Due to their electrical and optical properties, the integration of III-V semiconductors on Si would open the path to the combination of a various functionalities on the same chip, a potential solution to the challenges faced by CMOS components. The pionner studies by McKee and al have shown that SrTiO3 (STO) could be directly epitaxied on Si by MBE (Molecular Beam Epitaxy). Few years later, a Motorola team has shown that it is possible to epitaxy thin GaAs layers on STO/Si templates, hence opening a new path for III-V monolithic integration on Si. Based on this, the INL has undertaken the study of III-V semiconductors growth on STO. In particular, it has been shown that the weak adhesion specific to these systems favors a preferential accommodation mode of the lattice mismatch by breaking interfacial bonds rather than by plastic relaxation of an initially compressed layer. Hence, it is possible in spite of a strong lattice mismatch to grow III-V semiconductors without threading defects related to a plastic relaxation mechanism, which opens interesting perspectives for IIIV monolithic integration on Si. In this context, during this thesis, we have focalised in the beginning on optimising the growth of the STO/Si templates. In particular, we have shown that a relaxed and oxygen-rich STO layer favors undertaking InP growth. Next, we have studied systematically the InP growth on STO. The weak adhesion specific to this system leads to islands formation at the early stages of growth, as well as the observation of a competition between different crystalline orientations of the InP islands. We have worked out STO growth conditions and surface preparation strategies that allow obtaining purely (001) oriented InP islands. We have next optimised the islands coalescence step in order to form 2D InP layers on STO/Si. Based on a complete structural and optical study of these heterostructures, we have been able to analyse our approach’s potential and to point out cetain limitations of the STO/Si templates. On this basis, we have finally initiated the study of alternative templates for InP growth, by undergoing some preliminary studies of InP epitaxy on LaAlO3 substrates.
