ue to the constantly growing energy demand of power electronics and the need to reduce the size of electronic components like power modules for e-mobility, new challenges arise for ultrasonic wire bonding: the electrical connection must endure higher thermal and mechanical stress while the connecting partners become more sensitive or require more energy to get bonded. Past investigations have shown already that multi-dimensional ultrasonic bonding and welding yield the same or even better bond quality while reducing the load on the components. This contribution is intended to show whether multidi-mensional thick wire bonding is a promising concept to over-come the new challenges. The focus is on experimental investi-gations of different bond tool trajectories in ultrasonic wire bonding of aluminum and copper wire on DCB's and chips. The bond quality is analyzed by shear tests, microsections and, in the case of aluminum bonding, by a new machine learning method for an objective automated evaluation of the sheared area.

Waveguide-based methods can be used for the non-destructive determination of acoustic material parameters. One of these methods is based on transmission measurements of cylindrical polymeric specimens. Here, the experimental setup consists of two transducers, which excite and receive the waveguide modes at the faces of the cylinder. The measurement, as well as a forward model, are used to determine material parameters of the polymeric specimen in an inverse approach. 1-3 piezoelectric composites are used as an active element because they can be approximated by a thickness vibration only. This allows an easy identification of Mason model parameters to characterise the transducers’ vibration behaviour. However, sensitivity analysis shows a high uncertainty in the determination of the mechanical shear parameters due to the uniform excitation. To increase the sensitivity to these shear motions, arbitrary excitations were investigated by means of numerical simulation. In order to be able to realise the determined optimal excitation, new transducer prototypes were designed. By subdividing the electrodes of the active element, for example, ring-shaped excitation is feasible. Furthermore, it can be shown that modelling these transducers with a one-dimensional Mason model is sufficient.

Ultrasonic wire bonding is a process to form electrical connections in electronics well established industry. Typically, a clamping tool is pressed on the wire and forced to vibrate at relative high frequency 40 to 100 kHz. The ultrasonic vibration is transmitted through the wire into the interface between wire and substrate. Due to frictional processes, contamination like oxide layers are removed from the contact zone, the surface roughness is reduced, and with increasing bond duration an metallic connection of wire and substrate is established. It is known that the amount of ultrasonic energy over time directly influences the strength and reliability of the bond connection, but the determination of optimum bond parameters is still a challenging experimental task. For this, in the past different model approaches have been presented, to calculate the bond quality by simulation. Measuring the friction between wire and substrate to validate these models is a challenging task at ultrasonic bonding frequency. Therefore a versatile test rig for bonding experiments at frequencies lower than 1 kHz is setup to get detailed insight into the different phases of the connection process. It includes a piezoelectric force sensor for the measurement of the three-dimensional process forces, an electrodynamic shaker for the vibration excitation and a conventional tension-compression testing machine to apply the bond normal force. Using this test rig, it is possible to observe the different phases of bond formation in detail, validate and enhance existing models and finally optimize bond parameters for different processes.