Material and device characterization of indium phosphide/indium gallium arsenide avalanche photodiodes for multigigabit optical fiber communications

In this work we have conducted extensive studies of a planar separate absorption grading charge and multiplication (SAGCM) InP/InGaAs avalanche photodetector (APD) with a partial charge sheet in the device periphery.; The temperature and electric field dependences of ionization coefficients in InP were extracted using multiplied shot noise and gain-voltage characteristics of the APDs. This simplified the experiment and significantly improved our accuracy in comparison to other techniques. The experiments were conducted using devices with highly uniform two-dimensional (2D) gain profile. Low input optical powers were used to minimize gain saturation effects. It was found that the temperature and electric field dependences of the ionization parameters were different from previously reported results.; We have conducted extensive studies of low frequency noise in the planar SAGCM APDs. In a course of these studies we have investigated effect of different processing techniques and burn-in on the low frequency noise in the APDs. A direct relation between the process improvements and low frequency noise performance has been established. Also, it was found that for each studied processing technique there was a relation between device reliability and low frequency noise performance. A reliability screen based on the low frequency noise was developed for the optimized process. Although not perfect, it offered a non-destructive method to assess device reliability on a device-to-device basis. By taking a conservative approach we were able to eliminate all short lived devices.; Finally, we have conducted a detailed experimental investigation and a theoretical modeling of a temperature dependence of breakdown voltage in the SAGCM APDs. It was found that modeling of the temperature dependence of the breakdown voltage within accepted Okuto-Crowell theoretical framework requires an assumption of freeze-out within the multiplication layer below 150K. However, low temperature C-V studies showed that the doping profile of the device is temperature independent. Therefore, it appears that our experimental temperature dependence of the breakdown voltage can not be modeled within existing theory. As was previously pointed by Capasso, additional scattering mechanisms could be important for modeling of impact ionization in direct bandgap semiconductors.