The lift-off characteristics at the interface of thermosonic bond were observed by using scanning electron microscope (JSM-6360LV). The vertical section of bonding point was produced by punching, grinding and ion-sputter thinning, and was tested by using transmission electron microscope (F30). The results show that the atomic diffusion at the bonded interface appears. The thickness of AdAI interface characterized by atomic diffusion is about 500 nm under ultrasonic and thermal energy. The fracture morphology of lift-off interface is dimples. The tensile fracture appears by pull-test not in bonded interface but in basis material, and the bonded strength at interface is enhanced by diffused atom fiom the other side.

Fiber optic sensors (FOS) for vibration monitoring of smart structures have certain advantages over conventional strain gage based sensors due to electromagnetic environment insensitivity and high response bandwidth. During the present study, a spatially integrating fiber optic sensor was used for vibration monitoring. It is based on the concept that the optimal placement of the sensing element can be sought by using a priori knowledge of the mode shapes of the structure. The applicability of this approach to polarimetric optical fibers is described. The acquired integrating FOS signal was used to sense and control the vibration of an eleetrorheological adaptive structure subjected to a random external excitation.

Dynamical characteristics of PZT actuated transducer assembly are the key to reliable flip-chip bonding and basic understanding of metal interconnection. A systematic experimental observations and analysis clearly exhibit that, rich phenomena and effects like the negative cubic stiffness, dynamical coupling between transducer body and tool, velocity dropping and anomalous phase portrait of the die movement, may be explained by nonlinear features and non-parallel loading of transducer assembly onto its bonding area. It will be possible to explore their mechanism and effects to bonding quality by identifying and decoding them.

Aluminum wire bonding was performed on a lab test-bed with PZT transducer device. Vibration behavior of bonder transducer was monitored, and the correlations between bonding pressure, high frequency tool vibrations, and average bond strength were then demonstrated to determine the statistically significant differences among them. Wavelet decomposing, combined with statistics-based data processing, was utilized to explore the details of bonding process and evade inherent bonding uncertainties. Experimental results show that the method is practical and effective even if the interactions between transducer and its bonding target are very small. The obtained time-frequency plots were depicted for identifying un-modeled wire bonding dynamics.