Optimisation of III-V Buffer on Si Substrates

Monolithically integrating direct-bandgap III-V compound semiconductors onto silicon (Si) is a promising method for high-volume applications. However, when III-V materials are epitaxially grown on Si substrates, the inherent material distinctions introduce a high occurrence of crystalline defects including threading dislocations (TDs), antiphase boundaries (APBs), and micro-cracks. These defects act as non-radiative recombination centres, causing a significant decline in the performance and reliability of final devices. Furthermore, thermal cracks will significantly degrade the yield during the massive production. This thesis focuses on mitigating the impact of these critical defects in molecular beam epitaxy (MBE) grown III-V materials on Si substrates, while different buffer designs for alleviating these growth defects have been demonstrated. The ultimate goal is to develop practical on-chip light sources and low-cost III-Sb optoelectronic devices for Si-based PICs. To address the issue of micro-cracks without compromising the performance of the laser, a 300 nm Ge buffer layer was first investigated as a replacement for a portion of the thicker GaAs layer in Chapter 3. The optimisation of the Ge buffer layers involves experimenting with various temperatures for the low-temperature Ge layer, using Sb as a dopant with various doping density and different annealing temperature range to achieve a low TD density at 2.6 × 108 cm-2. This approach could save an 1100 nm thickness budget for the final laser devices, which has the chance to avoid the formation of thermal cracks. To further enhance the performance of InAs/GaAs QDs grown on Si substrates, a defect filter layer (DFL) was utilised to reduce the TD density. The thesis delves into the optimisation of DFLs, within Chapter 4 exploring and comparing the effect of the in-situ annealing at the different growth stages, different DFL structures, and different DFL materials. A novel InAlGaAs asymmetric-step graded DFL design with the optimised growth technique is ...