Messung der räumlichen Verteilung des Magnetfeldes oberhalb von Halbleiterbausteinen zur Erkennung fehlender Bonddrähte

This thesis is about an electrical fault detection method for production faults in semiconductors. The integrated circuit (IC) is electrically connected over several wirebonds to the package pins. Above a certain current level, two or more wirebonds have to be placed in parallel to minimize the power dissipation and to guarantee the output resistance of the chip. The device under test (DUT) is a two channel highside switch BTS–5120–2EKA from Infineon, which has three wirebonds per channel. In case of a fault at one or more of them, the remaining wirebonds have to carry the current and due to this, the thermal stress is increasing which is resulting in a shortened lifetime. Because the DUT is used in automobile applications, higher reliability requirements are applying (f.e. ISO26262). The probability of fault detection has to be sufficient high or redundant wirebonds have to be placed, what is increasing the production costs.

This work is a ”Proof of Concept” and shows the possibility of detecting wirebond faults by measuring the magnetic field distribution above the chip, caused by currents through the wirebonds. Therefore the sum of the currents through the wirebonds is equal for all fault conditions of the three wirebonds. This is the reason why the magnetic field distribution above the chip varies with the different fault conditions. Prototypes of GMR based sensor arrays were used for measuring the magnetic field. The sensors were produced for this purpose by Infineon. The first sensor array had 6 GMR–elements and the second improved one had 26 GMR–elements.

Magnetic fields in biomedical applications are either biological origin or technical origin. This magnetic field strengths were compared to the field strengths caused by the different fault conditions above the chip. A possible biomedical application of the sensors is discussed. Furthermore the measured magnetic field distributions of the fault conditions were compared with a simulation, which was developed on the JKU Linz during a PHD-thesis.