29. January 2026

BA: Development and Optimization of a Ball-Wedge Wire Bonding Process for Small Bond Pads

Fabian Schmid —

Contacting semiconductor chips is an essential prerequisite for the characterization and measurement of electronic components. In some cases, due to space constraints in test setups, only fixed wiring in the form of wire bonds is feasible. To achieve this, appropriate contact pads with sufficient area must be provided on the chip to enable a stable connection without short circuits. However, to minimize costs, as many components as possible need to be accommodated on a single chip, which means the contact pads should be as small as possible.

Therefore, the goal of this work is to develop a ball-wedge wire bonding process for small pads with dimensions of 100 ×100 μm. Experiments were conducted using 17.5 μm gold wire on SiC chips with platinum metallization and PCBs with ENIG-coated contact pads. Initially, the flame-off process was examined regarding the influence of time and arc current on the diameter of the free air ball (FAB) and the length of the melted wire. In the next step, a set of parameters was developed for the ball bond on ENIG surfaces, and the wedge bond on platinum was then optimized considering the influence of bonding time, ultrasonic time, and ultrasonic power on the yield. Following that, the process aimed to adapt first the wedge bonds and then the ball bonds on platinum to the DVS2811 standard, by minimizing the bonding force. In both cases, the goal was to achieve the highest possible yield under the given constraints. In addition, the bonds were electrically and mechanically tested. Finally, pin diodes were assembled using the developed parameters and analyzed focusing on their serial resistance. Also, a cleaning process for clogged bonding capil laries was developed.

The clogged bonding capillaries could be cleaned from gold residues using aqua regia, and from organic contaminants using piranha solution, without visible damage to the capillaries. The results of the EFO experiments revealed a linear relationship between the product of arc-time and -current and the length of the melted wire. Furthermore, a saturation effect was observed starting at a FAB diameter of 60 μm or a melted wire length of approximately 40 μm. For the ball bond, the bonding force was reduced from an initial 100 cN to 50 cN, achieving a yield of approximately 50% on both platinum and ENIG. For the wedge bond, the bonding force was reduced from 100 cN to 40 cN, achieving yields over 80% on platinum and over 70% on ENIG. A strong dependency of wedge bond yield on platinum was also observed with respect to the orientation of the bond loop. Electrical measurements showed that the bonds exhibited ohmic behavior, and the bond resistance was so low that its influence on the serial resistance of the pin diodes was not measurable. Mechanical tests using the available equipment could not detect any bonding force, and neither the ball nor the wedge bonds met the DVS2811 standard requirements using the developed parameters.

In conclusion, the process is technically feasible, but the bonding parameters require further optimization. However, it was demonstrated during the assembly of the diodes that a bond pitch of 160 μm is feasible.