Theoretical Study And Monte Carlo Simulation Of Avalanche Photodiode And Fabrication

In the long-haul optical transmission system, the photon starved becomes the inevitable restriction. Avalanche photodiodes(APDs) provide 5-10 dB higher sensitivity than PIN photodiode owing to its internal gain. High-speed avalanche photodiodes provide the sensitivity and speed needed for next generation fiber optical transmission. Although InP/InGaAs APDs of 10 Gbit/s and lower speed have been commercialized and widely applied in optical communication, the study on InAlAs/InGaAs APDs have been focused due to its favorable ionization coefficients and larger band gap, which makes less work in the study of 25 Gbit/s InP/InGaAs APDs. Compared with InAlAs material, InP is more commercial available with mature technology and high reliability. In this thesis, InP/InGaAs APDs for 25Gbit/s and higher speed and low noise tandem APDs are designed and characterized.For the device-level study of high speed APDs, we extend the random path length(RPL) model by Monte Carlo(MC) model. On the basis of the modified RPL model, we derive the impulse response of separate absorption, grading, charge and multiplication avalanche photodiodes(SAGCM APDs). Besides the multiplication layer, the ionization is also considered in part of charge layer, where the electric field is high enough for carrier multiplication. The impact ionization coefficients are adjusted using Monte Carlo simulation. The bit error rate(BER) was calculated based on the derived impulse response, considering tunneling current, inter-symbol interference(ISI) and bandwidth-limited gain. APD device‘s sensitivity is predicted and the factors that restrict the bandwidth are analyzed, which releases intensive calculation in MC simulations and reveals the ionization in charge layer should be considered in the thickness tradeoff of multiplication layer for high speed operation according to feasible technology.For the study of low noise tandem InP APDs, the physical mechanism is studied by MC simulations, and the feasibility is predicted. The dead space method theory(DSMT) model and RPL model are used to study the essential physics that governs the gain-noise characteristic and frequency response. Then the analysis and optimization of device‘s structure are addressed. Furthermore, the lock-in amplifier(LIA) based test system is used for the I-V and M-F measurement. The experimental results show that the tandem InP APDs have an improved noise figure.