Low-photon detection using InGaAs/InAlAs and InAs avalanche photodiodes

Single-photon avalanche photodiodes (SPADs) based on the InGaAs/InAlAs material system are designed, fabricated, and characterised for 1550 nm light detection. The two designed wafers reduce the electric field across the InGaAs absorber to a minimum in order to minimise dark current. The first wafer is designed to punch-through at the point of high breakdown probability (above breakdown voltage), while the second is designed to punch-through just under breakdown voltage. The first wafer is found to be unsuitable for single-photon counting due to an uncharacteristically fast rise in dark count rate, likely caused by the onset of punch-through during breakdown. Low photon levels are detected using diodes fabricated from the second wafer, however the diodes were found to not fully punch-through, preventing single-photon counting. Peak laser pulse detection probabilities at 150 K were 73, 71, and 46 % for 100, 30, and 10 photons, respectively. At room temperature, pulse detection probabilities were 39, 35, and 30% for the respective photon levels. This informs future SPAD designs; crucially that full punch-through must occur before breakdown voltage. A simulation model for the sensitivity of electron APDs (e-APDs) is developed and applied to InAs e-APD based optical receivers. The model simulates bit-error rate (BER), and captures the effects of inter-symbol interference (ISI), dark current, current impulse duration, avalanche gain, and amplifier noise. With a target BER of 10 −12 , the receivers’ sensitivities were -30.6, -22.7, and -19.2 dBm for 10, 25, and 40 Gb/s data rates. The simulated InAs APDs offer improvements over existing InAlAs APDs at 10 and 25 Gb/s, however SOA-PIN based receivers outperform both types of APD for 40 Gb/s for 1550 nm operation. Utilising the newly developed e-APD sensitivity model, and a previously developed sensitivity model for standard APDs, simulations are performed comparing InAs, InAlAs, and InP based optical receivers. The simulations utilise a common parameter set where appropriate, allowing for a direct comparison between the three avalanche materials for high-speed operation. Simulated InAs APDs achieved the best performance for 10 Gb/s operation (-29.4 dBm for InAlAs), while the simulated InAlAs APDs were found to perform better at 25 and 40 Gb/s, achieving a sensitivity of -23.5 and -21.0 dBm, respectively. InP APDs showed sensitivities of -27.9, -22.5, and -19.9 dBm, for 10, 25, and 40 Gb/s operation, respectively. These simulations demonstrate significant performance benefits to replacing InP with InAlAs as an APD avalanche region. Additional simulations of InAs APDs were performed, exploring how to further optimise InAs based optical receivers.