| الوصف: |
Group IV photonics is attracting more and more attention these days in order to realise large scale optoelectronic integration. Although many prototype devices have been demonstrated, few of them can be put into practical applications. One of the major challenges is the lack of the optimized design and theoretical models. In this project, a computational model based on 8-band k.p theory is built to simulate the Group IV quantum well based semiconductor, and in this thesis it is applied to two different types of devices, lasers and modulators. For lasers, we focus on the tensile strained structure as it can reduce the Sn content needed to reach direct bandgap structure, moreover, because of the splitting of light and heavy holes, the low density of states at the valence band top will be beneficial to lasing. Since Ge has indirect band gap, and the L valley may still be close to Γ valley even for direct gap GeSn, effective mass method is used for electrons in the L-valleys. The optimal range of well widths for transverse magnetic (TM) gain is found to be around 13-16nm. With constant well width (14 nm), the optimal choice of both inter-band gain and net gain with varying carrier density for different photon transition energy is found, by doing calculations throughout the parameter space of Sn and strain in the well. The inter-valence band absorption with split-off band was found to be an important loss mechanism that seems to be rarely discussed in the literature. A large inter- valence band absorption was found around 0.4-0.5 eV, when the bandgap is equal to the difference between the top valence band and split off band. And because of the influence of such loss mechanism the optimal choices of net gain have changed from that of the inter-band gain. Using the optimal Sn content and strain combined with waveguide design, the threshold current was estimated to be 1.19 kA/cm⁻², comparable to conventional III-V quantum well lasers. For modulators, a novel way using an intra-step quantum well was applied here to improve the performance of the GeSn quantum well electroabsorption modulator. Using SiGeSn as the barrier and GeSn as the material for the well layers, an intra-step well can be made by using different Sn contents in two intra-layers. The band structure is also calculated by the k.p method, and the exciton effect was considered by variational method. Without increasing the total well width, and compared to the square quantum well, a much larger quantum confined Stark effect can be realised with intra-step quantum well. By considering the figures of merit related to practical performance, Δα/F and Δα/F², the intra-step quantum well, compared to square quantum well, brings about 44% improvement on the bandwidth per unit applied voltage and 46% reduction on the power consumption per bit data transmitted. The model presented in this thesis can still be improved. It can include other realistic effects such as carrier transport, and on the other hand, the accuracy can also be improved by considering 'better' choice of parameters and perhaps higher order k.p theory. Other applications are also possible based on the existing code, e.g. light emitting diodes and photodetectors. |
