Non-local Impact ionisation coefficients in Al0.8Ga0.2As and Ga0.8In0.2N0.05As0.94Sb0.01 for GaAs-based APDs and SPADs

Abstract

This work attempts to develop AlGaAs/GaInNAsSb separate absorption and multiplication avalanche photodiodes (SAM APDs) and single photon avalanche diodes (SPAD). AlGaAs/GaInNAsSb is a promising structure, with significant potential benefits in detector and APD applications. The combination of the low-noise Al0.8Ga0.2As avalanche multiplication layer with the high detection capability of the lattice-matched GaInNAsSb, would introduce a new choice of high-speed and cheap GaAs-based APD for 1.3 and 1.55 µm telecommunication applications. In this work, experimental measurements as well as a comparison to stochastic models were carried out to investigate the characterisation of different AlGaAs and GaInNAsSb materials and structures. The use of GaInNAsSb, which is lattice-matched to GaAs, as an absorption layer in the SAM APD structure can exploit its narrow band gap energy to effectively detect wavelengths important to optical applications, such as 1.55 µm. When designing SAM APDs, it is crucial to consider if the absorber and avalanche materials have comparable electron-to-hole ionization coefficient ratios. Opposing coefficient ratios iv of the absorber and multiplication layer would produce the worst case for APD noise and response time. An initial evaluation of the impact ionization coefficients and threshold energies of GaInNAsSb has been undertaken. Even though these coefficients have some degree of uncertainty, they can be used as a sufficient foundation for further research on this material. On the other hand, these coefficients can offer essential knowledge, such as the fact that beta is much greater than alpha, which would negatively affect noise in a poorly designed APD. In addition, the approximate fields where multiplication initiates and breaks down were identified. In the future, more work needs to be done to optimize these coefficients. The large α/β ratio in bulk Al0.8Ga0.2As structure offers low excess noise and low dark currents. Also, the wide indirect band gap of Al0.8Ga0.2As enables fabrication of very thin multiplication widths without tunneling being a problem. Consequently, using Al0.8Ga0.2As in a SAM structure can take advantage of the thin multiplication layer, exhibit desirable characteristics suited to high-speed, low-noise avalanche photodiodes, and maximise the reduction in excess noise due to the non-local ionisation effects. In order to characterize thin APDs, the non-local electron and hole ionization coefficients and threshold energies using the hard dead space model were extracted in this work and compared with those of other materials. It is demonstrated that, despite the fact that previously investigated thin GaAs APDs can achieve a similar significance of dead space with the associated minimisation of excess noise, it is difficult to exploit this advantage due to the significant tunnelling in GaAs. On the other hand, it is possible to produce a thin Al0.8Ga0.2As APD with a dead space to ionisation path length ratio close to the fundamental limit, which leads to excellent low-noise multiplication without significant tunnelling. It is also found that in the case of the standard ideal p-i-n structure, these coefficients and threshold energies can effectively simulate multiplication in APDs as thin as 50 nm. This is substantially thinner than current state-of-the-art APDs. The optimal charge sheet conditions for AlGaAs/GaInNAsSb SAM APDs can be determined by initially using GaAs as the absorber layer. Five samples were studied, and samples 3, 4, and 5 were found to be unsuitable for SAM APD applications as they failed to punch through. Sample 3 is suitable for SPAD applications operating at a higher voltage than the breakdown voltage. Sample 2 showed punch through into the absorber but kept the electric field below the required level for multiplication in the planned dilute nitride absorber layer. Therefore, the structure of sample 2 should be used as the nominal design for preparing to use a GaInNAsSb absorber layer. The AlGaAs/GaAs structure can be successfully simulated and redesigned based on correct coefficients to control the charge sheet conditions and achieve the desired punch-through voltage. GaAs/AlGaAs SPADs with an extremely thin multiplication layer are found to exhibit a lower DCR than that in other common SPADs with relatively thicker multiplication layers, such as InGaAs/InP, InGaAs/InAlAs, and Ge/Si SPADs. On the other hand, these SPADs suffered from a higher afterpulsing probability than other SPADs. In the future, more work needs to be done to minimise the effect of afterpulsing, improving the SPAD’s performance. Despite the fact that this work provides the foundation for this structure's essential understanding, a lot of work must be done before it can be developed into products which improve on state-of-the-art APDs and SPADs. Using the extracted coefficients and threshold energies of AlGaAs and GaInNAsSb with a more sophisticated model such as a randomly generated ionization path length (RPL) can help to maximise the potential of the materials in new APDs and SPADs simulating important characteristics such as excess noise, impulse response, and breakdown probability. There is also scope for more open-ended simulation to explore the theoretical limits of APDs with multiplication widths which are substantially thinner than previously employed.