the adhesive strength of NCP under PCT which must be improved.