An increasing number of diverse applications rely on single photon detectors to detect weak optical pulses. Among various single photon detectors, planar InGaAs/InP SPADs are the most practical for NIR detection. Since SPADs are operated beyond their avalanche breakdown voltage, it is vital to suppress premature edge breakdown. To evaluate existing design rules for edge breakdown suppression, two rounds of planar InGaAs/InP SPAD wafers (referred to as Round 1 and Round 2) were designed, fabricated and characterised for 1550 nm wavelength single photon detection. This thesis investigated the design rules of double Zn diffusion and FGRs for edge breakdown suppression through the measurements of avalanche gain, DCR and SPDE as well as electric field simulation.
The Round 1 wafers were found to be unsuitable for SPAD operation due to the non-optimal electric field profiles caused by wafer growth uncertainties. Hence, measurements and data analyses (2 D electric field simulations) were limited to linear mode operation. Utilised actual Zn diffusion profiles, simulation data confirm the optimum difference to be d1 ≥ 1.5 μm and (d2 - d1) ≥ 0.5 μm. Both experimental and simulation data suggested that the FGRs are only effective when their spacing is ≤ 4 μm. Increasing the Zn extension margin (beyond 4 μm) is also advisable to gain further PEB suppression.
Using improved wafer and device fabrication for Round 2, planar InGaAs/InP SPADs exhibit SPDE of up to 50 % with DCR of 1 Mcps at 225 K. Devices exhibit similar DCRs regardless of various Zn diffusion extensions, FGR spacing, number of FGRs or the absence of FGRs. The 2 D electric field simulation results indicate that FGR spacing needs to be ≤ 4 μm. They also highlight how the deviation of Zn diffusion profiles could reduce the effectiveness of stepped Zn diffusion, such that edge breakdown is present regardless of FGR designs. Despite the poor performance, round 2 illustrates that achieving the desired stepped Zn diffusion profiles (within the tolerance afforded by Zn diffusion technology) is crucial in suppressing premature edge breakdown.
Finally, a new wafer structure for planar InGaAs/InP APDs/SPADs to minimise the impact of Zn diffusion depths’ deviation was proposed. In the proposed structure, although Zn diffusions are needed, the active region’s electric field are not affected by the deviation in Zn diffusion depths. This is attributed to adding a p charge sheet layer. 2-D electric field simulations suggest that the proposed structure does not suffer from PEB and so FGRs are redundant. Further simulations investigated how the proposed structure copes with ± 20 % variation in both p and n charge sheet doping densities. Overall, the simulation study demonstrated that the proposed wafer structure is promising in obtaining the desired electric field profiles regardless of the inaccuracies in Zn diffusion depths.
