Selective Epitaxy of Indium Phosphide and Heteroepitaxy of Indium Phosphide on Silicon for Monolithic Integration

Abstract: A densely and monolithically integrated photonic chip on indium phosphide is greatly in need for data transmission but the present day’s level of integration in InP is very low. Silicon enjoys a unique position among all the semiconductors in its level of integration. But it suffers from its slow signal transmission between the circuit boards and between the chips as it uses conventional electronic wire connections. This being the bottle-neck that hinders enhanced transmission speed, optical-interconnects in silicon have been the dream for several years. Suffering from its inherent deficient optical properties, silicon is not supposed to offer this feasibility in the near future. Hence, integration of direct bandgap materials, such as indium phosphide on silicon, is one of the viable alternatives. This thesis addresses these two issues, namely monolithic integration on indium phosphide and monolithic integration of indium phosphide on silicon. To this end, we use two techniques, namely selective epitaxy and heteroepitaxy by employing hydride vapor phase epitaxy method.The first part deals with the exploitation of selective epitaxy for fabricating a discrete and an integrated chip based on InP. The former is a multi-quantum well buried heterostructure laser emitting at 1.55 µm that makes use of AlGaInAs and InGaAsP as the barrier and well, respectively. We demonstrate that even though it contains Al in the active region, semi-insulating InP:Fe can be regrown. The lasers demonstrate threshold as low as 115A/cm2/quantum well, an external quantum efficiency of 45% and a characteristic temperature of 78 K, all at 20 oC. Concerning the integrated device, we demonstrate complex and densely packed buried arrayed waveguide (AWG) structures found in advanced systems-on-the-chip for optical code-division multiple-access (O-CDMA). We present a case of an error-free 10 Gb/s encoding and decoding operation from an eight-channel AWGs with 180 GHz channel spacing. Selective epitaxial growth aspects specific to these complicated structures are also described and guidance on design implementation of these AWGs is given. Mass transport studies on these AWGs are also presented.The second part deals with various studies on and relevant to epitaxial lateral overgrowth (ELOG) of high quality InP on silicon. (i) ELOG often encounters cases where most part of the surface is covered by mask. From the modeling on large mask area effects, their impact on the transport and kinetic properties has been established. (ii) It is known that ELOG causes strain in the materials. From synchrotron X-ray measurements, strain is shown to have large effect on the mask edges and the underlying substrate. (iii) The combination of strain and the influence of image forces when reducing the opening dimensions in ELOG has been modeled. It is found to be very beneficial to reduce openings down to ~100 nm where effective filtering of dislocations is predicted to take place even in vicinity of the openings. We call it nano-ELOG. (iv) By combining the modeling results of nano-ELOG and of a pre-study of ELOG on pure InP, a novel net pattern design is invented and experimented for nano-ELOG of InP on Si. PL measurements together with transmission electron microscopy observations indicate beneficial effects of small size openings (200 nm) compared to 1000 nm openings. (v) ELOG of InP on silicon-on-insulators together with a multi-quantum well structure grown on it has been demonstrated for the first time. This is particularly interesting for integrating silicon/silicon dioxide waveguides with InP.