To simulate the bow of wafers with integrated capacitors in the form of pit arrays, various approaches were pursued. After unfruitful attempts to reliably obtain the wafer bow directly from simulating part of the wafer, a multi-scale approach was used. In this approach, the layer with the integrated capacitors was replaced by a homogeneous material having the same properties. Small-scale simulations of representative parts of the layer were performed to determine its effective stiffness tensor. Inclusion of the intrinsic strains of the grown and deposited dielectric and conductive layers enabled the volume change to be calculated of the layer with the integrated capacitors upon fabrication. Finally, the structure obtained was used in a full-wafer-scale model to simulate the bow of the wafers. Even for uncalibrated values for the coefficients of thermal expansion, most simulations agreed well with measurements.

A ball bonding process was simulated over a high-voltage isolation structure. The removal of an inter-dielectric metal crack-stop layer was investigated through 3D simulation. Material properties for the bonded gold ball were obtained using nanoindentation and atomic force microscopy with a methodology from the work of Ma et al. This yielded both elastic and plastic material parameters. The methodology was then evaluated by using the parameters in a nanoindentation simulation. Although the topography simulated only roughly agreed with measurement, the simulated and measured indenter curves closely overlapped. The parameters were then used in the bonding simulation. The deformation of the bond ball was also measured so that the equivalent deformation could be simulated. This was achieved following the incorporation of both ultrasonic motion and softening in the simulation. Two bonding process geometries were then set up: one with the crack-stop layer present and the other without. Both were simulated and the output was applied within a failure theory to evaluate the risk to the isolation oxide.

Modelling was undertaken to investigate the role of bond wire size on reliability in power electronic converters. Experiments have shown that thin 125 μm Al wires used in place of 375 μm Al wires alleviate bond wire lift-off and further outlast other sources of failure such as solder degradation in a power module. To investigate the role of bond-wire size on wire lift-off, the effective plastic strain was estimated through thermo-mechanical simulation. Three-dimensional models were constructed for the thin and thick bond wires, respectively. For the critical deformation of the aluminium bond wires during thermal cycling, a temperature-dependent bi-linear plasticity model was used. The effect of a difference in yield strength for the thin wires was also investigated. Maximum as well as volumetrically averaged values of the effective plastic strain showed significant differences between the thick and thin wires and wires with different yield strengths. The modelling results show higher effective plastic strain for the thick wires - supporting the experimental findings.

For an assessment of the stresses occurring during ball bonding of high-voltage CMOS chips in a structure comprising a thin and a thick silicon dioxide layer below the bonding pad, a dynamic model of the process was set up and the materials parameters were calibrated. For a realistic result of the deformation of the bonding ball during the ultrasonic stage, up to 60 ultrasonic cycles were simulated. To reproduce the final height of the bonding ball, dynamically increased friction between the ball and the bonding pad as well as ultrasonic softening of the metals within the model had to be taken into account. For a more sensitive prediction of failure, the conventional failure criterion based on the ultimate tensile strength of brittle materials was complemented by an additional criterion suggested by Christensen which takes the combined effects of perpendicular tensile and compressive principle stresses into account. This yielded a prediction of earlier failure for the thinner oxide layer while no failure was predicted for the thick isolation oxide layer.