Experimental and modeling studies for thermosonic flip chip bonding

Thermosonic flip-chip bonding is a new, solderless technology for area-array connections. Its mechanism is the same as that of thermocompression bonding, but the bonding temperature and pressure are reduced due to the introduction of ultrasonic energy. This thesis work has been conducted to demonstrate this new technology and develop a knowledge base for the technology advancements. Five major contributions are (1) the successful assembly demonstration, (2) the review of studies on the ultrasonic softening effect, (3) the establishment of an assembly yield model, (4) the experimental verification of a random change of vibration modes, and (5) the establishment of a modal analysis and a vibration model for the bonding system.; A 64-I/O, GaAs-on-silicon test vehicle has been assembled with a substrate temperature of 180{dollar}spcirc{dollar}C, a pressing force of 30 gf/bump, and an ultrasonic power level of 15 W. The minimum shear strength of the assembly is larger than 5 gf/joint. The temperature and pressure required for the demonstration are low due to an ultrasonic softening effect on the joint. The studies on the effect have been reviewed. There is no unified theory for the effect; however, these studies provide a clear guideline: softening, or reduction in stress required to plastically deform a joint, is directly proportional to ultrasonic vibration amplitude.; To assist the technology development for higher number of I/O's, an assembly yield model has been established. The model predicts the yield as a function of force, planarity, bump height, bump diameter, and material properties. The process window for a 100% assembly yield can be estimated by assuming that 15-45% deformations result in good bondings.; A strong and stable softening effect is desirable for consistent deformations. Since the effect is related to an ultrasonic vibration amplitude, we have studied the transverse vibration amplitudes along a bonding tool and system impedances in various cases. A random mode change has been recorded experimentally, and according to our knowledge, this measured mode change for a thermosonic bonding tool is reported for the first time. The change has occurred when we had a wrong set of tool length, friction condition, and frequency window selected for the power generator. The change may affect vibration amplitudes and associated bonding quality.; To enhance our understanding of this mode change and other effects, we have developed a modal analysis and a vibration model containing an energy method. These models characterized vibration amplitude and system impedance as a function of tool length and mass with or without friction. Once calibrated by the experimental data, the models' accuracies have been verified by the data from different cases. The models have been successfully applied to explain the random mode change measured. In addition, the modeling results provide us with an insight into the selection of tool length and mass for an efficient bonding system.; This thesis work has been conducted for the thermosonic flip-chip bonding technology. However, the knowledge base and the models can also be applied to wire bonding and tape automated bonding (TAB) technologies.