Wafer level reliability testing of heterojunction bipolar transistors

A standard automatic prober was modified to keep the probe needles on the test pads during heating and cooling of the wafer. These modifications involved characterizing the chuck movement during a temperature cycle and writing a program to move the hot chuck to compensate for thermally induced movement. These modifications also required the use of Gravity probes that were able to keep the probe needles on the test pads while allowing moderate amounts of chuck movement. Nine gravity probes were placed on the prober so that 3 HBT devices could be tested/stressed at the same time. As part of the testing sequence the probe needles were placed on the test pads at the beginning of the testing cycle and not removed until after completion. It was believed that since the probe needles were never removed from the test pads that changes in the contact resistance between the probe needles and the test pads were essentially zero. Therefore, it was not necessary to use extra probes as Kelvin probes when testing/stressing the HBTs. This allowed maximizing the number of HBTs tested/stressed at the same time. A program was written in Lab View to automatically perform the testing, stressing, heating, and cooling of the wafer. Emitter resistance degradation measurements were then used to determine if this prober was usable for WLR testing of HBTs. A comparison between emitter resistance degradation results for device´s tested/stressed using this modified prober and device´s tested/stressed using a manual prober with Kelvin probes showed comparable results. The modified prober was then used to perform a three temperature lifetest on devices processed through TRW´s foundry GaAs/AlGaAs HBT process. The results showed that for emitter resistance degradation the MTTF of the HBTs at Tj=125°C is approximately than 1.5×10⁷ hours