T. Kiyan, Yildiz Technical University, Istanbul, Turkey; C. Brillert, Infineon Technologies AG, Munich, Germany; C. Boit, Berlin University of Technology, Berlin, Germany

Summary: Photoelectric laser stimulation technique in CW and pulsed mode is applied to a ‘real’ IC for the timing characterization. Two measurements units are employed: TAC in scan mode and high bandwidth oscilloscope in single point mode. The delay variations are measured for various supply voltages and compared. Any noise caused by PLS (e.g. the stimulation at filling structures or global delay shifts) may be misleading. Therefore, the structures that cause a change in the switching speed are located and investigated. It was shown that there are some secondary effects that cause delay shift. The continuous illumination creates a lot of electron-hole pairs in bulk Si and a spread of the carrier cloud is in the range of the diffusion length. The focused beam area creates a very dense profile of carriers within the wells that are extracted by the space charge regions of the active devices only within laser spot size. So, the photocurrent in the stimulated device is occurring only when the laser is hitting their active area. But those carriers created in bulk will be extracted from the well space charge region in a wide area, so the well gets charged up over the whole scanning time. Although we observed that local measurements on the transistors are still effective, the well charge causes a global delay to the delay chain. Therefore, CW laser operation is not very appropriate to stimulate a precisely localized delay variation in a circuit region with large well areas. In pulsed operation, the laser can be switched on just when the effective stimulation is required, minimizing the additional global delay shift and assuring resolution to the laser spot size. 1.064m laser beam stimulates also some filling (or dummy) structures as well as the buffers and the inverters. Those structures are p-implanted active areas and they are situated in an n-well forming a p-n junction. Since PLS creates a photocurrent on all p-n junctions including bulk-substrate junctions, the dummy structures may be stimulated. The best way to avoid these side effects is to pulse laser at the switching edges of the investigated signals. It limits the substrate current and therefore the stimulation on the secondary structures is avoided and the laser perturbations are more localized.