Modeling and optimization of heterojunction avalanche photodiodes: Noise, speed and breakdown

Avalanche photodiodes (APDs) are often the photodetectors of choice for high-sensitivity and high-speed applications such as modern optical communication systems, quantum cryptography, and medical imaging, to name a few. This preference is due to the internal optoelectronic gain that APDs provide, whereby each absorbed photon is converted to a multitude of electrons and holes comprising the gain. This gain is effective in combating the unavoidable electronic (or Johnson) noise and it is responsible for significant improvements in detection sensitivity in comparison to their PIN-photodetector counterparts, which do not exhibit any gain. The APD gain, however, is accompanied by an excess noise that arises from the randomness in the chain of electron and hole impact ionizations that give rise to the random gain in the first place. In addition to this intrinsic gain noise, the performance of APDs at ultra-high speeds is generally plagued by their limited bandwidth. This is due to their inherent avalanche buildup time, which is the time required for all the carrier impact ionizations to complete each time an incoming photon is absorbed. Moreover, the gain and buildup time are interrelated in a complex, stochastic fashion.; This dissertation provides analytical, stochastic models that advance our fundamental understanding of the behavior and limitations of APDs, and to develop optimization strategies for the design of cutting-edge APDs. Here, the modified dead-space multiplication theory (MDSMT) is developed incorporating several key physical phenomena observed in modern heterojunction APDs including the dead-space effect (deterministic or stochastic), softness of the ionization threshold, and variable electric-field and doping profiles in multilayer multiplication regions. The model predicts the low-noise performance in a class of heterostructure APDs for the first time to our knowledge. It is demonstrated that when hot (energized) carriers enter the multiplication region of an APD, the excess noise and the buildup time are significantly lowered. In addition, the breakdown probability distribution is also enhanced. Using this novel phenomenon and the accompanying model, optimal bandgap-engineering strategies are developed to design APDs structures with optimal excess noise, bandwidth, and breakdown-probability characteristics.