Search

We present and compare the existing methods of heteroepitaxy of III-Vs on silicon and their trends. We focus on the epitaxial lateral overgrowth (ELOG) method as a means of achieving good quality III-Vs on silicon. Initially conducted primarily by near-equilibrium epitaxial methods such as liquid phase epitaxy and hydride vapour phase epitaxy, nowadays ELOG is being carried out even by non-equilibrium methods such as metal organic vapour phase epitaxy. In the ELOG method, the intermediate defective seed and the mask layers still exist between the laterally grown purer III-V layer and silicon. In a modified ELOG method called corrugated epitaxial lateral overgrowth (CELOG) method, it is possible to obtain direct interface between the III-V layer and silicon. In this presentation we exemplify some recent results obtained by these techniques. We assess the potentials of these methods along with the other existing methods for realizing truly monolithic photonic integration on silicon and III-V/Si heterojunction solar cells.

Integration of active photonic components on silicon and silicon on insulator (SOI) would be versatile for nanophotonics since CMOS compatible processes are available for fabricating passive devices on Si/SOI. Selective area growth of III-V semiconductors is also attractive for realising periodic structures for nanophotonics. Here we report on the recent results of high quality InP on Si and InP on SOI achieved by means of nanopatterning. MQW structures have been realised on InP/Si and InP/SOI. We would elaborate routes for monolithic integration of active and passive devices for nanophotonics.

Epitaxial lateral overgrowth of (ELO) InP on (001) InP/Si substrate is explored in a low pressure hydride vapor phase epitaxy system under various growth conditions. The effect of gas phase supersaturation on boundary plane formation of ELO and the behavior of dislocations in the grown layers are investigated. We found that the growth rate on (1 1 1)A boundary plane is determined by Burton-Cabrera-Frank model, which predicts a parabolic relationship between gas phase supersaturation and growth rate. Formation of (1 1 1)A plane will cause stacking faults in the grown InP layer. They will interact and annihilate each other and introduce fresh dislocations during the growth. Gas phase supersaturation can also be changed by varying opening separation distance. Low gas phase supersaturation is obtained by decreasing the distance between two openings. It gives rise to a lower staking fault density due to the suppression of nucleation of { 1 1 1 } facet plane at the edge of ELO. Etch pit density (EPD) and X-ray diffraction (XRD) techniques are used to estimate the dislocation density. Full width at half maximum of rocking curve at (004), (115) and (117) reflections were used to calculate the dislocation density in ELO InPISi. Experimentally measured etch pit density is smaller than the dislocation density derived from XRD data. In general the dislocation density is dependent on gas phase supersaturation. In this work, we demonstrate that it is possible to grow high quality InP layer on silicon substrate by epitaxial lateral overgrowth technique under optimized growth conditions.